aluminium cast house technology xi : selected, peer reviewed papers from the international...

Aluminium Cast House

Technology XI

Aluminium Cast House Technology XI

Selected, peer reviewed papers from the international conference, organised by the CAST CRC,

on behalf of the aluminium industry. It was held from 13 – 16 September, 2009 on the Gold Coast, Queensland, Australia

Edited by:

J. A. Taylor, J. F. Grandfield, A. Prasad

TRANS TECH PUBLICATIONS LTD Switzerland • UK • USA

Copyright © 2010 Trans Tech Publications Ltd, Switzerland All rights reserved. No part of the contents of this publication may be reproduced or transmitted in any form or by any means without the written permission of the publisher. Trans Tech Publications Ltd Laubisrutistr. 24 CH-8712 Stafa-Zurich Switzerland http://www.ttp.net

Volume 630 of Materials Science Forum ISSN 0255-5476

Full text available online at http://www.scientific.net

Image on front cover courtesy of Furnace Engineering

Distributed worldwide by and in the Americas by Trans Tech Publications Ltd Trans Tech Publications Inc. Laubisrutistr. 24 PO Box 699, May Street CH-8712 Stafa-Zurich Enfield, NH 03748 Switzerland USA Phone: +1 (603) 632-7377 Fax: +41 (44) 922 10 33 Fax: +1 (603) 632-5611 e-mail: [email protected] e-mail: [email protected]

Printed in the Netherlands

Preface

Once again it has been a privilege to chair the Aluminium Cast House Technology, 11th

Australasian Conference and Exhibition. The meeting comes to fruition via the efforts of

many people and they are a great team to work with. CAST Cooperative Research Centre

makes a significant commitment in hosting the conference in terms of managing the finances

and contributing significant staff time and effort in editing and publishing the proceedings

and all the other aspects leading to a successful conference. On behalf of the Industry

Committee I thank CAST CRC for this commitment.

As always, the conference and exhibition rely heavily on the expertise and hard working

dedication of Caryn Morgan (Conference Organiser) and Gabor Horvath (Exhibition

Manager); their continued involvement in the conference is greatly appreciated.

Sponsorship is a key element in keeping registration costs low so that as many operations

personnel as possible can attend and ensures a high quality conference. Thank you to all the

conference sponsors.

Authors and reviewers go to considerable effort in preparing the high quality peer reviewed

papers appearing in these proceedings. They cover the latest process improvements in terms

of cost reduction, improved quality, safety, environment, and new products. John Taylor,

Arvind Prasad (Co-Editors), myself and the Technical Committee gratefully acknowledge the

authors’ and reviewers’ efforts.

The third leg that the conference rests on is the exhibition, which complements the technical

program, providing opportunities for suppliers and operations personnel to explore new

products, and processes. Thank you for the support of all our exhibitors.

The fact that the aluminium industry has continued major support of the conference during

the difficult times of the current global financial crisis speaks volumes regarding the value

that the conference delivers.

John Grandfield

Conference Chair

Organising Committee Lawrence Amato ............ Treasurer, CAST CRC John Grandfield .............. Conference Chair & Proceedings Editor, Grandfield Technology Pty Ltd Gabor Horvath ............... Exhibition Manager, Furnace Engineering Pty Ltd Sue Keay ........................ Conference Secretary, CAST CRC Michael Lee ................... Deputy Conference Chair, Olex Cables Australia Pty Ltd Caryn Morgan ................ Conference Organiser, CMA Events John Taylor .................... Proceedings Editor, CAST CRC, The University of Queensland Arvind Prasad ................. Proceedings Editor, CAST CRC, The University of Queensland

Industry Committee Hussain Al Ali ............... Aluminium Bahrain, Bahrain Stephen Augustine ......... Fives Solios, United Kingdom Chris Baker .................... Cometals, Australia Joseph Barressi ............... Rio Tinto Alcan, Australia John Christowitz ............ BHPBilliton, Mozal, Mozambique Chris Emes ..................... Mecatherm International, United Kingdom Peter Herd ...................... Pyrotek Ltd, Australia Michael Ison .................. Australian Aluminium Council, Australia David Latter ................... Hydro Aluminium, Australia Turina McClelland ......... Wagstaff Inc, USA Ray Morrow ................... Tomago Aluminium, Australia David Neff ..................... Metaullics Systems Co, USA Franz Neidermair ........... Hertwich Engineering, Austria Madhu Nilmani .............. Nilmani Consulting, India Peter Whiteley ................ Munimula Technology Pty Ltd, Australia

Technical Committee Hussain Al Ali ............... Aluminium Bahrain, Bahrain Fadi Awadhalla .............. Dubai Aluminium, United Arab Emirates Ian Bainbridge ................ CAST CRC, The University of Queensland, Australia Maria Vitoria Canullo .... Aluar Alumino Argentino, Argentina John Chen ...................... University of Auckland, New Zealand Corleen Chesonis ........... Alcoa Technical Center, USA Malcolm Couper ............ Rio Tinto Alcan, Australia Jan de Groot ................... Poly-Engineering & Consulting B.V., The Netherlands Claude Dupuis ................ Rio Tinto Alcan, Canada Mark Easton ................... CAST CRC, Monash University, Australia Dmitry Eskin .................. Delft University of Technology, The Netherlands Peter Forakis .................. Hatch & Associates, Canada Neil Hall ......................... BHPBilliton Bayside, South Africa Stephen Instone .............. Hydro Aluminium, Germany David Irwin .................... Portland Aluminium, Australia Dag Mortensen ............... IFE, Norway Joel Murrray ................... o.d.t. Engineering, Australia David Neff ..................... Metaullics Systems Co, USA Alan Peel ........................ Alutek MDY, United Kingdom Barbara Rinderer ............ Consulting in Partnership Pty Ltd, Australia Nivashni Vandayar ......... BHPBilliton Hillside, South Africa Mary Wells .................... University of Waterloo, Canada

Sponsors Platinum Sponsor

Rio Tinto Alcan

Gold Sponsors

Cometals Australia

Hertwich Engineering

Mecatherm International

Pyrotek Pty Ltd

Wagstaff Inc.

Corporate Supporter

Fives Solios

Silver Sponsors

Australian Aluminium Council

Tomago Aluminium

Trade Exhibitors ABB AB

Acme Fixtures Pty Ltd

Biolab Limited

Bruker BioSciences Pty Ltd

Dynaref Products

Furnace Engineering Pty Ltd

Major Furnace Australia Pty Ltd

o.d.t. Engineering

Oliver Footwear

Otto Junker GmbH

Sims Aluminium Australia

SLM Co., Ltd

Weston Aluminium Pty Ltd

Table of Contents

Preface

Committees

Sponsors

Industry Directions

A Historical Review of the Developments in Casthouse Technology and a Peek into theFutureP. Whiteley 3

Hydro Casthouse Reference CentreT. Furu and I.K. Steen 9

Safety & Environment

Recent Developments in Australian Aluminium Casthouse Personal Protective ClothingM. Lee 19

Greenhouse Emissions in Primary Aluminium Smelter Cast Houses - A Life Cycle AnalysisP. Koltun, A. Tharumarajah and J.F. Grandfield 27

Dross Generation & Handling

Predicting Dross Formation in Aluminium Melt Transfer OperationsJ.A. Taylor, M. Prakash, G.G. Pereira, P. Rohan, M. Lee and B. Rinderer 37

Furnace Operations to Reduce Dross GenerationA. Peel, J. Herbert, D. Roth and M.J. Collins 45

Dross Processing TechnologyA. Peel, J. Herbert, D. Roth and M.J. Collins 53

Automated Metal Siphoning and Cast House Energy ConsumptionJ. Locatelli and G.W. Liu 61

Technological Researches Concerning a Decrease in the Losses Due to the Oxidation ofRemelted Scrap from Aluminium AlloysI. Butnariu, I. Butnariu and D. Butnariu 71

Furnaces & Refractories

Furnace Operation: “A Gold Mine in your Casthouse”G. Girard, J. Barresi, C. Dupuis and G. Riverin 77

Homogenization Aspects, Continuous versus Batch in an Integrated Modern CasthouseF. Niedermair 85

Fives Solios Experience in Modern Secondary Aluminium Casthouse ConstructionL. Allen, P. Hipwood and B. Houghton 95

Dry Hearth Melting FurnacesP. Newman 103

The Benefits of Forced Circulation for Aluminium Reverberatory FurnacesP. Campbell 111

Launder System for Aluminium CastingM.V. Canullo, F. Daroqui, J. Ottaviani, M. Martín and R.A. Laje 119

Melt Quality & Treatment

b Aluminium Cast House Technology XI

The Impact of Rising Ni and V Impurity Levels in Smelter Grade Aluminium and PotentialControl StrategiesJ.F. Grandfield and J.A. Taylor 129

Development of a Phosphate - Free Reticulated Foam Filter Material for Aluminium CastHousesL.S. Aubrey, R. Olson and D.D. Smith 137

Results with a Multi Stage System of Filtration Employing a CycloneJ.H. Courtenay and F. Reusch 147

The Use of Electromagnetic Fields for the Detection of Inclusions in AluminiumS. Poynton, M. Brandt and J.F. Grandfield 155

Cast House Productivity

Estimating the Production Capabilities of Casthouse Equipment Configuration OptionsP.W. Baker 165

Direct Chill & Continuous Casting

Downstream Considerations of Fusion Clad CastingR.B. Wagstaff, T.F. Bischoff and D. Sinden 175

Billet Sump Modification of Al-Billets for the Prevention of Starting CracksM. Rosefort, T. Koehler and H. Koch 179

New Billet Mould Casting TechnologyC. Emes and R.J. Collins 187

Macrosegregation Mechanisms in Direct-Chill Casting of Aluminium AlloysD.G. Eskin and L. Katgerman 193

Organic Coatings to Prevent Molten Metal ExplosionsA.W. Lowery and J. Roberts 201

The Supply of Casthouse Equipment and Empowerment of Knowledge to Cast AerospaceAlloys the Almex WayS. Hamer 205

Grain Refinement and Hot Tearing of Aluminium Alloys - How to Optimise and MinimiseM. Easton, D.H. StJohn and L. Sweet 213

DC Casting Using a TF Combo Bag: 3D Modeling of the Distribution of Molten Metal andHeat TransferS.P. Tremblay, A. Arsenault, D. Larouche, F. Ilinca, J.F. Hétu and J.P. Dubé 223

Ingot Casting

Safety Enhancement in Ingot Casting at Tomago AluminiumV. Nguyen, J.F. Grandfield, P. Rohan and B. Todd 235

Technological Jump in Aluminium Ingot ProductionC.M. Brocato 243

Aluminium Bahrain - Benefits at ALBA from Use of Improved Graphite Rings forProduction of Extrusion IngotsM.A. Kadhim, H.A.A.J. Ghuloom, G. Martin and M. Jacobs 251

CHAPTER 1:

Industry Directions

A Historical Review of the Developments in Casthouse Technology and a Peek into the Future

Peter Whiteley

Munimula Technology Pty Ltd [email protected]

Keywords: furnaces, degassers, filtration, moulds, alloy management

Abstract

Over the past thirty (30) years or so aluminium casthouse technology has been driven by a number of factors which have variously included:

• Competition from alternative materials

• Lightweighting

• Market requirements for enhanced properties affecting gas levels, impurities, inclusions, physical and chemical properties, ease of downstream processing, reduced cost and improved delivery

• Reduction in conversion cost by various means including capacity creep, maximising asset utilization, minimizing scrap, reducing melt loss, labour, and energy costs

• The ever present need for improved safety performance.

This paper will explore how these, and related considerations have provided the stimulus for improved casthouse technology which has included developments in hardware, software and culture.

Introduction

In order to cover this topic with some logic, the story will start at solid or liquid metal supply to the casthouse; then into furnaces; followed by in line metal treatment, casting moulds and casting machines, heat treatment, and support plant such as the water system, with a final comment about culture.

The purpose will be to associate the technological developments to the needs that precipitated that change and to endeavour to look at where future developments may lead.

Metal Supply

We will consider very briefly two situations – an aluminium smelter and a remelt.

In the case of a smelter we have the classic problem of what are two mismatched processes – the continuous reduction facility interfaced with what is generally a discontinuous casthouse operation. To compound this is the serious issue of tapped bath (electrolyte). In bygone days of small smelters this interface was barely recognisable because of the small size and lack of complexity of smelters – however, nowadays we find smelters, for example like Aluminium Bahrain, with five potlines of various vintages, three casthouses, five customers for liquid metal, threfive ladle cleaning stations, an internal metal transfer station, five TAC/skim stations, and casting technologies which are batch, semi continuous and continuous. One can see the need for a very robust system of planning, scheduling and tracking liquid metal movements – as well as dealing with the tapped bath and control of metal temperature. This has led to the use of dynamic simulation modeling software; and in some cases use of GPS vehicle tracking, as well as consideration of autonomous vehicles. (Commonly used in other industries and covered in detail at the 10th Conference).

© (2010) Trans Tech Publications, Switzerlanddoi:10.4028/www.scientific.net/MSF.630.3

The bath problem has been tackled by improved potroom discipline and by installation of ladle skimming machines which permit every ladle to be skimmed.

The Alcan TAC process is being challenged by what at first sight is Heggset’s remarkably simple technology of injecting the aluminium fluoride into the tapping spout in the potroom in lieu of a separate operation. However the so called ICON technology is yet to be commercialised.

In the case of a remelt casthouse there are very different considerations. By a remelt I mean the casthouse accompanying a rolling mill or an extrusion plant which can have an amount of runaround scrap typically 40% of the mill output. A fabrication plant obviously has to purchase 100% of its metal needs. The conundrum is to decide in what format to buy this tonnage – whether as sheet ingot (or extrusion ingot), as P1020 remelt ingot, or as liquid metal.

The choice clearly has an enormous impact on the size and type of the remelt casthouse – and the choice is often not well made. The benchmark for extrusion remelt design was Alcan’s plant at Pickering in Ontario, which took extrusion scrap from across Canada, and stored, reclaimed, melted, cast, homogenized, sawed, and dispatched 30 000tpy of billet with a total of five operators per shift. This was possible only through the inline design of the plant and choices of technology which will be discussed in following sections.

Furnaces

The most common situation is to size the remelt to handle 100% of the runaround scrap plus a little P1020 ingot for dilution. So then what does the benchmark remelt melting furnace look like? Most desirably it is a circular, top charged, tilting melter with an EMS stirrer, regenerative burners, and possibly oxy fuel supplemented. Alcan have operated this style of furnace with a tap to tap time of three hours, permitted by the very rapid top charge feature and the burner configuration, which can be maximised on a circular furnace. In addition, the furnace would have excellent air/fuel ratio control and furnace pressure control and that will be covered in more detail later. The tilting feature enables an alloy change every melt if required; rapid metal transfer to the holder (included in the three hours); and is inherently safer because there is no heel when charging scrap.

From this stage onwards the smelter casthouse and the remelt casthouses are rather similar.

Going back in time, many smelter holding furnaces were stationary, small, brick lined, fitted with small doors and were filled by cascade pouring potroom ladles either from a crane or a tilt vehicle.

Many of this style of furnace still exist in some of the older smelters. They are typically energy intensive, very difficult to clean, difficult to change alloys, expensive and time consuming to reline. So what does our benchmark smelter casting furnace look like?

Well first of all it is as large as practicable consistent with the subsequent casting technology. Jumping ahead we will see that one way of maximising asset utilization is to maximise mould density on D.C. tables or increase casting speed on continuous casting lines. (Open moulds, wheel, or H.D.C.). This leads to the need for larger furnaces – witness Tomago successfully converting 35T holders to 48T, and Alcan routinely enlarging existing furnaces from 67T to 100T capacity. Another advantage of larger furnaces, is that in a D.C. context, it permits double drop which is the reason why Hydro at Kurri can comfortably do 14 drops per day from one D.C. station.

So our furnace is large, it has a full width single piece door with no returns on the endwalls, it is castable lined, and importantly has a single tilt cylinder. This feature permits the use of syphon metal transfer from the potroom ladle saving 0.5% melt loss every time. The payback from this initiative is about six months. If the casthouse is making primary foundry alloy with say up to 12% Si, the furnace would have an electro magnetic stirrer underneath to enhance alloying time and scrap melting. Alcan’s Alma smelter has four such furnaces.

4 Aluminium Cast House Technology XI

Like all modern furnaces it would be fitted with mass flow control for both gas and air for achieving best air fuel ratio, which would be combined with excellent furnace pressure control (so the furnace is slightly positive). By so doing, the highest scrap melt rates and best fuel economy can be achieved by maximizing flame temperature and thereby exploiting the T4 effect to exploit the dominance of radiation. Finally our furnace would be fitted with some type of automatic heel measurement in order to get the furnace batched on composition first time every time. There are a number of off the shelf systems now available which infer melt mass based on heel height measurement or another which directly measures mass.

In Line Metal Treatment

There are plants that, until quite recently, had no degassing or filtration and yet were selling billet into the Asian markets. Those with degassers and filters had box type degassers and single plate CFF’s.

Competitive pressure and the need for meeting much more demanding customer specifications have led to changes here. Importantly also has been the need to improve the operability of degassing and filtration equipment.

About fifteen (15) years ago I was with Alcan, and tasked with making a technology recommendation for five large degassers required by various casthouses in North America, Japan and Europe. This study covered every important aspect of metallurgical performance, capex, opex, safety, and so on. After rigorous benchmarking we finally, reluctantly, decided on Alpur D5000. Why do I say reluctantly? Because even though it was the best, it still had some major issues that all box degassers have of difficulty in alloy change, difficulty with temperature control at cast start, difficulty of cleaning, high cost, and unreliability of the heating systems. So while the recommendation was for Alpur, the report also carried with it the stronger recommendation for Alcan to develop a zero hold up degasser. The result of that is history. It is Peter Waite’s ACD – which has now replaced the five aforementioned Alpur D5000 degassers as well as most other box degassers in Alcan’s plants. Its advantages are ease of alloy change, ease of cleaning, and ease of temperature control. Additionally, the recent enhancements of a fully closed design, combined with flux injection, have overcome the earlier issues of excessive dross formation and have eliminated the use of chlorine.

Which leads logically to metal filtration.

The benchmark is still deep bed filtration so far as metallurgical performance is concerned (i.e. inclusion removal), however we have the same issue as with the old degassers – another furnace in the line with a major hold up of metal - typically two tonne, making alloy changes difficult and until recently making temperature control through the bed quite problematic. Moreover, this technology is costly and takes up a lot of real estate for the bed in which is in service, the bed dump station, the bed rebuild station and the bed preheat station. Pechiney’s engineering enhancements to the original Alcan design have provided more effective filtration area as well as use of Athermalu heaters below the bed for solving the temperature stratification issue.

But, we still have a high hold up box.

The intermediate answer currently in vogue is a dual stage CFF – however these are also not without problems of inclusion release if bumped, need for a significant priming head, requirement of excellence of preheat, and what can be awkward handling of the filter plates before and after use. Each of these issues can be overcome provided they are recognised and properly addressed. The larger question about CFF’s is the vulnerability to inclusion release with the addition of grain refiner, raised in various papers by Professor W. Schneider and N. Towsey. This work has led Hydro to develop a hybrid filter in which the CFF is protected by a preceding bed. My conclusion is that we still have a way to go on filtration.

J. A. Taylor, J. F. Grandfield, A. Prasad 5

My last comment on in line metal treatment has to do with launders. I say to my students at Sydney University. “Materials handling adds no value”. Indeed it frequently destroys value, as is the case with launders, with excessive temperature drop and requirement for manual intervention for cleaning and maintenance. So, the shorter the better. The benchmark is Rio’s Bell Bay DC four sheet ingot centre, where it is impossible to conceive of shorter launders – minimising the need for interdrop intervention and significantly reducing temperature drop.

DC Casting

One just has to go to the fundamentals of metal solidification to understand where the developments in DC casting have been made.

It is well known that in conventional aluminium water cooled moulds operated with a conventional metal head of say 60-70mm, you have the effect of primary cooling from the mould, followed some time later by secondary direct chill cooling by the water jets. As everyone here understands, it is this air gap creation between the two cooling zones which causes a shell formation, poor surface, and inverse segregation in the cast product. It has been amply demonstrated by Electromagnetic Casting or by Air Slip Casting, that if the primary cooling can be eliminated, there is an imperceptible shell zone combined with an almost perfect surface. The very best proof of this goes back a few years to Kaiser Trentwood who EMC cast sheet ingot for production of can body stock, and rolled these ingots without the need for scalping.

Air Slip billet moulds emulate EMC rather well, but Wagstaff’s admirable attempts to produce Air Slip sheet ingot moulds were not successful. This led to the fall back position of having a graphite surface and low head casting to reduce the primary cooling, maximise the advanced cooling effect and minimise the formation of air gap and shell which was the purpose of Wagstaff’s LHC mould. The net result of this is to reduce scalp depth and so provide the minimum melt loss due to scalper chip.

So we currently have a billet mould (Air Slip or Air Slip lookalike) which gives low shell and good surface – but a narrow operating window in air slip mode due to the high head; and a sheet mould technology which again gives low shell and reasonable surface (alloy dependent). One very important feature of the LHC graphite mould is that there is virtually no casting lubricant used. This therefore eliminates one of the major traditional problems of water contamination resulting from conventionally oil lubricated DC moulds where the changes in water quenchability due to oil levels higher than 5-10 p.p.m. result in a significant increase in cast scrap.

As to the DC machine itself, there have been some major developments over the years. In early machines the platen was moved up and down by cables over a drum, just as you see on a pit stripping crane. Pechiney enhanced this original design to have cables on either side of the platen to two synchronized cable drums each side of the pit. Indeed there are still many of these Pechiney pits still in service – and Tomago is a good example. In order to overcome the maintenance issues associated with cable drives and differential cable stretch, Alcan and others went to single hydraulic cylinder machines – first of all with two rails, then subsequently with four rails to guide the platen. The rails and the guide shoes were, and are, subject to metal spill with the subsequent issues of damage and rail adjustment. These problems were so serious as to lead Alcoa to develop an internally guided cylinder which removed the need for any external rails. Alcoa’s engineering company Nash, produced many of these cyclinders over the years and they have proven to be trouble free. The problem was however that the cylinder had a 100x100mm square key welded inside the full length of the cylinder to act as the guidance. This meant that the cylinder had to be wide enough for a person to get inside – and the result was huge expense.

Kaiser and others realised that there were other internally guided cylinders used on naval ships as rocket launchers, and worked with Remco Hydraulics to adapt these cylinders for DC pit use.


Subsequently Wagstaff, Hunger, Cylindrix and others now offer this style of internally guided cylinders – which like the original are more or less maintenance free (unless you happen to drop an ingot on the platen).

So where to in the future with DC casting?

For me it’s a “no-brainer”. We just emulate the steel industry, the copper and magnesium industries, and go for continuous VDC casting – especially for sheet ingot. This overcomes almost all of the engineered scrap associated with our dreadful batch process. You just have a saw with a blade in the horizontal position in lieu of vertical as you see on Hertwich three strand HDC Tee ingot machines.

Then we would have no more papers ever again at this conference on the subject of butt curl.

Heat Treatment

I will just make brief commentary on this topic of billet homogenization.

For many years batch homogenizing furnaces were absolutely the dominant technology – but then Mr Gunther Hertwich developed and marketed his continuous furnace and it has been dramatically successful with almost 100 installations worldwide. However, as always, the devil is in the detail. In his original brochure describing the continuous furnace Mr Hertwich was at pains to say that his technology was particularly appropriate for one heat treatment practice – i.e. homogenizing temperature and soak time – so it was best applied for a uni alloy plant with one diameter. He went on to say that if your product mix was such that you had a significant tonnage of alloys/diameters requiring different heat treatment practices, then you should have a batch furnace to complement the continuous furnace – otherwise you will lose very significant capacity during alloy changes. I can instance one plant in particular who purchased a continuous furnace with a “nominal” capacity of 100,000 tpy yet they have never achieved more than 60,000 tpy because of this product mix issue.

Bayside Aluminium recognised this problem and has been able to work with a supplier of continuous furnaces and so design the furnace with multiple zones and sophisticated controls so as to not lose capacity during changes in heat treatment practices.

Casthouse Water Systems

The changes in casthouse water systems have been quite dramatic over the years. There were many plants in Canada, USA and UK who used to have “once through” water systems. That is to say, you took water out of the nearest lake, river or canal, put it through your DC machine, and then simply deposited that warm, possibly oily, water back from whence it came. Those days have gone, and now we have more or less a closed system having a legal blowdown to maintain water composition within acceptable limits. This now means that all of the heat picked up in the DC process must be shed in evaporative towers, with the need for expensive fresh make up, appropriate water treatment and oil removal.

There are just four issues I want to discuss

• Cooling towers

• Oil removal

• Biocides

• Pit Depth.

It is well known in Australia that evaporative cooling towers are absolutely the perfect breeding ground for algae and bacteria. A number of people in this country die every year from legionella


bacteria arising from cooling towers. If you operate these towers you have to constantly monitor for this problem.

The mould lubricants most commonly used in DC casting are vegetable oils – canola, rapeseed or whatever. As the residual oil is washed off the ingot surface by the casting water it falls into the DC pit where it forms a colloidal mixture. Because of the oil colloid size and the similar specific gravity to water, the oil is exceptionally difficult to remove. The only successful technologies I know are dissolved air flotation and induced air flotation. Skimmers, lamella thickeners and so on are useless.

Biocides have traditionally been based in chlorine since it is such an inexpensive powerful oxidizing agent. However it has environmental issues as you will be aware and, raw chlorine will react with the vegetable oil colloids and form very waxy, sticky substances which will deposit in moulds, pipes, valves – to the extent that this shut down a number of Alcan’s DC pits some years ago. The better biocide is ozone, made by your own ozone generator. You don’t need to store ozone and it reacts with the oil to form innocuous products.

All I want to say about water depth in DC pits is that the most desirable depth is full pit. This lets you saw product after it is stripped due to improved cooling, and you do not need any water return pumps or level controls. This is the system employed in most modern plants.

Other

There have been many other areas of rapid development which are outside the scope of this talk but which importantly include:

• Saws (Radial versus band, and product strapping)

• Melt Loss (Methods of reduction)

• Quality Control (Limca, Alscan, on line composition analysis, etc.)

• Automation (Especially mould level control and auto abort)

• Continuous Casting (Twin Roll, Ingot chains, wheel)

• Safety.

I will conclude with one item mentioned in the Abstract, and that is culture – and I will specifically narrow this down to Kaisen.

Of all the management fads that I have been exposed to over the years, this has been the most sustainable, for the reason that it is common sense.

It goes back some years to when I visited NLM Nagoya sheet ingot remelt in Japan. I witnessed the last few minutes of a cast, the termination of the cast, watched removal of the distribution launder and combo bags, mould table up, platen up, three ingots stripped, platen all way up, stool base change, mould size change, CFF reconditioned, distribution launder and appertenances put in place, metal delivery commenced through the CFF, onto the stool caps, filling the moulds and platen descent (i.e. cast start).

The time from when the cast stopped, to the start of the next one, including the size change and pit stripping, was eighteen (18) minutes.

That’s poetry in motion, thanks to Kaisen.


Hydro Casthouse Reference Centre

Trond Furua and Idar Kjetil Steen

Hydro Research and Technology Development, P.O. Box 51, N-6601 Sunndalsøra, Norway a [email protected]

Keywords: Reference Casthouse Centre, aluminium, support

Abstract

Since October 2006 the Hydro Casthouse Reference Centre has been operating. The centre is a full scale state of the art pilot casting centre for extrusion ingot, sheet ingot and foundry alloys, consisting of a 17Mtons furnace with a metal loop, a launder system including modular in-line melt treatment units such as ceramic foam filters (CFF) and inline melt refining units (Hycast SIR) and a casting pit with the possibility to cast full size geometries and a casting length of 5.5m. A two strand horizontal casting machine further adds the possibility of continuous casting of extrusion ingot and foundry alloy ingot. The centre has a state of the art superior control system (SCS) and a lay-out, including control room facilities, well suited for training and demonstration purposes. In addition the centre has access to state of the art software codes for simulating the casting process (Alsim) and the as cast microstructure (Alstruc).

The present paper gives some examples on how the centre is operating and the support that is offered to casthouses in Hydro. This includes (i) simulation of the casting processes (hot tearing and as cast structures) applying the Alsim and Alstruc codes, (ii) pilot scale testing of casting and melt treatment equipment, (iii) testing of new parameters and procedures for melt treatment and casting (iv) production of trial orders of new alloys and (v) practical training of casthouse operators (basic for molten metal handling, emergency situations and response, casting principles and trouble shooting, etc.).

Introduction

The new Hydro Casthouse Reference Centre was opened October 2006. The Reference Centre came as a result of the modernization of the Hydro Sunndal metal plant and is now a part of Hydro RTD (Research & Technology Development) Centre. The centre is, as can be seen from Figure 1, located very close to the plant as well as it is next door to the metallurgical R&D site. This centre is a part of the “casthouse competence cluster” including (i) an extrusion ingot casthouse (≈375.000 Mtons/year) and a foundry alloy casthouse (≈120.000 Mtons/year), (ii) Hycast which is a supplier of Casthouse technology and (iii) Hydro RTD.

The old R&D casting centre was located inside the production casthouse of Hydro Sunndal, and had to be closed down when Hydro Sunndal was modernized. This old R&D facility had during many years proven to be an effective tool for development of casthouse technology and new alloys, and it had also been used as a training centre for casthouse personnel. Practical training of personnel before start up of Greenfield and brownfield casthouses (both remelt and primary) is an effective way of securing a high HSE level and fast ramp up of new production facilities.

The motivation for Hydro to invest approximately 50MNOK in a new casthouse reference centre was based on the experience from the old R&D centre and future needs related to: (i) R&D material science (alloy development), (ii) R&D Technology (developing new casthouse technology and new concepts), (iii) Training (tailor-made training programs on HSE and process operation) and (iv) Casthouse support (ad hoc support tasks, measurement campaigns, process validation etc).


Figure 1. Hydro Sunndal after modernization in 2003, Hycast and the Reference Centre.

The objective of the present paper is to give an overview of the various applications of a full scale pilot casthouse and the methodology that is being used when developing new casthouse technology and improved ingot/alloy qualities.

Equipment Installations

The Reference Centre building covers an area of 1200m2. Main equipment installations are: (i) 17 tons tiltable reverbatory furnace equipped with a 2MW cold air gas burner and porous plugs. The furnace is an old production furnace from the Hydro Sunndal metal plant, but modernized and modified for use as a pilot furnace, (ii) SIR inline filter from Hycast (drainfree inline melt treatment unit with 2 bottom installed rotors), (iii) DC casting machine from Hycast with internal guided cylinder. The machine is fully automated with a casting length of 5.5 m and equipped with various mould systems for casting of extrusion ingot and rolling slabs in several dimensions, (iv) 2-strand DC casting machine from Hertwich for extrusion ingot and primary foundry alloy ingots, (v) Special designed launder loop for melt treatment experiments, (vi) State of the art Superior Control system from Hycast, (vii) Control room with training facilities for operators, (viii) Homogenizing furnace and sawing system, (ix) Cast tooling for various dimensions of extrusion ingot, sheet ingot and foundry alloys and (x) Overhead crane, capacity 20 MT.

The “launder loop system” is based on a standard launder design from Hycast, and is equipped with a mechanical metal pump. When running melt treatment experiments the setup allows us to direct the flow of metal back to the furnace after passing through the filter systems. By this we can simulate continuous casting processes and high filter loads.

In addition the centre has equipment for measurement of melt quality (Alscan, Limca, Podfa, LAIS, etc). The laboratory part of RTD has state of the art facilities for evaluation of microstructures (optical microscopes, scanning- and transmission electron microscopes).


Figure2. Casthouse equipment installed at the Hydro Casthouse Reference Centre.

Methodology for Development of Casting Technology

The optimisation of existing casting processes and development of new casting technology in the Reference Centre are utilising various competences such as simulation competence, engineering competence, metallurgy competence and casting competence. This work takes place through close cooperation between RTD - Hycast - Hydro Primary and Remelting Casthouses. Main responsibilities for the involved partners are: (i) Hydro RTD: Concept development, prototype testing, optimizing of equipment and practises, (ii) Hycast: Design, engineering, automation and control systems, installation and commissioning of equipment and (iii) Casthouses: Pilot testing and validation, input for continuous improvement of existing equipment and practises.

The simulation codes that are being applied in the Reference Centre are the Alsim [1] and Alstruc [2] -codes, which have been developed during the last 2-3 decades in collaboration with SINTEF, Institute of Energy Technology (IFE) and Elkem.

Alsim: The casting simulation code Alsim has been used as a powerful tool the last 3 decades in developing and optimizing casting technology. Alsim calculations lead to the discovery of the”dimensionless temperature field” (the temperature field can be plotted generally for DC casting). This discovery leads to simple, but very predictive, models for optimal mould shapes and for critical casting speeds during billet and sheet ingot casting as well as casting of foundry alloys. The model is used on a regular basis within Hydro.

Alstruc: The Alstruc microstructure simulation code is built on standard solidification theory. The metal solidifies one part at a time with the concentration in the recent layer of solid-state aluminium almost proportional to the concentration in the liquid. The “constants” of proportionality are called distribution coefficients, and the values are found in the phase diagram. The microstructural input parameters are the composition, the dendrite arm spacing (DAS) and the grain size. The main output parameters are the temperature as a function of the fraction solid, the concentration profile in the solid-state from the dendrite centre to the dendrite boundary, the volume fraction of each type of particle, the temperature interval in which they form, and tentative particle sizes, the temperature-dependent thermal conductivity, density, specific heat and heat of fusion.

As an example, results of Alsim-simulations of DC-cast extrusion billets of the 6060 and 6082 alloys, are shown in Figure 3 where the temperature gradients during casting are presented (Figure


3a), and the grain size and the secondary dendrite arm spacing as a function of the billet radius (Figure 3b). The variations across the cross section are in accordance with values from industrially DC cast extrusion billets.

The grain size and the DAS-values are microstructural input parameters in the Alstruc-model for calculating particle size, size distribution of particles and the solid solution level. Examples of such simulations for the 6060 alloy are given in Figure 4a and b. In Figure 4a the solid solution level in the as cast condition is given for the various alloying elements. The first aluminium that solidifies is seen to be almost pure. During solidification the concentration of solid solution of the main elements Si and Mg is seen to increase steadily until approximately 80% fraction solid when it increases in a more exponential way. The size distribution (in terms of relative numbers of particles) as well as the mean size of the constituent particles for a 6060 alloy in as cast condition is given in Figure 4b). The Alstruc-model can handle the different particle types in 6xxx series alloys and for the present 6060 alloy it distinguishes between the iron-rich particles (the α and β-types), Mg2Si particles, Si-particles and π-particles.

0

50

100

150

200

250

0 0.02 0.04 0.06 0.08 0.1 0.12

Billet radius (m)

Grain size (m

icron)

0

5

10

15

20

25

30

35

40

45

50

Den

drite arm

spac

ing (m

icron)GS in 6082

GS in 6060DAS in 6082DAS in 6060

Center Surface

(a) (b)

Figure 3. Simulation of billet casting of the alloys 6060 and 6082 by the Alsim-code; a) Temperature gradients for the 6060-alloy and b) secondary dendrite arm spacing and grain

size as a function of the billet radius, for details see [3].

The Casthouse Reference Centre is used for validation of results from Alsim and Alstruc, and testing of full scale equipment prototypes. Since the Reference Centre operates with “real size equipment” (but with a reduced number of moulds) and a “state of the art” control system, scale up problems when going from a small pilot caster to a “production” casthouse are minimized. Examples of equipment developed and optimized by the involved partners are: (i) Hycast Gas Cushion: Casting system for extrusion billets, (ii) Hycast SIR: Drain-free inline melt refining unit with bottom mounted rotors, (iii) Hycast RAM: Equipment for crucible fluxing of pot room metal, (iv) Hycast Superior Control System: Automated control system for casthouse processes. The philosophy to develop and maintain inhouse technical and operational competence has proven to be a success for the casthouse technology development in Hydro.


(a) (b)

Figure 4. Microchemistry of a 6060 alloy in the as cast condition: (a) Level of solid solution of various elements versus fraction solid and the mean solid solution level (table), (b) size

distributions and average length (table) of various particle types (relative number), for details see [3].

Optimisation of ingot quality and introduction of new alloys to the market

The Reference Centre makes it possible to shorten the time from idea to implementation of both casthouse technology and new alloys/optimised ingot qualities. The 5.5m stroke length of the casting cylinder means that cast extrusion ingots can be heat treated in industrial continuous homogenizing furnaces and sheet ingots can be homogenised and hot rolled in rolling plants. Two examples of applying the Reference Centre for introducing new or refined alloy/ingot qualities to the market are given below.

Example 1: The price of Ti has increased significantly the last years and so has the price of grain refiner rod. Reducing the amount of grain refiner rod will therefore give a significant cost saving in the casthouses. In addition, for foil products it is important to avoid pinholes and the titanium borides from the grain refinement may lead to such pinholes. In order to test the effect of different grain refinement on foil quality three 5.5m long sheet ingots (Figure 5) with different grain refining practises were cast at the reference centre and shipped to the rolling plant Norf. In Norf the sheet ingots were homogenised and hot rolled according to standard practises before cold rolling in Hydro Grevenbroich.

(a) (b) (c)

Figure 5. Casting and rolling of sheet ingots for testing of impact of different grain refining practices on rolled product quality; a) the casting process, b) lifting of sheet ingot and c) as

cold rolled and coiled condition.


Samples were collected along the whole process chain in order to follow the effect of the different grain refining practises on resulting microstructures (particle structures, grain structures).

Example 2: The Reference Centre plays a central role in developing new alloys for various applications. An example of the methodology is given below for the development of alloys towards the extrusion market. Based on introductory work involving reviewing of literature and application of various available modelling tools alloy variants are cast in a downsized casting centre for small batches (Fig.6a) in order to test different alloy combinations, casting practices and homogenization treatments. The influence of these variations on thermomechanical processing can further be evaluated in the Gleeble machine (physical simulation tool available at Hydro RTD) or in the 800 tons laboratory extrusion press at SINTEF/NTNU in Trondheim (Fig.6b). Extruded profiles are evaluated with respect to relevant properties for the actual product. When the recipe of optimised alloy chemistry is available full scale casting of billets is carried out in the reference centre (Fig. 7a) and homogenised either in the Reference Centre or in an industrial homogenization furnace. The billets are further shipped to industrial extrusion plants for validation trials of various extruded products (Fig.7b).

Figure 6. a) The downsized casting unit for casting of laboratory/small scale extrusion trials, b) extrusion of billets in an 800 tons vertical laboratory extrusion press at SI6TEF.

(a) (b)

Figure 7. a) The unit for casting of full-size billets in the Reference Centre and b) examples of some extruded products that can be optimised by utilizing the Reference Centre.

Operator Training

In modern high productivity casthouses many of the process steps involving handling of molten aluminium are PLC controlled. Process parameters are recorded for documentation and advanced software is used to analyse both process parameter variation and product properties. Even in the most modern casthouse, operators need to be trained to a high level to be able to utilize the equipment in a safe and effective manner. Even if many process steps are controlled by computers, we still have manual operations where things can go wrong. From the yearly incident statistics from


the Aluminium Association we are reminded about the potential hazards that exist in an aluminium casthouse. Even the most advanced process control systems can fail, and casthouse personnel need to know how and when to react and eventually use the emergency shut down procedure. Tailor made training sessions for casthouse operators within casting, melt treatment and safety for handling of molten aluminium are therefore an important area for the Reference Centre.

The training courses are normally divided into “Basic” and “Advanced” level of training. Both levels include class room training on the important HSE principles for correct handling of molten metal, correct use of PPE, emergency stops etc. It has also proven to be very valuable to use reported industrial incidents as a basis for the training.

The training programs include also theoretical and practical demonstration on how casting parameters will affect surface quality, crack sensitivity etc. Trouble shooting is normally not included in the Basic training, but is an important element of the advanced training session.

Especially the practical sessions where parameters like casting speed, metal temperature and cooling water flow are varied are popular among the casthouse personnel. At a high productivity casting centre you are not allowed to do such experiments, but at the Reference centre we can demonstrate why it is important to follow the established standard practices and the consequences when parameters are outside the established window.

An important issue for all operator training is how to handle emergency situations. If a situation appears during casting or metal transfer, it is very important that everyone involved knows how they should react and how the emergency system will react. At the Casthouse Reference Centre we have adapted the safety philosophy from Hycast with the following main items: (i) All casting centres shall have a central emergency stop, (ii) All local emergency stops shall be hard wired to the central emergency stop, (iii) All process modules must be designed to “fail safe” in case of a power failure, (iv) Risk assessment is mandatory for all casting centres. This shall include a detailed emergency stop philosophy.

A “live” demonstration on how the process equipment reacts when the central emergency stop is engaged during casting is therefore a central part of operators training. This includes a detailed description of the safety philosophy for the casting centre, and how each of the equipment parts will end up in a fail safe mode.

One of the main health issues for operators in modern casthouses are hearing impairments. Handling of large volumes of scrap and finished goods combined with powerful gas burners has destroyed the hearing ability of several operators. The classic way to help out the problem is mandatory use of personal hearing protection. Hearing protection will reduce the possibility for a permanent hearing impairment, but could also lead to other dangerous situations in a busy casthouse with moving vehicles. The noise guard should therefore be on the machine and not on the people. The Reference Centre has therefore developed a one day training course in modern principles for noise control. Different techniques for how to reduce noise in the casthouse are explained and practical examples are demonstrated in real life.


Figure 8. Practical casting demonstration performed at the pilot casting machine.

Concluding Remarks

The possibilities and advantageous of having a full-scale pilot casting centre with industrial-relevant technology has been demonstrated. The Hydro Casthouse Reference Centre is very valuable for development of new alloys, new casthouse technology, improving the knowledge in casting practices among casthouse operators and especially in training the operators on critical safety situations.

The strong link between industrial casthouse experience, R&D and an engineering company manufacturing casthouse equipment in addition to the very close collaboration with the university in Trondheim and SINTEF makes the presented cluster to be one of the strongest competence centres in the world regarding casthouse and materials technology.

References

[1] Mortensen. D. (1999) A mathematical model of the heat and fluid flows in direct-chill casting of aluminium sheet ingots and billets. Metallurgical and Materials Transactions B, Vol. 30B, 119-133.

[2] Dons, A. L., Jensen, E. K., Langsrud, Y., Trømborg, E. and Brusethaug, S. (1999) The Alstruc Microstructure Solidification Model for Industrial Aluminium Alloys. Metallurgical and Materials Transactions A, Vol. 30A, 2135-2146.

[3] Furu, T., Johansen, A., Sæter, J.A., Dons, A.L., Pedersen, K., Berstad, T., Lademo, O.G., Holmedal, B., Marthinsen, K., Hopperstad, O.S., Nes, E. and Mortensen, D. (2004) Through Process Modelling of Extrusion: Evolution in microstructure and mechanical properties through the whole process chain from as cast and homogenized condition to forming of profiles, ALUMINIUM, 6.


CHAPTER 2:

Safety & Environment

Recent Developments in Australian Aluminium Casthouse Personal Protective Clothing

M.J. Lee

Olex, 207 Sunshine Road, Tottenham, VIC 3012, Australia

[email protected]

Keywords: casthouse safety, protective clothing, ppe

Abstract

With the recent release of the Victorian WorkSafe Authority Foundries Compliance Code [1] it is appropriate that Australian Aluminium cast houses, die casting shops and foundries review the status of their personal protective clothing/ equipment (PPE) and practices. Since the issuing of the Foundry Code of Practice [2] in 1986 and the issuing of the new Compliance Code in September 2008 there has been a significant change in the range of PPE utilised in cast houses. This change has been brought about as a result of the advancement in the design and development of the materials used, extensive industry experience and collaboration. The choice of appropriate PPE is also guided by the range and impact of injuries sustained in cast houses. This paper aims to highlight the number of PPE advancements and range of experiences gained between the writing of the Foundry Code of Practice and Foundry Compliance Code as well as to serve as a reference for future improvements for the protection of cast house personnel.

Introduction

Occupational Health and Safety Regulations in Victoria, Australia were reviewed and updated by the Victorian Government in 2004 to 2007. One of the results of this review process was the move from Industry Codes of Practice to a more descriptive and educational format classified as Industry Compliance Codes. The first group of Codes of Practice converted to Compliance Codes included the Foundry Code of Practice. The resulting Foundries Compliance Code represents the latest state of knowledge with respect to occupational health and safety issues relevant to foundries – both ferrous and non-ferrous. The review process drew heavily on input and guidance from a number of industry associations (AFI, ADCA, AAC), technical expertise (SIMS Aluminium, CAST CRC, CSIRO, consultants) and publications from the US Aluminum Association [3], Rio Tinto Alcan [4] and the Queensland Government [5]. The Compliance Code issued in September 2008 is viewed as the pre-eminent Foundry OH&S document within the Australian WorkSafe network.

Injuries in the Foundry

Australian Industry has a long involvement with and input to the Foundry Code of Practice/ Foundries Compliance Code. This involvement was intensified after the tragic explosion [6] at SIMS Aluminium in 1986 which resulted in the loss of four lives. The following 22 years although not incident free have been (molten metal) fatality free in the Victorian Foundry Industry. The one hundred foundries operating in Victoria recorded [7] just over 1100 injuries (Figure 1) over a four year period with a median claim cost of $3,170. Musculo-skeletal injuries accounted for just over half of all claims with fractures and open wounds accounting for close to one third. Deafness related injuries have been noticeably increasing over the last decade, possibly due to increased monitoring. Burn related injuries account for approximately three percent of claims followed by stress (which includes both heat stress but more typically psychological related stress such as bullying and discrimination). Electrocution, poisoning, respiratory, skin and eye related injuries (predominantly hot dross related) round out the range of claims made by Victorian Foundries.


Musculo-Skeletal52.5%

Fractures/Contusions/Open Wounds29.3%

Deafness9.7%

Burns2.8%

Stress1.6%

Skin0.9%

Eyes0.5%

Poisoning0.5%

Respiratory0.5%

Eelctrocution0.1%

Other1.6%

Other5.7%

Figure 1. Victorian Foundry reported injuries 2003/04 to 2006/07.

On reviewing the range of injuries occurring in the Victorian Foundries the following three key OH&S issues account for over two thirds of the injuries (Figure 2):

1. Manual handling

2. Plant and machinery (both stationary and mobile equipment such as forklifts)

3. Noise

Molten metal is ranked fifth below slips, trips and falls, accounting for a few percent of all injuries.

Manual Handling50.1%

Plant & Machinery19.7%

Noise8.7%

Slips/trips/falls7.3%

Molten Metal7.1%

Airborne Contaminants2.0%

Hazardous Substances1.9%

Bullying/Harassment0.9%

Hot Conditions0.9%

Vibration0.9%

Stress0.5%

Electricity0.0%

Other7.1%

Figure 2. Key OH&S issues in Victorian Foundries.

Foundry Shirts

With the above statistics in mind it is understandable that the new Compliance Code addresses a number of issues associated with joint and muscle sprains/ strains; slips, trips and falls; mechanical hazards associated with moving equipment (entanglement/ crushing) and non moving equipment (noise and vibration); noise in the work environment; respiratory disorders; effects on skin from exposure to gases, vapours and fumes; and heat stress. These issues impact on the choice of PPE in


a number of ways. However, the choice of PPE for use within the cast house environment as discussed within the Compliance Code focuses heavily on the protection from molten metal, acknowledging the serious effects of molten metal and the important role that PPE plays as the last line of defence with respect to molten metal.

Cast House clothing developments over the last two decades have focused largely on the more widespread usage of wool based clothing replacing Proban treated cotton. Additionally there has been a move away from cotton overalls within the smaller die casting shops and foundries to shirt and trouser combinations. In the last five to ten years there has been a dramatic re-clothing of Victorian foundry workers driven largely by the need to meet industry requirements for high visibility clothing. The incentive for high visibility clothing is to reduce the number of manufacturing industry fatalities associated with forklift/ pedestrian collisions/ entrapments/ crushings. The ten year period 1985 to 2005 saw 54 fatalities involving forklift trucks in Victoria. The incorporation of high visibility reflective tape and coloured sections of clothing has been taken up widely by the cast house industry, yet there is no clearly accepted industry ‘norm’. The ‘Bell Bay’ smelter shirt with orange upper chest/ back sections and reflective tape strips (Figure 3) is in use at a number of non Aluminium smelter cast houses such as SIMS Aluminium, Weston Aluminium, Nissan Casting and Olex.

Figure 3. Range of high visibility cast house PPE (‘Bell Bay’ smelter shirt is on far right) [8].

While the more widespread use of a common item of cast house PPE would be expected to be more cost effective for all involved there is a wide range of similar cast house shirts in use around Australia. The wide range of shirts approved for use can be attributed to two main reasons: firstly individual corporate branding and colour choice; secondly individual site approval and interpretation of high visibility and molten metal protection standards [9]. The noticeable trends in the larger aluminium cast houses during the last few years are a move to more and more reflective tape being used (larger and wider strips), the use of yellow/green pigments in place of orange, and the incorporation of air vents on the back and armpit sections of the shirts. Over the coming years it is anticipated that these trends will find their way into the PPE of the smaller die casting shops and foundries in a similar manner to the uptake of PR97 type wool viscose clothing replacing Proban treated cotton or untreated cotton clothing.

Foundry Trousers

The move to heavy wool viscose trousers from either cotton overalls or cotton/ Proban treated cotton trousers has been common in aluminium cast houses over the last twenty years. The weight of the woollen material is typically twice as heavy (measured in grams per square metre: gsm) as the material used for shirts and half as heavy as the wool blend material used for lightweight casting jackets. A common observation and cause for replacement of foundry trousers are numerous small burn holes on the lower sections of the trouser leg. In order to address this issue one of three things are typically done:


• The use of spats over the top of the trouser leg (figure 4). More widely the use of spats has diminished in aluminium cast houses, die casting shops and foundries. Spats are still reasonably prevalent in ferrous foundries as are lace up boots which have all but disappeared in the aluminium industry. Debate about the positioning of the spats (above or underneath the foundry trouser) is not easily resolved as it is intimately involved with the choice of boot (low or high sided) and the tightness of the seal at the top of the spat. When used spats are more commonly placed underneath the trouser leg.

• Incorporation of a double layer of wool material on the lower shin section of the foundry trouser – thus making what is often referred to as a ‘heavy trouser leg spat’.

• Use of iron on patches to repair damaged sections of foundry trousers (this is only a recent innovation).

Figure 4. Example of spats. Figure 5. Wool and aluminized casting jackets

Furnace Coats

Although there is some form of standardisation across the industry taking place with respect to cast house shirts and trousers this can not be easily said for furnace coats/ jackets. This is largely due to the variety of tasks requiring furnace jacket protection in conjunction with the time required to carry out the relevant task. Although jackets may come in a (limited) one size fits all format, furnace operators notoriously do not. Additionally furnace jackets appear to be typically a shared item of clothing used by a number of crew members and crews. Hence, furnace jackets tend to be custom made/ ordered for particular tasks such as manual furnace skimming or spectrometer sample collection. It is not uncommon to view multiple jacket types at one site (figure 5) – heavy ‘welders’ wool jackets which offer excellent splash runoff and aluminized coats which offer good radiant heat protection but poor splash runoff. On a number of sites it is also possible to have multiple variants of the same jacket design where safety committee/ employee feedback has been sought with the result of expensive tailor made jackets being purchased catering for multiple staff needs (such as different staff heights). Figure 5 also highlights a number of trends more noticeable now when compared to 1986 – greater chin/ throat protection from radiant heat, more liberal use of Velcro straps and a move away from ‘soft’ leather aprons (although the later is still to be seen in some foundries).

Gloves

The range of gloves available to cast house staff is extensive (figure 6). Similar to there being no one furnace coat to suit all tasks a range of gloves can be found within a typical cast house as different gloves are often utilised for specific tasks. Furnace mitts are one such example where a glove is used for specific tasks (such as man handling hot ingots/ cast bar). In the one cast house mitts can be found alongside fingered Kevlar gloves which are used for furnace skimming and spectrograph sample casting where more dexterity is needed. Similarly leather riggers gloves can


be found for general cast house work such as strapping (but should not be used when handling molten aluminium).

It must be remembered that molten metal interacts quite differently with these three examples of gloves [10]. The ability of the glove to withstand the high temperatures coupled with the ability or otherwise to shed molten metal are important factors when determining the most appropriate glove for each cast house situation.

Figure 6. Range of cast house gloves: (from left) mitts x2, Kevlar woven palm, ‘welders’ leather, Kevlar flat weave and riggers leather.

Foundry Boots and Socks

There is a clear distinction between Aluminium smelter cast houses and die casting shops/ foundries when comparing safety footwear. The larger smelter environments tend to employ metatarsal protection and high sided boots (figure 7) while the smaller cast houses tend to utilise slip on steel toe capped work boots (Figure 8). It is common for both types of safety boot to have a high temperature sole [11]. The larger boots typically incorporate Velcro fasteners which have been the source of much debate at individual sites – questioning their flammability, whether molten metal will adhere or runoff, whether spats are required and the variability in tightness of fit. The smaller boots raise discussion relating to molten metal entrapment and whether the elasticised side can ignite if splashed with molten metal. The resulting evolution of the ‘smelter’ boot over the last twenty years has been dramatic with ‘dip’ testing (Figure 9) of boots into molten metal demonstrating the protective nature of the modern designs and addressing the concerns of different cast houses. Figure 9 shows a smelter boot and a woollen sock after dip testing in molten metal. The long multi layer woollen sock [12] is finding wider acceptance and usage in Victorian cast houses and is expected to find greater use over the coming years.

Figure 7. Metatarsal ‘smelter’ boot.

Figure 8. Standard slip on foundry boot.

Figure 9. Dip test sock


Safety Glasses and Face Shields

The 1986 Foundry Code of Practice refers to the need for safety glasses to have ‘side shields’ (figure 10). Polycarbonate safety glasses available in 2009 are nearly universally of the ‘wrap around’ variety (figure 11) and are far more ‘snug’ fitting than 1986 variants. The need for specialised side shields is still a necessity for prescription glasses.

Figure 10. Side shields. Figure 11. Tinted ‘wrap around’ safety glasses.

Another vision related safety issue which is of more concern/ emphasis in the new Compliance Code compared to the former Code of Practice is the recommended banning of contact lenses in the foundry environment. As contact lenses are more common in the general population and because there is a risk of hot dust/ dross entering the eye there is a risk of the contact lens partially fusing onto the lens of the eye resulting in a tear when flushed at an eye wash station.

The Compliance Code places a lot of emphasis on the use of face shields. This has been prompted by a number of significant injuries resulting from molten metal splash to the face, injuries which would have been avoided if face shields were in use. Polycarbonate face shields of 2 to 4mm thickness have been shown to offer good protection from molten metal splash. The more widespread usage of face shields in aluminium cast houses is expected over the coming years. Specialist Kevlar helmets with thick polycarbonate face shields designed for fire-fighters are in use in some cast houses, however the more simplistic design shown in the centre of figure 12 is by far more common.

Figure 12. Range of available face shields: (polycarbonate after molten metal explosion; common polycarbonate face shield; Kevlar helmet with heavy duty polycarbonate face shield

– @Z fire brigade helmet/ shield).

Mobile Phone Batteries

Although not an item of clothing a mobile phone, radio or pager can be regarded as an item of PPE. The 1986 Code of Practice did not address the safety issues (both good and bad) associated with such battery operated equipment. In 2009 an aluminium cast house needs to address the potential safety issue of metal hydride batteries exploding when exposed to excessive heat – such as


accidental immersion in molten metal. Although to the authors knowledge no aluminium cast house fatalities can be attributed to exploding batteries the warning messages printed on the batteries are quite explicit. Procedures to restrict the use and or address the potential risk need to be put in place. Such new technology (i.e. post 1986) also serves as a warning to evaluate all new materials and equipment introduced into the cast house environment in the post 2008 Compliance Code environment.

Conclusion

A number of advances in clothing design coupled with a healthy collegiate attitude has seen the improvement in personal protective clothing standards for all staff involved in molten metal activities in Victoria’s cast house, die casting shop and foundry environments. The recently released Foundry Compliance Code addresses the range of potential incidents and injuries associated with cast house environments in a comprehensive manner. The Code also acts as an informative and educational document written in a manner that is easily used across the whole shop floor. The Code emphasises the use of wool/ wool blend clothing, safety glasses/ face shields, foundry/ smelter specific footwear and gloves. The ongoing development of these items of PPE will doubtless continue however, cast house personnel can confidently reference the 2008 Foundry Compliance Code as the benchmark expectations.

References

[1] WorkSafe Victoria: Foundry Compliance Code, September 2008: http://www.worksafe.vic.gov.au/wps/wcm/connect/WorkSafe/Home/Forms+and+Publications/Compliance+Code/Foundries+CC .

[2] WorkSafe Victoria: Foundry Code of Practice, November 1986.

[3] US Aluminum association: Guidelines for handling molten aluminium, 3rd edition, 2002.

[4] Molten Aluminium Safety Guide for Foundries: Rio Tinto Alcan

[5] Workplace health and Safety Act 1995: Foundry Industry Code of Practice: Queensland Government: June 2004.

[6] Coroners Report Summary 3685/86 December 3, 1987: Public Record Office Victoria, Australia.

[7] WorkSafe Victoria data reported to the die casting and foundry industry in 2007/2008 via a series of industry association/ public talks.

[8] PR97 clothing (Barden clothing/ Melba Textiles): http://barden.com.au/edge.php?brand=edge

[9] AS/NZS 4602/1999: High visibility safety garments.

[10] Lee, Jones & Forakis (2005) Molten metal splash testing of cast house personal protective clothing, Aluminium Cast House Technology, Melbourne, Australia.

[11] Oliver boots web site: http://www.oliver.com.au/asp/show_products.asp?CatId=6.

[12] Leigh Kelly web site: http://www.leighkellyagencies.com.au/leighkelly.swf.


Greenhouse Emissions in Primary Aluminium Smelter Cast Houses- A Life Cycle Analysis

Paul Koltun1, Ambavalavanar Tharumarajah1and John Grandfield2,a

1 CSIRO Material Sciences and Engineering, Victoria, Australia

2 Grandfield Technology Pty Ltd, Brunswick West, Victoria, Australia CAST CRC, Brisbane, Australia

a [email protected]

Keywords: cast house technology, greenhouse gas emissions, life cycle assessment

Abstract

The aluminium industry is reducing its carbon dioxide emissions and environmental footprint. In order to identify and prioritise areas in the cast house where greenhouse gas emissions can be reduced it is necessary to quantify CO2e (CO2 equivalent tonnes) emissions for the various cast house operations. In this study two typical cast house layouts are examined. In one case, 22kg 99.85% aluminium remelt ingots are produced using chain conveyor ingot casting machines. In the second case, wrought alloy extrusion and rolling slab direct chill cast products are made. Both plants are sized at 500ktpa. The various process inputs in terms of energy and materials were identified and typical usage rates assigned. The results show that general electricity consumption, dross generation and furnace energy consumption are the three biggest areas of CO2e and should be targeted for improvement. Magnesium consumption also has a large effect in the case of the wrought alloy plant.

Introduction

Australia has very high greenhouse gas emissions per capita and is the thirteenth largest emitting country [1]. This is partly attributable to contributions from the energy intensive aluminium industry. The Australian aluminium industry’s CO2e emissions derive substantially from the reduction process either as PFC emissions, carbon anodes oxidised during reduction, and consumption of coal fired electricity. The industry is taking steps to reduce energy consumption per tonne of aluminium by going to larger more efficient reduction cells, examining waste heat recovery, using lower CO2 per GJ sources such as hydro-electric, natural gas and nuclear [2].

While primary smelter cast house operations are generally considered to emit only a small fraction of the total aluminium industry emissions, they are not insignificant compared to other industries. Furnace energy efficiency during melting operations has been the subject of study for many years and is receiving renewed attention in the light of aims to reduce CO2 emissions and increasing energy prices (see for example [3]). It is generally assumed that burning of fossil fuels in furnaces is the biggest CO2e emitting operation in the cast house, however there are other operations which may contribute significantly. Particularly in the case of primary smelter cast houses where there is only a small amount of metal melted it is not clear if furnace energy is the main CO2 source. In order to reduce carbon footprint it would be useful to identify and quantify the key activities in primary cast houses which emit CO2.

The main objective of this study was to examine and determine possibilities for reducing environmental impact of primary aluminium cast house operations within the constrained opportunities of meeting societal demands for aluminium products. This paper evaluates the carbon footprint of aluminium casting operations based on a LCA (life cycle analysis) activity based approach for two “typical” primary smelter cast house layouts. Case 1 - is of a simple plant producing 500,000tpa of aluminium 23 kg remelt ingot in pure and Al-Si alloy. Case 2 is of a more complex plant producing 300,000 tpa of extrusion alloys, and 200,000 tpa of rolling slab alloys


(Table 1). The work described in the paper is conducted following the international framework for life cycle assessment [4].

Table 1. Product mix for each cast house examined.

Case Products and tonnage output 1 22kg remelt ingots, 350ktpa of 99.85% Al, 150ktpa of 7% Si alloy 2 300ktpa of 6xxx alloy extrusion billets in diameters 178-330mm, 100ktpa

of 3004 alloy rolling slab, 100ktpa of 5182 alloy rolling slab

Goals and Scope of LCA Study

The main goals of the presented LCA study are: identification of processes, materials and systems that are major contributors to environmental impact (GHG Green House Gas emissions) and compare environmental performance of two different cases under consideration. This comparison is also extended to determine the major contributors of GHG emissions from cast house operations. The study includes only operations within the cast house plants and recycling of dross and transportation of dross between them.

The impact categories that have been selected to report the study are global warming potential (GWP) that deals with the emission of GHG emissions and energy consumption due to their importance for the aluminium industry. The life cycle impact and any comparison is expressed in terms of annual operation for each cast house and on a per tonne basis.

Product System, System Boundaries and Assumptions

For case 1, the assumed layout is 8 holding furnaces feeding 4 ingot casting conveyors (Figure 1). Liquid metal is delivered from the reduction cells to the furnaces. For the production of Al-7% Si alloy ingots silicon is added to the furnace. The Al-Si alloy product is assumed to be degassed with argon and filtered inline.

Figure 1. Generic “gate-to-gate” operations and material flows for aluminium ingot casting cast house 1.


For case 2 a range of wrought alloy products is produced by direct chill casting (Figure 2). Four DC casting units and 10 furnaces are assumed. All these products are degassed and filtered. Sodium removal is carried out using furnace fluxing with MgCl2. Typical alloy compositions were assumed for the 6xxx alloys i.e. a mix of soft and hard alloys.

Figure 2. Generic “gate-to-gate” operations and material flow for cast house 2 producing wrought alloy products.

The generic cast house life cycle systems (shown in Fig. 1 and 2) are considered to be applicable for casting aluminium ingots and extrusion wrought aluminium alloy within cast house. The emissions from production primary aluminium are not counted, as the aluminium is the primary product of the cast house operations. In line with the goal of the study only emissions due to activities to produce aluminium products within the cast house are counted including emissions from consumption of other materials such as lost alloying additions and auxiliary materials like refractory (for making furnaces), steel and iron, etc. are also counted. All energy required for production these materials and energy consumed within cast house are counted, as well as, required water (so called “virtual” water). The only transportation emissions are included only for transportation of aluminium dross to the dross processing plant and transportation back recycled aluminium, as dross processing is included into the system boundary.

Assumptions

To conduct the study, some simplifications have been made due to availability of data and the complexity of processes itself. It was assumed that all electricity is coming from coal fired power stations. Energy consumption is the only input assumed for recycling the aluminium from dross. Only internal transportation within plant and transportation for dross recycling operations has been taken into account with assumption of transport distance to dross recycling facilities of 300km (two ways) made by road trucks. The following emissions within the study are considered to be negligible: a) emissions due to inputs of auxiliary materials (such as chemicals for water treatment, etc.); and b) emission created from the construction, maintenance and renewal of facilities and equipment, except for refractory for furnace maintenance and replacement of casting moulds. Other assumptions are;


1. Magnesium alloy recovery is 85% and Chinese Pidgeon process magnesium is used. Silicon recovery is assumed to be 95% and 98% recovery is assumed for Mn, Cu and Cr.

2. Cast iron mould life on the ingot conveyors is 18 months. The moulds weigh 50kg and there are 260 moulds per casting line.

3. Siphoning is used to transfer metal from reduction crucibles to the furnaces.

4. Dross generation is assumed to be 1.5% of production for case 2 and 1.0% for case 1. The higher dross amount for case 2 is due to the high magnesium content in the wrought alloys, the additional fluxing operations and the higher scrap remelting operations. See [8] for typical data.

5. Metal recovery from the dross is 50% of the dross weight i.e. a typical figure reported in the industry.

6. For Case 1 the furnace energy consumption is assumed to be 300MJ/tonne. This would be a reasonable typical figure for the industry [5] considering a 1% scrap return. The theoretical figure for melting Al is 1.1 GJ/tonne. For Case 2 a higher figure of 500MJ/tonne is assumed because of the higher 7% scrap return and the longer residence in the furnaces due to alloying and fluxing operations.

7. Steel strapping is assumed for the remelt ingot product.

8. In the degassing operations, a typical figure of one litre of argon per kg of melt treated is assumed.

9. Refractory consumption is assumed to be higher for case 2 because of the presence of magnesium in the alloys, more furnaces with more furnace operations and more refractories used in the casting process. Reference [6] is used as a guide.

In computing of the emissions, each process and material involved in production is considered bounded (i.e. treated in isolation) and the environmental burden from one to the other is assumed to occur inside the boundary. Most of the data for cast house operations, materials production, energy and water consumptions have been obtained from published data and directly from cast houses. All emissions have been obtained by using cast house operation models in SimaPro database [7].

Results

The total GHG emissions and energy consumption, as well as GHG emission sand energy consumption by cast houses themselves are presented in Table 2 and Table 3. We assume 0.3kg CO2e from coal fired electricity. Case one producing only remelt ingots has about 1/3 the GHG emissions of case 2 wrought alloy cast house. Dross generation is the biggest greenhouse gas emitting activity for the Case 1 cast house because the energy consumption in the furnace is lower. The greenhouse emissions associated with production of the liquid aluminium are the reason dross plays a significant role. For case 2, furnace energy, dross and plant power are the three major GHG emitting activities with magnesium consumption also contributing significantly.


Table 2. Annual material consumption for the two cast house examples.

Materials/Energy (Input/Output) Case 1 Remelt ingots Case 2 Wrought alloys Liquid aluminium (Al), t 489,466 487,236

Magnesium (Mg), t 882 9882 Silicon (Si), t 11,052 2,210 Copper (Cu), t — 300

Manganese(Mn), t — 2,322 Chromium (Cr), t — 200 Flux (Mg Cl2), t — 2,000 Argon (Ar), m3 150,000 500,000 Lubricants, m3 50 1,000 Refractory’s, t 100 2,000

Cast iron (moulds), t 265 265 Steel (packaging), t 1,700 —

N.Gas, TJ 200 1,400 Electricity, TJ 117 240

Water, Ml 281 240 General waste, t 100 100

Dross, t 5,000 7,500 Recycled Al (from dross processing) 2,500 3.750

Energy for dross processing, TJ 10.8 16.2

Table 3. GHG emissions (kg), embedded energy and water consumption per production of 1t of aluminium alloy product due to cast house activities.

GHG emissions/Embodied Energy and Water Remelt Ingot cast house

Wrought alloy cast house

GHG emissions due to furnace heating 25.7 179.8 GHG emissions due to other activities 6.6 44.9

GHG emissions due to electricity consumption 70.2 144.0 GHG emissions due to aluminium melt loss 94.3 141.9

GHG emissions due to magnesium consumption 8.3 93.5 GHG emissions due to other materials consumption 16.6 19.1

Total GHG emissions 221.7 623.2 Primary energy consumed within cast house (including

electricity), MJ 1269 5056

Primary energy embodied in lost aluminium, MJ 1176 1769 Primary energy embodied in consumed magnesium, MJ 66 744

Other primary energy (embodied in other then Al and Mg materials, dross recycling, etc.), MJ 345 419

Total embodied primary energy, MJ 2856 7988 Water consumed within cast house, kl 0.56 0.48

Total embodied water, kl 12.48 33.53


Furnace heating

12%Other

activities3%

Electricity consumption

32%Aluminium

melt loss42%

Magnesium consumption

4%

Other materials

consumption7%

Case 1

Furnace heating

29%

Other activities

7%

Electricity consumption

23%

Aluminium melt loss

23%

Magnesium consumption

15%

Other materials

consumption3%

Case 2

Figure 3. Proportion of GHG from various activities with cast house case 1.and case 2.

Discussion

The analysis shows that although the use of fossil fuel for furnace heating is a large contributor to GHG it is not the largest. For case 1, the melt loss is the biggest contributor (Figure 3). Interestingly the electricity used in the cast house emits more GHG than the fuel used in the furnaces.

Limitations and Variations

The analysis is based on indicative data from the industry and knowledge of typical practice. The actual proportion of emissions coming from different activities for a specific cast house could easily vary considerably from these calculations.

If hydro electric power is used together with electric furnaces then melt loss and magnesium consumption would make up ~2/3 of the total cast house emissions.


Opportunities for CO2 Emission Reduction

It appears from this analysis that cast house activities can make up to 5% of total GHG emissions associated with primary aluminium production.

Reduction in dross generation and an increase in melt recovery seem to be a high priority in terms of reducing GHG for cast house operations for both case and case 2. Increasing furnace energy efficiency will also lead to reduced GHG emissions. For case 1 type cast houses another option is to cast directly from crucibles to the casting process i.e., eliminate the furnaces and the energy and materials consumed there. Waste heat recovery might also be considered for example using launder cooling and of course using electricity sourced from renewable energy sources would also significantly reduce GHG. Reducing electric power consumption in general across the plant also deserves examination. For case 2 type cast houses increasing magnesium recovery would also reduce GHG significantly.

Benchmarking the industry for furnace energy consumption, dross generation and magnesium recovery would enable better analysis and identify GHG reduction opportunities.

Sequestering Opportunities

Another approach to reducing GHG would be for the cast house to sequester the CO2 emissions.

CO2 could be captured from furnace off gas using the same technology under development for the coal fired power stations. The use of evaporative coolers in the cast houses results in production of briny water and the necessity to take in clean water for makeup. The briny water presents a potential CO2 fixing opportunity whereby carbonates are precipitated out of the water thus fixing the CO2 and reducing the need for make up water.

The cast house activities make a small but significant contribution to overall GHG emissions to produce the liquid aluminium (~22 t/t). For case 1 it is about 1% and for case 2 it is about 3%.

Conclusions

1. Cast house activities make up a small but significant fraction of the total emissions associated with production of aluminium products.

2. The three main GHG emitting activities in cast house operations are furnace energy consumption, dross generation and plant power.

3. Magnesium consumption makes a significant contribution in the case of the wrought alloy cast house.

References

[1] Garnaut, R. (2008) The climate change review, Cambridge University Press.

[2] Evans, J. & Kvande, H. (2008) Sustainability, Climate Change, and Greenhouse Gas Emissions Reduction: Responsibility, Key Challenges, and Opportunities for the Aluminum Industry, JOM, Vol. 60 (8) 25-31.

[3] Eckert, C.E., Osborne,M. Peterson, R.D (2009) Improving Energy Efficiency in a Modern Aluminum Casting Operation, Energy Technology Perspectives, TMS, p149.

[4] ISO 14040 International Standard ISO 14040: Environmental Management – Life Cycle Assessment – Principles and Framework, Geneva, Switzerland: International Organization for Standardization, 2006.


[4] "Energy efficiency best practice in the Australian aluminium industry", Industry, Science and Resources, Energy Efficiency Best Practice Program May 2000, Australian Commonwealth Government

[6] Krüger, J.Rombach, G. (1997) Understanding of sustainable primary and secondary aluminium production. In: 8th International congress of ICSOBA; Milan, Italy; ed.by AIM; S. 290-298

[7] Pré Consultants (2008) SimaPro, Amersfoort, The Netherlands, http://www.pre.nl/simapro/, accessed 30 January, 2008.

[8] A.G. Clark, Furnace dross prevention, melt loss reduction and dross recycling: a review of best practice, 3rd International Melt Quality Workshop, Dubai, UAE, 14-16th November 2005


CHAPTER 3:

Dross Generation & Handling

Predicting Dross Formation in Aluminium Melt Transfer Operations

J.A. Taylor1,a, M. Prakash2, G.G. Pereira2, P. Rohan3, M.J. Lee4 and B. Rinderer5

CAST Cooperative Research Centre (CAST CRC)

1 School of Engineering, The University of Queensland, Brisbane, QLD 4072, Australia

2 CSIRO Mathematical & Information Sciences, Clayton, VIC 3169, Australia 3 CSIRO Materials Science & Engineering, Clayton, VIC 3169, Australia

4 formerly Swinburne University of Technology, Hawthorn, VIC 3122, Australia 5 formerly Rio Tinto Alcan Pacific Technology Centre, Thomastown, VIC 3074, Australia

a [email protected]

Keywords: aluminium, oxidation, dross, modelling

Abstract

Aluminium melt transfer operations can lead to significant amounts of dross formation as a result of chemical oxidation and physical entrapment processes. It has been suggested that these activities may contribute up to 50% of the total metal loss of ~1% in a typical primary aluminium smelter (i.e. 2,500 tonne/annum (tpa) in a smelter of 500,000tpa output). This is a large financial loss to any company, and also, in the new CO2-conscious era, it also represents a significant carbon footprint to ameliorate. A significant proportion of this metal loss may be prevented by adopting more efficient melt transfer strategies that reduce splashing and turbulence thereby resulting in reduced oxide and therefore dross formation. Optimisation of such systems is normally achieved by trial-and-error approaches, however a clear opportunity exists for rapid optimisation by employing computational modelling to explore the effects of changed equipment design and process conditions, such as tilt speed, spout height, spout geometry, etc. In the present paper, the Smoothed Particle Hydrodynamics (SPH) modeling method is used to predict the amount of oxide generated during molten metal transfers from a 500kg capacity tilting crucible furnace into a heated sow mould. Various conditions were tested. An oxidation model based on skimming trials performed in a laboratory-scale (8kg) oxidation rig is employed in the simulation. The predicted oxide from the simulations is compared against those of the experimental pours. It is anticipated that the validated model will be used for modifying the design and optimizing the operation of various melt transfer operations occurring in the aluminium industry.

Introduction

Oxidation of aluminium, dross formation and melt loss in aluminium smelter casthouses have been discussed previously in a general overview paper by Taylor [1]. That review dealt with a broad range of issues from the science of oxidation of aluminium and its alloys through to practical steps that may be taken to reduce dross formation through careful melt practices. It is not the intention of the current paper to restate these matters, and therefore the reader is referred to that paper if a basic understanding of these issues is sought.

This paper instead aims to describe a research approach taken by CAST CRC to more fully understand the amount of oxidation, and hence dross formation, that occurs during melt transfer operations. Clark and McGlade [2] carried out a survey of the Australian and New Zealand primary aluminium smelters and found that it was accepted amongst casthouse practitioners that 80% of melt losses occurred within the casting furnace, and that 60% of those losses were generated during furnace filling procedures, typically “cascading” events (Figure 1). Consider the case for processing 50tonnes of aluminium and assuming a typical 1% dross level formed during processing, then 240kg (or 48% of the total dross) would result directly from the furnace filling activity. Further,


assuming all of this dross formation resulted from the full conversion of Al to oxide, this represents a nett loss of 127kg metallic Al and 113kg weight gain due to oxygen uptake1. Since some of the dross is likely to be unrecovered metal, the above calculation is an over-estimate.

0

10

20

30

40

50

60

70

80

90

100

Furnace Casting Crucible Launder

Mel

t lo

ss c

ontr

ibution (%

)

0

10

20

30

40

50

60

70

Filling

Treatment

Stirring

Dross removal

Alloying

Scrap additions

Sludge removal

Cleaning

Mel

t lo

ss c

ontr

ibution (%

)

Figure 1. (a) Graph showing the estimated breakdown of melt losses occurring within the

casthouse; (b) Graph showing the estimated breakdown of the furnace-based melt losses [2].

Using 0.48% dross as typical for a pouring event, it can be predicted pro-rata that the transfer of smaller amounts of metal (e.g. 300kg) by cascading over a significant vertical fall (e.g. 1metre) should result in measurable amounts of oxide being formed during the process. The upper estimate of the overall weight gain due to oxygen uptake is estimated at 0.67kg (based on 1.44kg dross).

CAST CRC is carrying out a project to understand oxide formation during melt pouring events with the ultimate aim of developing technologies to minimise dross formation and melt loss. The project involves several aspects, of which a few are reported here in general terms. These are:

1. Establish a reliable oxidation algorithm for molten aluminium exposed to air;

2. Build a medium-scale (500kg max) instrumented melt transfer rig that can detect expected weight gains associated with oxidation during a turbulent pouring event;

3. Develop a modelling method that can adequately predict oxide formation on flowing melts;

4. Validate the model against experimental data;

5. Use the modelling methodology for complex industrial-scale problems and develop appropriate solutions.

Developing an Oxidation Algorithm

After an extensive survey of past experimental approaches of measuring oxidation rates, it was decided to employ a method used and partially-described by Freti et al [3] in 1982. This involved constructing a bath that could hold molten aluminium at a constant temperature (Figure 2a) and that could be skimmed across a significantly-sized exposed surface area (0.05m2) quickly and cleanly on a repeated cyclic basis from intervals ranging from a few seconds to several hours (Figure 2b). The molten metal surface was fully exposed to air. The skimming blade produced a folded skin of oxide with entrapped metal that was removed immediately and cooled to prevent further oxidation. Several skims were collected for each time interval, allowing the representative exposed surface area to be increased several fold. Typically ten skims, representing 0.5m2, were collected for each of the shorter time intervals (up to 1hr), with fewer samples taken for longer intervals.

1 Melt loss and dross levels are defined and calculated in various ways depending on the particular casthouse concerned. Clarke and McGlade [2] acknowledge this in presenting their industry data. The values calculated in this section are therefore based on our own assumptions and may not necessarily match the assumptions of others.


(a)

(b)

Figure 2. (a) General arrangement of the skimming rig (with optional lid shown) constructed at University of Queensland; (b) Oxide skim being collected by the pulling the skimming

blade across the melt surface at regular time intervals.

These skims were weighed and then analysed for metal content using a specially-formulated molten salt mix to separate the oxide from the aluminium. The molten mixture was then chill solidified, the crystallised salt dissolved away in warm water and the metallic droplet(s) residue was dried and weighed. A comparison between the original skim weight and the recovered aluminum weight gave the amount of oxide formed for a given exposed surface area for a given exposure interval, thus giving an oxidation rate. Various test conditions were used and the data collected and analysed to determine oxidation rate as a function of time.

The experimental data obtained at The University of Queensland (UQ) for commercially pure Al (99.7%) melt between 750 and 850ºC yielded a general equation for overall oxidation rate, OOR, (i.e. for a given exposure time, t) of the power law type (Equation 1), where a and b are fitting constants derived from the straight line obtained on a log-log plot of OOR versus t.

[ ]12 −−−⋅= smgtaOOR b (Eqn. 1)

For this equation to be useful for subsequent computational modelling (see below), it was necessary to extrapolate the relationship to much shorter times (i.e. small fractions of a second) than it was possible to collect data from experimentally. This would clearly present a serious problem unless the extrapolation could be verified. However, two different forms of data were found that correlated very well with the necessary time-scale extrapolations. Longer-time scale data was obtained from published data for oxidation of pure Al melts for up to 16hrs for temperatures ranging up from melting point to 1000ºC. Shorter-time scale data was estimated from oxidation rates for fine-scale aluminum powder particles of various sizes that were calculated based on their measured oxygen contents and known production parameters [confidential source].

Medium-Scale Pouring Trials

The availability of a 500kg tilting crucible furnace at CSIRO Materials Science & Engineering was considered to be ideal for exploring oxidation in medium-scale sized cascading events. A sow mould capable of holding the same amount of Al was constructed to receive the metal while being suspended from a supporting sling. Three high sensitivity load cells were installed under the tilting crucible frame, and a single load cell was installed in the supporting system for the sow mould. These four load cells were connected to a high performance data gathering and processing device. The basic equipment is shown in Figure 3.


Figure 3. The 500kg tilting crucible furnace and slung sow at CSIRO that was used for the medium-scale pouring experiments.

From the outset, it was found that there were issues of signal noise and drift that would make mass change measurements difficult. A considerable amount of time and effort were expended in reducing these problems, and it was found, using static and transferred solid weights, that the accuracy expected during a liquid pour trial was approx. ± 25g on either side of the system (i.e. the crucible with metal to be transferred, and the sow with received metal) or ± 50g across the whole system. It was also evident that real-time measurement would not be possible (due to errors induced by furnace and sow movement) and would have to be replaced with simple static start and finish measurements.

Aluminium ingots (99.7% Al purity) were charged into the crucible furnace and melted. The coated sow mould was pre-heated in an oven to 500ºC (to prevent premature solidification of transferred metal). When the melt temperature was stabilised, the heated sow mould was moved into position and suspended from the sling. Once load cell outputs had settled and start weights recorded, the pouring operation commenced. Various trial parameters were used: temperature, fall height, total mass transferred and pouring rate. When the transfer was complete, the furnace was returned to full back position, the sow was steadied, and the finish weights for both sides were recorded. The sow was then skimmed in order to remove and collect the accumulated dross for future reference/analysis (if required). Once the metal was skimmed, steel dividing bars were placed into the solidifying mush to make it easier to later cut up the solid sow for remelting.

It soon became apparent during the trials that there was no measurable weight gain resulting from the cascading transfer process across the full range of testing conditions. This was in stark contrast to expectations (as described above). The implications were that either the experimental system was unreliable, or that the actual amounts of oxide produced were quite small, i.e. below our detection level of ~50g.

Since, dross skims had already been collected from each of the trials, it was decided to analyse them for metal content. Most collected skims were approx. 0.5 – 1kg mass, far too large for analysis at UQ, so they were sent to Rio Tinto Alcan for molten salt assay analysis. When the results came back, it was surprising to see that the skim samples typically returned 96 - 97% metallic content and that therefore they each only contained around 15 - 45g of actual aluminium oxide. This amount of oxide was clearly below the detection limit of the experimental rig and was also one to two orders of magnitude less than expected. The next step was to ascertain what amount of oxide formation our selected modelling technique (see following section) would predict for the simulated CSIRO test scenarios.


SPH Modelling Approach

Smoothed Particle Hydrodynamics, SPH, is a Lagrangian (grid-free) computational method of modelling heat and mass flows that is well suited to simulating free surface flows such as molten metal transfer. In SPH, materials are approximated by particles that are free to move around rather than by fixed grids or meshes. The particles are basically moving interpolation points that carry with them physical properties, such as the mass of the fluid, its temperature, enthalpy, density and any other relevant properties, such as degree of oxidation. The inter-particle forces are calculated by smoothing the information from nearby particles to ensure that the motion of the model is consistent with that of the real fluid.

One particular advantage that SPH has over other competing methods for modelling industrial heat and mass flows is that complex free surface and material interface behaviour, including fragmentation, can be modelled easily and naturally and can produce accurate predictions with no mass lost or numerical diffusion of the interfaces. SPH has been applied successfully by CSIRO Mathematical & Information Sciences on behalf of CAST CRC to model filling of high pressure die casting cavities [4] and to optimise the design of an ingot caster filling system to reduce oxide formation during aluminium ingot casting [5].

In the current case, a detailed 3D model of the CSIRO furnace and sow were constructed. Various cases were simulated to match the experimental conditions used in the tipping experiments. A sequence of selected images from one of those simulations is shown in Figure 4.

Figure 4. Sequence of SPH images simulating a metal transfer using the CSIRO test rig with a fall height of 920mm, cast pouring rate of 5.5kg /s, and a total mass transfer of 275kg.

The SPH modelling technique keeps track of the free surface area that is created whenever a particle is exposed to air. When a particle is initially exposed at a melt surface, it is considered free of oxide; however, from that point on, oxidation occurs and the amount of oxide associated with the particle is tracked as a function of time. Note that particles stop oxidising if they cease to be exposed on a free melt surface, but recommence oxidation (with the appropriate time rate at which they ceased) when/if they re-emerge onto a free surface again. The oxidation model applied to the SPH simulations is the power law relationship as determined in the UQ study (see Eqn. 1). In this

T=0 sec T=27.5 sec

T=32.5 sec T=47.5 sec


way, SPH is able to keep track of the total amount of oxide formed in the system as a function of time.

Initial Model Validation

Examples of the output delivered from the SPH simulations, using the various CSIRO test conditions and the UQ oxidation algorithm are shown graphically in Figure 5. In these graphs, it can be seen that in each test case (i.e. represented by the differently numbered lines) the oxide formation tends to develop in three distinct stages:

1. Slow oxidation of the melt surface exposed during the standing and initial tilting of the crucible, before the melts starts to cascade;

2. Rapid oxidation rate during the actual cascading flow of metal from crucible to sow;

3. Diminishing oxidation rate of the metal as it is gradually settles into quiescence in the sow.

(a) 720mm fall height; low sow weight

(b) 720mm fall height; high sow weight

(c) 920mm fall height; low sow weight

(d) 920mm fall height; high sow weight

Figure 5. Graphical output showing predicted oxide mass (kg) versus time for two fall heights and two sow weights (low 80kg, high 300kg). The different numbered lines on the plots reflect

other test variables such as melt temperature and pouring rate that are not crucial to the present discussion.

These values are compared with the corresponding experimental data from the CSIRO pouring rig samples (i.e. the collected dross skims analysed for metal/oxide content) for the low sow weights in Figure 6. It can be seen that the SPH simulations predict oxide mass values that are of a similar order of magnitude to those obtained experimentally rather than the much higher values expected from industrial dross experience (i.e. based on typical ~0.5% melt losses associated with furnace filling, with drosses containing only 30-80% recovered metal content). Not only are SPH simulated


oxide values of the same order of magnitude as experimental values, they are also consistently under-predicted by an average of 40% (full range 6 - 71%).

Under-prediction of simulated values is not a particular problem in this instance, since the simulations end with a quiescent melt and do not simulate the complex act of collecting the dross skims from the melt. This can involve several passes with the skimming tool resulting in the generation of a large amount of fresh molten Al surface that re-oxidises, thus increasing the amount of oxide present. Additionally, the collected skim contains significant levels of entrapped metal which when removed from the melt as dross may be subject to further oxidation until the mass finally solidifies. It is not surprising therefore that experimental oxide levels are consistently higher than the simulated oxide levels.

720 mm fall height; low sow weight

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0 1 2 3 4 5 6Cast Rate (kg/s)

Oxi

de

Conte

nt (

kg)

Experiment

Simulation

920 mm fall height; low sow weight

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0 1 2 3 4Cast Rate (kg/s)

Oxi

de

Conte

nt (

kg)

ExperimentSimulation

Figure 6. Plots of oxide content (simulation-predicted and experimental) versus casting rate for low sow weight trials (~80kg) at two different fall heights; 720mm (left), 920mm (right).

What is surprising though is that both simulated and experimental values are much smaller than indicated by industrial expectations. This indicates that the actual physical activity of “cascade” pouring several crucibles into a furnace in order to fill it for casting is probably not the most significant oxide contributor to dross generation in the overall furnace filling process; certainly not if Clark & McGlade’s estimate of 48% of total melt loss is correct [2]. Most furnace filling scenarios (even those using siphoning techniques) involve punctuated activity with periods of prolonged oxidative exposure in between crucible transfers. This exposure involves several different aspects:

1. Large general furnace bath surface area with normally-growing oxide skin; 2. Regions of convoluted, folded oxide skins formed from each transfer that float and drift around

the furnace as oxidisable “islands”; 3. Patches of oxide skin that have developed breakaway oxidation characteristics over time; 4. High near-surface temperatures (and sometimes complex atmospheres) in operating furnaces,

compared to constant source of ambient fresh air above the test rig melt surface.

Future Directions

Initial indications are that CAST CRC has developed a useful oxidation algorithm that when used in conjunction with the SPH computational modelling technique is capable of predicting the approximate amount of oxide that is generated in a simple, single, medium-scale pouring event. It is clear however that much more needs to be done in order to take this tool and use it to predict dross levels in industrial furnace situations.


It is our intention to further refine the oxidation algorithm so that other parameters such as melt temperature, alloy composition and furnace atmosphere can be included. It is also intended that the SPH modelling technique will be improved by the incorporation of greater detail and resolution in the plunge/splash region where the metal stream impacts the bath. Additionally, further medium-scale pouring trials will be required to validate the predictive capability of such changes.

Up-scaling the SPH modelling technique to simulate industrial furnace filling by multiple crucible transfers is some way off, although verification of the method using a single pouring event followed by capture and analysis of the oxide formed from that event should be possible. Clearly a much deeper appreciation of the complexity of oxidation within the furnace environment is required before this computational modelling method can be used to optimise equipment and processes.

Acknowledgement

Dr Paul Cleary of CSIRO is acknowledged as the principal developer of the CSIRO SPH modelling code, and the authors are grateful to him for his advice concerning this particular application.

References

[1] Taylor, J.A. (2007) Oxidation, dross and melt loss issues involved in the aluminium cast house, Proceedings of Aluminium Cast House Technology, 6-9 August 2007, Sydney, Australia, 47-55.

[2] Clark, A. and McGlade, P. (2005) Furnace dross prevention, melt loss reduction and dross recycling: Review of best practice, 3rd Inter. Melt Quality Workshop, Dubai, UAE, 14-16th Nov 2005, 8 pp.

[3] Freti, S., Bornand, J.D. and Buxmann, K. (1982) Metallurgy of dross formation on aluminum melts, Light Metals 1982, TMS, 1003-1016.

[4] Cleary, P.W. and Ha, J., Ahuja, V. (2000) High pressure die casting simulation using smoothed particle hydrodynamics, Int. J. Cast Metals Res., 12, 335-355.

[5] Prakash, M., Cleary, P.W., Grandfield, J.F., Rohan, P. and Nguyen, V. (2007) Optimisation of ingot casting wheel design using SPH simulations, Progress in Computational Fluid Dynamics, 7, 101-110.


Furnace Operations to Reduce Dross Generation

Alan M. Peel1,a, James Herbert2, David Roth2 and Martin J. Collins1

1 ALTEK-EUROPE Ltd, Station Road, Castle Donnington, Derbyshire, DE74 2NJ, UK 2 ALTEK-MDY, 825 Springdale Dr, Exton, PA 19341, USA

a [email protected]

Keywords: aluminium, dross, cast house, furnace

Abstract

While it is generally acknowledged that dross generation should be kept to a minimum, too often,

the importance of maximising the aluminium content of the dross is overlooked. Some mistakenly

believe that a low metal content is a good thing and that the aluminium is being kept in the furnace.

In reality, this metal is most likely being lost due to insufficient cooling and thermiting.

Much can be gleaned from looking at the dross that is generated in a casthouse; in fact, the quality

of dross can provide a good indication of the overall efficiency of the operation. Even with the very

low aluminium prices of today, of circa. US$1400 per tonne, a reduction in dross generation within

the furnace can provide huge savings per year. Effective dross management also results in better

metal quality, improved fuel efficiency, prolonged refractory life and improved yield in the entire

facility.

This paper will look at how dross is generated within the furnace in the first place, followed by

ways to minimize the dross generation within the furnace using continuous and sub-surface

circulation which can also provide significant energy and CO2 reductions. A separate paper will

discuss dross processing options and possibilities.

In summary, by careful attention to the equipment and process techniques around the furnace and

the follow-on dross management significant cost savings and environmental benefits can be realized

by cast house operations.

Dross Generation within the Furnace

Furnace Bath Analysis

As we know the predominant heat transfer mode from the combustion space in a furnace to the

surface of the aluminum bath is via radiation, and because the aluminum bath is opaque, then the

predominant mode of heat transfer from the bath surface to the bath itself is via convection and

conduction. Hence any analysis of the heat transfer of the bath has to include the hydrodynamic

analysis (molten metal flow) of the bath.

To illustrate the heat transfer dependency between the combustion space and the bath, Figure 1,

shows a typical cross-section of a reverb furnace, where heat is supplied to the bath surface by a

chemical flame, and the dominant heat transfer mode in the furnace is the thermal radiation. The

thermal radiation is transferred both directly from the flame (combustion gases), and from the

heated refractory walls.

This link between the various furnace zones and physics are further complicated by the fact that the

aluminum bath surface heat transfer characteristics change with time, furnace oxidizing atmosphere,

amount of heat transferred to the bath surface, and amount of impurity in the charged scrap (if any).

These changes in the heat transfer characteristics are triggered by a chemical change in the

composition of the aluminum bath at the bath surface layers. This chemical change is commonly

known as aluminum oxidation or dross layer formation. As is well known, this oxidized layer,

despite the fact that it is thin in its size, impacts on the heat transfer to the bath significantly, and at


its extreme can cause significant reduction to the furnace overall thermal efficiency and even

damage to the furnace refractory due to overheating.

Q_rad_wall

Q_rad_gas Q_rad_reflec

ted

Figure 1. A typical cross-section of a reverb furnace.

During the melting cycle (in a scrap melting furnace) the surface temperatures and the factors

defining the heat transfer to the bath change continuously and in a non-linear way. It is well known

that the molten metal bath, as it circulates inside the high temperature and oxidizing atmosphere of

the furnace, forms a thin oxide layer that covers the aluminum bath surface. This thin oxide layer

(typically the first surface layer forms in milliseconds) as it grows in thickness changes completely

the optical properties of the molten aluminum surface. [1]

The prolonged exposure of the bath surface to heat causes the thin flexible oxide layer to start

undergoing a physical change from its amorphous flexible structure to the crystalline rigid structure.

Unlike the amorphous structure the crystalline structure does not allow for flexibility in its

microstructure and during that phase the oxide layer starts to rupture (in microstructural terms, as

the crystal lattice slips along preferred fault lines) and further exposes more of the pure aluminum

metal below to the oxidizing atmosphere, and hence transforms it also to alumina (Al2O3) in a

process that is continuous. This transformation problem arises mainly when the bath receives

excessive heat from the flame, over a prolonged exposure time in the presence of oxygen

molecules, which is the case when the bath is stagnant in a furnace. The presence of impurities

(from melting coated and dirty scrap for example), which already have crystalline oxide structure

causes further acceleration to this transformation process (i.e. acts as a seed for the crystalline

transformation). Hence the oxide dross layer keeps getting thicker and accordingly the thermal

resistance keeps steadily increasing (see Figure 2). This increase in the thermal resistance along

with the increase in the surface emissivity and the reduction in the surface reflectivity causes even

larger heat deposits into the oxide surface, which in turn causes further oxidation. This process of

increased thermal resistance causes less heat dissipation into the bath depth, which in the absence of

effective circulation yields to a large temperature gradient, where the bath surface temperature is

considerably higher than in the bath depth. This causes a reduction in the furnace thermal

efficiency, along with refractory temperature rise.

The lack of bath circulation causes a cycle that continues in one direction, and as follows; higher

bath surface temperatures lead to excessive dross generation, which lead to higher thermal

resistance, which lead to a higher temperature gradient, which cause further dross generation at the

surface. This cycle can only be broken via furnace surface dross skimming. However, using

skimming tools for bath stirring involves door opening, interruption of the melting cycle, furnace

cooling, and entrainment of air (oxygen) into the furnace. Hence a process that is wasteful in

energy, time, and manpower.


Thermal Resistance vs Residence Time

0

4

8

12

16

20

0 5 10 15 20 25 30 35

Residence Time

Th

erm

al

Res

ista

nc

e [

C/W

]

Increase Bath

Circulation

Figure 2. Bath surface thermal resistance vs. residence time.

Continuous bath circulation metal stirring devices like the one shown in Figure 3 below reduce the

intensity of this cycle, and if applied correctly, could extend the early melting phase with reduced

contribution from the later (thick oxide layer) phase [1]. The processes by which the bath

circulation could potentially improve the melting cycle are:

1. The bath circulation ensures that the region of the bath surface that receives high intensity

heat flux constantly moves, and accordingly, the molten bath does not get exposed to

excessive heat for prolonged periods of time.

2. The bath circulation also ensures that the lower depths of the molten bath constantly move

and aids in the dissipation of the heat, hence minimum heat accumulation at the bath surface

will occur.

Figure 3. Bottom mounted EM Stirring.

Technologies for Reducing Oxidation and Dross Generation

Opportunities for reducing dross can come in a multitude of areas depending on the generator‟s

existing technology and practices. It should be recognized that furnace design, burner location and

type all can have a major influence on dross generation and the appropriate style of furnace should

be selected for the particular scrap types that will be melted or re-cycled. This is relevant for the re-

melt operation, the re-cycler and the primary cast house alike. Furnace and burner designs are a

topic in their own right and it is not the intention of the author to provide an in-depth review on this

subject within this paper.


Other Contributors to the Generation of Dross

Scrap Type and Quality

There is an old rule in the aluminum industry that for every 1% of contamination charged into a

melting furnace, there will be at least an equivalent 1% melt loss [2]. The scrap type and makeup

will therefore make a significant difference with regard to dross generation. While it is not always

possible to choose the type of scrap that is charged into the furnace, it should be recognized that the

level of contamination (water, oil, paint, plastic, dirt, etc.) will hinder the melting process and

reduce the recovery of metal that is present. This paper does not intend to discuss the methods of

scrap preparation and there are many reference papers that have discussed this in greater detail.

Furnace Charging

Charging a furnace is the next important step in control of the dross generation. It is always an

advantage to be able to submerge lighter scrap directly under the molten metal. Depending on the

types of scrap and the furnace being utilized, this is not always possible; however, as a general rule,

light scrap should be protected from direct burner impingement.

Burner Technology and Burner Control

The selection and type of burners used in a furnace and their control during the cycle is very

important and there is often a fine balance between providing sufficient heat transfer to meet the

demands of production while at the same time minimizing oxidation of the metal.

All burners will produce some quantity of dross which will be generated from one of two sources;

direct impingement of the flame with either the charge or the molten metal or, from the creation of

hot spots below the flame on the molten metal surface. The area directly below a burner can be

subject to overheating and wicking of the molten metal into surface dross, this generates further

oxidation and thus more dross. The increase in dross buildup insulates the metal from the heat and

will require harder firing; this will again generate more dross. It‟s a vicious cycle!

As previously stated, movement of the molten metal through stirring will also help prevent these hot

spots.

Furnace Skimming

Skimming removes most of the dross from the furnace allowing for more efficient melting and

temperature control. When and how to skim is very important and will affect the overall aluminium

recovery that is experienced from dross that is removed. Timing is everything, skimming too late

can cause dross to build up, reducing melting efficiency and causing overheating of the molten

surface which ultimately generates more dross.

Traditionally, forklift trucks equipped with homemade tools are used in the aluminum industry to

carry out routine furnace operations such as skimming and cleaning. While functional, the results

are inconsistent and reliant on the judgment and skill of the operator. Improper skimming technique

can pull a significant amount of aluminum out of the furnace. It is far more economical to keep as

much of the aluminium in the furnace than it is to remove it with the dross and recover it in other

ways.

Dedicated furnaces skimming machines can provide a far more economical solution to the

traditional forklift truck. Such systems are specifically designed for the arduous casthouse

environment and have a useful life in excess of 10 years. Precise control is possible with such

technology enabling the operator to minimize the amount of metal removed from the furnace.


To Flux or not to Flux

Fluxes are used for a variety of purposes in the aluminium melting and holding furnaces. There is

much debate and opinion however, on whether to use fluxes in the reverberatory melting furnace or

side well melting furnace to aid aluminium recovery from the dross. In general, a flux is typically

used to separate the oxides and dirt from the free metal. Exothermic fluxes were widely used in the

past to “heat-up the dross” and make the aluminum flow back into the bath. It is now generally

recognized that the opposite actually occurs. The capillary action of the aluminum seeking the low

pressure area where the reaction is occurring “sucks” up the aluminum adding additional fuel to the

reaction. On the most part, endothermic fluxing is now generally used in the industry although

exothermic fluxes are still predominantly used in Asia. Operator training and understanding of the

use of flux is critical to making any material achieve the intended result.

Furnace Temperature

The temperature of the metal is the single most important controllable factor that determines the

level of dross generation in a furnace. Once the temperature of the metal exceeds 782°C, dross

generation increases exponentially [3] as shown in Figure 4.

Figure 4. Dross Generation [3].

When the molten metal temperature is not properly regulated, dross can begin to thermite. Any time

a thermite reaction occurs, metal units are being lost and thus every very effort should be made to

prevent this from occurring. The fuel in a thermite reaction is the aluminum; the excessive

temperatures generated by this reaction can cause the surface temperature of the molten bath to

increase rapidly above the 782ºC mark causing further oxidation of the surrounding metal. In

addition to excessive metal loss, thermite reactions can cause damage to refractory, shortening the

life of the furnace lining.

Figure 5. Effect of stirring on metal temperature.


As introduced in Section 1 of this paper, stirring a furnace can reduce the temperature gradient

between the top and bottom of the furnace in a matter of minutes (see Figure 5) and reduce dross

generation by as much as 25%.

There are a variety of stirring technologies on the market today. Electromagnetic systems have

increased in popularity in the last 10 years (20 years if we consider the Japanese casthouses) and

although more expensive to install, there are no moving parts and many are non contact, so allow a

non intrusive stirring possibility to the furnace, providing minimal maintenance requirement and

risk of failure or breakdown to the furnace operation. Permanent magnetic pumps and stirrers are a

new comer to the industry and have been quite extensively used in China and Japan during the past

several years. They give single, directional stirring.

In general, any form of „sub-surface‟ stirring technology is advantageous in helping to reduce metal

surface temperature and reducing dross generation. Keeping the surface temperature as low as

possible will also improve refractory life and reduce energy consumption. There are also other

consequences to circulating a furnace to minimize dross generation, the positive impact it can have

on energy reduction and therefore CO2 also.

Impact of Circulation on Energy Consumption and CO2

When considering the use of electromagnetic stirring devices within the furnace it is also important

to consider the contribution that the stirrer itself adds to the energy footprint. As has been

previously discussed at earlier conferences there are many factors one needs to consider when

choosing the correct device for circulating a furnace [4]. There are several new technologies that are

promoting minimal CO2 footprints while achieving the significant objectives of continuous furnace

circulation.

Figure 6 shows the air cooled side mounted EM Stirrer which has been widely adopted throughout

the Russian primary industry over the past 15 years. Its unique design promotes very low energy

consumption reported at 1 - 2 kW/tonne, based upon 40kW power.

Figure 6. Side Mounted EM ‘air cooled’ stirrer.

Additional benefits are clearly the lack of necessity for water, as the inductor is air cooled providing

lower installation cost and high reliability due to the solid construction of the inductor‟s coil. The

energy operational cost for a typical cycle at a primary cast house would then be $US 2 per kWh

per tonne.

Clearly capital costs, the overall TIC (Total Installed Cost) and payback calculation become

important in determining which furnaces can justify such an investment. Typically any furnace that

has a large proportion of scrap added or produces a high alloy product (for example, 5xxx) would

benefit from investment in Electromagnetic Stirring based upon production increase (reduced cycle

time, energy and dross reductions and also the less tangible „qualitative‟ benefits of chemical and

temperature homogeneity and improved metal cleanliness).

However, the advent of Permanent Magnet Stirring used extensively throughout China has opened

up a low capital and operating cost stirring possibility, particularly for the smaller casting furnaces


that are predominantly tilting types with basements. This technology can be adapted to furnace sizes

up to 40 tonne capacity and provide very effective results similar to those promoted by

electromagnetic stirring.

The Table below provides an overview of the two technologies and appropriate selection choices

depending upon furnace size and operational requirements.

Table 1. An overview of the two technologies and appropriate selection choices depending

upon furnace size and operational requirements.

EM Stirring

(Water Cooled)

EM Stirring

(Air Cooled)

Permanent

Magnet Stirring

Energy Consumption Range (kWh/tonne) 6 1 - 2 1 - 2

Investment Cost (US$ 000‟s) $450 - $1,000 $400 - $500 $80 - $250

Operational References 200 + 150 + 80 +

Typical energy saving per cycle 10 to 15% 10 to 15% 10 to 15%

Additional to the energy savings achieved through continuous circulation of melting and casting

furnaces is the carbon footprint reduction for the cast house. By far the largest contributor to carbon

emissions is energy production. Any technology that reduces energy consumption is therefore

extremely advantageous. The CO2 released from energy production using natural gas is 0.185 kg

per kWh used [5].

If we assume a conservative figure of 10% to 15% as the energy reduction from having continuous

circulation applied to the melting or casting furnace, this equates to a saving of approximately

2770kWh per furnace cycle assuming a 50 tonne melting furnace operating a 6 hour melt cycle

equipped with regenerative burners (assuming a 550kWh/tonne efficiency). Clearly the cycle time

will also reduce and it is possible to operate at a higher roof temperature without overheating the

surface of the bath due to the improved heat transfer.

However it is not just the energy saving from a reduced cycle time and improved energy efficiency

of the furnace due to the reasons explained earlier that are important, relative to CO2 contribution.

One also needs to consider the impact of the stirring device that is used as these devices do consume

electrical energy. One therefore needs to consider the nett figure. Table 2 below shows the impact

of this.

It is reported for every 1kWh of electricity purchased it generates 0.537kg of CO2 [5]. This supports

the need to use low energy consuming devices to achieve the energy reduction and production

improvement objectives from any investment.

If we take the following assumptions we can then assess the savings that can be obtained from

energy and CO2 contribution from furnace circulation.

Assume:-

50 tonne „melting‟ furnace

Regenerative burners @ 550 kWh/tonne

Cycle time – 6 hours (including 1 hour alloying/holding @ 200kWh)

Cycles per day – 4

Operating days per year – 330

Cost of electricity - $0.10/kWh

Cost of natural gas - $0.055 per kWh


Table 2. Evaluation of CO2 impact of different stirring technologies on a 50t

melting furnace.

EM Stirring

(Water Cooled)

EM Stirring

(Air Cooled)

Permanent

Magnet Stirring

Energy Saving % 10% 10% 10%

Energy Saving per year (kWh) 3,656,400 3,656,400 3,656,400

CO2 Saving per year (Kg) 676,434 676,434 676,434

Energy Saving (US$) 201,102 201,102 201,102

Operating Energy (Stirrer) per year (kWh) –

See Note 1

396,000 99,000 90,000

CO2 contribution from stirrer energy

consumption

212,652 53,163 48,330

Nett CO2 saving 463,782 623,271 628,104

Note 1 – Based on kWh/tonne from Table 1 above.

Conclusions

Commercial and environmental pressures will continue to make the aluminium industry ever more

competitive. Those companies that focus on effective dross management will not only minimize

their unit cost of production but will also benefit from the many process and environmental

advantages of a well managed casthouse. This, when done correctly, translates to a better work

environment and major cost savings.

Focusing on keeping the aluminium in the furnace is the key practice through techniques discussed

above as these not only benefit reduced metal losses but also can provide significant benefits in

energy and CO2 reductions.

Each stage of the aluminium production process within the casthouse should be carefully evaluated

to determine the potential for improvement versus the required capital investment. Typically, return

on investments for dross management projects are extremely short since metal units are extremely

valuable.

References

[1] Peel, A., Alchalabi, R. and Meng, F. (2002) Furnace Operation Optimisation with EMP

System, Light Metals 2002.

[2] Guest, G. and Evans, R., Stein Atkinson Stordy Ltd, “The Aluminum Decoating Handbook”.

[3] Rossel, H. (1990) Fundamental Investigations about Metal Loss during Remelting of

Extrusion and Rolling Fabrication Scrap, Light Metals 1990, TMS, 721-729.

[4] Peel, A. (2003) A Look at the History and Some Recent Developments in the Use of

Electromagnetic Devices for Improving Operational Efficiency in the Aluminium Casthouse,

Proc. 8th

Australasian Aluminium Casthouse Technology Conference.

[5] Carbon Trust – Greenhouse Gas Conversion Factors 2009.


Dross Processing Technology

Alan M. Peel1,a, James Herbert2, David Roth2 and Martin J. Collins1

1 ALTEK-EUROPE Ltd, Station Road, Castle Donnington, Derbyshire, DE74 2NJ, UK 2 ALTEK-MDY, 825 Springdale Dr, Exton, PA 19341, USA

a [email protected]

Keywords: aluminium, dross, cast house, processing technology

Abstract

While it is generally acknowledged that dross generation should be kept to a minimum, too often the importance of maximizing the aluminium content of the dross is overlooked. Some mistakenly believe that a low metal content is a good thing and that the aluminium is being kept in the furnace. In reality, this metal is most likely being lost due to insufficient cooling and thermiting.

Much can be gleaned from looking at the dross that is generated in a casthouse; in fact, the quality of dross can provide a good indication of the overall efficiency of the operation. Even with the very low aluminium prices of today of about US$1400 per tonne, a recovery improvement of just 3% for a facility producing 500t of dross per month can provide savings in excess of $250.000 per year. Effective dross management also results in better metal quality, improved fuel efficiency, prolonged refractory life and improved profitability in the entire facility.

Over the years, as facilities have focused on better dross cooling and handling techniques, dross recoveries have improved. Today, dross recoveries should be in the range of 60 – 70%. These numbers will raise debate but 30 years of experience give us deep insight into these results.

The paper looks at the different techniques of handling the dross that is produced within the melting/casting operation with the objective of maximizing aluminium recovery. This paper will consider both the initial dross handling within the cast house but then also how secondary processors should be evaluated to maximize the value of the dross being processed. A company can lose as much dross recovery opportunity here as in their own facility. .

In summary, by careful attention to the equipment and process techniques around the furnace and the follow-on dross management, significant cost savings and environmental benefits can be realized by cast house operations.

Historical Perspective on Dross Handling in the Casthouse

Floor Cooling

Floor Cooling was the first and is the most basic form of dross management, and this practice can still be found in casthouses around the world today. This method of increasing metal recovery is accomplished by spreading hot dross over aluminium ingots or a steel slab floor. After the dross is sufficiently cooled, workers will pick out the visible “chunks” of aluminium. Typically, this method will provide a total metal recovery of approximately 20-30% [1]. While this is a significant improvement over doing nothing, it is a long way from the potential recoveries achievable when using today’s available technology. This method of dross management is dusty and hazardous to the environment, equipment and plant personnel.

Dross Stirring

Stirring first appeared in the industry in the late 1960s / early 1970s. The basic principle is that dross is skimmed directly into a refractory lined container. The container is then transferred to the stirring machine which contains a paddle that stirs the dross. After a period of time (4 - 6 minutes),


the container is tapped into a sow mold. This method of dross management was the first to provide a considerable in-house drain (20 – 30%) due to the agglomeration of aluminium droplets [1]. The stirring action however promotes oxidation and thermiting which results in a fine dusty material that is difficult to process at the secondary processor. Although this form of dross management was a step in the right direction, these systems suffered significant downtime and maintenance costs. On average, stirring will provide a total metal recovery of 40%.

Rotary Dross Coolers

Rotary Coolers made their first appearance in the 1970s. The cooler consists of four main components: a large drum which is externally water-cooled, a charging device, a trammel screener and an air pollution control system. The dross is skimmed into skim pans constructed with drain holes to allow for natural in-house drain. The pans are placed into the charging device and tipped into the drum. The drum rotates and is either sprayed with water, submerged in water, or built with lifters that pour water over the shell. Typically, in-house recoveries are lower than those experienced with stirring machines; however, the cooling characteristics of this technology can provide for an improved secondary and overall metal recovery, with a well designed system, typically ranging from 50 – 60%. Losses in the system come from two areas, real oxidation through an incorrectly designed or applied pollution control system and hood placements. In a properly designed system the dust hoods are at both ends of the cooler drum and dust is collected from each hood. The interior of the drum is free from any rush of air that might promote oxidation. In a poorly designed system there is one hood at the exit end of the unit controlling all the dust. This draws large quantities of air through the drum. This acts like a bellows and has a significant effect on lowering recoveries by oxidizing the smaller particles of aluminum. You can picture this material falling through an air path with hot fine particles of aluminum. The losses can be very high. This was a common problem with some rotary systems.

The second loss occurs because of poor secondary processing practices. In the late 70s and 80s it was a common practice to screen off the fines (-1/4” or 6mm) before secondary recycling in a rotary furnace. Before rotary dross coolers, this material was nearly all oxide with very little metal content. It was a waste of energy to put it into the furnace. After rotary cooling however a significant quantity of the aluminum to be recovered was in the -1/4” (6mm) fraction. When it was screened off operators did not realize for a long time that they were losing valuable metal units recoverable in a good tilt type rotary furnace. Most recyclers soon learned that the entire dross lot had to be put in the TTRF without the removal of the fines, after which recoveries went up dramatically.

Rotary coolers also have the advantage of being able to handle thermiting dross. Disadvantages of this technology include high capital maintenance costs and dependency on a good pollution control unit. Although relatively efficient, rotary coolers are seldom sold today due to safety concerns of hot dross and molten aluminium being processed over a pool of water. There have been instances with this type of dross cooler where the shell of the drum has cracked and allowed water to contact the dross, thus causing explosions. Several companies are moving away from this technology for this very important reason.

Inert Gas Dross Coolers

Inert Gas Dross Cooling systems became commercially available in the early 1990s. The system consists of heavy steel skim pans and enclosed cooling stations. In these stations, the atmosphere is replaced with argon gas or, in some cases, nitrogen which prevents further oxidation of the dross. Generally, the performance of these systems is similar to that of the Rotary Cooler; however, due to the slow nature of cooling (typically 12 – 24hrs) users typically require many skim pans and cooling stations. This takes up significantly more floor space. It could be argued that the cooling characteristic of the skim pans is more effective in preserving metal units than the addition of the inert gas, and it is not uncommon to find the skim pans being used without the cooling stations due


to maintenance issues. The skim pans provide a small amount of in-house recovery typically in the range of 5 – 10%. Secondary recoveries generally range from 40% to 50%.

Dross Presses

The Dross Press became commercially available in the early 1990s and today there are several manufacturers that supply different versions of this technology [1]. Dross pressing technology is based on the principle that a liquid placed under pressure will separate from a solid and flow to the areas of least pressure. The press system consists of a steel frame, hydraulic unit, a pressing head and a set of skim pans. Once skimmed, the dross is transferred into the press and the head is slowly lowered. The pressure forces metal out into the sow mold which is located under the skim pan, and agglomerates the fine particles of aluminium on the outside surface of the dross. This encapsulates the oxides thus preventing dusting and thermiting (see Figure 1 below). Dross presses were the first technology to physically reshape the dross, improving the casthouse environment and recoveries at the secondary processor. The system not only rapidly cools dross but can also provide the highest in-house drain. Overall dross recoveries can range from 60 – 70%. Two conditions where the dross press may be less effective are if the dross is too cold to press or if the dross is heavily thermiting. Thermiting dross can be processed but it requires revised practices, longer cycle times and special cooling techniques.

Figure 1. Dross after pressing. Figure 2. ALTEK Dross Press.

Figure 3 shows the results of a study conducted at Cressona Aluminium in the mid 1990s comparing various dross management techniques. It is important to recognize that these tests were conducted using the same type of dross, i.e. dross from the same facility skimmed from the same furnaces managed by the same operators. This is the only way to compare technologies. We have assumed an aluminium price of $US 1400 per tonne for RSI.

Figure 3. Dross recovery study at Cresona Aluminium, USA.


Cooling Heads

Some of the latest technologies have been specifically designed for smaller operations that do not generate enough dross to justify a dross press. Cooling Heads are very simple methods for improving the value of dross and consist of a skim pan and a matching cast steel Cooling Head (see Figure 4). The head is positioned onto the dross using a forklift which can provide enough pressure to generate approximately half of the in-house drain experienced when using a dross press. Cooling heads can be effective for cooling thermiting dross since the heads are typically more massive than those used in dross pressing. The heads can also be left in place for longer periods of time to get the desired cooling. Overall metal recoveries are typically in the range of 40-60%.

Figure 4. Patented cooling head designs.

Hot Dross Processing

A quantum leap in dross recoveries can be made if dross is processed in house using a tilting rotary furnace. This method of dross management has been predominately used in Asia, as well as some locations in Europe and the USA. Once skimmed, the dross is immediately charged into the rotary furnace, which is essentially used to tumble the dross and coalesce the molten aluminium droplets. After a period of time, the molten metal is poured out of the rotary furnace and, since it is the same alloy, can be put straight back into the melting furnace making use of the available energy. The rotary furnace can be operated with minimal salt additions by charging clean scrap together with the dross. This helps cool the material and controls potential thermiting. This method of dross management can provide an additional 5 – 10% of metal recovery compared to Dross Pressing.

Through studying the process, significant improvements can often also be made to the overall melt loss at a facility. This can equate to millions of dollars of potential savings per year. Cooling the oxide residue can become the major challenge in this process. Cooling pans, cooling heads, and rotary coolers are all options used to cool oxide residues.

Secondary Dross Processing

This section highlights the importance and value of developing a comprehensive aluminium dross management program. While it is important to pay attention to the in-house management of a generator’s dross, it is also equally as important for an aluminium facility to understand the important role played by the secondary dross processor [2]; it is here that the value of the dross will be ultimately decided.

Dross recovery values are primarily the responsibility of the dross generator since good furnace practices and cooling techniques will have a greater effect on overall recoveries than what happens at the secondary processor. While all secondary processors will always promise the highest recoveries, there are many subtle practices and process techniques that can affect the overall quality


and recovery of the dross. Dross generators should have the right to inspect and audit their processor to observe their material being processed and feel comfortable that good practices are in place and that they are getting the true value from their dross.

Although the dross has perhaps left the aluminium operator’s (dross generator’s) premises, they still have a moral and environmental responsibility for what happens to that dross. With many pressures to reduce or eliminate landfill, in general, and salt-containing landfill, in particular, dross generators are increasingly insisting upon environmentally sound and green technologies and processes being adopted by their secondary dross processors.

The Evolution of the Rotary Furnace

The centerpiece of most dross processing operations is the rotary furnace. Rotary furnaces have been used to recover aluminium from dross for at least 50 years. A conventional rotary furnace is a round steel vessel with a refractory lining mounted horizontally on a set of trunnions (see Figure 5). The operation of the furnace involves melting a salt flux and then charging the aluminium dross. The flux coats the aluminium particles, helping to avoid metal loss through oxidation and assists in aluminium droplets coalescing to form a liquid bath. The rotation of the furnace forces the dross beneath the surface of the flux and away from direct impact of the burner flame. The furnace is operated in cycles that will last for several hours. Each cycle will include:

• Charging and melting the flux;

• Charging and melting the dross;

• Tapping out the recovered metal and removing the spent flux or salt-cake.

Figure 5. Traditional fixed-axis rotary furnace.

Until the mid-1980s, many dross processors utilized relatively crude rotary furnace systems compared to today’s standards. It was not uncommon for a dross processor to manufacture their own furnaces using old kilns or drums salvaged from concrete tucks and other second hand parts.

Today, with the ever increasingly stringent air and solid waste environmental regulations, along with the competitive need to maximize production and recoveries, rotary furnace technology has evolved. Dross processors are able to take advantage of furnace manufacturers supplying tilting type rotaries [8] with sophisticated oxy-fuel combustion systems, advanced computerized control and SCADA systems, and the ability to operate using dry flux melting techniques. An example of such a system can be seen in Figure 6.


Figure 6. Modern dross processing.

Salt-Free Dross Processing

There has been much focus over the past several years on salt-free secondary dross processes. With ever tighter environmental control being imposed by various government agencies on landfill the ability to find suitable land fill sites for salt slag is fast reducing. If no salt reclamation plants exist in the vicinity then secondary dross processors are being forced to look for alternatives to using salt in their secondary dross processing. While this is possible with dross, it is not as feasible for contaminated scraps, another typical charge material for many rotary furnace operators due to the high amount of contamination on the aluminium scrap. New multi-chamber furnaces with after-burner type systems or an up stream de-coating and scrap preparation system will become more prevalent for these types of scrap, as is being witnessed particularly in Europe and Asia. It is not just the salt that causes issues in the land fill though. The aluminium nitrides [4], carbides and fines formed at the various stages of the dross production handling and management life cycle (from melting or holding furnace, management within the casthouse, through to secondary and post secondary dross processing) also cause problems. These are the elements that when combined with water release ammonia, methane and hydrogen gases in landfills and cause a lot of the environmental problems.

While it is possible to process contaminated scraps and drosses in a rotary furnace without the addition of salt fluxes, the reduction in aluminium recovery levels are substantial and may make this option cost prohibitive. There are several mechanical based processes on the market today that allow for the processing of aluminium drosses directly received from the dross press or furnaces. One such process is the TUMBLER® which is a semi-batch system that breaks down the large blocks of dross and separates the large pieces of free aluminum from the oxides that remain in the primary breaking chamber, muller chamber and autogenous milling chamber. These large pieces (+15 mm) are typically 90% - 96% aluminum after the predetermined processing cycle. They are discharged from the end of the unit into a tub or exit conveyor. Various sizes of machine exist allowing for small to large scale dross processing throughputs.

Mechanical impact separation is based on the premise that the aluminum is mixed into an oxide base material. The aluminum is not intimately joined into or with the oxide. By impacting the oxide and aluminum particles the aluminum separates and the oxides which are fine particles can be screened off. The Tumbler uniquely does this in one processing step. Materials that are a solid mass of oxide and aluminum can not be separated by impacting. Such materials are typically cold dross rich in silicon, some foundry drosses, floor scrapings, and single action pressed drosses where there is a poor drain strategy and very quick cooling.

The dross press as an intermediary step between the cast house furnace and Tumbler is ideal for rapid cooling of the drosses within the casthouse and ensuring minimization of oxidation. The


correct choice of press that gives good aluminium plating within the dross skull, ensures that the maximum amount of aluminium can be recovered in the Tumbler process. As previously mentioned the type and quality of dross skulls become very important for this technology to ensure that the skulls can be processed in an efficient way and maximum aluminium is recovered.

There are several advantages of this technology over conventional mechanical processing technologies like primary crushing, screening, and secondary crushing/screening. These include a significantly smaller footprint for the processing plant, lower capital and operating costs and lower maintenance costs.

The remaining (– 15mm) material runs through the Tumbler’s recirculation flights to remove the fine (-2mm) oxides from the system. The –15mm to +2mm size is recycled through the system again and concentrated to a level of 40 - 50% aluminum and then semi-continuously discharged through the front of the system.

The images in Figure 7 show such a system. Recent trials on a variety of different white dross, black dross and also salt slag have verified expected recovery results. The new design of the Tumbler has significantly reduced maintenance costs compared to previous Tumbler systems installed in the past 10 years.

Figure 7. The ALTEK ‘TUMBLER’ with impact crushing chamber.

Material Handling and Segregation

There are several measures a generator can take to ensure that he will get the most value from his dross. At times, the dross pile at an aluminium facility is treated as a refuse dump, and it is not uncommon to find floor sweepings and other scrap mixed in with the dross. It should be recognized that keeping the material dry, alloys segregated, and the dross free from trash will ensure a better quality of dross and therefore, better recoveries.

It is just as important for the dross processor to recognize the importance of material handling and segregation. If the dross is stored outside (see Figure 8) and gets wet, metal recoveries will be adversely affected. Generally speaking, every 1% of moisture will result in 1% of aluminum recovery loss. It is therefore important that the generator be able to walk through the processor’s facility to ensure material is being stored in an appropriate location.


Figure 8. Poor material handling.

Conclusions

Use of the appropriate technology within the cast house to handle the inevitable dross that comes from the furnaces can provide significant cost savings to an aluminium operation. Through the appropriate choice of technology it is possible for the cast house to maximize its in-house recovery of aluminium from the dross before it leaves their premises for secondary dross processing.

There has been a growing focus on the need to remove salt from the secondary dross processing process due to the environmental issues associated with salt slag disposal. Governments are legislating for these changes and large aluminium producers are focusing ever more strongly on the whole life cycle effect of their aluminium production. Consequently more and more countries are banning such land fill which is forcing the secondary dross processing industry to modify or even totally change their technologies and processes to remove salt and minimize other unwanted elements. This trend is likely to continue and it is not unforeseeable that eventually the rotary or fixed axis salt furnace will become a technology of the past.

References

[1] Herbert, J., Roth, D. and Collins, M. (2008) “The Art of Dross Management – Maximising Dross Values and Minimising Dross Generation”, TMS 2008.

[2] Shemwell, L. (2000) “A Dross Processors View”, ALTEK International – 3rd Annual Dross Seminar.

[3] Daley, J. (2000) “The Evolution of the Rotary Dross Furnace”, ALTEK International – 3rd Annual Dross Seminar.

[4] Hryn, J.N. (2007) “Salt Cake Management at Secondary Aluminium Smelters. A Case Study of Best Practice”, TMS 2007.


Automated Metal Siphoning and Cast House Energy Consumption

Jerry Locatelli1,a and Guangwei Liu2

1 Millennium Metals Pty Ltd, 35 Charlton Street, Norwood, Tasmania, 7250, Australia 2 Major Furnace Australia Pty Ltd, 100 Fairbank Rd, Clayton, Victoria, 3169, Australia

a [email protected]

Keywords: transfer of molten aluminium, siphoning, dross reduction, holding furnace, potline crucible

Abstract

The transfer of molten aluminium from the potline crucible to the holding furnace is receiving renewed attention as pressure to reduce energy wastage in the form of dross is heightened by the prospect of Emissions Trading. Siphoning has long been recognised as the most loss efficient method of metal transfer and now with the availability of fully engineered, automated systems which are safe and reliable in addition to reducing melt loss by 75%; this method is set to become the industry standard. This paper presents some examples of automated siphon transfer systems and demonstrates the benefits of siphoning in applications ranging from the largest smelters to the smallest of foundry applications.

Introduction

The main focus of this paper is to consider the benefits of employing siphoning as an alternative to conventional pouring in the process of transferring aluminium from the potline crucible to the holding furnace. Firstly this process will be considered in terms of energy consumption for comparison with the overall usage of energy in the cast house and with the aluminium smelting process as a whole.

In the smelter environment, energy usage in the cast-house has tended to receive relatively scant attention due to the very high energy involved in the smelting process itself. In future, the Climate Change debate will require cast house energy usage to come under closer scrutiny.

Simply put, energy management in the cast-house involves mainly:

• the control of heat

• the minimisation of oxide generation (dross)

A common view is that the generation of dross is unavoidable or inevitable and similarly, the large amount of heat consumed in the cast-house is an inherent requirement. The writers prefer to keep an open mind on these issues.

The fact is that there is still much which can be done to reduce dross generation, particularly in some of Australia’s older smelters. Opportunities also need to be taken to better utilise the 200 degrees of superheat which molten aluminium possesses when it enters the cast-house.

Cast House Energy Consumption

In this paper, the principle used to assess cast-house energy consumption is as follows:

1. Aluminium is received into the cast-house as a liquid at ~950oC and leaves it as a solid at

~30oC. The heat produced by any fuel used in the process is accounted for as energy

consumed.


2. Some aluminium is converted to dross during the various cast-house handling processes. The net loss of aluminium in this conversion is counted as an energy loss (as well as a material loss, due to its irreversible nature) by accounting for the cumulative energy consumed in producing the aluminium from bauxite.

The energy involved in producing aluminium has been estimated as:

Bauxite to alumina - 23.3 GJ/tonne

Alumina to aluminium @ 950oC - 55.7 GJ/tonne

Total aluminium reduction energy - 79.0 GJ/tonne [8]

If we consider a simple cast house model involving purity ingot production, then the energy used in a cast house can total in excess of 2.0 GJ/tonne. The energy consuming processes can be broken down into 4 broad areas as follows:

• Crucible transport and marshalling

• Crucible transfer to the holding furnace

• Furnace holding and melting

• Casting into solid product

For each of these process areas, the energy consumed has been estimated in the form of heat loss and melt loss from oxide generation. The data presented here is based on limited available data. The oxide losses from crucibles, furnaces and ingot casting have been calculated from generic oxidation rate data taking into account typical temperatures, durations and exposed surface areas. The melt loss data comparing pouring with siphoning is based largely on actual measurements.

Figure 1 illustrates the energy consumption of a typical purity ingot casting line with cascade poured crucible transfer and twin 50 tonne holding furnaces:

T ypic al C as ting L ine E nerg y C ons umption

0

500

1000

1500

2000

2500

En

erg

y (M

J/t)

heat los s MJ /t 54 45 554 621 1274

melt los s MJ /t 32 790 34 40 894

c ruc ible trans port

c ruc ible trans fer

furnac e holding

ingot c as ting

total energy c ons umed

Figure 1. An estimate of typical casting line energy consumption.

Note: The above estimate excludes the crucible superheat energy which must necessarily be lost to achieve a casting temperature of ~700

oC.

It is clear from Figure 1 that the melt loss energy during cascade poured crucible transfer is much greater than all the other melt loss areas combined. The furnace heat energy losses are nearly equal to the heat lost during cooling the cast product to ambient temperature.


Mechanism of Oxide (Dross) Formation

The main factors contributing to oxide formation in molten aluminium are temperature, turbulence and oxygen availability. The temperature of metal arriving from the reduction lines is normally well above 900

oC. At this temperature the oxidation rate can be up to 10 times higher than when the

metal is at casting temperature, typically 700 to 750oC. The other major contributor to oxide

generation is turbulence, which provides a constant source of fresh molten surfaces to stimulate oxide formation. At these high temperatures the formation of oxide on these fresh surfaces progresses at an extremely rapid, characteristically parabolic rate. Amongst the four energy consuming areas shown in Figure 1, poured crucible transfer to the holding furnace stands out as the largest energy loss contributor. The combination of high metal temperature and turbulence resulting from cascade pouring wastes a huge amount of energy in the generation of dross.

Why Does Cascade Pouring Generate so Much Dross?

In addition to the operation taking place above 900oC, cascade pouring a crucible into the holding

furnace results in high flow velocities, creating unavoidable turbulence and splashing of the metal. In the worst cases when the metal falls through some distance into the molten furnace bath, a further doubling of the rate of dross generation can be expected [6]. This metal flow also occurs in the open air, where oxygen is available to allow the maximum rate of oxidation to take place. Measurements have shown that metal loss from cascade poured crucible transfers alone can amount to between 0.8% and 2 % of the metal transferred [2,8].

Minimising Melt Loss by Siphoning

To minimise dross generation, any form of cascading or turbulent metal flow must be completely eliminated. The metal flow must be sufficiently quiescent to avoid disturbance to, or breaking of the protective oxide layer which naturally forms on the metal surface. For best results the authors recommend metal flow must occur entirely ‘sub-surface’, preferably producing surface velocities of less than 30mm/sec.

Transferring molten aluminium by siphoning has been found to virtually eliminate dross generation and consequential melt loss. In a well designed system, transfer by siphoning takes place with almost complete absence of turbulence and with minimal disturbance to the protective oxide layer in both the crucible and the receiving furnace. Further, because the transfer occurs within an enclosed tube and under a strong vacuum, the availability of oxygen for oxide formation is minimal. Available data suggests metal loss from filling a holding furnace by siphoning can be less than 0.2%, i.e. a 75% reduction when compared with that typically lost during cascade pouring. [2]

Automated Siphoning

The Automated metal siphon works in a different way to a conventional siphon. It operates in ‘Weiring mode’, such that the siphon pipe is never completely filled with metal. Weiring mode is created by applying a controlled suction at the siphon ‘head’ which causes the molten metal to rise up both legs of the pipe. When metal reaches the weir point it begins to flow down the outlet leg of the pipe (Figure 2). Because of the height difference between inlet and outlet, the metal in the inlet leg reaches the siphon weir first and flows over the weir and down the outlet leg. Flow is maintained in Weiring mode by gradually increasing the suction under PLC control. Flow rate is unaffected by the level of metal in either the crucible or the furnace.


Figure 2. Basic automated siphoning process.

The resulting system is safe and easy to operate compared with earlier, manually controlled siphon systems. The effects of variations in crucible contents, furnace contents and siphon pipe cleanliness are largely eliminated, resulting in a smooth, steady, repeatable flow of metal. Typically, the metal flow rate during automated siphoning is maintained between 3 and 5 tonnes per minute, depending on the application.

Background to automated siphoning

Siphoning has been used to transfer aluminium from crucibles to furnaces for more than 40 years. A manual system was originally devised by Alcan, which performed adequately in the hands of skilled operators but lacked operational refinement and user friendliness needed to encourage a wider acceptance.

As part of its dross reduction programme, Comalco Aluminium carried out siphoning trials at its NZAS smelter in 1992. Following very positive results, in 1994, Comalco (now Rio Tinto Alcan) developed an automated siphon system for use on the continuous atomising process at its powder plant at Bell Bay, Tasmania. The success of this system encouraged the installation of a system which married the Comalco developed siphon automation to a specially engineered pipe handling system designed by Major Furnace Australia for the third potline expansion at the Boyne Smelter, Queensland. This system was commissioned in 1997 and involved the conversion of 12 existing holding furnaces from cascade pouring to siphoning. The Boyne smelter currently siphons 550,000 tonnes of aluminium annually through 14 furnaces. [3]

In 2003 the powder plant at Bell Bay, Tasmania, now owned by Ecka Granules GmbH installed a second automated siphon system on its new atomising line and in 2008 the ‘green field’ smelter at Sohar in Oman was commissioned with four automated siphon systems designed by Major Furnace.

The Rio Tinto Alcan - Major Furnace Siphon System

After 11 years of operation and ongoing refinement, the Rio Tinto Alcan – Major Furnace (RTA- Major) automated siphoning system has demonstrated proven reliability and robustness. The automated siphoning process is regarded as inherently safer [3] than other transfer methods, mainly due to the ability to instantly stop the metal flow by releasing the vacuum in the event of a problem.

As the process is intentionally ‘quiescent’ in normal operation, siphoning provides very little natural feedback to the operator, so the automation includes a control interface designed to inform the operator of the system status and identify problems if they occur.

(CRUCIBLE

WEI POIN SIPHO PIP

META FLOW

(CRUCIBLE

WEI POIN SIPHO PIP

META FLOW

SUCTIO

(HOLDING


The key functional parts of the system are: the siphon pipe handling & pre-heating system; the siphon vacuum control system; the operator interface; pipe transport and maintenance equipment and the furnace interlocks.

The RTA- Major siphon pipe handling system has been designed to suit a typical smelter cast house arrangement with holding furnaces aligned adjacent to a hot metal delivery passage. Siphon pipes are parked overhead in the hot metal passage, in line with each furnace (Figure 3). While the siphons are parked on standby, gas fired pre-heat burners maintain the pipes at temperature to eliminate moisture, reduce metal chill and minimise thermal shock to the cast iron siphon pipes.

Figure 3. Smelter siphoning arrangement showing the RTA - Major pipe handling system.

When a crucible of metal arrives on a hot metal carrier vehicle, the siphon pipe is lowered to simultaneously enter the crucible and the holding furnace, by a simple pushbutton operation. The crucible is tilted to about 5

o to concentrate the metal under the pipe inlet and so minimise the

residual metal heel. Once the siphon pipe is in position (Figure 4), the siphon vacuum sequence is started, which automatically empties the crucible at a controlled rate. When the crucible becomes empty, the vacuum is automatically released and the siphon pipe is returned home to its parked position.

Figure 4. Smelter automated siphoning in progress.


The operator is provided with a closed circuit TV to monitor the metal level in the crucible. This visual feedback, along with real time display of process variables and alarm messages ensures the operator is kept constantly informed of the process status from a safe location at the operator control station. The RTA - Major system also facilitates safe and efficient removal, replacement and transport of siphon pipes to and from the pipe cleaning bay. Regular cleaning and maintenance of siphon pipes is essential for reliable siphon system operation.

Experience has shown that before starting to siphon metal the following conditions must be met:

• There must be sufficient clearance under the pipe at both ends for metal flow.

• The siphon pipe must be pre-heated sufficient to eliminate moisture and minimise risk of a metal explosion as well as to reduce thermal shock to the cast iron.

• The pipe should be clean to avoid the risk of blockage. Pipes are cleaned regularly according to a pre-determined number of siphon cycles counted by the PLC.

• The holding furnace must have sufficient metal heel to fully submerge the outlet of the siphon and the metal temperature should be above 700

oC and with no sludge beneath the pipe outlet.

• The crucible temperature needs to be above 800oC and free of excessive sludge.

The basic steps involved in siphoning a crucible of metal are: Position the crucible vehicle; Tilt the crucible; Lower the siphon pipe into the furnace; Initiate the vacuum cycle; Return the siphon to its parked position.

This sequence can normally be completed in 8 to 10 minutes for a 10 tonne crucible.

The Ecka Granules Siphon System

The siphon system at Ecka Granules Australia Pty Ltd employs a fully automated siphoning system similar in principle to the RTA - Major Furnace systems at Boyne Smelters; however this installation also takes the concept of energy management a stage further. In addition to automated siphoning, the facility uses a patented system known as ‘Charging During Casting’ (CDC). The holding furnace is able to be re-filled from a potline crucible without the need to stop casting. Each of the two atomising lines at Ecka Granules routinely cast continuously from their respective furnaces for several days without stopping. (Figure 5)

Figure 5. Charging during casting at Ecka Granules.


The implications of this arrangement with regard to energy and financial savings are quite dramatic: the cost of operating a second holding furnace (as would normally be the case in a conventional smelter ingot line) has been completely eliminated.

Metal deliveries from the adjacent Rio Tinto Alcan smelter to the Ecka Granules facility are scheduled to maintain the two 10 tonne furnace contents between 4 and 9 tonnes. Furnace contents are estimated from the furnace tilt angle by a process computer, which automatically keeps the smelter updated via an electronic data link. Casting takes place uninterrupted, 24 hours per day for several days at a time. The process is stopped for several hours to enable siphon pipe change-out after about 40 siphon transfers. A spare siphon pipe which has been previously cleaned is exchanged with the active unit. The furnace is usually also manually skimmed during this stoppage, although skimming can be carried out whilst casting is in progress, provided the furnace contents are above 7 tonnes (i.e. the furnace is at a low angle of tilt).

Additional features such as using the furnace flue gas to pre-heat the siphon pipe and highly insulated, lidded launders further reduce the process energy consumption.

This system has been designed to suit a smaller operation with 10 tonne furnaces. The siphon pipe handling system is furnace mounted enabling the furnace to tilt freely so as to permit casting and siphoning to occur simultaneously. A simple purpose designed push trolley is used to remove the siphon pipes for cleaning.

How Much Could Cast House Energy Consumption be Reduced?

Let us now consider the simple purity ingot casting line model and estimate what impact the following suggestions could have on cast house energy consumption:

• Crucible transport: Fit sealed lids and blanket crucibles with nitrogen before they leave the potline. This would reduce oxidation, particularly at the metal line where slopping takes place during transport. Possibly there is no need to improve crucible insulation as some superheat needs to be lost to match holding furnace temperatures. Monitor temperature and marshal the crucibles so that they can be emptied at precisely the right temperature to suit the furnace.

• Crucible transfer to furnace: Use siphoning • Furnace holding: How about operating a single furnace configured for ‘Charging During

Casting’ similar to the furnaces at Ecka Granules, Bell Bay. This would entirely eliminate the energy involved in operating a second furnace. Further this furnace would not need to be a 100 tonne monster, 40 or 50 tonne capacity should be quite adequate, resulting in reduced surface area available for oxidation and heat loss. Avoid disturbing the metal surface by frequent dross skimming and allow a protective oxide film to form. Optimise burner combustion and maintain slightly positive pressure control.

• Ingot casting: Use narrow profile, low heat loss, lidded launders with run lengths as short as possible to reduce temperature drop and enable lower furnace operating temperatures. Use full under-pour mould filling systems [5] to minimise oxide generation. Recover the waste heat from ingot cooling by using a counter-flow cooling conveyor to increase the exit water temperature, then use the waste heat to power heat pumps for building heating or air conditioning, etc.

Figure 1 estimates the ingot line energy losses with cascade poured crucibles and twin 50 tonne furnaces to total 2.17 GJ/tonne or about 2.75% of total aluminium smelting energy, whilst if the above suggestions were to be implemented, Figure 6 shows the energy consumption reduced to a total of 0.78 GJ/tonne which is less than 1% of total aluminium smelting energy. On this basis a reduction of energy of over 50% would seem feasible. Perhaps some of these suggestions seem radical; however they are all based on existing proven technologies, whilst several have already


been suggested by others. With the current focus on carbon emissions and energy conservation, maybe they deserve further consideration.

R educ ed C as ting L ine E nerg y C ons umption

0

500

1000

1500

2000

2500

En

erg

y (M

J/t)

heat los s MJ /t 27 54 137 320 538

melt los s MJ /t 16 197 9 21 244

c ruc ible trans port

c ruc ible trans fer

furnac e holding

ingot c as ting

total energy c ons umed

Figure 6. Estimated energy consumption with suggestions incorporated.

Conclusion

This paper has attempted to articulate the well documented bogies of melt loss from oxide generation and process heat loss to present a view from an energy loss perspective on a series of typical smelter cast house processes. The purity ingot casting model was chosen to avoid the complexities involved in alloying operations. Clearly the production of alloys always involves additional oxide and heat energy losses.

Two different automated metal siphoning systems have been described. The RTA - Major Furnace system is suitable for larger smelter based operations whilst the system in use at Ecka Granules can be adapted and if necessary, simplified for use in smaller operations such as foundries.

It is clear from the data presented that filling a holding furnace with molten aluminium is no longer the ‘tall poppy’ of energy wastage when siphoning is employed and now with the availability of fully engineered, automated systems, this method is set to become the industry standard.

Finally, suggestions have been made which could enable the overall energy consumption in a cast house to be reduced by some 50%. There is still much fertile ground for those prepared to accept the challenge of reducing net energy consumption in the cast-house, whether by tackling melt loss or by reducing fuel consumption.

,otes: In Figures 1 and 6, energy consumption values have been estimated as follows:

Heat Losses: Crucible transport - typical rate of crucible temperature loss from ref. [6], converted to heat energy for 60 min. (Fig 1) and 30 min. (Fig 6) Crucible transfer - heat loss through a siphon from ref. [2] and author's own cascade pouring data Furnace holding - typical heat energy consumption for two furnaces from ref. [4] (Fig 1) and. author’s own data for single furnace with improved insulation (Fig 6) Ingot casting - typical ingot casting temperatures converted to heat loss (Fig 1) and author’s estimate of 50% saving from heat recovered from cooling water Melt Losses: Crucible transport - melt loss at temperature data from ref [1,7], using typical surface areas for 60 min (Fig 1) and 30 min (Fig 6) Crucible transfer - melt loss through a siphon (Fig 1) and for cascade pouring (Fig 6) from ref. [2] Furnace holding – melt loss at temperature data from ref [1,7], using typical furnace surface areas for two furnaces (Fig 1) and one furnace (Fig 6) Ingot casting - oxide rate data from ref [5] converted to melt loss for typical (Fig 1) and improved ingot casting systems(Fig 6)


References

[1] Fox, M.H., Nilmani, M., (1993) Dross Minimisation – Its Relationship to Melting and Melt Handling Practice, 3rd Australasian Asian Pacific Conference.

[2] Locatelli, J.R., (1997) Dross Reduction Through Improved Metal Transfer From Crucible to Furnace, 5th Australasian Asian Pacific Conference.

[3] Hannah, R.C., (1999) Conversion of the Casthouse at Boyne Smelters Limited from Crucible Tipping to Siphon Transfer, 6th Australasian Asian Pacific Conference.

[4] Meadows, L., Ingot Line Productivity, Aluminium Cast House Technology 2005.

[5] Grandfield, J., Rohan, P., Nguyen, V., Oswald, K., Prakash, M., Cleary, P., Development of an Ingot Caster Filling System, Aluminium Cast House Technology 2005.

[6] Whiteley, P., The Potroom/Casthouse Interface, Aluminium Cast House Technology 2005.

[7] Taylor, J., Oxidation and Melt Loss involved in the Aluminium Cast House, Aluminium Cast House Technology 2007.

[8] Green, J.A, (2007) Aluminium Recycling and Processing for Energy Conservation and Sustainability, ASM International.


Technological Researches Concerning a Decrease in the Losses Due to the Oxidation of Remelted Scrap from Aluminium Alloys

Ilie Butnariua, Ioana Butnariu and Dana Butnariu

University “Politehnica” of Bucharest, 313 Spaiul Independentei, Bucharest, Romania

a [email protected]

Keywords: elaboration, oxidation, quality, technologies

Abstract

The present paper refers to the remelting of light waste. The research aims to provide a substantial decrease of the oxidation losses and a decrease of the noxious expulsion resulting from the production of aluminium alloys.

The fundamental solution found consists in achieving a stable flux layer. For this the wastes are introduced directly in the metallic bath, without a previous preparation.

Data about oxidation losses, the content of gases and the correlation between the casting properties and the quality of remelting secondary alloys are presented.

Several samples are analysed in order to point out the quality of the alloy and the relationship between the chemical composition, the content of gases and the free linear contraction of the alloy.

Introduction

Worldwide, reuse of non-ferrous scrap has become a very important issue, taking into account that the quantity of non-ferrous ores and concentrates is decreasing and that the price of non-ferrous metals is increasing.

The term “non-ferrous scrap” refers to the remaining non-ferrous metals and alloys which result from lamination, forging, stamping and splintering (chips, splinters), the refuse resulting from the technological processes of casting or plastic formation, as well as other materials resulting from the thermal and chemical processing of non-ferrous metals (ashes, oxides and slag).

Old non-ferrous materials are defined as being the assets and materials (parts, sub-assemblies) which are worn or broken. [1]

The non-ferrous scrap and the old non-ferrous materials are classified and put into different categories and sorted according to their chemical composition and to their degree of contamination. In the case of waste and scrap, impurities and other foreign elements may be present such as humidity, oil, the remains of dirt, adherents and insulation, as well as iron and steel. [2] [3]

Experiments

This paper presents a simple and efficient method to recover scrap aluminium. This method is based on eliminating several elements present in the aluminium scrap, such as impurities (humidity, oils), which generate toxic gas, and on decreasing scrap losses due to oxidation.

Experimental studies have been conducted using an induction oven with a capacity of 30kg and a frequency of 50Hz, with silica-aluminium lining, as well as using an oven with flame and graphite crucible with a capacity of 8kg.

Aluminium cuttings produced from the processing of cast parts of AlSi10Cu3Mg were used as wastes.


At the beginning of the experiment, a liquid bath was made using Si = 10.05%, Mg = 1.2%; Mn = 0.4%; Cu = 3.8%; Fe = 0.56%. This bath was protected by a liquid salt flux layer.

In this liquid bath, cuttings of aluminium alloy were introduced, having a density between 0.12 ÷ 0.25kg/dm3 and a chemical composition close to that of AlSi10Cu3Mg alloy.

The experimental results obtained in the two types of ovens are presented in Table 1.

Table 1. The experimental results obtained by using classic technology.

Type of oven Oven with induction Oven with flame and graphite crucible

Characteristics Charge 1 Charge 2 Charge 1 Charge 2 Liquid alloy (kg) 10.8 11.3 5.217 4.835 Wastes of cuttings introduced (kg) 21 20 0.98 0.98

Resulted alloy (kg) 25.34 24.57 4.36 4.146 Losses (%) 20.3 21.5 29.7 28.7

Table 1 shows that by using standard technologies to recover aluminium alloys from scrap (cuttings), the losses are relatively big, between 20-22% in the case of the induction oven and between 28-30% in the case of the oven with flame and graphite crucible. Losses are calculated using the relationship:

100 wastesalloy liquid

alloy resulted - wastesalloy liquid×

+

+

In order to decrease the losses due to the oxidation of the elements and in order to eliminate some impurities, a new technology is proposed, which can be used with both induction oven and an oven with flame and graphite crucible.

The experimental installation used is presented in Figure 1. It mainly uses the same devices and principles found in the oven with flame and crucible having, in addition, a system made of a steel pipe, with diameter 80mm and height 740mm. This system contains aluminium cuttings having the density of 0.23kg/dm3, mixed with a deoxidizing flow having the composition 40% NaCl + 40% KCl + 20% NaF. The pipe is painted both inside and out with a graphite-based paint in order to avoid contaminating the bath with iron.

The pipe continues with a joint that has a filter used to stop the small particles and which is connected to a vacuum installation.

After introducing aluminium scrap in the pipe, the pipe is sealed with an aluminium lid. The pipe is then heated and immersed in the bath containing the liquid alloy. The pipe’s joint is connected at the same time to a vacuum pump.

The decrease in pressure achieved using the vacuum pump varies between 0.006-0.008 atm and not more, because a higher value would decrease the efficiency of the melting process.

The aluminium cuttings are hardened by the pressure created and the gas resulting from the hardening is evacuated using the same vacuum pump.

The experimental results obtained when using the installation shown in Figure 1 are presented in Table 2.


1. metallic casing; 2. fireproof walls; 3. graphite crucible; 4. aluminium bath; 5. lid; 6. wedge of fireproof brick; 7. oven propping support; 8. fan; 9. metallic pipe; 10. steel pipe; 11. aluminium

cuttings; 12. aluminium lid; 13. filter; 14. joint pipe; 15. vacuum pump

Figure 1. Experimental installation for the efficient recovery of alloys from scrap (cuttings).

Table 2. The experimental results obtained by using the new technology.

Type of oven Oven with flame, graphite crucible and vacuum system

Characteristics Charge 1 Charge 2 Charge 3 Liquid alloy (kg) 5.068 5.122 5.080

Wastes of cuttings (kg) 0.865 0.850 0.880 Mass of the ring (kg) 0.010 0.010 0.010 Resulted alloy (kg) 5.099 5.014 5.158

Losses (%) 14.2 16.2 13.6

Based on the experimental results shown in Table 2, it is clear that, when using the technology we propose, the losses are between 13-16%, much less than in the case of remelting aluminium scrap (cuttings) in ovens with induction or ovens with flame and graphite crucible. Losses are calculated using the relationship:

100ring of mass wastesalloy liquid

alloy resulted -ring of mass wastesalloy liquid×

++

++

At the same time, the quantity of gas released by the liquid alloy decreases and is evacuated by using the vacuum pump.

Figure 2a presents the microstructure of the AlSi10Cu3Mg alloy obtained from wastes (cuttings) by using the new technology and Figure 2b presents the microstructure of the same alloy obtained by using classical methods.


2a (x 200) 2b (x200)

Figure 2. Microstructure of the AlSi10Cu3Mg alloy obtained from wastes by using a new technology (a) and by using classical methods (b).

By analysing Figure 2 we notice that a finer structure is obtained, having a smaller element segregation degree, when using the new technology, than when using standard technologies.

The element segregation degree was determined by analysing the microstructures of the AlSi10Cu3Mg alloy by using the electron microprobe.

In conclusion, when using this new technology for remelting aluminium scrap (cuttings), the alloys obtained have a homogenous chemical composition, a finer structure, a smaller element segregation degree and emanate a smaller quantity of gas. All these properties are achieved along with smaller losses due to oxidation of elements.

Bibliography

[1] Sofroni, L. (1975) Alloys Smelting and Casting (cart iron, steels, non-ferrous alloys), Bucharest – Didactical and Pedagogical Publishing House.

[2] Ienciu, M; Moldovan, P; Panait, N. (1985) Smelting and Casting Non-ferrous Special Alloys, Bucharest – Didactical and Pedagogical Publishing House.

[3] Altman, M.B. (1984) Improving the Properties of Aluminium Alloys for Cast Parts, Moscow, Metalurghia.


CHAPTER 4:

Furnaces & Refractories

Furnace Operation: “A Gold Mine in your Casthouse”

G. Girard1,a, J. Barresi1, C. Dupuis2 and G. Riverin3

1Rio Tinto Alcan PTC, 15 Edgars Rd., Thomastown, Vic, 3073, Australia 2Rio Tinto Alcan ARDC, 1955 Boulevard Mellon, Jonquière, Qc, G7S 4K8, Canada

3Rio Tinto Alcan Aluval, 725 Rue Aristide Berges, Voreppe, 38341, France a [email protected]

Keywords: furnace, dross, operation, melt quality

Abstract

Scrap rate, throughput, alloy recovery and raw materials are all areas of the casthouse which usually get substantial attention as means of either increasing profits or reducing costs. However, furnaces, which are often overlooked by the casthouse, can also deliver surprisingly high savings. Moreover, these potential savings can only magnify as energy costs and pressures to reduce carbon footprint increase. This paper gives an insight into where savings can be achieved by a casthouse with proper furnace operation. Knowledge of how a furnace should be operated is often neglected but as this paper tries to highlight, developing this knowledge can be extremely worthwhile.

Introduction

Furnaces are the first processing step in the casthouse but more often than not they are the last step to be looked at in terms of optimisation. This paper describes some areas where substantial gains can be achieved when using good furnace practices. The areas covered in this paper include:

• Melt charging techniques

• Scrap additions & melting

• Alloying

• Stirring

• Fluxing

• Skimming

• General furnace energy management

For each area considered, opportunities are highlighted followed by a section summarising the benefits which can be achieved by following best practices.

The calculated potential savings are based on a hypothetical 250ktpa primary aluminium smelter which has gas fired furnaces1 and an internal return scrap rate of 5%. As will be shown, the potential gains can quickly reach in excess of ½ Million $/year.

Melt Charging

Metal transfer is the first step of furnace preparation and has more impact on the process and on casthouse revenues than is often understood. The method in which metal is added to the furnace is critical as it can introduce oxides into the melt and generate dross. When metal falls from a height, air is drawn below the melt surface, the same way that your beer ends up with a large head when you pour it straight into a glass (Fig. 1).

1 The industrial price of natural gas is assumed to be 0.35$/m3 as given by reference [1] for July 2008.


Figure 1. Schematic diagram showing effect of metal fall height on air entrainment in the melt [2].

The cascading action increases the metal exposure to oxygen resulting in the formation of dross and oxides that can remain suspended in the melt. With any melt transfer operation the aim should be for the metal to experience low turbulence. Tipping metal straight into a furnace from a crucible at a height is the worst situation as it generates turbulence and entrainment of electrolytic bath which is well known to increase dross formation and melt loss [3, 4].

The dross generation can be minimised by simply reducing the pouring height. Pouring metal along a transfer launder is also an effective way to reduce metal turbulence and fall height and to reduce dross formation – the same way as a beer head can be minimised by tipping the glass to pour that beer. By far the best melt transfer method is an under pour system such as siphoning where there is minimal turbulence but more importantly, virtually no contact with oxygen. One drawback from reducing turbulence and metal exposure to the atmosphere is that Na losses from transfers by vaporisation and oxidation are reduced so upstream treatment such as TAC may be required to reduce Na.

Here’s what you can gain….

Potential savings from minimising dross are significant. Even a 0.1% dross reduction can result in savings greater than >100k USD/year.

Scrap Additions and Melting

Part of filling a furnace often includes a scrap addition. As is shown in Figure 2, scrap aluminium comes in many different forms which can greatly affect the amount of melt loss and therefore the amount of metal recovered. The main variable affecting melt loss is the thickness of the scrap with the amount of melt loss drastically increasing for scrap having a thickness smaller than a critical thickness which has been determined to be ~1.5 mm (Fig. 2b) [5]. This critical thickness is established by the weight of the liquid inside a chip equalling the strength of the oxide skin which contains it. In other words, for scrap having a thickness smaller than the critical thickness, the weight of the liquid aluminium inside the chip is not sufficient to break the oxide skin and the resulting melt loss is drastically increased. The effect of this critical thickness on melt loss is highlighted in Figure 2 c) where briquettes containing swarf of ~0.25mm (Arrow 1) were melted in a small furnace. As is shown, the liquid metal within the oxide skin could only be released by physically breaking the oxide skin (Arrow 2).


0.3% 1.0%3.5%

25.9%

0%

5%

10%

15%

20%

25%

30%

Dross

Gen

erated

Heavy Scrap Thin GaugeScrap (~2mm)

CompactedCoarse Swarf

CompactedThin Swarf

a)

b)

c)

Figure 2. a) Wet hearth melt loss percentage, as measured by total weight of skim, for different scrap types, b) Melt loss of AA1100 – AA3003 vs. sheet thickness as obtained by Stewart and McCubbin [5], and c) photo showing aluminium entrapped within swarf as

revealed by breaking the oxide skin (critical thickness effect).

To minimise metal losses from scrap additions, scrap should preferably have a low surface area to volume ratio. If light gauge scrap is added, it should also be added first before heavier gauge so that it remains submerged in the melt. Any swarf should be compacted to a high density to minimise floating and exposure to the oxidising atmosphere. Metal recovery can also be improved by the addition of fluxes, although this has cost and environmental implications, or by devices such as mechanical or electromagnetic stirring. Fortunately for a smelter application, most of the scrap added in the furnace is rather thick and yields high recovery.

When considering the energy required for melting the scrap, significant savings can be achieved by using superheat from the incoming reduction cell metal as the energy source. In most cases, potline metal will reach the furnaces at a temperature between 800°C and 850°C and will allow melting of up to approximately 5% of solid scrap for “free”. This can be achieved with good furnace

2

1


management by charging the liquid and the scrap at the right time while limiting furnace doors opening times.


Applying the practices mentioned above can reduce melt losses. If our hypothetical smelter was to produce and melt 200t/year of the thin compacted swarf shown in Figure 2a rather than the coarser type, an equivalent of $25k would be lost each year.

On the melting side, consider a 40 tonne furnace load at 800°C and a target casting temperature of 700°C. Using the “free” heat will theoretically allow to melt approximately 2.5 tonnes of scrap before reaching the target casting temperature, allowing for some heat loss. Achieving the same just with furnace burners would cost ~ 9$/t of scrap or in our example save you $110k/year in fuel costs.

Alloying

Once the metal is in the furnace, alloys need to be added to create the desired composition. Here again, cost effectiveness will depend on the techniques employed to charge the material. One comment which applies to all alloying elements is that the melt should be skimmed before charging to avoid entrainment of inclusions in the melt and alloys becoming trapped in the dross layer resulting in poorer recoveries. Apart from this general rule, the addition technique which will give the optimum recovery will be different for almost every element and should therefore be added according to supplier recommendations. Magnesium (Mg) is probably the most sensitive element in terms of recovery and impact on metal quality. It should be added below 750°C and submerged as quickly as possible to prevent excessive burning and inclusion formation (spinels).


Consider our hypothetical smelter producing 100kt of a 6XXX series alloy having an average Mg composition of 0.7%. By following proper addition practices, approximately $15k/year will be saved for every 1% gain in recovery (Price of Mg taken as 2000 $/t).

Stirring

Stirring is necessary in order to assist alloying ingredient dissolution and to achieve temperature and composition homogeneity of the melt. Improper practices have the potential to increase dross generation and to reduce melt cleanliness by introduction of inclusions below the surface. Whether stirring is performed by forklift equipped with stirring tools, dedicated furnace tending vehicles, impellers or other means, the same principle required for metal transfer applies; turbulence and air entrainment must be minimised. Stirring tools should be kept below the melt surface and moved so that there is minimal disturbance of the melt surface. The utilisation of forklift tools is still probably the most widely used method although the efficiency of this technique is highly dependant on operator skills and, as it is performed through an open furnace door, there is a high energy loss. Electromagnetic stirring (EMS) is a very powerful, non intrusive, subsurface stirring device; however, this equipment is very costly and if not used correctly can increase dross generation. Utilisation of rotary impellers is another very efficient approach to stirring with performances similar to EMS. Savings here are difficult to quantify but can be significant.

Fluxing

Furnace fluxing is the main step during furnace preparation to remove impurities such as alkali elements and inclusions and will therefore greatly affect metal quality. It can also assist in recovering metal from dross. This operation involves injection of a chemically active ingredient such as gaseous chlorine or salt flux mixture with a carrier gas. Injection with static lances is an extremely turbulent process that generates a major amount of dross and suffers of poor performances. Utilisation of rotary injection processes such as the Rotary Gas Injection (RGI) or


Rotary Flux Injection (RFI) will minimise the surface turbulence and increase the overall process efficiency [6] (reduce turbulence, reduce temperature loss and reduce time).

Consequences of improved furnace metal quality impacts on filter performance and final product quality. All filters whether they are Ceramic Foam Filters (CFF), Deep Bed Filters (DBF) or Porous Tube Filters (PTF) work on a percentage removal of inclusions. A high incoming inclusion level will be reduced by filtering but the level will be higher than if the incoming metal had a lower inclusion level.


Increased metal cleanliness has several benefits including less cast aborts due to blocked filters, less casting defects and a lower potential for customer complaints resulting from poor product quality. For multi use filters such as DBFs or PTFs an additional benefit of cleaner metal is a longer life between rebuilds which has maximum benefits in long order runs. Figure 3 shows, for a given alloy, the relationship obtained between furnace metal quality and the useful life of a DBF. Potential savings from improved metal quality can be thousands to 100’s of thousands of dollars, not to mention harm to reputation.

0.5

0.6

0.7

0.8

0.9

1

0 1 2 3 4 5

Inclusion Level

Relative DBF Life

Figure 3. Correlation between increasing furnace inclusion level and DBF life for a given alloy.

Skimming

Any melt surface will contain a dross layer which forms while preparing the metal for casting (Fig. 4a). The thickness of this layer will greatly affect the efficiency of the furnace as dross will act as a blanket and prevent the heat from reaching the melt. Figure 4b shows the extra energy required to melt submerged scrap as a function of the dross layer thickness [7]. It can be seen that while it is important to minimise the dross from forming in the first instance, when it is present in a significant amount, skimming must be performed. The skimming practice itself is also important as it should minimise removing unnecessary metal with the dross.


Consider our hypothetical casthouse where the nominal thickness of the dross layer would increase from 2.5cm to 5cm following a stir or scrap charge. According to Figure 4b, the energy required to melt submerged scrap would be increased by 1000 kJ/kg. This would be equivalent to an extra cost of approximately $10 to melt each tonne of submerged scrap. Assuming that 75% is submerged scrap, this would be equivalent to ~$100k/year.


a)

0

500

1000

1500

2000

2500

3000

3500

0 50 100 150

Dross Thickness (mm)

Extra Ene

rgy (kJ/kg

)

b)

Figure 4. a) Compacted swarf bales floating on top of the melt and b) extra energy required to melt submerged scrap as a function of the dross thickness [7].

Furnace Energy

Energy Utilisation

No matter if they use natural gas, heavy fuel oil or electricity; furnaces are the main energy user in a casthouse. Surprisingly enough, even with their high operating costs, the efficiency of the furnaces is something rarely tracked by the casthouse. As is shown in Figure 5, making the effort of monitoring the furnace energy efficiency can expose potential savings. In this example, the consumption of two identical furnaces in terms of layout, amount of scrap processed and alloy type was compared over 6 months. As this data illustrates, furnace B used significantly more energy than Furnace A. Without tracking this data you would not know there was an issue.

Here’s what you can gain…

An investigation into the root cause of this excessive usage showed that inadequate maintenance was the primary cause for the 30% higher fuel usage and greenhouse gas production of furnace B. Depending on the size and usage of your furnace, this can quickly reach in excess of $100k/year.

0.40

0.50

0.60

0.70

0.80

0.90

1.00

Jan Feb Mar Apr May Jun

Relative Energy Consu

mptio

n

Furnace AFurnace B

Figure 5. Monitoring energy consumption can result in significant savings.

Floating Swarf Bales


Energy Recuperation

With furnace flue gases reaching temperatures as high as 1000°C in the chimney stack, this represents a lot of energy which is wasted to the atmosphere. Using this energy in recuperative/regenerative burner systems can result in significant energy savings. Generally recuperator systems will preheat the combustion air around 400°C to 500 °C which can increase the fuel efficiency by 20%. The regenerative burner technology will preheat the combustion air above 800°C; in this case the fuel efficiency will be improved by 40% as graphically shown in the Figure 6a. While savings can be significant, these systems require high capital. A simpler way to make use of this available energy is to preheat scrap. Figure 6b shows the temperature profile for 1t of ingot which were preheated in an oven using exhaust gases from a 2t melting furnace as the energy source. As is shown, the ingots reached a temperature of 300-350°C after approximately 2 hours. The heating potential from a 40t furnace will be much higher.

Here’s what you can gain…

Let’s consider that our smelter has an exhaust gases system that can preheat scrap to 350°C. This is equivalent to an approximate saving of 3$/tonne of scrap or a saving of $35k/year in fuel costs.

a)

0

50

100

150

200

250

300

350

400

0 1 2 3 4 5Time (h)

Tem

perature (°C

)1st Ingot Set2nd Ingot Set3rd Ingot Set

b)

Figure 6. a) Effect of combustion air preheat temperature on fuel consumption and b) temperature profile of ingot bundles in an oven using waste gases as the heat source [8].

Conclusion

As was highlighted in this paper, there are a number of opportunities for significant savings and reduction of greenhouse gases in the operation of your furnaces. Opportunities for a more efficient use of available energy are numerous within the casthouse and making small efforts in different areas of furnace operation will quickly show on product quality and the bottom line. Whether it’s related to furnace operation, operator practices, scrap additions or metal quality, the savings can quickly reach ½ Million $/year. Here is a summary of what you should look for:

• Minimise turbulent melt transfer and stirring operations

• Reduce exposure to air for scrap with a low surface area to volume ratio

• Use superheat to melt scrap making use of the “free” energy

• Opt for fluxing practices which give good efficiencies and minimal dross generation

• Ensure that your alloying practices are optimised for the element considered


• Ensure that the dross level on the melt surface is not excessive

• Train operators so they understand how they can affect the process

• Monitor furnace consumption and understand how they work

Acknowledgements

This paper would not have been possible without the contribution of the following colleagues: Vincent Goutière and Claude Harvey from the RTA Arvida Research and Development Centre, Barbara Rinderer, Cathy Smith and Shane Charles from the RTA Pacific Technology Centre, Phil Austen and Michael Bishop from the Bell Bay smelter.

References

[1] U.S. Energy Information Administration (2008) Official Energy Statistics from the US Government, http://www.eia.doe.gov

[2] Campbell, J., (2003) The New Metallurgy of Cast Metals - Castings, Second Edition, Elsevier Science Ltd, 337pp.

[3] Goutière, V., Gariepy, B. and Dupuis, C., (2007) Mapping of Bath Carryover from Cell Tapping to Casting in Smelter Operations, Aluminium Cast House Technology, 231-238.

[4] Williams, E. et al., (2008) Evaluation and Reduction of Potroom Bath Carryover to the Casthouse, Light Metals, 557-561.

[5] Stewart, A.L., McCubbin, J.G., Sulzer, G., (1977) Melting Aluminium and Aluminium Alloys, Light Metal Age, Vol. 35, n 11, 13-15.

[6] Béland, G., Riverin, G. and Dupuis, C., (1998) Rotary Flux Injection: Chlorine-Free Technique for Furnace Preparation, Light Metals, 843-847.

[7] Reed, R.J., (1985) Pouring the Heat to Aluminium, Light Metal Age, 10-13.

[8] Cumerford, B., (2007) Visit to Mullins Wheel 2/2/07 – Examination of Heating Rates in Preheat Furnace, Rio Tinto Alcan Internal Document.


Homogenization Aspects, Continuous versus Batch in an Integrated Modern Casthouse

Franz Niedermair

Hertwich Engineering, Weinbergerstr. 6, Braunau, 5280, Austria [email protected]

Keywords: continuous homogenizing, batch homogenizing, ultrasonic testing, sawing & packing

Abstract

Continuous homogenizing for extrusion billets has become the accepted standard and state of the art

technology. Hertwich Engineering (HE) has, to date, built more than 120 such plants worldwide,

which are processing some two thirds of the worldwide extrusion billet production.

Batch type homogenizing is still the best suited technology for various alloys which require long

holding times, e.g. 4 – 12 hours, as well as for large-only log diameter production. Again, HE has in

recent years supplied several batch homogenizing plants.

Topics:

Process technology, heating, holding, different types of coolers

New log support saddle design, exit car, no rolling of logs with focus to good billet surface

finish

Automated sawing, packing

High efficiency batch furnace and cooler with reversing air and precise temperature regime

Introduction

Developed in the 1970s by Hertwich Engineering (HE), continuous billet production plants have

become the accepted standard for the production of extrusion billet and have since revolutionised

modern casthouses. Today more than 60% of worldwide production of extrusion billet pass through

HE continuous homogenizers.

This extraordinary success stems from a number of clear advantages:

Perfectly uniform metallurgical property of billet due to precise and reproducible temperature

regime during heating, holding and cooling

Combination of various production steps into one fully automated, continuous process, hence

substantial labour savings

All equipment components have been developed, designed and built by HE

Plants built to latest state of the art technology, 120 such plants in operation


Description of a Modern Plant; Typically 40,000 to 150,000 Tons Per Year, with the Emphasis on Technical Features, Novelties and Specific Options

Figure 1. Typical plant layout.

The fully automated production sequence starts with laydown of cast billets on the entry table. From

there billets are moved over a storage conveyor to the inspection station, where billets are visually

inspected for surface defects and ultrasonic inspected (UT) for center cracks and inclusions. For

extrusion billets a linear UT station is adequate, while billets destined for aircraft or automotive

components are inspected with the helical method, whereby the entire billet volume is inspected.

Billets are rotated while several equidistant probes move along the billet surface, each probe

inspects only its section of the billet. This arrangement is necessary to achieve the required

throughput. Fault sizes as small as 1.2mm FBH can be detected. Data of UT-inspection-results for

each individual billet is recorded and remains “attached” to the billet throughout the system (log-

tracking database). For instance, a center crack detected near the foot end of a billet causes the

integrated billet saw to automatically remove and scrap the faulty section of the billet.

Figure 2. Linear testing. Figure 3. Helical testing.

A storage conveyor provides adequate billet storage between (UT) inspection and continuous

homogenizing furnace. For reduced noise and to avoid surface damage logs are lifted and lowered

during transfer, no rolling is permitted. Incidentally, this applies throughout the billet production

plant.

UT Furnace Entry Conveyor Air Cooler Briquetting Press

Stacker

+ Packing

Saw

Saw

Log Laydown Storage Conveyor Continuous Homogenizing Furnace Saw Entry Conveyor


Heating and Holding

Heating in a continuous homogenizing plant

In the heating zone of the furnace (extrusion quality) billets are heated to homogenizing temperature

within 1.5 to 4 hours, holding times may range between 2 to 6 hours. However, certain plants have

been built to meet different, special heating requirements.

Figure 4. Heating in a continuous homogenizing plant.

Figure 5. Heating curves in a continuous homogenizing plant.

For instance the plant at Raufoss Automotive processes hard ZnMg alloys. This furnace is capable

of slow heating in the critical temperature range to avoid development of cracks due to inner

tensions. (red line (2) vs. standard blue line(1)) Throughout holding all logs are kept precisely on

the set temperature, over the entire log length. Temperature deviations are within 2°K, and

guaranteed within + 3°K. Actual log temperature for each log is verified with several sting

thermocouples on entering the holding zone as well as on the last place before transfer to the

cooling station. Temperature control is supported by an intelligent algorithm.

Heating in a batch homogenizing plant

Some 20% faster heating is achieved with the new generation of furnaces due to reversing air. This

is latest technology in batch furnaces and allows very narrow temperature deviation between outer

and inner logs, which is no longer significant.

1 2

2 3 4 5 1

1 entry accumulator/ conveyor

2 rapid heating zone(s)

3 holding zone

4 cooling station

5 exit accumulator / conveyor


Figure 6. Log A: Outermost position in a batch; Log B: Log positioned inside a batch.

Figure 7. Batch type billet heating – Comparison single flow/reversing flow.

Log Diameter 254mm.

Cooling

Cooling in a continuous homogenizing plant

Cooling after homogenizing is as important as heating and soaking to achieve best possible

extrudability and mechanical properties of billets. Many different cooling regimes are in use,

depending on metallurgical requirements and clients recipes, hence a wealth of different coolers

have to date been built. Log temperature uniformity in a continuous cooler is typically superior to

that in a batch type cooler.


Tem

pera

ture

(C

°)

Time (min)

Line “a”: cooling by water quench. With AlMgSi alloys, Mg2Si thereby remains completely dissolved.

Line “b”: cooling in very intense air flow, with cooling rates 600 to 800° C/hour.

Line “c”: the log temperature during cooling in a moderate airflow, whereby cooling rates can be adjusted 350 to 550° C/hour. The majority of plants built to date are designed and equipped for such cooling rates, as these meet the mainstream requirements of extruders.

Line “d”: step-cooling”. In practice Al-Extruders cannot find – so far – a real advantage vs. our standard quality including hard to soft degressive or quasi step-cooling, whereby similar structures are generated, too. Therefore the extra technical investment for step cooling is hardly required.

Line “e”: extremely slow cooling in the range 75 to 150° C per hour, which is required for certain special alloys. Such slow cooling rates cannot be achieved simply, not even with standing air, but require closed cooling chambers with several zones of different air temperatures.

Figure 8. Cooling curves in a continuous homogenizing plant.

In accordance with this variety of cooling regimes, actual cooling stations vary much in design and

construction. We distinguish between open and closed type coolers, even coolers with built in

heating devices. For certain countries with cold climate we have built cooling stations with

integrated waste-heat recovery. Thereby cooling air is re-circulated until it is suitable for heating of

buildings.

Cooling in a batch homogenizing plant

Cooling in a batch plant naturally results in differing log temperatures, depending on the position of

a log in a batch. However, with modern plants with reversing air, such log temperature difference is

only marginal and quite insignificant.

inner logs outer logs

Figure 9. Cooling curves in a batch homogenizing plant.

Figure 5: Cooling Curves


Automatic Log/Billet Handling for Batch Homogenizing Plants

Modern plants include automated stacking and unstacking equipment for lean operating labour and

enhanced safety, comprising spacer manipulator, spacer magazines and a crane for log layers.

Spacers made of heat resistant steel are well suited for automation of the process, and are extremely

durable – spacers supplied in 1990 are still in use today. In combination with a moveable bottom in

furnace and cooler single spacers are placed in optimized distances for best possible log support and

to avoid overhanging of log ends (HE variable spacer system). Heat expansion or contraction is

compensated to avoid surface marks on logs. The lower weight of single spacers also contributes to

lower energy consumption.

Sawing, Marking, Packing, Weighing, Strapping

After cooling billets are transported to the saw plant, passing over a combination of storage

conveyors, shuttle cars, roller table lines etc. Once cut at the saw, all billets are marked. The

introduction of a pin-stamping unit in recent years is a small but essential innovative contribution to

full automation. Today’s pin stamping unit receives marking information directly from the control

system, eliminating time consuming manual changing of stencils and potential mistakes by inserting

incorrect numbers. HE has developed its own pin stamping unit, as other systems available on the

market were too slow.

Head and butt ends and scrap pieces are gripped in sawing position by a scrap-manipulator and

deposited in scrap-containers. Swarf is extracted directly from the saw and pneumatically

transported to the briquetting press.

Cut billets now travel to the packing area on roller table lines. The packing plant shown is designed

for creating long and short billet stacks simultaneously. This type stacker is particularly suited for

bulk production of short billets.

Today’s standard strapping center includes an automatic wooden runner magazine and –positioning

device. Number and position of wooden runners and strapping bands are applied fully automatic

according to customer specification, no operator intervention or adjustments are required. Actual

strapping is with steel band, or more recently with PET-band. Weighing of stacks is done prior to

and after strapping, to generate net- and gross weight. Ancillary equipment may comprise stack

manipulator and label applicator.

Figure 10. Sawing and packing.

The label applicator attaches the label in a defined position on the bundle. The manipulator stacks

completed bundles in dedicated positions for removal by fork truck, or places stacks directly onto

trailers.


Summary of Improvements and Essential Developments Achieved within the Last 10 Years (Continuous Homogenizing Plants)

Increase of operational reliability

With extensive, ongoing development many improvements were achieved in recent years.

Plants are of sturdy construction, and highest quality. Plant availability is 99 %.

Careful handling for improved billet surface

New generation of billet supports with special inlays (patented) developed for the furnace’s

billet transport system. Thereby surface defects on billets can be largely eliminated. Removal

of billets from the furnace is by a shuttle car, replacing earlier roller table lines. All rolling and

banging of logs has been eliminated!

Improved heat transfer

By intensifying hot gas convection and modifying air guide channels, the heat transfer rate

onto billets has been increased substantially (without increasing the number of ventilators).

HE always prefers to rely on a limited number of high-quality powerful ventilators rather than

a “forest of ventilators” on top of a furnace. The achievement directly results in overheat

temperature (heathead) in the heating zone of max. 5 to 10° K above set homogenizing

temperature.

Improved direct measurement of billet temperature inside the furnace and cooler

Increased production of the furnaces has prompted the desire for increased number of

measuring points, even more accurate temperature control and refined data processing.

Consequently direct temperature measurement is further on done at more positions and with

higher accuracy. The intelligent analyses – software with trend analyses, provides an

appropriate tool for early recognition of possible changes to the heating regime.

Implementation provides the means for even tighter quality control. Changes of billet diameter

and homogenizing temperature can be executed with higher safety and great flexibility.

Improved, innovative control software for operator- and maintenance friendly production

Fully integrated, automated production plants need an excellent control and monitoring

system to ensure safe and reliable production and high availability. All single movements of

the plant are monitored for correct execution. The refined fault diagnose system efficiently

guides operators and maintenance personnel to the causes of malfunctions, thereby supporting

easy and quick corrective action. Malfunctions are stored and can be queried in statistic and

trend form - a potent tool to preventively fix weak points.

Automatic restart programs efficiently and reliably bring the plant into a predefined condition,

eliminating potential human error. With Log/Billet tracking relevant data on each log or billet

is tracked throughout the system. Temperature evolution is available for every log in furnace

and cooler. The operator friendly graphic display of billets within the system comprises all

associated data like batch no., diameter, UT-inspection data.

Host connection of the plant

Relevant production data including homogenizing parameters, sawing and stacking

dimensions, positioning of wooden runners, strapping specifications etc, are downloaded from

the clients production planning and order process host system, directly to the plant control PC.

On the other hand the plant provides data to the clients host system:

o Measured data of homogenizing time and temperature (for quality assurance) for each

and every log, for modern paperless production

o Quality inspection results, i.e. center cracks, inclusions, surface defects


o Produced quantities, weights, charge No., log No., time stamp etc.

Increased throughput of large furnaces

The past 10 years has shown a trend to ever greater throughput of furnaces and plants. To date

the highest throughput rate is achieved with a plant (Dubal IV), where one billet passes

through the furnace every 82 sec. Throughput is 22 t/hour or 1,100 logs per day.

Discussion: Continuous Vs. Batch Type Homogenizing

Both types have their justified place in modern casthouses. The continuous homogenizer though is

the front runner in terms of mainstream production of extrusion billet.

Continuous or batch homogenizing

For 6,000 series alloys and holding times 2 to 6 hours, as well as for extended campaigns of the

same log diameter, the continuous homogenizer is the best suited choice, e.g. lowest energy

consumption, best billet straightness, lowest labour costs etc. The batch type homogenizer is

typically favoured for large billet diameters and special alloys with rather long holding times, as

well as for frequent temperature changes or intermittent production. The homogenizing quality is no

longer a valid argument since differences are only marginal in today’s generation of modern

equipment.

Figure 11. Layout continuous + batch homogenizer.


Rational coexistence for continuous and batch homogenizing

At many prominent aluminium producers, a combination of continuous and batch homogenizing is

a common sight for ultimate flexibility. In case obsolete batch homogenizing equipment is replaced

with continuous homogenizers, most customers retain some of the existing batch plants to

specifically process certain special alloys.

Operating labour for a combined plant is typically limited to:

Data storage on log entry

Quality control at visual/UT inspection station

Billet saw

Conclusion

HE production plants for extrusion billet have been steadily improved to now present fully

developed and automated plants – a must in a modern billet casthouse to hit target values! HE

draws from a wealth of experience with 120 such plants built to date, to provide tailor made

solutions in terms of functionality and available space. All plants of HE are designed and built

exclusively upon proprietary know-how by HE itself, including ancillary equipment like UT-

inspection, strapping, swarf briquetting.

Fifty highly qualified engineers are available for development of control software and data

management to clients host systems. Our customers have the certainty to buy thoroughly

engineered, top-quality equipment, which ensures reliable production throughout an exceptionally

long service life. To address our customers future needs and to provide extended service assistance,

comprehensive after sales and spare part support is offered.


Fives Solios Experience in Modern Secondary Aluminium Casthouse Construction

Lee Allena, Paul Hipwood and Barry Houghton

Solios Thermal Ltd, Heath Brook House, Heath Mill Road, Wombourne, Wolverhampton, West

Midlands, WV5 8AP, UK a [email protected]

Keywords: furnace technology, aluminium recycling, secondary smelting, melting furnace

Abstract

In early 2006, Fives Solios, namely Solios Thermal, was selected as a key cast house supplier by a major aluminium producer for a new 130,000te/yr re-melting facility. The project scope included not only the supply of modern melting and holding furnaces plus ancillary equipment, but also pre-heating furnaces, an air pollution control system, basic cast house engineering, cooling water schemes and complete turnkey project management. This project presented particular technical and logistical challenges and was required to meet exacting European environmental / efficiency standards for all major contractors involved. The facility was commissioned in March 2008 and the first production was on schedule.

This paper will discuss the challenges that the Solios Thermal project team had to overcome, from layout of equipment to conclusion of the project; also how good communication and close collaboration between the customer and all major subcontractors resulted in a successfully implemented plant.

Introduction

A requirement for a new 130,000te/yr casting facility had been identified in Europe. The strategic placement of the site in central Europe would present major benefits for the client including;

• Creation of a centre of excellence for cast products in a major market area.

• Use of an existing rolling plant adjacent to the new casting facility.

• To shorten the supply chain duration for special cast alloys to this market area, the supply for which had been traditionally from the USA.

Cast House Challenges

Specifically for the cast house, the client had important design criteria that had to be respected in the engineering phase. To conform to the need for short production runs of various alloys; a high level of flexibility had to be provided in the design layout and equipment choices made. [1]

Cast house cooling water supply would utilise existing available water coming from a combination of local town and existing plant resources. The design concept would need to also consider maximised efficiency and a reduction in operational waste. Material flow and personnel interfaces had to be designed to comply with European HSE standards and Environmental regulations would need to be respected.

The date of the first metal cast was fixed to be in March 2008.

Cast House Engineering

Solios Thermal, was chosen to be part of the clients engineering team because of its extensive experience in cast houses. The engineering contract was for the preparation and development of the


complete cast house layout, water cooling system, basement, casting pit, and building design with sufficient detail to allow local civil engineering contractors and other suppliers to accurately bid for detail civil design work and ancillaries.

The physical working envelope of each major piece of equipment was used in conjunction with maintenance requirements and the plant process flow description to generate a model layout and determine the size of the building. Other important factors such as safety, ergonomics and efficiency were incorporated into the final scheme.

Service requirements for the proposed equipment were estimated, so that a total plant flow and consumption of services could be evaluated. This resulted in a tabulation of cast house loading to define the total incoming services. Several iterations of these calculations were required before an optimum building outline was finally produced. [2]

Optimised Cast House Layout

Solios Thermal took an active role in developing the cast house layout with the client, including the placement of major items such as furnaces and the casting machines.

An extensive basement was included for the placement of various pieces of process equipment including, cooling water components, hydraulic power units, gas stations, electrical and pneumatic service equipment. Service routes and placement for these facilities were then outlined. Many plans were tried and this led to re-positioning equipment to optimize placement.[2]

Removing equipment from the cast house floor enabled the reduction of the overall building envelope by some 20m in length from the original plan and provided an open plan layout.

Figure 1. Cast House layout showing pre-heating furnaces and main melting furnaces.

To accommodate the plant HSE requirements, personnel access routes and material flow were separated as much as possible within the building envelope. Great care was taken to ensure that material flow took the quickest and most practical path through the plant to reduce plant lead times. (Figure 1).


Cast House Furnaces

A strict regime was employed to ensure that moisture does not enter the cast house melting furnaces, this being accommodated by a number of pre-heating furnaces (Figure 2). These furnaces also aid the melting process by pre-heating the solid metal prior to melt. The pre-heaters were placed locally to the main cast house furnaces to reduce charging times (Figure 1). Specially designed charging vehicles are used to collect heated scrap from the pre-heaters and convey it to the melting furnaces. Access for the charging vehicles had to be accommodated in the cast house layout. (Figure 5)

Figure 2. Actual pre-heating furnace arrangement.

The operating requirements for the clients casting process imposed the need for a high level of flexibility to enable various quantities of alloys to be produced simultaneously. This resulted in a ‘multi-furnace, various capacity’ approach. A selection of six melting and holding furnaces ranging from 20te to 65te capacity are used to meet many combinations of production runs. The latest 3-D design technology was used in combination with parametric sizing programs to quickly access and iterate to the optimum furnace designs. (Figures 3 and 4)

High melt rates of up to 20Te/hr were needed to fore fill plant productivity wishes. To achieve this the furnaces required a high power input and it was important to ensure that this was done efficiently to reduce fuel usage costs. Twin regenerative burner systems were used on the melting furnaces running at up to an efficiency of 80% to comply with the projects specified fuel consumptions of 600kWh/te or less. The NOx emissions were kept below 250mg/Nm3.


Figure 3. 3-D models were used to design the 65te melting furnaces.

Figure 4. Installed 65te melting furnace.

Cooling Water System

The cooling water system was a unique design to the plant; Existing water supplies were used and were routed under ground. It was decided to create two large tanks in the basement to act as reservoirs for the water supply; one receives the cold water and is of sufficient volume to complete a full cast and the other holds the hot water from the process. A complex system of piping and control items for this process is also contained within the basement of the plant. The process water is cooled between the casts using the incoming cold water supply. The temperature of any excess water flowing into the adjacent river is tightly controlled to reduce any environmental impact.

Space Heating and Cooling

A heating and ventilation system is integrated into the building design to make use of all available plant energy and to minimize running costs.

During winter low-grade heat from the hot water tank is used to heat the building and during summer the cold water supply is used to provide some cooling. Any resulting rises in the water temperature are too small to affect the VDC casting process.

To reduce noise emissions from the building it was decided that the ventilation system would not include normal open roof vents or panels in the fabric of the building. A balanced forced ventilation system was used. It was important to ensure that the building pressure was balanced correctly, high negative pressure in the building could lead to possible down flow in the furnace exhaust stacks. Air extracted from the roof is exhausted to the outside through sound absorbing ducts. Fans are used to introduce air into the building through sound absorbing boxes that also contain radiators. [2]

Environmental System

The local legislation with regard to emissions was very exacting and had a large influence on decisions made during the project. As is normal with any project in Europe, an extremely detailed Environmental Impact study had to be prepared by the client.

To reduce the risk of heavy metal and particulates being discharged to atmosphere from the furnaces, a bag house filter was provided by Solios Thermal in the cast house furnace exhaust system. (Figure 6).


Figure 5. Solid charging into melting furnaces. Figure 6. Bag filter for furnace waste gas management.

Turnkey Contract

In addition to involvement in the cast house engineering process, Solios Thermal were contracted to supply key cast house equipment.

The maximum capacity of 130,000 tonnes was carried out in two-phases. The new building and services were to be capable of housing all the equipment required for the two phases.

Scope of Supply

Phase 1

Turnkey supply included the following melting, casting and ancillary equipment:

• 3 x 20 tonne Pre-heating furnaces.

• 1 x 65 tonne Melting Furnace.

• 1 x 60 tonne Tilting Holding Furnace.


• Furnace to furnace transfer launders

• Waste gas bag house filter and associated ducting network.

Phase 2

Turnkey supply included the following melting, casting and ancillary equipment:

• 2 x 20 tonne Pre-heating furnaces.

• 1 x 65 tonne Melting Furnace.


• 1 x 30 tonne Tilting Holding/Melting Furnace.

• Furnace to furnace transfer launders.


Execution

The engineering phase of the project was only half of the story; as with any major construction project the success can only be assured if the correct equipment is specified, delivered, installed and commissioned in time and within budget. Once all the unique designs for the project had been taken into consideration then a carefully structured implementation phase became active.

The complexity of this project required a core project team to be assembled by the customer that included representation from all of the major sub-contractors. The team was established to respond quickly, both to any omission / oversight in the original concept or to changes from external pressures. Solios Thermal was identified as one of the major partners who could satisfy this role for the cast house area.

Communications between the various project partners was by means of regular progress meetings that were managed by the client; this approach resulted in the issue of mutually agreed schedules.

Logistics and Construction

During the construction phase of the project, there was a requirement for deliveries to be made using road transport. This meant equipment was limited in size to the vehicles that could access the various tunnels and passes leading into to the plant.

Furnace casing panels were designed to the largest allowable component sizes suitable for such transportation. Once delivered the furnace casing was assembled directly onto their final foundations (Figure 7). This required a considerable amount of fabrication and welding capability.

Once the casings were erected, fabricated and fully welded, refractory lining commenced. A monolithic refractory lining system was used for these particular furnaces. Special measures were taken for refractory material stored outside, pre-heaters were used to raise refractory temperatures prior to implementation.

Other ancillary components were delivered, where possible in pre-assembled form, then stored and issued for use as required (Figure 8).

Initial furnace construction started whilst the cast house building was not complete. Precautions had to be taken to protect the welding and construction processes from the harsh local elements. Temporary tents and portable heaters were used to allow the work to continue.

Once material arrived on site it was placed directly into its ultimate position wherever possible. Where this was not possible, areas were allocated for each contractor for the management of deliveries. This resulted in double or triple handling of material delivered. Co-operation between the major subcontractors was actively encouraged and scheduling of deliveries had to be done in such a manner that this did not cause progress obstruction on the site. This was facilitated by regular site meetings. [2]


Figure 7. Melting furnace under construction in final position.

Figure 8. Modular construction of furnace exhaust stacks.

Plant Construction

From the cast house engineering exercise, sufficient information was given to the civil engineers to start the detail design. This was an iterative process due to ongoing changes in equipment or process.

Shared routes between the various vendors for utilities and site wiring were encouraged for reducing potential conflicts during the installation phase. A good example of this was in sharing cable trays to eliminate duplication of installation work, a process that Solios Thermal actively participated in.

Vendors were also approached to provide the minimum installation requirements for major items so that the installation schedule could be organised efficiently. For example, assembly of the furnace casing was started by Solios Thermal as soon as the roof over the foundations were completed.[2]

Conclusions

The various challenges that faced the project were met successfully. The official opening of the cast house occurred on time.

To achieve the complex requirements of the project including, unique project designs, compliance to European regulations, tight timescale, etc… the following characteristics became critical.

• Core teams with multidiscipline skills and application expertise are essential.

• Integrated planning of the project.

• Team spirit to encourage a “can do” attitude.

• Realistic targets that are challenging but are agreed and achievable.

Without these attributes the project would not have been a viable undertaking. Fives Solios were very much part of this approach to the plant construction and firmly believe in the project managements decision to encourage and support the communication and co-operation between all major responsible subcontractors. The lessons learned only reinforced good project management practices such as planning, communication and efficient design with strict progress monitoring and control, which are crucial for project success.


Acknowledgments

The authors would like to thank members of the project engineering and installation teams for their support and work involved in construction of this paper.

References

[1] Pierre Le Brun (2008) Melt Treatment – Evolution and Perspectives, Light Metals 2008, p621.

[2] Paul Hipwood (2009) Fives Solios Experience in Constructing a Modern Day Secondary Aluminium Cast House – Aluminium International Today March/April 2009, p19.


Dry Hearth Melting Furnaces

Peter Newman

Furnace Engineering Pty Ltd, 50 Howleys Road, Notting Hill, Victoria, 3168, Australia [email protected]

Keywords: furnace technology, charging, dry hearth, furnace design

Abstract

This paper outlines various aluminium melting furnaces arrangement alternatives and their related benefits as well as the physical and practical challenges of the aluminium melting process using fuel fired reverberatory furnaces. Performance comparisons are made between dry hearth and wet hearth furnaces to highlight the benefits of dry hearth melting as well as the impact of melting practice on ultimate equipment performance. Both single chamber and twin chamber dry hearth furnaces are described in various configurations including the unique benefits of each design.

Introduction

The single purpose of an aluminium melting furnace is to convert the metal from its solid phase to liquid. Whilst this is a very simple goal, the industry has many versions of the melting furnace, each with its own benefits and (of course) limitations. Aluminium furnaces must be engineered to suit the specific application of the user, whether it is for:

• Primary smelter remelting;

• Continuous casting;

• Die casting;

• General scrap recycling; or

• Extrusion scrap recycling.

Within the budgetary limitations, consideration must be given in the final furnace design to maximizing performance in the following areas:

• Heat transfer and melting rate;

• Oxidation loss;

• Energy consumption;

• Utilisation;

• Versatility; and

• Safety.

Aluminium Characteristics

Density

Solid aluminium is approximately 15% more dense than molten aluminium and will therefore sink under the surface of the bath, significantly reducing heat transfer. Pure aluminium will absorb approximately 1150MJ/tonne to convert from solid metal at room temperature to liquid metal at 750oC. Approximately 55% of the energy is absorbed before the melting point and over 30% is absorbed during melting as latent heat of fusion. Therefore 85-90% of the energy is being transferred to the metal whilst it is in a solid form, potentially submerged beneath the molten bath.


Emissivity

Aluminium in both solid and liquid forms does not readily emit or absorb radiant energy. It will generally reflect 80-90% of infrared radiation. Maximising the differential temperature between the charge and the furnace is therefore very important for heat transfer, as is maximising the surface area of dense charge material exposed to the hot furnace gases and to the radiant furnace surfaces.

Thermal Conductivity

Liquid aluminium has a thermal conductivity approximately 50% lower than solid aluminium. Energy movement from the surface of the bath down to the solid charge material will be limited by internal resistance to energy flow. Metal temperature variation from the top to the bottom of the static bath can vary by 1oC per centimetre or 60-70oC in a typical furnace. Any solid material resting on the floor of the furnace will therefore melt at a much slower rate due to this temperature gradient.

Oxidation

Oxides form on the surface of the bath and this effect rapidly increases with temperature. These oxides tend to form an insulating blanket that will reduce the heat transfer coefficient of the bath surface, further limiting the energy available for melting submerged solid materials

Single Chamber Reverberatory Melter

The single chamber reverberatory melter is the most common aluminium melting furnace concept presently used within cast houses.

Solid aluminium charge is placed directly into the furnace holding/melting area. Smaller furnaces can be charged with a forklift directly from a chute placed in front of the door. With larger furnaces, this presents some difficulties and may require a dedicated charging machine. Often there will be a remaining heel of molten aluminium in the furnace. In this case the charge material must be completely dry and free of liquid contaminants which otherwise may result in an explosion within the furnace. The advantages of this furnace solution are simplicity, low cost and versatility. However, the physical characteristics of aluminium described earlier highlight some of the limitations of this furnace.

Wet Hearth Single Chamber Furnace Heat Transfer and Melting Performance

Figure 1. Heat transfer through a molten bath.


Figure 1 shows solid aluminium submerged beneath the surface of the molten bath. Once submerged, heat transfer to the solid metal can only be indirectly through the surface of the liquid bath. Heat transfer to the bath heat is by a combination of radiation from the refractory surfaces and from the burner flame envelope, as well as forced convection from the products of combustion that circulate through the chamber.

Heat transfer rate to the bath will increase with:

• Furnace temperature;

• Burner flame temperature and burner envelope size;

• Bath stirring;

• Bath area to depth ratio;

• Reduced dross levels; and

• Increased insulation.

The rate will generally range from 150-250MJ/hr.m2 If all melting had to be accomplished below the surface of a molten bath, the specific melting performance of a furnace with conventional firing and no stirring would be in the range of 140-170kg/hr.m2 When specifying furnace melting performance, it is important to ensure that the conditions under which this melt rate is achieved are well qualified.

A single chamber furnace melting rate can be as high as 300kg/hr.m2; however, this will only be realised under a very specific set of conditions with a certain type of charge material. If the metal is charged into a substantial heel, then the melt rate per m2 is likely to be less than 200kg/hr.m2. Selection of burners with heat recovery systems and application of stirring devices such as EMS, PMS or mechanical pumps will considerably increase thermal efficiency or improve heat transfer from the furnace to the aluminium, however, this comes at additional capital and maintenance cost.

The following chart in Figure 2 shows the typical performance of an aluminium melting furnace where 4,000kg of ingots or sows with minimal preheat have been charged directly into a molten bath of approximately 10,000kg.

Figure 2. Single chamber furnace with metal charge directly into bath, typical operation.


Four regions can be defined from Figure 2.

0-20 minutes: Initial thermal input to the furnace is very high to recover heat lost from the door opening and because some solid material will remain exposed above the metal line. Bath temperature drops rapidly as the charge material cools the molten metal to a temperature that is close to solidification. Stored latent heat of the molten metal is released as the cooling occurs, and this tends to prevent the bath temperature from dropping below 650oC.

20-35 minutes: Flue temperature and roof temperature rapidly rise to the set point. The bath temperature is maintained at approximately 650oC due to the latent heat of melting aluminium and the phase change continues to absorb energy from the bath. Thermal input must reduce because the refractory surfaces have reached the operational limit. Heat transfer at this stage is primarily through the surface of the molten metal. The rate of heat transfer can be increased by raising the set point of the roof and flue temperature, but will however result in reduced refractory life and increased fuel consumption.

35-50 minutes: Bath temperature increases because the solid metal within the furnace has reduced. Rate of energy transfer through the surface of the bath now exceeds the cooling effect of melting aluminium and casing heat loss.

50-75 minutes: Thermal balance must be maintained and so the burner system output is reduced accordingly. This last part of the melting cycle delivers the final 7-10% of the total energy to the metal but requires more than 30% of the cycle time. Heat transfer in this final stage is constrained by the surface coefficient of the molten bath.

Single Chamber Dry Hearth Furnace

Figure 3. Single chamber dry hearth melting furnace.

In its most simple form, the melting area may simply be an extended and elevated charge ramp or sill that enables the solid metal to either be preheated or partially melted without coming into contact with the liquid bath (Figure 3). The melting area does not have a separate burner and is completely open to the holding area.


Operation and Performance

Figure 4. Single chamber dry hearth furnace with recuperative Low !Ox burner system.

The 16 tonne holding capacity furnace shown in Figure 4 has a charge hearth capacity of four ingot stacks, each weighing 1,000kg. Burners are Ultra Low NOx with preheated air at 400oC. Metal placed on the charge hearth rapidly absorbs energy from the furnace by radiation and convection. This arrangement is ideal for melting on a semi continuous basis where only a proportion of the metal in the bath is tapped, leaving a substantial liquid metal heel.

Metal remains on the dry hearth until the bath temperature has recovered back to set point. At this time the preheated and semi-molten charge is pushed into the bath and cold metal is placed onto the dry hearth.

Figure 5. Single chamber dry hearth furnace typical operation.

In comparison with the single chamber furnace shown in Figure 1 the melting technique is far superior to charging cold metal into a wet hearth. From Figure 5 it can be established that:

• Average melting rate improve by 25-30%;

• Energy consumption reduced by 10-15%; and

• Furnace utilisation is at maximum.

The principal reason for the improvement is that approximately 50-70% of the energy is being transferred to the ingots before they are submerged into the molten bath. The total energy required to be absorbed through the bath surface has now been reduced by around 50%. Energy consumed during melting is approximately 2800MJ/tonne


Dual Chamber Dry Hearth Melting Furnace

Figure 6. Dual chamber dry hearth melting furnace.

Figure 6 shows a furnace with a separate chamber that is dedicated to melting. Liquid metal can flow from the metal chamber to hold chamber through ports in the dividing wall. This wall may be common to the two chambers or they may be quite separate with a refractory lined port (or ports) joining the two chambers.

The benefits of this design over the single chamber dry hearth furnace are as follows:

• Combustion systems can be specifically designed with a single purpose for each chamber. High power and high velocity for the melting chamber; high turn-down and medium velocity for the hold chamber;

• Increased thermal efficiency;

• High equipment utilisation;

• Increased safety;

• Easier to clean; and

• Reduced energy loss from door opening.

Of the total thermal input rating of the furnace, approximately 80% is directed into the melting chamber and 20% to the hold chamber. It is important therefore to ensure that no significant quantities of solid aluminium can enter the hold chamber otherwise the metal temperature will fall excessively and the recovery period will be quite long.

Heat Transfer in Dual Chamber Furnace

Burners in the melting chamber are located in the roof of the furnace to fire directly downwards, with the charge of ingot stacks directly underneath. Selection of burners in this case is very important to avoid direct flame impingement. High velocity burners are used to promote convective heat transfer from the products of combustion to the charge material. Convective heat transfer rates will increase at approximately 0.5-0.7 the power of the velocity increase due to the reduction of the laminar film around the charge material. For this benefit burners should be selected with a discharge velocity exceeding 130m/s.

An important design feature of the dry hearth furnace is the flue system as shown in Figure 7. Flue exit ports should be located to maximise the contact time between the products of combustion and the charge material.


Figure 7. Flow of products of combustion within dry hearth furnace.

Reducing the temperature of the exiting gases is the most significant factor in the thermal efficiency of any aluminium melting furnace. For a combustion system operating at 10-15% excess air (2-3% O2) and an exhaust temperature of 1100oC, the exhaust gases will contain about 60% of the energy provided by the combustion of fuel. A well designed dry hearth furnace will operate at an exhaust temperature 100-200oC lower than a conventional wet hearth furnace. This will reduce the total energy required by 10-15%.

The hold chamber only receives liquid metal so the amount of energy required to be transferred through the surface of the bath is less than 15% compared to that with a submerged solid charge.

Operation and Performance

Figure 8. Dual chamber dry hearth furnace typical operation.

Operation of the dual chamber dry hearth furnace is illustrated in Figure 8. Four regions can be defined.

0-20 minutes:

Furnace has been fully charged. Thermal input into the melt chamber is at maximum. The furnace temperature increases slowly due to the high mass and surface area of charge available. Flue temperatures are low because the products of combustion transfer a high proportion of their energy to the charge before leaving the furnace.


20-40 minutes: Metal is flowing from the melt chamber to the hold chamber at a relatively low temperature which begins to reduce the temperature of the metal in the hold chamber. The hold chamber thermal input increases according to the new demand.

40-75 minutes: A large proportion of the metal has been melted and the rate of total energy absorption by the remaining metal reduces. The melt chamber combustion system thermal input is reduced so as to not exceed the chamber and flue set temperature limits.

75-100 minutes: As the rate of metal flowing into the hold chamber reduces, the thermal energy demand on the hold chamber also reduces. The furnace is ready for a clean and another charge. Thermal performance of this style of furnace will be similar to a single chamber furnace with a recuperative heat recovery system. In this case the charge material provides the heat exchange. Specific energy consumption during melting is approximately 3,000MJ/tonne.

Conclusions

• The concept of melting aluminium on a dry hearth offers significant benefits in terms of energy efficiency, safety, and furnace utilization.

• Charge material is directly exposed to high velocity products of combustion and furnace chamber radiation throughout the melting cycle High thermal efficiency is possible without elaborate combustion and heat recovery systems by maintaining the solid surface area to mass ratio throughout the melting cycle.

• Liquid metal is constantly drained into an adjacent, holding chamber for superheating and holding. As the solid charge is never submerged under molten metal, the melting rate per unit of area of hearth is much higher than for a wet hearth furnace.

• Solid material can be completely separated from the molten bath, virtually eliminating the risk of a steam explosion caused by wet charge material.

• Molten metal in the holding section is inherently clean and is not exposed to the harsh conditions of melting.

• If the two chambers are independently heated. Metal can generally be tapped at any stage of the melting cycle.

• Design parameters are unique for dry hearth melting and careful consideration must be given to all aspects of the design to ensure that the benefits are realized. Burner style, burner placement, exhaust system, refractory selection and chamber dimensions will all affect the ultimate performance of the furnace.


The Benefits of Forced Circulation for Aluminium Reverberatory Furnaces

Paul Campbell

Metaullics Systems Div., Pyrotek Inc., 31395 Aurora Road, Solon, Ohio, 44139, USA

[email protected]

Keywords: molten metal pumps, mechanical pumps, EMP pumps, aluminium reverberatory furnace, furnace circulation

Abstract

The benefits of circulating molten metal in an aluminum reverberatory furnace are well documented and include higher productivity, reduced fuel consumption, and excellent metallurgical and temperature homogeneity. Forced circulation can be provided by many different means, but two of the most popular are mechanical and electromagnetic pumps. This paper discusses each benefit in detail to allow the reader to better understand how they are achieved by the use of pumps.

Introduction

With the current state of the world wide economy it is more important than ever for companies to operate at the highest possible efficiency. In the last year many companies in our industry have faced lay offs, declining sales, customers going out of business, and reduced or negative profits. As companies struggle to cope with these circumstances they must do everything possible to reduce operating costs, maintain their customer base, and develop new markets. Failure to do this could mean failure to survive. Now, more than ever, companies must look at existing technologies to ensure that they are using the latest, most efficient processes.

Technology for circulating aluminum furnaces has existed for more that 50 years, but in the last 10 years there have been significant improvements in the reliability and effectiveness of the equipment used for this purpose. This paper will not address these new technologies, but rather, will address what they can do for the aluminium casthouse.

One of the most important results from forced circulation is product quality, an important component to customer retention. While the economic benefits of quality can be difficult to quantify, the benefits of improved productivity, reduced cycle time and better thermodynamics can be readily translated into dollar savings.

Forced circulation, which could also be expressed as forced convection, provides the major benefits of reduced metal temperature stratification in the bath, improved melting rates, enhanced metal quality through alloy homogeneity, reduced energy consumption as a result of improved heat transfer, longer refractory life, less dross formation with attendant lower furnace tending requirements, and accelerated alloy dissolution rates. Where possible, this study will attempt to quantify the typical results which can be attained.

Today, the application of centrifugal pumps represents the majority of forced circulation systems used on reverberatory furnaces in the United States. In other parts of the world EMP electromagnetic pumps are also widely used. The results reported here are mostly based on experience from these two types of pumps, however, any method which achieves similar circulation patterns and velocities would produce similar results. Figure 1 illustrates a mechanical pump installation in a typical side well reverberatory furnace. Figure 2 schematically shows an EMP pump installed on a direct charge furnace.


Recent studies have found that even with furnaces that already have circulation, additional benefits can often be obtained by increasing the amount of circulation. This has been confirmed in several cast houses in the last few years.

Figure 1. Mechanical pump in side well furnace.

Figure 2. EMP pump and charge well on a reverberatory furnace.

Reduced Temperature Stratification

In most large facilities, the conventional means for melting and holding aluminum is a gas-fired reverberatory furnace [1]. In this arrangement, radiation from the burners, roof and sidewalls heats the bath and solid charge. Forced convection through stirring or pumping overcomes a number of process limitations. Without forced circulation, the concentration of heat at the bath surface readily produces undesirably high temperatures. The high heat flux generated by direct radiation at the bath surface is typically much greater than conductive and natural convective forces which distribute the heat through the bath. Temperature differentials in a 915mm deep bath with a clean surface and without forced convection can typically reach 50 to 85 degrees C. Without effective heat transfer to the metal, other quality and furnace performance factors may be less than optimum.

Bath temperature is a key factor which impacts overall product quality. Forced circulation improves bath product quality by insuring the bath temperature will be consistent throughout the furnace. Typically reductions in temperature differentials for a 915mm bath are from 50 to 85oC in an uncirculated furnace to 3 to 7oC for stirred furnaces. Circulation ensures that the temperature of the metal exiting the furnace is virtually the same as that indicated by the furnace thermocouple. Production parameters can be set and more easily controlled when the metal delivered is at a constant temperature. Additionally, there will be less superheat of the metal required because there will be less variance in the furnace temperature.

Increased Melt Rate

Constant circulation of the furnace will increase the melt rate of the furnace as well as improve the thermal profile. The high velocity stream from a circulation pump will increase the melt rate of large submerged solids in the furnace. Insulating boundary layers that are formed around large solids as they melt are swept away by the hot metal stream. Such a device equipped with gas injection can provide bath fluxing and degassing during the melting cycle, eliminating or greatly reducing the time required at the end of the cycle for metal treatment. The same holds true for


temperature control. Continuous circulation provides excellent temperature control so temperature adjustments are usually not required prior to transfer.

Cunard, using a direct charge furnace in a billet casting process, reported in a study over a five month time period that furnace cycle times increased between 8.1 and 20.6% when their pump was not in operation [2]. He estimated that using a pump produced a conservative value of 10% cycle time savings. Charging time reductions of 25% to 50% have often been observed in side well furnaces when a pump is used. Savings may be obtained by increasing output with existing equipment or, in some cases, by maintaining production levels with fewer furnaces.

A major U.S. manufacturer of foundry furnaces estimates that the melting capacity of a given furnace will increase by 10 to 15% when forced circulation is added to one of their uncirculated furnaces.

Alloy Homogeneity

Results from an Alcan developed pump installed in a top charge 70-ton round melter were documented by Thibault [3]. The results were obtained during the preparation of an AA4000 series alloy with a high silicon content. The silicon specification for this alloy is 7.15% to 7.45%. The silicon content was monitored in the trough during the cast and compared with samples taken from the furnace at the end of the batch casting process. A significant difference was observed between the two samples for several batches when the pump was not used.

For eight runs without the pump, the average silicon composition in the furnace at the end of the cast was 7.28%, with values ranging from roughly 7.22 to 7.3%. In contrast, the average for trough samples during the cast was 7.22%, but the individual values ranged from close to 7.1 to almost 7.4%. Two ingot batches were actually outside of the specification. In five runs where stirring was used, no significant difference between furnace and trough samples was observed and all ingots were within specification.

In foundry applications similar results can be expected when the furnace is circulated. Without forced circulation, alloy additions must be stirred in mechanically to ensure uniformity. Since mechanical stirring is done over a relatively short period of time, it is difficult to ensure that the alloy is mixed uniformly throughout the furnace. Some additions will also begin to sink to the bottom or float to the surface if not continually stirred.

When a stirring device such as a pump is used, alloy elements are constantly mixed with the rest of the metal in the furnace. In a typical foundry application, the volume of metal in the furnace passes through the pump every 10 to 15 minutes. Since the pump is operated continuously throughout the process cycle, samples taken from any point in the furnace will show uniform chemistry.

One other consideration with foundry alloys is the formation of sludge. This is an iron-chromium-manganese compound that precipitates out of the metal if the temperature drops below a critical point. This is discussed in detail by Jorstadt [4]. With a temperature differential of 50 to 85oC between the top and bottom of the bath, it is easy to see how the metal at the bottom of the bath can cool below the critical point if the furnace temperature is monitored and controlled by a thermocouple located near the top of the bath. As previously discussed, temperature is uniform throughout the bath when the furnace is constantly stirred. This allows for accurate temperature control and the avoidance of sludge formation.

Reduced Energy Consumption

The rate in which energy is transferred from the burners and refractories to the metal is referred to as the heat flux (q). Heat flux in a reverberatory furnace is determined by the following equation [5].


q = Ah[(t1)4 - (t2)4]

Where: A = surface area of the bath inside the furnace h = heat transfer coefficient of the area inside the furnace above the bath t1 = temperature of the heat source t2 = temperature of the bath surface

From this equation it can be seen that the rate in which energy is transferred into the metal is dependent on the difference between the surface temperature of the metal and the temperature produced by the combustion system, to the fourth power.

Without circulation, energy moves away from the surface of the bath by conduction and natural convection. Heat flow in these two modes is directly proportional to the difference in temperature between the surface and bath. Since the heat flow is much higher by radiation, the bath surface can become superheated if nothing is done to help transfer the heat away from the surface and into the bath.

Constantly stirring the furnace allows heat to be removed from the surface to the rest of the bath by forced convection. As heat flows more efficiently away from the surface of the bath, the bath surface temperature is reduced. As the surface temperature drops, the temperature differential between the heat source and the metal increases which translates into much higher heat transfer rates. This subject is covered in more detail by Campbell [6].

Since stirring occurs during the melting process and with the doors closed, energy that would have been lost out the door when the bath is stirred manually is saved. Additional savings are possible from the elimination of through the door fluxing operations if the circulation pump is equipped with gas injection capability.

Actual furnace efficiency improvements vary from operation to operation. Experience has shown that a minimum of a 15% improvement can be expected. A 25 to 30% improvement is possible, depending on furnace and operation specific variables. In the study done at VAW in Ellenville, New York2 savings ranged from 10.3 to 20.2% over a five-month period.

Longer Refractory Life

By reducing the metal surface temperature, more heat is transferred to the metal and less is absorbed by the refractories. Roof and stack temperatures are usually reduced by 80 to 115 degrees C when a furnace is circulated. These lower operating temperatures translate into longer refractory life. Lower temperatures will also result in less corundum formation at the metal line. Less build-up means less need for mechanical cleaning and therefore less opportunity for damage from cleaning tools. While specific information on savings from improved refractory life is not widely shared, dramatic improvements have been reported by some companies utilizing circulation pumps. Thornton [7] reported in 2007 that by increasing the circulation on one of his melting furnaces with a high capacity J-50/LOTUSS system he increased his refractory life on that furnace by at least six months.

Decreased Dross Formation

Another aspect which should not be ignored is melt loss. As shown in Figure 3, oxidation increases with temperature. At temperatures above 775°C the rate at which aluminum oxidizes is much higher. Since aluminum oxide is thermally insulative, the differential between surface and bath temperatures increases with the growth of the oxide layer. Significant melt loss is incurred due to ever increasing dross formation under these conditions. Reducing the bath surface temperature


can significantly reduce the rate of dross formation. Minimizing surface agitation with submerged circulation removes another source of dross formation.

5

10

15

20

25

30

650 675 700 725 750 775 800 825

Wt. Gain, g/cm2hr

Temperature, oC

Figure 3. Oxidation of aluminum versus temperature.

Many foundries do not pay enough attention to melt loss that results from dross formation. Every kilogram of aluminum that is lost to oxidation in the form of dross is lost to the manufacturing process and must be replaced. While not widely documented, typical dross reduction in the hearth when using a centrifugal pump for forced circulation is about 0.5%. Using a conservative value of 0.25%, a foundry that melts and casts 500 tons of aluminum a month would save the replacement cost for 1,250kg of metal every month.

Easier Alloy Adjustments

The benefits associated with alloy additions can manifest themselves in three areas: alloy addition dissolution, treatment during melting, and improved melt quality before downstream treatments. By continually providing forced circulation, alloy additions and adjustments are easier to control since the bath will remain homogeneous. Silicon and other alloy additions go into solution much faster when exposed to the discharge stream of a pump.

In secondary smelting operations producing low magnesium foundry alloys a gas injection pump is an extremely effective tool for reducing magnesium in scrap materials high in magnesium when chlorine can be used [8]. The dispersion and subsequent reaction of chlorine gas as it exits the pump in the molten metal stream creates a very effective kinetic environment for removal of magnesium from the melt. Magnesium combines with chlorine in a two step chemical reaction to form magnesium chloride (MgCl2). A 100% efficient reaction would require 2.95kg of chlorine for each kg of magnesium removed. In actual practice, secondary smelters routinely operate at ratios of 3:1 or better.

Results from tests conducted at Golden Aluminum, Fort Lupton, Colorado in 1995 show that with a mixture of 95% argon and 5% chlorine, a gas injection pump can significantly reduce sodium and calcium levels without affecting magnesium content [9]. Figure 4 shows the results that were obtained in a 27 MT melting furnace. Sodium levels dropped quickly followed by the calcium. These tests also showed an improvement in PoDFA results and a reduction in hydrogen levels while the gas injection pump was in operation. By removing inclusions and hydrogen in the melting furnace the load is reduced on downstream operations such as degassers and filters.


Alkali Metal Reduction with 95/5 % Mix of Ar/Cl2

1.7

1.8

1.9

2

2.1

2.2

2.3

2.4

2.5

2.6

2.7

0 60 120 180 240 300 360 420

Time (minutes)

Magnesium Content (%)

-0.100.10.20.30.40.50.60.70.80.911.11.21.31.41.5

Ca and Na Content

(% x 1000)

Mg

Ca

Na

Figure 4. Alkali metal reduction with a gas injection pump.

Benefit of Increased Flow

Historically it was felt that proper circulation required the volume of metal in the furnace pass through the pump 3 to 4 times per hour. In 2001, computer modelling of a furnace showed that increasing the circulation rate to 7 turns or more per hour had a significant positive impact on the melting rate of the furnace. Subsequent tests on actual furnaces showed melting rate increases of 7 to 17% when circulation rates were increased from traditional levels to more than ten turns per hour.10 Foundry alloy producers in the United States often use furnaces with a holding capacity of 100 tons or more. In order to circulate one of these furnaces ten times per hour, the pump will need a capacity of almost 17 tons per minute. The J-50 gas injection pump from the Metaullics Systems Division of Pyrotek shown in Figure 5 has a pumping capacity of 17.5 tons per minute at 350 rpm.

Figure 5. J-50GI gas injection pump.

In 2008, Laminazione Sottile, a slab producer in Caserta, Italy upgraded their circulation pumps. They have three 70 ton side well melting furnaces. The previous pumps operated at a rate of 9 tons per minute which turned the furnace over 7.7 times per hour. The new pump is operated at a rate of 13 tons per minute which increases the number of turns to more than 11. The result has been a 20 % increase in melt rate.


Similar results have been reported by secondary smelters in Germany and the United States. By changing to a larger capacity pump they have seen improvements in furnace output, metal recovery, homogeneity, and lower energy consumption.

Conclusions

When a full picture is developed examining the relative benefits and drawbacks of imparting forced circulation in a furnace, several conclusions can be reached. Any form of continuous forced circulation will improve the performance of the furnace.

While some production parameters may be difficult to fully quantify, clearly forced circulation improves many key aspects which affect the total process. Forced circulation has been demonstrated to reduce energy consumption by 15% to 25%, extend refractory life at least 25%, improve recovery rates by a minimum of 0.25%, and reduce melt cycle times by 10% to 50%.

Quality improvements are difficult to quantify, but the benefit to having a uniform temperature and chemistry during casting is real and will certainly result in better control of the process and end product.

References

[1] Henderson, Chandler, and Brown (1996) A New Electromagnetic Circulation Pump for Aluminum Reverberatory Furnaces, Light Metals, TMS, 869-876.

[2] Cunard, T. (1987) Application of Graphite Pump to Direct Charged Aluminum Melting Furnace, Aluminum Industry Energy Conservation Workshop X, Aluminum Association, October 20-21, 159-185.

[3] Thibault, M.A., Tremblay, F. and Pomerleau, J.C. (1991) Molten Metal Stirring: The Alcan Jet Stirrer, Light Metals, TMS, 1005-1011.

[4] Jorstad, J. (1986) Understanding Sludge, Die Casting Engineer, NADCA.

[5] Mangalick, M.C. (1973) Reduced Fuel Consumption Through Mechanized Molten Metal Circulation, The Society of Die Casting Engineers, Vol. 17, No. 5, Pgs. 22-31.

[6] Campbell, P.S. (1990) Use of Molten Metal Pumps to Reduce Energy Consumption in Reverberatory Furnaces, Aluminum Industry Energy Conservation Workshop XI, Aluminum Association, November 1-2, 537-546.

[7] Thornton, Hammond, van Linden, Campbell, and Vild (2007) Improved UBC Melting Through Advanced Processing, Presented at 2007 Annual Meeting of TMS, Orlando.

[8] Neff, D.V. (1985) Use of Gas Injection Pumps in Secondary Aluminum Metal Refining, Recycle and Secondary Recovery of Metals, Metallurgical Society of AIME, December 1-4, 73-95.

[9] Hopkins, Beasley, Henderson, and Campbell (1995) Quantification of Molten Metal Quality Improvements Using an L-Series Gas Injection Pump, 3rd International Symposium on Recycling of Metals and Engineered Materials, TMS, November 12-16.

[10] Bright, Chandler, and Henderson (2007) Advances in Molten Metal Pump Technology Expand the Capability of Aluminum Reverberatory Furnace Production Rates, Light Metals, TMS, 603-607.


Launder System for Aluminium Casting

María Victoria Canullo1,a, Fernando Daroqui1, Julio Ottaviani1 , Mauricio Martín1 and Rodolfo Acuña Laje2

1 Aluar Aluminio Argentino SAIC, C.C. 52, Puerto Madryn, Chubut, 9120, Argentina

2 Infa S.A., Ruta 4 s/n, Puerto Madryn, Chubut, 9120, Argentina a [email protected]

Keywords: launder, metal transfer, refractory, metal level control

Abstract

In 2000 Aluar installed its first in-house designed launder system for liquid metal transfer from the

furnaces to their continuous casting machines. Since then, this type of launder system has been

successfully installed in 9 casting lines, both batch and continuous, with flow rates ranging from 2

to 44 tonne/h, for conventional open mould casting, Wagstaff vertical DC billet casting , and

Properzi rod casting.

The Aluar system is designed to ensure an effective metal temperature control as well as a constant

steady “under skin” metal flow, so avoiding turbulence and oxide generation.

The highly adaptable system consists of a metal structure, which houses the pre-moulded high

durability refractory launders of standardized shapes and dimensions required for the geometry of

the desired lay-out: Y shaped, V-shaped, straight, elbowed launders, etc. In this way, the launders

enable the connection of the furnaces to the casting machine optimizing the new or existing lay-out,

and can be fully integrated to the degasser unit and the filter box without dross generation. The pre-

moulded refractory pieces can be easily and quickly removed when their life time is over, thus

reducing maintenance costs.

Each module is insulated by a thick layer of material to reduce temperature loss (approx. 1ºC per m

in continuous casting).

Another important feature of the system is the thermally insulated heated lids that can be

commanded by a PLC to ensure excellent metal temperature control.

Efficient launder level control systems ensure an effective achievement of consistent process

control parameters. Control flow devices with laser commanded pins allow handling metal height

differences up to a range of 200mm, without turbulence.

This paper describes the main features of the launder design and presents some case studies in

continuous and batch casting lines to show their performance in aluminium alloyed and unalloyed

products. Maintenance performance indicators showing typical refractory life time are also

included.

INFA, an Engineering company of the Aluar group, with a vast experience in engineering works

for the aluminium industry, was responsible for the installation and commissioning of these systems

and is planning to market this product worldwide in the near future.

Introduction

Many years ago launders were simply a way of conveying metal between furnaces and casting

machines. Nowadays in order to produce a wider range of alloys and different product formats,

launders need to do more than this. Effective control of temperature and metal level in the casting

machine is of paramount importance. Strict control of metal cleanliness is also vitally important.

Low inclusion content is necessary to comply with the metal quality specification of the final


products. Launders must be designed to link the casting equipment and prevent turbulent flow in the

metal transport in order to minimise dross formation.

The Aluar launder system [1,2,3] was designed to address all these issues. They have now been

used for over 8 years very successfully, in nine casting machines, both continuous and batch: two

HDC Hertwich horizontal continuous machines for T ingots and small (prismatic) ingot (6 tonne/h);

three continuous rod casting machines (2.7 tonne/hr, 5.25 tonne/hr and 2 tonne/hr), two open mould

casting machines (6 and 14 tonne/h, respectively), and two Wagstaff vertical casting machines for

billets (27 and 44 tonne/h, respectively). The main characteristics of the system are described in the

following sections.

Basic Launder System

Launder Modules

The launder system mainly consists of a set of modules with standardised shape and dimensions

with tight tolerances, to meet the geometry of the desired lay-out.

Standard modules: „Straight‟ launder, 2.40m long; „Elbowed‟ 120º and 150º launders; „Y‟ shaped

launder; ‟T‟ shaped launders; „V‟ shaped launders, designed to drain the launder metal at the end of

the cast and can be lifted by a pneumatic system.

Special modules, designed for connection of different equipment: furnace spout, in/out spout of

filter box, in/out spout of degassing system, connection to casting machine, etc.

Structure of Modules and Lay Outs

The modules consist of a metal structure which houses a pre-moulded high durability refractory

launder, insulated by a thick layer of low density materials to reduce temperature loss, typically to

1ºC per meter in continuous casting.

The joints between modules take into account thermal expansion of metallic pieces. In this way

metal leaks between ceramics modules are avoided. Thermal expansions are minimised because the

metal structure is below 60°C in stationary conditions thanks to insulation properties of the

launders. Additionally, the improved insulation also benefits the work environment for the operator.

The launder system is used to link furnaces, casting machines, degasser and filter box equipment

layouts previously installed. Figure 1 shows one of the layouts currently in operation in the Aluar

casthouses.

Figure 1. Aluar launder systems applied to an HDC machine.

Furnace 1 Furnace 2

Casting machine


Heated Lids

The straight modules have thermally insulated heated lids that can be automatically commanded by

a PLC to take one of the following three conditions: a) lid closed with heating, b) lid closed without

heating and c) open lid. Figure 2 shows the straight module cross section.

The main purpose of upper lid system is to preheat the installation before casting as well as holding

and controlling metal temperature during the cast.

The temperature control of the launder system is based on two criteria. Firstly, the possibility to

control the lids one by one, with or without power supply, and secondly the insulation properties of

the launder leading to a temperature loss of approximately 1 or 2°C per meter, depending on the

metal flow.

During the setups, when there is no metal, the heating of the lids is controlled by thermocouples

located in the refractory pieces. In this way the launder can be preheated without using gas burners,

to a particular uniform temperature without hot zones which can degrade the refractory pieces and

reduce durability.

Figure 2. Straight module cross section.

During casting, the system detects that there is metal in the launders, and heating of the lids is

controlled by two thermocouples submerged in the molten metal. One thermocouple is close to the

furnaces (at the entrance of the launder system) and the other is close to the casting machine (at the

exit of the launder system).

According to the temperature difference measured by these thermocouples and the target value set

in the casting machine, the control system actuates automatically on each lid individually, setting

one of the three conditions already mentioned.

Filter Boxes

Filter boxes for ceramic foam filters are available in a range of sizes, from 15 x 15‟‟ to 23 x 23‟‟

depending on the metal flow required by the casting machine and metal cleanliness desired

(inclusion content).

Depending on the required filtering area, single boxes and double boxes are used. Twin boxes

systems (two boxes in parallel) are usually installed for continuous casting. For maximum filtration

efficiency two boxes are used in series, for instance one # 30 ppi filter in the first place followed by

a second # 50 ppi filter downstream.


Pre- heating of filters prior to the start of each cast is carried out with a high air flow gas burner.

Hot convection air ensures a uniform and even pre- heating through the total area of the ceramic

foam filter, avoiding any „cold spot‟ or channeling of the filter that would cause premature

blocking.

Control of Metal Flow and Level

Control of metal feeding from furnace to casting machine is carried out mainly by two devices:

launder level control and flow control devices.

Launder Level Control for Tilting Furnace

A laser sensor takes a measurement of the launder level metal near the furnaces. A signal from a

PLC actuates a proportional valve in the hydraulic system, which makes the furnace tilt upon metal

demand. The control system varies the tilting velocity in line with the geometry of the furnace,

maintaining a constant metal level in the launders. The launder metal level control has a precision

of approximately ± 1mm during casting and it is possible to obtain height between 100 and 230mm

from the furnace pivoting point.

Flow Control

A common way to control the flow from a launder at a certain height to another launder in a lower

position is by using a down spout with an internal stopper rod (pin).

Control flow devices with laser commanded pins allow handling metal height differences up to a

range of 200mm, without turbulence.

In high accuracy applications like horizontal continuous casting processes, it is recommended to

have two stages of level regulation: a first loop to control the flow from the furnace and a second

one to set the final level close to the casting machine (metal box). A typical level control loop

consists of a level sensor, PLC controller, actuator and spout/stopper device for flow control.

Results

In this section results with the Aluar launders system installed in one of the Hertwich horizontal

casting machines (HDC) are shown. The analysis was made for the temperature control, the metal

level control and maintenance performance.

Temperature Control

Figure 3 shows metal temperature versus time in HDC. The upper curve shows the metal

temperature measured near the furnaces (just before the entrance of the launder system) and the

lower curve represents the metal temperature measured in the tundish (the exit of the launder

system), both in the same time interval. The black centre lines show the averages and standard

deviations 3 for both temperatures.

The average furnace temperature is 705ºC and the standard deviation (3) is 18ºC while for the

same metal but after the launder system the average temperature is 660ºC 6ºC along the five day

period shown in Figure 3. In other words, the system can dampen temperature variations of

approximately 35ºC reducing them to 10ºC. The exit metal temperature is kept within acceptable

values in spite of the large oscillations of the entrance metal temperature, particularly when the

furnaces tilt starts or ends. The elimination of thermal spikes at furnaces change over is of

paramount importance to help minimize premature stops during casting campaigns.


Figure 3. Temperature near the furnace (upper curve) and at the exit of the

launder system (lower curve).

Metal Level Control

In Figure 4 the upper curve shows the launder metal level control and the lower curve shows the

casting machine metal level control during 8 hours, when two furnace changes take place at HDC.

The left y- axis represents launder metal height referenced to the furnace pivoting point and the

right y- axis represents casting machine metal height referenced to the same point. As it can be

seen, during the cast the launder metal control is very accurate (3 = 1mm), but furnace change

overs introduce perturbations in the range of 10mm. However, the casting machine metal level

control does not show variations during the furnaces changes, reducing the height deviations

produced in the launders at the furnaces exit.

The Aluar launder system allowed metal level and the temperature parameters to remain stable. As

a result of the installation of this technology and improvements in the lubrication and in the moulds,

an increase in the casting time was achieved, from an average casting time for HDC of 2 days with

the old launder system to 7 days average and 15 days maximum casting time with the Aluar launder

system.

Figure 4. Launder and casting machine metal level during two furnace changes in HDC.

Maintenance Performance

Each module can be easily and quickly removed and replaced by the corresponding spare part,

when the pre-moulded refractory is damaged or its life is over, reducing maintenance time and cost.

710

730

13-Feb 14-Feb 15-Feb 16-Feb 17-Feb Date

T ºC

670

650

690

00:00 02:24 04:48

150

140

160

170

180

Time

Lau

nde

rs m

eta

l le

ve

l m

m

-50

-30

-10

10

30

50

70

90

Castin

g m

achin

e m

eta

l le

vel m

m

Furnaces changes


The re-lining of each module is carried out in the refractory shop and requires no more than

standard trained operators.

Operating times for replacing modules and refractory re-lining are carefully recorded, and are

available upon request. With this data the customer can evaluate in advance maintenance and

operating costs of the launder system. Typical times for refractory relining are:

Straight launder: 2 operators x 4 hr; other launders (elbow, „Y‟, „T‟, „V‟, etc): 2 operators x 3 hr, no

drying is necessary.

Typical times for refractory relining for filter boxes are: Single 15 x 15‟‟ filter box: 2 operators x 8

hr +1 hr forklift; Single 23 x 23‟‟ filter box: 2 operators x 10 hr + 1 hr forklift; Double 23 x 23‟‟

filter box: 2 operators x 16 hr + 2 hr forklift.

Relined filter boxes require 3 days drying.

At the moment there are 25 filter boxes installed in 13 casting lines at Aluar Casthouse supplied by

INFA.

The following results are based on the data from the different casting machine currently in

operation, after and before the Aluar launder system installation, showing substantial improvement

in ceramic piece lifetimes for casting similar products.

In the old launder system, it was necessary to repair the launder system almost every week,

equivalent to less than 1000 tonne of aluminium produced.

Since the Aluar system was implemented, this rate has dramatically decreased to one repair per

month, corresponding to one repair for every 5000 tonne for a billet casting line, and every 3000

tonne of aluminiun produced for continuous rod casting line.

Conclusions

The Aluar launder system is a versatile piece of equipment, which can be installed in different

casting technology machines, either in new lines or pre-existent ones. Due to its modular nature, it

can be used to accommodate different configurations and layouts. Modules are very easy to change,

resulting in lower maintenance time and costs.

From the point of view of quality and process control, the efficient temperature and metal level

control ensure that these critical parameters remain stable, even during furnace changes. As a result

an increase in productivity is achieved.

All these characteristics result in a high quality final product, increasing productivity and

decreasing maintenance costs. After 8 years of successful performance, the Aluar launder system

has been installed in most of the production lines in our three casthouses. INFA SA., an

Engineering company of the Aluar group, with a vast experience in engineering works for the

aluminium industry, has been responsible for the installation and commissioning of these systems

and is planning to market this product worldwide in the near future. This technology is being

patented.

Acknowledgements

The authors wish to acknowledge Aluar Aluminio Argentino SAIC and INFA S.A. for authorizing

this paper and the kind assistance of J.P.Mercado to obtain process data.


References

[1] Daroqui, F. Dispositivo para el control de Desniveles de utilización en colada de metales.

Patent # P040101255 Argentina.

[2] Daroqui, F. Disposición para transferencia de metal líquido, canal empleado en dicha

disposición y método de control de transferencia empleado en dicha disposición. Patent #

P040101254 Argentina.

[3] Daroqui, F. & Bellomio, P. (2004) Launder system for aluminium continuous casting, in

Proceedings of Drache Seminar.


CHAPTER 5

Melt Quality & Treatment

The Impact of Rising Ni and V Impurity Levels in Smelter Grade Aluminium and Potential Control Strategies

John Grandfield1,a and John A. Taylor2

1 Grandfield Technology Pty Ltd, Brunswick, Victoria, 3055 Australia

2 CAST CRC, School of Engineering, The University of Queensland, Brisbane, QLD 4072, Australia

a [email protected]

Keywords: nickel, vanadium, boron treatment, trace elements

Abstract

The technology for controlling smelter metal impurities post reduction has steadily improved. For example, control of sodium has seen the reduction and, in some plants, the elimination of chlorine gas from the casthouse. However, changes in the purity of cell feed materials such as anodes are giving rise to new challenges in impurity control; vanadium and nickel levels are an emerging problem. This paper briefly reviews the important impurities and their effects on downstream casting, forming and final application properties. Particular emphasis is given to nickel and vanadium. Strategies for controlling these impurities are also discussed and areas where new technology is needed are also highlighted. In some cases it is not known what the tolerable limits of impurities are. There are a plethora of metal refining techniques used in the extraction of other metals which could be investigated for the control of impurities in smelter grade aluminium.

Introduction

The various impurity species present and their concentrations in primary aluminium produced from the Hall-Héroult electrolytic process depend on cell operation, the composition of the feed materials and the cell construction materials that the molten metal comes in contact with. Concentration ranges vary considerably depending on feedstock sources and modes of cell operation [1], typical examples being shown in Table 1.

Impurities can take the form of dissolved elements, solid particles or liquid phases. Particulates of borides, carbides, nitrides, oxides and fluorides can be present. Solid particles such as oxides can cause defects and die wear during extrusion, pin holes in foil and sometimes failure in service of castings. Carbon in solution normally precipitates as aluminium carbides when the liquid metal cools [2]. Chlorides can be present downstream where either chlorine gas or magnesium/potassium chloride additions are used for sodium removal and melt cleaning, respectively. Some impurities such as phosphorous and sulphur can actually be harmful to good cell operation [1], while other elements can cause problems downstream with casting, heat treatment, rolling, extrusion and in-service corrosion.

The effect of impurities is briefly reviewed in this paper; however, the main focus is on rising levels of V and Ni due to rising levels in the petroleum coke feedstock. Both nickel and vanadium come predominantly from the anodes (i.e. from the pet coke) but in the case of vanadium, a small contribution is also made from the alumina. Studies on the partitioning of V between the metal and the duct in the cell give a wide range of values [3-5] with 15% to almost 100% of the total V reporting to the metal in the cell. Nickel tends to partition in a similar way to vanadium.

As demand from the industry has outstripped the supply of low V/Ni green coke sources, the average levels of these metal impurities in anodes and in smelter metal is creeping up, see Figure 1 [8]. With anode consumption being around 0.45kg per kg of metal, and assuming a 50% partition


we would expect vanadium levels to exceed 150ppm soon. Depending on cell practice some plants may already be exceeding certain specifications.

Table 1. Typical concentration ranges of impurity elements in primary aluminium metal [2, 6-7].

Element Concentration range (ppm) Iron 400 - 3000

Silicon 200 - 1500 Copper 5 - 100

Magnesium 5 - 60 Zinc 10 - 200 Nickel 1 - 80 Titanium 10 - 100 Vanadium 10 - 200 Sodium < 1 - 500 Boron < 2 - 60 (as borides) Calcium < 1 - 50 Lithium 1 - 10 Oxygen 1 - 100 (as oxides) Nitrogen 1 - 60 (as nitrides) Carbon < 1 - 100 (as carbides) Barium < 1 - 10 Bismuth 0 - 10 Chromium 2 - 50 Lanthanum < 1 - 10

Lead 1 - 50 Manganese 5 - 50 Phosphorus 1 - 30 Sulphur < 1 - 20 Tin < 1 - 30

Zirconium 10 - 40

Impurities and Their Effects

Iron and silicon are normally the highest concentration impurities in potline aluminium. Sources and control of iron have been extensively studied elsewhere and are not discussed here; see for example [9] for an extensive coverage and [10, 11] on the important effects of iron in Al-Si foundry alloy castings.

Silicon along with zinc and titanium are generally not considered to be a problem downstream and are usually beneficial (for example, titanium assists grain refinement) although Grjotheim et al [2] do report that Zn above 100ppm is detrimental. In many alloys, Si and Zn are deliberately added as major elemental additions.

Li, Na, and Ca are known to cause problems with hot cracking during casting and hot forming operations. Some high magnesium alloys have specifications of < 1ppm Na but the industry has good control over these elements. Use of Cl2, AlF3 or MgCl2 as reactants to remove alkali earth metals is well established in casthouse technology, fulfilling all processing and operational requirements and will not be discussed here. Na and Ca levels in coke are also rising [12] and while


little of the Ca reports to the metal, a significant amount of the Na does. This will put an extra load on sodium removal technology in the casthouse.

Figure 1. Increase in the vanadium content of coke over time; from Vogt et al, 2004 [8].

Gallium originates predominantly from the cell alumina. It may affect corrosion resistance adversely. It also raises the recrystallisation temperature of aluminium [13] and changes its anodisation response [14]. Gallium also presents the possibility of localized formation of a very low melting point (~27 ºC) grain boundary eutectic phase [13].

Although sulphur causes problems with cell operation and emissions it only reports to the cell metal at very low proportions of total input sulphur and the resulting concentrations are generally < 1ppm. Downstream issues are generally insignificant although in alloys containing iron, manganese or nickel, higher sulphur levels can cause accelerated corrosion [13].

Phosphorus is known to cause problems in hypoeutectic Al-Si foundry alloys where it can act as a poisoner to Sr and other modifier additions. Critical concentrations are typically around 8ppm but if excess modifiers are present, the P effect is ameliorated.

Chromium and zirconium can also act as poisoners of titanium diboride-based grain refiners; however they are generally present at very small levels in cell metal.

A general issue with all impurities is the generic maximum “wt% - other elements” (each and total) used in many alloy specifications. For many wrought alloys the total specification is 0.15%, while the each specification is 0.05%. As Fe, Si, V, Ni and other element concentrations rise, the generic specification levels may be exceeded.

The use of alumina that has been used in dry scrubbing processes as cell feedstock further increases impurity levels of many elements in the metal [2] since elements can only leave the cell via the metal or the off gas.

Some alloys have specifications on Ni and V of 500ppm maximum each. Most smelters operate with much lower internal specifications, e.g. 100-150ppm Ni because of corrosion concerns, while electrical conductivity grade aluminium alloys may have a total V and Ti specifications as low as 100ppm.


Effect of Vanadium

The price of vanadium has generally precluded its use as an alloying element. Vanadium is known to have a small beneficial grain refining effect [13] although in the presence of the conventional titanium boride-based grain refiners it seems to have no effect [15]. There seems to be little beneficial effect on mechanical properties of aluminium from vanadium additions, although the modulus of elasticity is also increased by 2250MN/m2 per 1% V addition which could be a useful effect. Vanadium is known to raise the recrystallisation temperature and this is a potential problem during subsequent annealing heat treatment during rolling. The solubility of vanadium (V) is quite high so V-containing intermetallic particles are unlikely to form at concentrations below 1000ppm [16].

The main detrimental effect of vanadium is the reduction in electrical conductivity that occurs. The electrical resistivity increases 0.4-0.5x10-8Ω.m for every 0.1% V added. For metal that is destined for production of electrical grade alloys this necessitates vanadium removal using boron treatment (discussed below). The electrical conductivity of the liquid metal also falls as V levels rise [17]. There is presumably a corresponding drop in the thermal conductivity as vanadium level rises (the two properties being correlated) which could be an issue with alloys used in thermal applications such as radiators and may also begin to change the solidification behaviour of alloys during casting.

A study of vanadium additions at ~0.1% level was carried out with AlMgSi alloys [18]. As with Mn, V was found to promote formation of cubic α−AlFeSi intermetallic phase during homogenization, a useful effect. The grain size after heat treatment was smaller, formability slightly decreased and strength increased around 10% in the AlMgSi material with V addition. Vanadium was also found to affect the colour of the anodized sheet; potentially a useful effect but one that could give rise to unwanted variation in product.

Effect of Nickel

Nickel can be used as an alloying addition but only for relatively specialist alloys; for example, to impart strength, creep resistance and hardness at high temperature in some silicon and copper-containing casting alloys or to impart resistance to corrosion from high pressure steam [13, 19]. Young’s modulus is increased by nickel addition. Again this may be a beneficial effect offering new design opportunities. Increasing Ni content results in a small reduction in the thermal expansion coefficient [14].

Nickel presents a particular concern with respect to its effects on corrosion rates [13]. High nickel content also precludes use of the aluminium as an alloying addition to magnesium-aluminium based alloys where nickel causes accelerated corrosion. Nickel can also have some effect on “fir tree” grain structure in 1000 series alloys and give unwanted variation in anodizing response.

Nickel and V may of course have effects on the behaviour of other elements that negates the positive effects of these elements in alloy design. For example, nickel reduces the solid solubility of Mn which may then have flow-on effects in alloys such as the Mn-containing 6000 series type.

Nickel and Vanadium Control Strategies

The industry has already responded to higher V/Ni grades of coke by the blending of high metal content coke with low metal content coke [20]. Another option is to source lower metal content carbonaceous material to make coke. Possibilities include coal [21-24] or trees [25]. Using trees to make bio-anodes would offer lower ash, metal and sulphur contents than coke from coal and in addition be a carbon-neutral source of anode carbon. Problems include the high water content of the starting material, however the possibility is under active examination by CSIRO [26]. Inert anodes


would of course be expected to eliminate the pickup of Ni and V, unless they specifically contain Ni or V; as some of the candidate electrode materials do.

Another approach might be to split the process flow and run some cells (or even smelters within large companies) with solely low V/Ni to make those metal products that require lower V and Ni levels and then charge a premium for these products. Meanwhile, the high V/Ni coke would be used in alloys where these impurities are not a problem. Another possibility is to deliberately make a high V/Ni product that is then more amenable to treatment to remove V and Ni. Some of this liquid could go to nickel-containing aluminium alloys or aluminium-containing Ni-based super alloys.

Removal of Impurities from Coke

It may also be possible to treat the raw materials to remove the impurities before they are used in the cell. There is considerable financial incentive as both vanadium and nickel are valuable metals and there has already been a lot of work on extracting V from coke and oil sands [27-35]. However, most of these processes destroy the carbon values and do not result in a useable low V/Ni coke product. A number of patented acid leaching processes [36] have been investigated for V removal from carbonaceous materials. Some of these processes burn the coke and then leach the ash remaining. Others inject the coke into an iron bath resulting in nickel pickup in the iron and a V-rich slag layer [37]. Room temperature acid leaching of coke can recover 30-40% of the metal value [28]. A partial review is given by Armas [38]. The treatment of coke in a fluidized bed variant of the Flexicoke process with gaseous HCl to recover Ni and V as chlorides and produce a low V/Ni coke [38] also seems promising. Of course, it may actually make more sense to work with a high V-content Flexicoke with these processes in order to make them more financially viable. Studies on the coking process and the role of V and Ni in this may point the way to a means of producing lower residual levels [15, 39-41].

Melt Refinement Methods

The other approach is to refine the potline metal in the cast house. The standard process for producing high purity metal is the 3-layer electrolysis process which of course could be used to remove impurities but the high cost would preclude this option for run-of-the-mill products.

A standard melt refining method is to introduce another element or compound to create a new phase that preferentially contains the impurity element (solid, liquid or gas) that can then be separated (decanted, skimmed, filtered, settled, etc.) from the purified melt. A classic example is the Kroll-Betterton process to remove bismuth from lead by adding magnesium and calcium to form calcium-magnesium-bismuth compounds that float out of the lead and are skimmed off. In aluminium, another example is the addition of AlF3 in crucible treatment systems (e.g. TAC & RAM) [42, 43] to remove sodium by forming NaF3 that is skimmed off.

For electrical conductivity grade alloys, V, Cr and Ti removal is achieved using an addition of Al-B master alloy (containing aluminium borides) to preferentially form borides of each of these impurity elements. The heavy borides formed then settle to the bottom of the furnace rendering the alloy EC-grade [15, 44]. A typical treatment time is 40 minutes to allow the borides to settle out. If scheduled carefully, the treatment need not hold up production; however, it does add an extra cost (~US$5/tonne) in terms of the aluminium-boron master alloy required and the cleaning of furnaces to remove settled materials post EC-grade metal production. Problems can occur with cross contamination in other alloy products since the borides cluster readily and can cause filters to block or act as inclusions causing defects in thin foil, for example.

If all potline metal were to be given this treatment it may well be possible to conduct the Al-B additions as part of a crucible treatment step to reduce the amount of boride carryover. Unfortunately, the boron treatment only works on elements that form a peritectic with Al and therefore nickel is not affected.


Thermodynamic studies are needed to identify a purge element/compound that could be reacted with nickel to remove/reduce it. Some progress has been made in the removal of iron using sodium borate, Na2B4O7 [45]. Such a method for precipitating iron compounds might also collect nickel, due to the common co-substitution of Fe and Ni, etc, in many complex intermetallic phases.

Alloy and Process Adjustments

Another strategy is to adjust the alloy compositions to counter the detrimental effects of V and Ni and also make use of their beneficial effects. There seems some scope to make use of the beneficial effects of V for example. However, to determine optimum alloy compositions and limits on V and Ni, extensive (and expensive) metallurgical studies will be needed. Process adjustments, e.g. in homogenization and forming, may also be needed. These may also be expensive.

Establishing Product Limits

It is clear from this review that there are many unknowns on what are the tolerable levels of V and Ni for a given final application. Studies are needed to determine what the tolerable levels of nickel and vanadium are for given alloys. Such studies will require considerable resources.

Conclusions

While there are no major problems with Ni and V levels at present, the trend of rising Ni and V levels will continue and some internal plant specifications will be exceeded. At this time, the response to the problem is essentially one of monitoring the V and Ni levels and checking for any adverse effects. Metal products customers may need assurance that that there are no adverse property changes as levels rise by conducting some property testing. In the longer term however the critical levels of V and Ni where significant property changes become unacceptable need to be established. A strategy to control V and Ni also needs to be formulated. Such a strategy might involve changes in alloy composition and downstream processing practice, treatments to purify coke, and/or removal of V and Ni from liquid metal. A coordinated approach is needed to optimize impurity pickup and control across the process chain. Further study to quantify the effects of rising levels of Ni and V on metal properties is needed.

References

[1] Grjotheim, K., Krohn, c., Malinovsky, M., Matiasovsky, K., Thonstad, J. (1982) Aluminium electrolysis: Fundamentals of the Hall-Heroult process, Aluminium Verlag.

[2] Rødseth, J., et al. (2002) Solubility of carbon in aluminium and its effect upon the casting process, Light Metals: Proceedings of Sessions, TMS Annual Meeting (Warrendale, Pennsylvania), 883-887.

[3] Augood, D.R. (1980) Light Metals, TMS, 413-427.

[4] Sparwald, V. (1973) Volatilization of Impurities during Fused Salt Bath Aluminium Electrolysis. Beitrag zur verfluechtigung der begleitelemente bei der aluminium-schmelzflusselektrolyse, 26(11), 529-533.

[5] Danek, V. et al. (2004) Material balance of vanadium in aluminium reduction cells. Industrial and Engineering Chemistry Research, 43(26), 8239-8243.

[6] Simensen, C.J. (1985) Sampling and analysis of impurities in aluminium, International Seminar on refining and Alloying of Liquid aluminium and Ferro-Alloys, Aluminium Verlag.

[7] Taylor, J.A. (1995) Metal related castabilty effects in aluminium foundry alloys. Cast Metals 8(4), 225-252.


[8] Vogt, F. et al. (2007) A preview of anode coke quality in 2007, TMS Light Metals, 2004, 489-493.

[9] Belov, N.A., Eskin, D.G. and Aksenov A.A. (2002) Iron in Aluminium Alloys: Impurity and Alloying Element, Taylor and Francis.

[10] Taylor, J.A. (1997) The role of iron in the formation of porosity in Al-Si-Cu alloy castings, University of Queensland, PhD Thesis.

[11] Mbuya, T.O., Odera, B.O. and Ng'ang'a, S.P. (2003) Influence of iron on castability and properties of aluminium silicon alloys: Literature review. International Journal of Cast Metals Research, 16(5), 451-465.

[12] Edwards, L. (2007) Responding to changes in coke quality, Australian Smelting Conference, Terrigal, NSW.

[13] Mondolfo, L.F. (1976) Aluminium Alloys: Structure and Properties, Butterworths.

[14] Hatch, J.E. (1984) Aluminium properties and physical metallurgy, American Society of Metals.

[15] Cook, R., Kearns, M.A. and Cooper, P.S. (1997) Effects of residual transition metal impurities on electrical conductivity and grain refinement of EC grade aluminium, Light Metals: Proceedings of Sessions, TMS Annual Meeting (Warrendale, Pennsylvania), Minerals, Metals & Materials Soc (TMS), 809-814.

[16] Mannweiler, U. et al. (1989) High vanadium Venezuelan petroleum coke, a raw material for the aluminium industry?, Light Metals: Proceedings of Sessions, AIME Annual Meeting (Warrendale, Pennsylvania), Metallurgical Soc of AIME, 449-454.

[17] Auchet, J. and Terzieff, P. (1997) The effect of Ti, V and Cr impurities on the transport properties of liquid aluminium. Journal of Alloys and Compounds, 261, 295-298.

[18] Maitland, A.H.R. (1987) 8th Internationale Leichtmetalltagung, 423-425.

[19] Belov, N.A., Eskine, D.G. and Aksenov, A.A. (2005) Multicomponent phase diagrams: applications for commercial aluminium alloys, Elsevier.

[20] Edwards, L., Vogt, F. and Wilson, J. (2001) Coke blending at Anglesey aluminium, Light Metals: Proceedings of Sessions, TMS Annual Meeting (Warrendale, Pennsylvania), 689-694.

[21] Chaudhuri, P.D. (1996) A Procedure for the Production of Ultra-Pure Precursors from Coal for the Manufacture of Value Added Carbon Products. Fuel Science and Technology International, 14(10), 1433-1446.

[22] Goodes, C.G., Eady, J.A and Nixon, J.C. (1987) Production of anode grade carbon from coal, Light Metals: Proceedings of Sessions, AIME Annual Meeting (Warrendale, Pennsylvania), Metallurgical Soc of AIME, 459-464.

[23] Renganthan, K.Z., Mintz, J.W. Kneisl, E.A. and Stiller, A.H. (1988) Preparations of an Ultra-Low Ash Coal Extract under Mild Conditions. Fuel Processing Technology, 18, 273-279.

[24] Gala, H.B.S., Kee, K.H. and Hucko, R.E. (1989) An Overview of the Chemistry of the Molten Caustic Leaching Process. Coal Preparation, 7, 1-28.

[25] de Vries, M. (2008) 'Biocoke’—sustainable materials for carbon anodes. [cited; Available from: http://www.csiro.au/news/newsletters/Metals/0803_metals/html/LMF-eNews_Biocoke.htm.

[26] Mousa, A., de Vries, M., Lovel, R. and Tassios, S. (2007) Aluminium Production in a Carbon Constrained Society: A Way Forward. PACE Petroleum Coke Quarterly, p. 31.

[27] Queneau, P.B. et al. (1992) Processing of petroleum coke for recovery of vanadium and nickel. Hydrometallurgy, 5(1), 3-24.


[28] Jack, T.R., Sullivan, E.A. and Zajic, J.E. (1980) The release of vanadium from Athabasca oil sands coke and coke ash. Fuel, 73 (May), 151-156.

[29] Jack, T.R., Sullivan, E. A. and Zajic, J. E. (1979) Leaching of vanadium and other metals from Athabasca oil sands coke and coke ash. Fuel, 58(8), 589-594.

[30] Lehmann, R., et al. (1985) Concepts for heavy metal handling in heavy oil upgrading processes, UNITAR/UNDP Information Cent for Heavy Crude & Tar Sands, 2053-2077.

[31] Schneider, L.G. and George, Z.M. (1981) Recovery of vanadium and nickel from oil sands coke ash, Extraction Metallurgy '81., Inst of Min and Metall, 413-420.

[32] Haas, A. (1981) Method for recovering vanadium from petroleum coke, US Patent 4,243,639.

[33] McCorriston, L.L. (1983) Process using sulphate reagent for recovering vanadium from cokes derived from heavy oils, US Patent 4,389,378.

[34] McCorriston, L.L. (1985) Process using carbonate reagent for recovering vanadium from cokes and ashes derived from heavy oils, US Patent 4,536,374.

[35] Thornhill, D.H. (1992) Process and apparatus for recovering heavy metal from carbonaceous material, US Patent 5,277,795.

[36] Schemel, R. (1984) Method for leaching and recovering vanadium from vanadium bearing by-product materials, US Patent 4,539,186.

[37] Malone, D.P. and Holcombe, T.C. (1981) Recovering vanadium from petroleum coke as dust, US Patent 6,241,806.

[38] Davila Armas, C.E. and Monhemius, A.J. (1987) Recovery of vanadium and nickel from petroleum coke by chloride volatilization, Pyrometallurgy, The Institute of Mining and Metallurgy.

[39] Belov, N.A. and Zolotorevskiy, V.S. (2002) The effect of nickel on the structure, mechanical and casting properties of aluminium alloy of 7075 type, Materials Science Forum, 935-940.

[40] Escobar, A.S., et al. (2006) Role of nickel and vanadium over USY and RE-USY coke formation. Applied Catalysis A: General, 315, 68-73.

[41] Wallenstein, D., Kanz, B. and Haas, A. (2000) Influence of coke deactivation and vanadium and nickel contamination on the performance of low ZSM-5 levels in FCC catalysts. Applied Catalysis A: General, 192(1), 105-123.

[42] Leinum, T. and Rasch, B. (2001) Crucible fluxing of potroom metal in a Norsk hydro cast shop effect on dross reduction and increased metal recovery, Light Metals: Proceedings of Sessions, TMS Annual Meeting (Warrendale, Pennsylvania), 1049-1052

[43] Dube, G. (1984) Removal of alkali metals and alkaline earth metals from molten aluminium, US Patent 4,470,846.

[44] Karabay, S. and Uzman,I. (2005) A study on the possible usage of continuously cast aluminium 99.6% containing high Ti, V, and Cr impurities as feedstock for the manufacturing of electrical conductors. Materials and Manufacturing Processes, 20(2), 231-243.

[45] Gao, J.W., et al. (2007) Effects of Na2B4O7 on the elimination of iron from aluminium melt. Scripta Materialia, 57(3), 197-200.


Development of a Phosphate - Free Reticulated Foam Filter Material for Aluminium Cast Houses

Leonard S. Aubrey1,a, Rudolph Olson1 and Dawid D. Smith2

1 SELEE Corporation, 700 Shepherd Street, Hendersonville, NC 28792, USA 2 Consultant for SELEE, 700 Shepherd Street, Hendersonville, NC 28792, USA

a [email protected]

Keywords: ceramic foam, SELEE, filtration, phosphine

Abstract

Filtration of molten aluminium using porous reticulated ceramic foam was developed in 1974 by

SELEE Corporation. Since that time, there have been significant technical advances in filter bowl

design and construction, preheat systems, development of fine pore size filters, development of

compact two-stage filtration systems, filter gasket materials, and equipment automation. One area

that has remained relatively unchanged has been the refractory filter material technology.

The refractory material utilized by all of the major cast house filter suppliers is based on an alumina

aggregate grain bonded with aluminium phosphate (AlPO4). This filter material, commonly

referred to as “PBA” in the aluminium industry, has become an industry wide standard and accounts

for nearly 99% of the filters supplied to cast houses worldwide. There are significant technical

limitations of PBA filters in terms of refractory performance, as well as potential environmental,

health and safety concerns.

This paper describes the development of a cost effective replacement filter material that overcomes

the limitations of PBA filters. The new material utilizes a low expansion – low modulus aggregate

material and a non-phosphate bond. The result is a significant reduction in mechanical failures

during use, improved filtration performance and elimination of the issue of potential environmental,

health and safety concerns.

Introduction

Ceramic foam for use in aluminium cast houses was developed in 1974 by SELEE Corporation.

During the 1980’s, there was rapid market acceptance of ceramic foam filter technology for a broad

range of fabricated aluminium products including rigid packaging materials, lithographic sheet,

aerospace products (sheet, plate, forgings and extrusions), bright finish trim, condenser tubing, foil,

architectural extrusions, foundry alloys and electrical conductor cable and wire. The subsequent

rapid market acceptance by aluminium cast houses of all types and levels of sophistication was due

to the following reasons: 1) Ease of use and operator acceptance, 2) Operational flexibility – drain

after every cast, 3) Low variable operating cost, 4) Low capital installation cost, 5) Effective

inclusion removal and 6) Small footprint requirements (minimal floor space required for

installation). In the last ten years, there have been significant advances in ceramic foam filtration

technology in terms of filter bowl design and construction, filter heat systems, development of fine

pore size filters, development of compact two-stage filtration systems, filter gasket materials, and

equipment automation [1-7].

One area that has not seen significant technical advancement is the refractory filter material. The

original SELEE

filter was based on a chrome-alumina (Cr2O3-Al2O3) aggregate grain and an

aluminium phosphate binder [8,9]. This chrome-alumina filter material was relatively expensive

and had potential environmental problems due to potential hexavalent chrome issues. After several

years, it was replaced with an all alumina grain material and incorporated refractory ceramic fibers

(RCF), but still utilized the aluminium phosphate binder [10]. This material is commonly referred

to in the aluminium industry as phosphate bonded alumina (PBA). Today nearly all (98-99%) of


the filters used in aluminium cast houses are based on aluminium phosphate-bonded alumina. High

purity (silica-free) sinter bonded filters are commercially available and exhibit excellent corrosion

resistance, but have notoriously poor thermal shock resistance. The high cost of reactive grade

alumina combined with high temperature sintering cost, make sinter bonded cast house filters cost

prohibitive, except in the case of specialty Al-Li-X alloys.

There are several significant problems with aluminium phosphate bonded alumina filters.

Aluminium Phosphate Bond Reaction

The aluminium phosphate bond is reactive with the magnesium in aluminium alloys.

3AlPO4 + 3Mg Mg3P2 + 3MgO + AlP (1)

The above reaction has been confirmed by x-ray diffraction analysis of used PBA filter material

after immersion testing in 5182 alloy. X-ray diffraction analysis confirmed the presence of

magnesium phosphide and magnesia. The magnesium and aluminium phosphide reaction products

in used PBA filters can then react with water or atmospheric water moisture resulting in the

formation of phosphine gas.

Mg3P2 + 6H2O 2PH3 (g) + 3Mg(OH)2 (2)

AlP + 3H2O PH3 (g) + Al(OH)3 (3)

The release of phosphine gas from spent aluminium phosphate bonded alumina filters has been

confirmed with a short term Dräger tube system and a hand held photoionization detector (PID).

Pure phosphine is a colorless and odorless gas. If diphosphine (P2H2) impurities are present, an

odor of garlic or rotting fish can be detected. The European Union classifies phosphine gas as

highly flammable (F+), very toxic (T+) and dangerous for the environment (N) [11].

Magnesium reaction with the aluminium phosphate binder results in intergranular corrosive attack

and weakening of the filter microstructure and loss of inclusion retention capability. Figure 1 is a

photomicrograph of used PBA filter after filtering AA5454 alloy for 90-minutes showing

intergranular metal penetration and the release of calcined alumina grain.

Figure 1. Photomicrograph showing intergranular attack of an aluminium phosphate bonded

alumina filter. Alloy Type: 5454, Molten metal contact time: 90-minutes.

The extent of the Mg-AlPO4 reaction is dependent on the magnesium level, temperature and time.

The extent of the reaction was documented by analysis of metallurgical samples from controlled

dynamic molten aluminium corrosion tests where the reaction could be systematically evaluated as

a function of magnesium content, metal temperature and contact time. Table 1 contains the results

of these tests conducted on PBA filter material with 6061 (1.01 wt. % Mg) and 5182 (4.83 wt. %

Metal


Mg) alloys as function of metal temperature (675, 700, 725 and 750 C) and contact time (30, 60,

120-minutes). After the corrosion tests were completed, the filters samples were mounted,

polished and evaluated with a stereomicroscope using cross-polarized light. Un-reacted filter

material is white in appearance. The first stage of corrosive attack results in an orange/brown

discoloration that starts at the filter-metal interface and extends inward. The second stage is metal

wetting and infiltration (intergranular attack) into the alumina matrix, which results in a dark grey

discoloration, also starting at the filter-metal interface and extending inwards. Figure 1 shows the

greyish discoloration zone in reflective light microscopy.

Table 1. Results of Dynamic Aluminium Resistance Testing of PBA Filters in 6061 and 5182

Alloys a Function of Metal Temperature and Time.

Alloy Type: 6061 (1.01 Wt. % Mg)

Metal

Temp., C

Test Duration,

Minutes

%

Un-reacted

% Orange/Brown

Discoloration

%

Wetted/Infiltrated

675

30 90 10 0

60 63 35 2

120 5 90 5

700

30 85 15 0

60 5 90 5

120 2 88 10

725

30 15 80 5

60 10 82 8

120 2 88 10

750

30 10 78 12

60 5 80 15

120 1 79 20

Alloy Type: 5182 (4.83 Wt. % Mg)

Metal

Temp., C

Test Duration,

Minutes

%

Un-reacted

% Orange/Brown

Discoloration

%

Wetted/Infiltrated

675

30 85 15 0

60 10 70 20

120 5 65 30

700

30 15 70 15

60 2 38 60

120 0 5 95

725

30 5 25 70

60 0 20 80

120 0 5 95

750

30 0 17 83

60 0 15 85

120 0 5 95

Lateral Compressive Edge Crushing Failures

PBA filters are susceptible to compression-type failure due to the magnesium reaction and the high

thermal expansion and compressive modulus of the filter material. These factors in combination

with the heat expandable gasket, creates a high lateral compressive stress in the filter when seated in

the filter bowl. During use, the loss in compressive strength due to the magnesium softening

reaction results in a compressive failure (crush) parallel to the filter edge. Figure 2 shows an edge


section of an edge-crushed PBA filter where the edge is deformed inwards. Figure 3 shows the

fragmented filter structure inside the crush zone immediately adjacent to the filter edge. When edge

crushing occurs, filter debris can be released from the filter and end up in the ingot or billet. Figure

4 shows a reacted fragment of PBA filter material in rigid container sheet as a result of Mg-AlPO4

reaction and lateral compressive edge crushing of the filter.

Figure 2. Photograph of a sectioned PBA filter

showing lateral compressive edge crushing.

Note filter edge is deformed inwards.

Figure 3. Backscattered electron image

showing fragmented filter structure

within the crush zone.

Figure 4. Fragments of PBA filter material in

can end sheet as a result of lateral

compressive edge crushing.

Figure 5. Thermal expansion of aluminium

phosphate bonded alumina showing a

berlinite phase transformation at 80 to 180

C resulting in a 2 – 3% volume change.

Poor Thermal Shock Resistance

PBA filters have poor thermal shock crack resistance due to the high thermal expansion of the

alumina grain, as well as a low temperature phase transformation in the aluminium phosphate

binder. Figure 5 shows the thermal expansion behavior of PBA and a 2 – 3% volume expansion

due to berlinite phase transformation around 80 to 180C. During preheat with hand held high-

pressure gas-air inspirator burners, PBA filters can be easily cracked. Although the use of preheat

systems using air modulation controlled medium velocity burners can eliminate the cracking

problem, a lot of casting pits still utilize high-pressure gas-air inspirator burners to preheat filters.

Filter Edge

Deformed Edge


Hydrogen Pickup from Aluminium Phosphate Bonded Filters

Aluminium phosphate bonded filters may contain residual aluminium phosphate hydrates. During

use, hydrates can be reduced by magnesium and release hydrogen into the metal, typically in the

range of 0.005 to 0.030ml of H2 per 100-grams of aluminium. This is not an issue for most

products, but can sometimes be a problem in the case of hard alloy heat-treated plate products.

Non-Phosphate Filter Material Development

The following technical and commercial objectives were set for the development of the next

generation aluminium cast house filter to replace the current PBA filter:

1. Filter material must have a better “environmental footprint”.

a. Eliminate the use of phosphate bond and the potential for phosphine gas release.

b. Eliminate the use of refractory ceramic fibers (RCF).

c. Eliminate use of energy intensive materials (calcined and reactive aluminas).

d. Utilize energy efficient high-speed roller hearth firing cycle.

2. Reduce reactivity in Mg-containing aluminium alloys (more non-wetting).

3. Eliminate the problem of lateral compressive edge crushing and release of filter debris.

4. Improved thermal shock resistance and resistance to preheat cracking.

5. Eliminate problem of hydrogen release.

6. Must be cost effective replacement for PBA filters. Raw materials must be locally sourced

and not subject to global market price-supply volatility and foreign exchange rate variations.

7. Must be a direct substitution for PBA filters and an operationally transparent replacement to

cast house operators.

The following approach was taken:

1. Utilize locally available alumino-silicate aggregate material with low thermal expansion and

compressive modulus.

2. Utilize a more stable boron glass bond to protectively encapsulate the alumino-silicate grain.

Boron glass bonds can be sintered at conventional roller hearth firing temperatures.

The thermal stress generated in a brittle ceramic body during heating is dependent on the material’s

physical properties [12].

th = ET/(1-) (4)

Where th = thermal stress, E = Young’s modulus, = coefficient of thermal expansion, T =

temperature difference and = Poisson’s ratio. Lowering both the Young’s modulus and the

coefficient of thermal expansion can substantially lower the generated thermal stress responsible for

lateral compressive edge crushing and improve the thermal shock resistance. Introducing well-

dispersed micro-porosity into a brittle ceramic material can further improve thermal shock

resistance by improving the materials resistance to crack growth. Boron glasses are utilized in rigid

media tube filters to bond tabular alumina grain material, as well as for filter cloth [13]. During the

firing process, the boron glass fluxes and reactively bonds the alumino-silicate grains. Figure 6

shows the two-phase microstructure of the boron glass coating, the alumino-silicate grain and well-

dispersed microporosity that further improves thermal shock resistance.


The new filter material has been given the commercial designation CS-X. Table 2 provides data

on the general physical properties of the CS-X filter.

Table 2. Physical Properties of CS-X Filter Material.

Filter Material: Boron glass bonded mullite

Microstructure Phases: Mullite + kyanite

Apparent Density: 2.95 grams/cm3

Thermal Expansion Coefficient: 5.33 X 10-6 /C (heating to 750 C)

Bulk Macro-porosity: 70 – 80%

Nominal Chemical Composition by ICP Analysis – Wt. %

Al2O3 SiO2 B2O3 CaO MgO Na2O

48.7 42.0 5.49 2.1 0.36 0.21

Figure 7 shows the thermal expansion of PBA and CS-X filters materials showing the reduced

thermal expansion and the elimination of the low temperature berlinite phase transformation. Table

3 shows the calculated relative thermal stress of CS-X and PBA filter materials using the thermal

expansion data and text book value for Young’s modulus and Poisson’s ratio. Table 3 indicates that

relative thermal stress is reduced 77% with the new filter material.

Table 3. Calculated Relative Thermal Stress Values for CS-X and PBA Filter Materials.

Filter

Type CTE ()

l/dl /C

Youngs’s Modulus

(E), GPa

Poissons

Ratio ()

Calculated Relative

Thermal Stress, (th/T)

PBA 8.87 X 10-6 354 0.25 0.004187

CS-X 5.44X 10-6 137 0.23 0.000968

Figure 8 shows the average cell and window size as a function of filter pore size for CS-X filters.

Cell and window diameter were determined by optical measurement using a stereomicroscope and

digital imaging system. Table 4 contains window size specifications for CS-X filters.

Table 4. Window Size Product Specification for Standard Pore Size CS-X Filters.

Pore Size Grade Average Window Size - Microns

Maximum Minimum

10 1,875 1,578

20 1,368 1,071

30 1,013 795

40 795 611

50 611 464

60 534 401

The filter priming head is a critical operational characteristic that affects startup flow rate and filter

life. Detailed knowledge of the filter priming head is required for the filter bowl design and

installation. A detailed investigation was conducted to determine priming head as a function of the

filter pore size and alloy type. Immersion priming tests were conducted at 700C in a 10-kg

capacity electrically heated crucible furnace with 99.5% purity aluminium and 2024, 3104, 5182,

6063 and 7075 alloys. A total of 173 immersion priming head tests were conducted to determine

the critical metallostatic breakthrough pressure as a function of filter pore size. The results were

analyzed using a multi-regression analysis software package. The best fit of the priming head data


was obtained as a function of the reciprocal of the filter window size, which follows the general

behavior predicted by a Washburn equation [14].

P = -2.

.cos/r (5)

Where P = pressure required for infiltration, = liquid surface tension, = liquid-solid contact

wetting angle, and r = pore radius. The metallostatic priming (hp) is then obtained by:

P = .g

.hp = -2

.

.cos/r (6)

hp = -2.

.cos/

.g

.r (7)

hp = -4.

.cos/

.g

.dp (8)

Where hp = metallostatic priming head, = molten metal density, g = Newton’s gravitation constant

and dp = filter window diameter = 2.r. The alloy magnesium content was determined by the

regression analysis to significantly increase the priming head. The following regression equation

was obtained.

hp = 68.10 + 2.37* Wt. % Mg + 83,337/dp (9)

Figure 9 shows the filter priming head as a function of filter window size and alloy type (99.5 and

5182). Figure 10 shows the recommended filter priming head as a function the filter pore size grade

number. The recommended priming head has an additional 25-mm to insure reliable filter priming

at casting startup.

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

5,000

0 10 20 30 40 50 60 70

Pore Size Grade

Dia

me

ter

- M

icro

ns

Window

Diameter

Cell

Diameter

100

120

140

160

180

200

220

240

260

280

300

0.0 1.0 2.0 3.0

1000/Window Size - Microns -1

Prim

ing H

ea

d - m

m o

f A

l

99.5

5182

Figure 8. Cell and window diameter as a

function of pore size grade for CS-X filters.

Figure 9. Filter priming head (hp) as a

function of window size (dp) showing the

effect of magnesium content.

hp = 68.1 + 2.37

X

%Mg + 83,337/dp


0

100

200

300

400

500

600

500 600 700 800 900

Test Temperature - C

Ho

t M

OR

- k

Pa

PBA

CSX

Figure 10. Recommended priming head

after adding a 25-mm safety factor for

reliable priming.

Figure 11. Hot modulus of rupture strength

of CS-X and PBA filters.

The room temperature modulus of rupture and compressive strength of CS-X and PBA are very

similar. Figure 11 shows the modulus of rupture strength over the range of 600 to 800C for PBA

and CS-X . Overall CS-X has a modulus of rupture about 20% higher than PBA.

The corrosion resistance of the new filter material was initially validated using controlled

experiments in a laboratory crucible furnace. A dynamic aluminium resistance test (DART) was

developed to fully simulate the molten metal environment in a wide range of cast house casting

processes. The test was run in a 300-mm deep electrically heated crucible furnace and allowed two

filter test samples to be run simultaneously so head-to-head comparison in material performance

could be directly made. The following parameters were fully controllable in DART procedure:

1. Molten metal immersion temperature: Corrosion tests were conducted at standard

temperatures between 675 and 800C using immersion temperature control ( 3C).

2. Test Duration: 30-minutes to 2-hours.

3. Alloy Type: The one metric ton master alloy lots were obtained in the form of 1kg ingots to

provide consistent alloy chemistry for the corrosion-testing program: 99.5% purity, 6063,

6061, 3104, 2024, 5182 and 7075.

4. Filtrate Speed (velocity): Average filtrate speed was fully controllable over a range of 10 to

35cm per minute.

At the completion of the test, the corrosion samples were solidified against a chill plate.

Metallographic samples were then prepared for examination using reflective light and scanning

electron microscopy to judge the degree of metal-filter reaction and wetting. The corrosion samples

were evaluated using a rating system that considered discoloration, wetting/intergranular

penetration, strut thinning and infiltration into the foam precursor void space. Head-to-head testing

of PBA and CS-X filters confirmed that the boron glass system was superior to the phosphate

bonding system.

Field Testing and Pre-Production Qualification

Prior to commercial introduction in May 2008, an extensive field-testing program was carried out in

aluminium cast houses in North America. Initial field-testing was conducted in billet cast houses

casting 6XXX alloys followed by ingot casting of 5XXX and 7XXX series alloys. Operationally


there was no difference in filter priming, metal flow rate or filter head loss. Extensive field-testing

confirmed the technical approach of using a lower thermal expansion and compressive modulus

filter material to solve the problem of lateral compressive edge crushing. During cast house trials,

when PBA filters were aggressively preheated with hand held gas-air inspirator burners, nearly

100% of the PBA filters were noted to develop full-length transverse cracks. CS-X filters

preheated under the same condition did not crack during preheating. The results of LiMCA and

PoDFA inclusion testing indicated the filtration performance of CS-X filters to be equal or better

than PBA run under similar conditions. The performance difference between CS-X and PBA

was more apparent in the more severe filtration environment (increased Mg content, metal

temperature and time). Based on AlSCAN measurements taken upstream and downstream of the

filter bowl, no hydrogen pickup was detected with CS-X filters.

Table 5 lists the casting processes and end products where CS-X filters have been qualified and

are in routine commercial use.

Table 5. Casting process and end products where CS-X filters are in commercial use.

Casting Process Alloy Type(s) End Product

Rolling Ingot 3104 Can body

Rolling Ingot 5182 Can end

Forging Ingot 2XXX, 5XXX, 6XXX, 7XXX Aerospace, defense forging stock

Rolling Ingot 2024, 7075 Aircraft plate

Billet Casting 6XXX Architectural extrusions

Twin Strip Roll Caster 1XXX Food packaging foil

Belt Caster 5054 Sheet

In addition, process and product design failure modes and effects analyses (FMEA) were

completed. Multiple full-day manufacturing production runs were made to develop a robust filter

manufacturing process prior to commercial production. Process capability studies and control plans

were also completed prior to product introduction.

After an 18-month period of pre-production manufacturing and field-testing, commercial sales were

officially started in May 2008. After 10 months, a total of 37,000 filters have been manufactured

and sold. It is anticipated that at the end of 2009, 90% of SELEE’s cast house filter customers will

have converted from PBA to CS-X filters.

Conclusions

There are several technical limitations with the use of the aluminium phosphate bonded alumina

filter used in aluminium cast houses. These limitations arise due to the reactivity of the aluminium

phosphate binder with magnesium and the high thermal expansion and modulus of the alumina

grain. The reaction of the aluminium phosphate bond results in wetting and intergranular attack,

weakening of the filter and the potential for phosphine gas release after use. The high thermal

expansion and modulus of the alumina grain material results in poor thermal shock resistance and

tendency for the filters to crack during preheating. Phosphate bonded filters can also contain

aluminium phosphate hydrates that can react with magnesium and result in hydrogen pickup.

A cost effective replacement filter material has been developed to replace aluminium phosphate

bonded alumina filters. The new filter utilizes a low thermal expansion and modulus grain that

significantly reduces the thermally generated stresses in the filter and improves the thermal shock

resistance. The new filter uses a boron glass bond that is chemically more stable and resistant to

wetting and reaction. Field-testing of the new filter confirmed it has superior thermal-mechanical

properties relative to PBA filters. Metallurgical analysis of spent PBA and CS-X filters run under


similar conditions in the aluminium cast houses indicated the boron glass bonding system to be

more stable.

Acknowledgments

The authors would like to acknowledge the dedicated hard work of Niki Rhodes and Dr. Feng Chi

for the filter development work and to Mark Topolski for metallographic sample preparation work

and metallurgical analysis.

References

[1] Butcher, K. R., Rogers, D. B. (1990) Update on the Filtration of Aluminum Alloys With Fine

Pore Ceramic Foam, in Proceedings of the TMS Annual Meeting and Exhibition: Light

Metals 1990, Anaheim, USA, 797-803.

[2] Aubrey, L. S., Martins, L. C. B. (1995) Advances in Ceramic Foam Filter Bowls and Preheat

System Design, In the Proceedings of the 4th

Australian Asian Conference on Aluminium Cast

House Technology, Sydney, Australia, 185-197.

[3] Aubrey, L. S., Oliver, C. L., MacPhail, B. T. (1997) Dual Stage Ceramic Foam Filtration

System and Method, Patent Number 5,673,902, United States of America.

[4] Aubrey, L. S., Cummings, M. A., Oliver, C. L, Chow, M. (1996) The Development and

Performance Evaluation of a Dual-Stage Ceramic Foam Filtration System, in Proceedings of

the TMS Annual Meeting and Exhibition: Light Metals 1996, Anaheim, USA, 845-855.

[5] Smith, D. D., Aubrey, L. S., Miller, W. C. (1998) LiMCA II Evaluation of the Performance

Characteristics of Single and Staged Ceramic Foam Filtration, in Proceedings of the TMS

Annual Meeting and Exhibition: Light Metals 1998, San Antonio, USA, 893-915.

[6] Barbis, D., Smith, D. D., Aubrey, L. S., Miller, C. W. (1998) Performance of a Staged

Filtration System at Norandal USA Inc to Filter Continuous Cast Converter Stock, in

Proceedings of the TMS Annual Meeting and Exhibition: Light Metals 1998, San Antonio,

USA, 917-937.

[7] Niedzinski, M. M., Williams E. M., Smith, D. D., Aubrey, L. S. (1999) Staged Filtration

Evaluation at an Aircraft Plate and Sheet Manufacturer, in Proceedings of the TMS Annual

Meeting and Exhibition: Light Metals 1999, San Diego, USA, 1,019-1,030.

[8] Pryor, M. J., Gray, T. J. (1976) Ceramic Foam Filter, Patent Number 3,947,363, United States

of America.

[9] Yarwood, J. C., Dore, J. E., Preuss, R. K. (1976) Ceramic Foam Filter, Patent Number

3,962,081, United States of America.

[10] Brockmeyer, J. W. (1982) Ceramic Foam Filter, Patent Number 4,343,704, United States of

America.

[11] EU Dangerous Substances Directive (67/548/EEC).

[12] Lee, W. E., Rainforth, W. M. (1994) Ceramic Microstructures – Property Control and

Processing, Chapman and Hall, 109p.

[13] McDonald, H. A., Snyder, R. C. (1970) Filter Medium for Molten Metal, Patent Number

3,524,548, United States of America.

[14] Washburn, E. W. (1921) Proceedings of the National Academy of Sciences, USA, Vol. 7,

115p.

Alumino-

silicate

grain


Results with a Multi Stage System of Filtration Employing a Cyclone

John H Courtenay1,a and Frank Reusch2

1 MQP Limited, 6, Hallcroft Way, Knowle, Solihull West Midlands, B93 9 EW

2 Drache Umwelttechnik GmbH, D-65582-Diez, Germany a [email protected]

Keywords: filtration, cyclone

Abstract

The development of a new prototype multi stage filter was described at TMS 2008 in which a ceramic foam filter was applied in a first chamber operating in cake mode; grain refiner added in a second chamber and a cyclone deployed in a final chamber to ensure removal of any oxides or agglomerates arising from the grain refiner addition or release events from the foam filter. The first industrial prototype was installed at Trimet Aluminium at Essen in Germany and demonstrated that liquid metal could pass through the cyclone successfully without excessive turbulence or splash. The further development of the prototype based on new water modeling work is described.

Introduction

A definite need to develop an efficient, low hold up volume, filtration process capable of treating high flow metal rates has been recognized (Waite, 2002 [1]).

The object of the current work was to develop a filter that could deliver the high efficiency performance of a deep bed filter but with low hold up volume, low floor space requirement and the ability to be used economically in conjunction with frequent alloy changes.

As an initial consideration the phenomena of enhanced filtration efficiency in ceramic foam filters that could be achieved by adding the grain post filter reported by Towsey and others [2] provided a good starting point.

This phenomenon was first reported by Kakimoto and others [3] in 1996 in relation to the operation of porous tube filters.

Kakimoto concluded that bridges of CaO particles that tended to form at the pores at the surface of a tube filter were “suppressed” by the addition of boron containing grain refiners. That is the addition of titanium diboride particles prevented the formation of a stable filter cake which is initiated by the formation of bridges as a first stage to support the subsequent cake formation. This conclusion was reached on the basis of metallographic examination of spent tube filters.

In 2002 Towsey and others [4] reported the results of an extensive study on the effect of addition of various grain refiner compositions on the performance of ceramic foam filters with the conclusion that grain refiner addition, via a suspected agglomeration behaviour and alteration of filtration mechanism, prevented bridge formation in finer pore ceramic foam filters thus reducing the hitherto observed very high filtration efficiencies reported in earlier work [5].

In 2005 Instone and others [6, 7] described a new design of filter unit named the XC filter which gave superior filtration efficiency achieved by the combination of ceramic foam filtration and deep bed filtration. Importantly this design comprised a three chamber unit with a ceramic foam filter in the first chamber, grain refiner addition in the second chamber and a small bed filter in the third chamber.


The design concept is shown below:

Figure 1. XC Filter design, Instone and others [6, 7].

Performance data was compiled using LimCa during extensive industrial casting trials and showed that excellent filtration efficiency could be achieved.

In a further objective of the current work it was decided to, in addition to employing the principle of adding grain refiner in a second chamber after a ceramic foam filter, to concentrate on developing an alternative to the third chamber proposed above [6,7] in order to simplify the operation and maintenance of the filtration unit.

In 2005 Katgerman [8] described work on water modeling and computer modeling of flow control devices. It is well known that dams and weirs placed in a launder section or a chamber, such as that forming part of a degassing apparatus, contribute to the removal of inclusions. However, Katgerman concluded that, although this could be effective for small concentrations of particles, this technique suffered from the drawback that small fluctuations in the flow behaviour may reintroduce the sunken particles (collected at the base of the dams due to settling out by virtue of their higher density relative to liquid aluminium) into the metal flow.

Instead, based on flow calculations, the concept of a cyclone was conceived and subsequently proven in terms of effectiveness by further numerical and water modeling experiments.

This then was taken as a potential basis for the development of a system, which would be relatively simple and have low operating maintenance by virtue of having no replaceable media required in the third compartment.

Development of a Three Stage Reactor Design

Then the design concept for the current work was determined based on the above considerations and comprised three chambers:

• A first chamber containing a ceramic foam filter;

• A second chamber for the addition of grain refiner;

• A third chamber containing a cyclone.

In all cases it was considered that a third chamber was necessary to ensure that any titanium diboride agglomerates, other indigenous particulates and/or any oxide inclusions or stringers arising from the grain refiner rod manufacturing process were removed.

In the first chamber, the absence of grain refiner particles would allow the appropriately pore sized ceramic foam filter element to operate in cake mode so as to achieve high levels of filtration efficiency.


In the second chamber, grain refiner rod would be added contra to the metal flow to ensure rapid and adequate mixing.

With respect to the operation of the ceramic foam filter it has been demonstrated [7] that it can work satisfactorily in cake mode with 1xxx alloys without blocking off over a period of up to three casts when casting relatively clean pot room metal. The tendency of the filter to block will depend on the nature of the filter cake formed which in turn will depend on the type and quantity of inclusions in the liquid metal. The tendency of the filter cake to block is a function of whether the cake is compressible or incompressible and in reality will be so to a greater or lesser extent. What is clear is that, under the operating conditions envisaged, the cake should be sufficiently incompressible to allow metal flow without impedance; at least over the period of one or two casts.

Collaboration Partners

To realize the above concept, collaboration was established with partners as follows:

• TU Delft: Under Professor Katgerman flow modeling was conducted to define the correct design and performance for a cyclone to be incorporated into a filtration unit which would fit into the available space at a casting pit;

• Trimet Aluminium agreed to act as the alpha test site and to install the unit on a production casting pit;

• Drache Umwelttechnik provided the expertise and manufacturing facilities for the design and building of the filter unit including the filter elements, cyclone and refractory lining.

Design of the Prototype

A key requirement of the design was that it should be able to work effectively with a maximum available head height at the casting pit of 1000mm. The cyclone itself was required to fit into maximum external space of 1000mm x 1000mm x 1000mm which meant in practice, after allowing for the metalwork and refractories, that the maximum internal height for the cyclone would be 740mm.

As reported elsewhere [9] the efficiency and flow capacity of the cyclone was determined to be sufficient to meet the project requirements based on the flow modeling studies.

The final dimensions of the cyclone were:

0.44m x 0.36m x 0.74m.

The flow rate range for satisfactory operation was: 0.01m/s – 1.0m/s corresponding to up to 15t/h at the casting pit.

The efficiency predicted is shown below. The figure depicts the % particle removal efficiency for different flow velocities and two types of cyclone design, type 1 without a container for inclusion capture at its base and type 2 with a container for inclusion capture.

Type 1 Type 2


In the modeling, one of the important parameters, the coefficient of restitution, was assumed at unity since no accurate data was available but this assumes then that if a particle hits a wall it bounces back into the flow stream at the same velocity. This assumption probably means that the predicted removal efficiency is conservatively stated since some attraction between the refractory surfaces and the inclusions can be anticipated.

Figure 2. Predicted efficiency for particle removal by cyclone.

The model shows an expected removal efficiency of approximately 50% for particles >60micron with a flow velocity of 0.5m/s rising to 80% for particles > 100 microns. However, the true removal efficiency can only be determined by actual operation of the unit in practice.

With these criteria in mind the final chamber design was arrived at and is shown below:

Figure 3. Final design of three stage “OptiFilter” filtration unit.

As can be seen, because of space constraints it was necessary that the metal flow enter and exit the unit from the long side.

Separate drain plugs are provided for each chamber and in particular it is planned to remove inclusions from the base of the cyclone via the left hand taphole shown. Calculations based on the known metal cleanliness indicate that only a very small quantity of inclusions would need to be removed after each cast, of the order of micrograms.

First Casting Trials

The Optifilter unit was delivered to Trimet Aluminium in January 2008 for installation at casting pit 11 and the first casting trials took place on February 18th.


Table 1. Trial conditions.

Alloy 1000 series Billet No 32

Billet length 2000mm Cast size 13,000kg

Casting speed mm/min 115mm/min Casting speed kg/min 200kg/min Casting temperature 790 ºC

Some difficulties were experienced with the preheating of both the first chamber containing the ceramic foam filter and the cyclone chamber. An electric air heater had been designed to provide the preheat in the cyclone chamber. This method of heating was selected because of the difficulties associated with heating such a fully enclosed chamber with normal gas heating. In the event the electric air heater proved inadequate and a gas flame was used although this was not very effective.

Despite these initial problems a heat was successfully started at the second attempt and metal flowed through the filter for some 20 minutes before freezing off on the casting table.

No problems were experienced with splash or turbulence during start up and metal flowed through the filter very smoothly throughout the short casting period.

The objective of the first cast had been to verify that metal could flow through the unit and no attempt was made to determine the cleanliness of the metal passed though the filter, however as a purely visual observation the surface of the metal exiting the filter was very bright and clean in appearance with no evidence of surface scum.

Figure 4. The prototype “Optifilter” filtration unit under pre heat at Trimet Aluminium.

Improvement of Pre Heating

Following the first trial it was necessary to take the filter unit out of service and carry out modifications to increase insulation, reduce the thermal mass and improve the preheating system for the cyclone.

The first prototype had the cyclone cast in a single refractory block and it was considered that this had represented a significant thermal mass contributing to the start up problems and it was decided to change the construction to a thin walled cone in fused silica set inside fibrous insulating material.

The Wiedemann DSB pre heater burners for the ceramic foam filter chamber had functioned well and therefore were not changed. The preheat arrangements for the cyclone were reviewed and


despite the previous difficulties it was decided to remain with the electric air heater principle because of its suitability for heating the fully enclosed space created by the cyclone however the thermal energy was increased so that a theoretical temperature of 1000ºC could be reached.

The temperatures reached ranged from a minimum of 397.5ºC at the cyclone entry slot to a maximum of 535.5ºC at the cyclone exit. 403ºC was achieved at the middle of the cyclone and 474ºC at the bottom of the cyclone.

These results were considered sufficient to proceed to a second liquid metal trial. Temperature measurements during preheat confirmed that the box was being adequately pre-heated with a temperature of 705ºC being reached at the exit tube from the cyclone.

Second Casting Trial

Casting was attempted using the same conditions as in the previous test however although liquid metal passed successfully through the cyclone it was observed that the flow rate was very low – the exit tube only filled to approximately 10% of the exit cross section and it was estimated that the flow rate being achieved was of the order of 3t/h instead of the 15t/h target. Efforts were made to increase the head height in the input launder to the maximum – to the extent that there was virtually no free board but despite this no increase in the exit flow rate could be achieved. As a consequence after 10 minutes of casting the casting table froze off and the cast was aborted. This procedure was attempted a further two times but no improvement was seen with in each case the highest flow rate achieved being restricted to 3t/h.

Figure 5. View of the exit flow of the cyclone showing that the exit flow tube is

only partially filled.

From these second series of trials it was concluded that:

• The measures to improve insulation and increase preheat temperature had been successful;

• Nonetheless despite this the unit still froze off and this was due to the outlet flow rate having been restricted to only 3t/h.

It was considered that restriction to the outlet flow must be due to either insufficient head height to overcome the resistance to flow of the cyclone and or an insufficient cross sectional area at the cyclone inlet slot to allow adequate flow through the cyclone.

Further Water Modeling

It was decided to re validate the flow modeling and conduct further water model tests at Delft University to verify the model.


The water model was set up with a facility to use a variable inlet cross sectional area and the following observations were made [10]:

• The ratio between the inlet and outlet cross-sections decreases in the sequence of the cyclone designs used in Turchin’s thesis [11], in pilot setup, and in the water model as 1.53, 1.43, and 1.2.

• For a given water head the throughput is controlled by the cross-section of the inlet.

• Analytical calculations confirmed that about 40mm of the metal head should be sufficient to provide for the inlet velocity reflecting the metal flow rate 15t/h. This means that the metal head of 145-150mm achieved in industrial trials should be sufficient. As this is not the case, other reasons for the decreased throughput should be sought.

• Measurements performed during experiments with the water model (water head, flow rate, inlet cross-section) allowed us to construct a dependence of the hydrostatic head on the squared velocity, which is similar for the case of the watermodel, Turchin’s simulations [11] and the cyclone used in industrial trials. These dependences are shown in Figure 2. For the given geometry with the given inlet

Figure 6. Estimated dependences of the hydrostatic head on the inlet velocity for three cases: “numerical” (Turchin’s model [11]), experimental (“water model”), and data from pilot cyclone (“alum”). Light gray areas shows the range of velocities corresponding to melt flow rates from 10 to 20t/h with the current size of the inlet in the pilot cyclone, darker gray areas shows the inlet velocities in the case when the cross-section of the inlet is increased two times.

Conclusions

Subject to the final results the preliminary conclusions are:

• A unique three stage reactor containing a cyclone as a third stage has been built based on flow modelling calculations and installed at Trimet Aluminium;

• Molten aluminium passed successfully through the reactor however the outlet flow rate achieved was insufficient to achieve satisfactory casting and was estimated to be 3t/h;

• There were no issues with metal turbulence or splash at the exit from the cyclone;


• High rates of heat loss experienced initially in the cyclone due to the high thermal capacity of the refractory have been overcome by modifications to the construction and the heating system;

• Based on the water model experiments it was concluded that the cyclone inlet cross-section should be increased to accommodate the melt flow rate and to decrease the required metal head so that the melt flow rate reaches the target 10-15t/h;

• Detailed measurements of the filtration efficiency for several alloys and casting rates will be made.

Acknowledgments

The authors would like to thank the casthouse management team and the technical and casting personnel of Trimet Aluminium for their support and participation in this collaborative programme.

References

[1] Waite, P. (2007) A Technical Perspective on Molten Aluminium Processing, Light Metals, 2002, 841-848.

[2] Towsey, N, Schneider, W., Krug, H-P. Hardman, A. and Keegan, N.J. (2001) The Influence of Grain Refiners on the Efficiency of Ceramic Foam Filters, Light Metals, 973 -977.

[3] Kakimoto, K., Yoshida, T., Hoshino, K. and Nishizaka, T. (1996) The Filtration of Molten 1xxx Series alloys with Rigid Media Filter, Light Metals, 833 – 838.

[4] Towsey, N., Schneider, W. and Krug, H-P. (2002) The Effects of Rod Grain Refiners with Differing Ti/B Ratio on Ceramic Foam Filtration, Light Metals, 931-935.

[5] Keegan, N., Schneider, W. and Krug, H-P. (1999) Evaluation of the Efficiency of Fine Pore Ceramic Foam, Light Metals, 1031-1041.

[6] Instone, S., Badowski, M. and Schneider, W. (2005) XC Filter - A Filter for Increased Filtration Performance, Proceedings of Aluminium Cast House Technology, Melbourne, Australia, 259 – 267.

[7] Instone, S., Badowski, M. and Schneider, W. (2005) Development of Molten Metal Filtration Technology for Aluminium, Light Metals, 933 – 938.

[8] Katgerman, L. and Zuideman, J. (2005) Upstream Fluid Flow Particle Removal, Light Metals, 927 – 931.

[9] Turchin, A.N., Erskin, D.G. Courtenay, J.H. and Katergerman, L. (2007) Novel purification concept in molten metal processing, Proceedings of Aluminium Cast House Technology, Sydney, Australia, 225-230.

[10] Private communication Jafari, A. and Eskin, D.G. Delft University of Technology, Department of Materials Science and Engineering, Mekelweg 2, 2628 CD Delft, Netherlands

[11] Turchin, A.N. (2008) Proefschrift ter verkrijging van de graad van doctor aan de Technische Universiteit Delft, op gezag van de Rector Magnificus Prof. Dr. Ir. J.T. Fokkema, voorzitter van het College voor Promoties, in het openbaar te verdedigen op diensdag 20 mei 2008.


The Use of Electromagnetic Fields for the Detection of Inclusions in Aluminium

Steve Poynton1,2,a, Milan Brandt1,2, John Grandfield1,3

1 CAST CRC, UDP No 055, The University of Queensland, St Lucia QLD 4072 Australia

2 Swinburne University of Technology, PO Box 218 Hawthorn VIC 3122, Australia 3 Grandfield Technology Pty Ltd, 37 Mattingley Cr, Brunswick West, VIC 3055, Australia

a [email protected]

Keywords: inclusion, detection, aluminium, K-Mold, Podfa, Lais, Prefil, LiMCA, ultrasound

Abstract

Management of inclusions is an important part of quality control within the aluminium cast house. Inclusions have a detrimental effect on many aluminium cast products. The ability to reliably detect inclusions in a timely fashion is an essential part of this process. There are a number of tools available for inclusion measurement based on different principles. Techniques for inclusion detection such as metallurgical analysis, K-Mold, Podfa, Lais, Prefil all have a delay before detailed results are available. Ultrasound provides a possible technique for an online sensor, however has not as yet managed sufficient sensitivity. LiMCA, based on the Electrical Sensing Zone, has provided the most sensitive online detection to date, but other electromagnetic techniques such as a multiple voltage array sensor offer promise of a sensor which can be built for lower cost and can sample a larger portion of the melt.

Introduction

Monitoring molten metal cleanliness is an essential part of the quality control process required to maintain a consistent product and measurement and control of inclusions is an important part of maintaining the melt cleanliness [1-3]. The most common inclusions found in the aluminium melt include aluminium oxides, magnesium oxides and spinel, carbides, refractory particles and silica particles [1,4-6]. Any of these inclusions not removed from the molten metal prior to solidification can be harmful to physical and surface characteristics of cast products, causing a variety of defects including pinholes in foil, streaks, tears in deep drawn forms, surface defects in sheet products, provide nucleation sites for hydrogen gas leading to blistering, increased porosity, increased breakage rates when drawing wire and poor machinability, and influence the fatigue response of metals. [2,3,6-10]. The effects of inclusions on the final product, and inclusion detection and removal impose a substantial cost and industry is therefore keen to improve inclusion detection methods.

Building sensors to operate in molten aluminium has a number of challenges because the corrosive nature and high temperature of the melt provide a hostile environment to any devices. The opaqueness of the melt and very low inclusion levels (<30 ppm) required for most aluminium products add to this challenge [11]. This review examines the advantages and disadvantages of each method, and discusses opportunities to improve inclusion detection, keeping in mind factors such as cost, the time required before obtaining any results, the sensitivity of the system, the range of inclusion sizes detected, the level of information reported and the size of the sample.

Non-Electromagnetic Techniques

There are several principles used for inclusion detection in molten aluminium. Non-Electromagnetic techniques discussed here use metallographic analysis, ultrasound and chemical analysis.


Metallographic Techniques

There are a number of techniques based around metallographic principles. Classic metallography involves the physical examination of ingot slices to determine the presence and type of inclusions. The ingot is cut and polished, and the surface is examined for trapped inclusions. Metallographic analysis can give a very good idea of the exact nature of inclusions in the sample, but it has a very small sample size, is time consuming, costly and requires an experienced technician. [4,7,12,13].

A standardized metallographic analysis, the K-Mold, is a simple shop-floor test for detecting larger inclusions, in the range of 60-80µm. The K-Mold produces a small casting consisting of a flat plate with four notches which serve as fracture points. The test consists of casting one or more plates, which are fractured at the notches to provide a number of fracture surfaces. The fracture surfaces are examined for inclusions, and the inclusion level expressed as a ratio representing the proportion of faces with inclusions [1,2]. The advantages of this test include low cost, low required level of operator skill, sampling flexibility and sample retention [2], however this test is an offline test and has a very small sample volume.

The next step up in metallographic analysis is to use pressure filtration which works by forcing aluminium sample under pressure through a fine filter which will trap the inclusions on the surface. The inclusions are concentrated 5000-10000 times on the surface of the filter, and can later be analysed metallographically [12]. Alcan’s Porous Disk Filtration Analyser (PoDFA) and Union Carbides Liquid Aluminium Inclusion Sampler (LAIS) and Prefil are commercial tools using this principle. PoDFA and LAIS use the same principles, differing only in the operation [13-15]. The Prefil Footprinter expands on the simple pressure filtration test by measuring the rate at which metal flows through the filter which gives an indirect measurement of the build up of inclusions trapped on the surface of the filter. The rate of metal flow can be compared with reference curves to give a quick indicator of the overall inclusion level in the melt [12,13].

These tests are cheap to set up, but have very small sample volumes up to 2000g [4] and the samples are time consuming and expensive to analyse [7,13]. Although Prefil does give an immediate indication of melt cleanliness, detailed results are still time consuming.

Ultrasound

Ultrasound has been an attractive option for inclusion detection in molten aluminium due to it’s ability to probe the interior of materials non-destructively [16,17] and because it offers the capability of a high sampling rate of 100s of samples per second and a large sample volume[18].

Most current research is looking at ultrasonic systems which detect the signal reflected from individual particles [18-20]. Although it is theoretically possible to use changes in attenuation or velocity of the ultrasound signal to indicate the presence of inclusions in the melt, due to the very low level of inclusions found in molten aluminium, typically less than 30ppm, the variation in velocity and attenuation are not sufficient for monitoring the melt to a sufficient quality [11,19,20].

A typical probe will consist of an ultrasonic transducer, linked to the melt via an aircooled buffer rod, usually between 200 and 300mm in length, acting as a focused waveguide [3,4,11,16,17,19-21]. A spherical concave acoustic lens may be machined into the end of the buffer rod to focus the ultrasonic signal, increasing the spatial resolution of the probe, allowing the detection of smaller inclusions than a flat ended probe. [3,11,17,19,20] The probe may be used in a Pulse Echo configuration, using a single probe as both a transmitter and receiver [11,16,19-21] or in the pitch catch configuration, using two probes, one as a transmitter and the other as a receiver [11,19-21].

Buffer rods are required as current transducers cannot operate at the high temperatures of molten aluminium, however the use of buffer rods results in problems caused by spurious echoes within the rod, and signal loss through the walls of the rod, which degrades the signal to noise ratio (SNR) and


reduces the sensitivity of the probe. [3,4,11,16,17,19-21]. The tapering and cladding will also help reduce unwanted echoes and improve wave guidance of the buffer rod, resulting in a better SNR [3,11,16,17,19-21], and new materials may offer better acoustic coupling and high corrosion resistance. [3,11,16,19,20], however more work is required to overcome these problems before a commercially useful probe can be developed.

Chemical Analysis

Chemical analysis and Optical Emission Spectrography measure the chemical composition of a very small sample, however they suffer problems due to the low concentrations in which inclusions occur in molten aluminium, the non uniform distribution of the inclusions, and the very small sample size of these techniques. For example, research has shown a poor correlation between oxygen concentration and inclusion levels due to the presence of variable background oxygen levels in surface oxides and sub-micron films [4,8,22-24].

Electromagnetic Inclusion Detection Methods

The large difference in electrical properties between the non-conducting inclusions and the highly conductive aluminium melt in which they are contained means there are a number of electromagnetic techniques which can be used to detect inclusions in molten aluminium, however the high conductivity of the aluminium melt can limit the sensitivity of these techniques, increasing the technical challenges of producing a useful measuring device. The three techniques discussed here are the Electrical Sensing Zone, as implemented in LiMCA, the use of an induced Lorentz force and a multiple electrode voltage probe.

Coulter Counter/Electrical Sensing Zone (ESZ)

The Electrical Sensing Zone (ESZ) or Coulter Counter involves a current path being provided by a conductive liquid carrying a current flowing through an aperture. The impedance across the aperture is monitored. When a non-conducting particle passes through the aperture a resistance change can be detected proportional to the volume of the particle [25].

The Electrical Sensing Zone technique can be used for detecting small inclusions in molten aluminium. Molten aluminium is drawn through a small aperture in the presence of a large DC current. The non-conducting inclusions in the melt are detected as they pass through the aperture with the aluminium. The particle displaces it’s volume of conducting fluid causing micro-ohmic changes in resistance which can be detected as a voltage pulse across the aperture. The LiMCA, LiMCA II and LIMCA CM analysers are implementations of this system [7,26,27].

The detector in the LiMCA analysers consists of two electrodes separated by an insulating borosilicate tube (See Figure 1). A small orifice, of around 300µm in diameter, in the tube provides a current path between the electrodes. A large current in the order of 60 Amps is passed between the two electrodes, through the orifice in the insulating tube. Molten aluminium is drawn through the orifice into the tube via negative pressure. This draws inclusions in the melt through the orifice, which can be detected as a change in the voltage between the electrodes of between 20µV–10mV, providing a real time count of the inclusion level in the sample. Roughly one sample a minute can be taken, each sample being 17.5g [2,4,7,18,26,28-32].

The resolution of LiMCA is limited by the background electrical noise. This limits detection of particles down to 20µm using an orifice diameter of 300µm. [4]. The system can resolve particles up to 155µm [32]. An estimated 20-60% of a melt’s total inclusions may be too large to pass through the orifice [2].


Figure 1. Schematic diagram of the LiMCA system [30].

The main advantages of the LiMCA system are that it can provide continuous sampling over the duration of the melt, can measure inclusions down to 20µm and that it provides user independent results. The disadvantages to the LiMCA system are that it is a high cost system, limiting it’s use within research or the foundry industry and due to the narrow orifice of 300µm the sample size is small [4].

Electromagnetic Detection using the Lorentz Force

A method of concentrating and detecting inclusions based on the Lorentz force (Archimedes electromagnetic force) was proposed by Makarov, Ludwig and Apelian [4,10]. A DC current can be applied to the melt via two electrodes. This current will induce a magnetic field which will induce a Lorentz force in the metal. This force will not be induced in the non conducting inclusions, which will result in the inclusions moving in the opposite direction to the Lorentz force. By arranging the electrodes to produce a downward pointing Lorentz force this effect can be used to force inclusions to the melt surface where they can be inspected visually [4,10,33,34].

This method could provide continuous monitoring of the aluminium melt, taking up to 200 samples per minute, sampling a volume up to 2cm3 and detecting particles down to 10µm [4,10].

The surface tension of the melt will tend to prevent the inclusions breaking the surface of the melt, which prevents the identification of the inclusions, so a method of overcoming the surface tension is required [10].

Multiple Voltage Probe Sensor

Makarov and Ludwig have proposed the use of a multiple voltage probe sensor to detect non-conducting particles in a conducting media [35].

The Multiple Voltage Probe Sensor uses principles similar to resistivity arrays in geophysical prospecting and Electrical Image Tomography (EIT) used in medical imaging [36-38].

Argon

Atmosphere

Differential Amplifier

Signal Processing

Data Display

A

Insulating Tube

Electrodes

Vacuum


For both geophysical prospecting and medical imaging EIT a current is injected into the body being analysed, and a voltage response is measured by electrodes at the surface. These measurements are used to generate an image of the resistivity distribution under the surface [36].

Calculating a resistivity distribution from surface voltage measurements is computationally expensive, and very susceptible to error as the small variations in voltage correspond to much larger variations in the distribution of conductivity [36,37,39,40]. For inclusion detection, voltage profiles can be calculated to indicate the presence of inclusions, which gets around this problem. Instead the main challenge in this technique will be achieving sufficient sensitivity to detect small inclusions.

The proposed multiple voltage probe sensor consists of an array of voltage electrodes located over a surface. An electric current is applied to the melt via two of the electrodes, and the surface voltage response measured via the other electrodes [35].

A single non-conducting inclusion passing under the sensor should be seen as a standard two peak response on the sensor, one the inverse of the other, as shown in Figure 2. The peak magnitude is related to the inclusion size, whilst the peak separation distance is proportional to the depth of the inclusion [35].

Figure 2. Standard 2 peak response.

The surface voltage response for a small inclusion (R<100µm) is small, however it is distributed over a large sensor area so that even if the noise for individual pins is high, the signal present on multiple pins can still be recovered [35].

Molten aluminium is ideal for multiple voltage probing because it has good electrode contact, high electrical homogeneity, low concentrations of inclusions and the metal acts as a Faraday shield reducing the background noise to a level of around 5µV [35].

The multiple voltage probe sensor may not achieve the same sensitivity as the LiMCA system, but it can scan larger melt volumes and does not require the artificial aperture or the vacuum pump to draw liquid metal through the aperture so it should be much cheaper to implement [35, 36].

Conclusion

There are a number of techniques used for inclusion detection in molten aluminium. Metallographic techniques, including classic metallography, and pressure filtration can give good information about


what types of inclusions are present, however they are slow to carry out, limiting their usefulness for process control. Measuring the overall chemical composition of the melt is also time consuming, and the chemical composition of the sample has been a poor indicator if inclusion levels.

Ultrasound has potential to provide online monitoring, however issues with the signal to noise ratio, and finding appropriate materials for the buffer rod will need to be solved.

This leaves techniques based on the use of electromagnetic fields providing the best methods for the online monitoring of inclusion levels.

The Coulter Counter or Electrical Sensing Zone technique implemented in LiMCA currently provides the only real time detection of individual inclusions, and is capable of measuring inclusions down to 20µm in diameter. Both the use of a Lorentz force induced by an electromagnetic field to force inclusions to the surface of the melt, and the use of an electrode array to measure a surface voltage response caused by an inclusion in an electric field, have the potential to provide online monitoring of inclusion levels and sampling a larger volume at a lower cost than LiMCA.

References

[1] Neff, D.V. (2004) Evaluating Molten Metal Cleanliness for Producing High Integrity Aluminum Die Castings, in Die Casting Engineer, 2-6.

[2] Rasmussen, W.M. (1996) Aluminum Melt's Cleanliness, in Modern Casting, 45-48.

[3] Ihara, I., Aso, H. and Burhan, D. (2004) In-situ observation of alumina particles in molten aluminum using a focused ultrasonic sensor. JSME International Journal, Series A: Solid Mechanics and Material Engineering, 47(3), 280-286.

[4] Makarov, S., Apelian, D. and Ludwig, R. (1999) Inclusion Removal and Detection in Molten Aluminum: Mechanical, Electromagnetic and Acoustic Techniques (99-150). TRANSACTIONS-AMERICAN FOUNDRYMENS SOCIETY, 727-736.

[5] Simensen, C.J. and Berg, G. (1980) A survey of inclusions in aluminum. Aluminium (Isernhagen, Germany), 56(5), 335-40.

[6] Keles, O. and Dundar, M. (2007) Aluminum foil: Its typical quality problems and their causes. Journal of Materials Processing Technology, 186(1-3), 125-137.

[7] Guthrie, R.I.L. and Li, M. (2001) In situ detection of inclusions in liquid metals: Part I. Mathematical modeling of the behavior of particles traversing the electric sensing zone. Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science, 32(6), 1067-1079.

[8] Doutre, D., et al. (1985) Aluminum cleanliness monitoring: methods and applications in process development and quality control. Light Metals (Warrendale, PA, United States), 1179-95.

[9] Tian, C., et al. (2002) Effect of melt cleanliness on the formation of porosity defects in automotive aluminium high pressure die castings. Journal of Materials Processing Technology, 122(1), 82-93.

[10] Makarov, S., Ludwig, R. and Apelian, D. (1999) Electromagnetic visualization technique for non-metallic inclusions in a melt. Measurement Science and Technology, 10(11), 1047-1053.


[11] Ono, Y., Moisan, J.F. and Jen, C.K. (2003) Ultrasonic Techniques for Imaging and Measurements in Molten Aluminum. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 50(12), 1711-1721.

[12] Enright, P.G. and Hughes, I.R. (1995) A new shop floor technique for quantitative measurement of molten cleanliness. Proceedings of the International Conference on Molten Aluminum Processing, 4th, Orlando, Fla., Nov. 12-14, 1995, p. 431.

[13] Simard, A.A. et al. (2000) Cleanliness measurement benchmarks of aluminum alloys obtained directly at-line using the prefil-footprinter instrument. in Light Metals: Proceedings of Sessions, TMS Annual Meeting (Warrendale, Pennsylvania), Nashville, TN.

[14] Rinderer, B., Danilova, N. and Couper, M. (2005) A comparison of LAIS and PoDFA analysis of metal cleanliness. in Proceedings of the Australasian Conference and Exhibition - Aluminium Cast House Technology, Melbourne.

[15] Dion, S. (2004) Device and method for measuring metal inclusions (Alcan International Limited, Can.), WO, 20 pp.

[16] Jen, C.K., Legoux, J.G. and Parent, L. (2000) Experimental evaluation of clad metallic buffer rods for high temperature ultrasonic measurements. NDT and E International, 33(3), 145-153.

[17] Ihara, I., et al. (2002) Ultrasonic in-line sensors for inclusion detection in liquid metals. in Proceedings of the IEEE Ultrasonics Symposium, Munich, Germany.

[18] Aubrey, L.S., Smith, D.D. and Martins, L.C.B. (2001) New product developments for aluminum cast houses. in Proceedings of the Australian Asian Pacific Conference on Aluminium Cast House Technology.

[19] Ono, Y., et al. (2004) An On-Line Ultrasonic Cleanliness Analyzer for Molten Light Metals. JOM, 56(2), 59-64.

[20] Ono, Y., et al. (2002) Development of ultrasonic techniques with buffer rod in molten aluminum. in Proceedings of the IEEE Ultrasonics Symposium, Munich, Germany.

[21] Ihara, I., Jen, C.K. and Franca, D.R. (2000) Ultrasonic imaging, particle detection, and V(z) measurements in molten zinc using focused clad buffer rods. Review of Scientific Instruments, 71(9), 3579-3586.

[22] Falk, H. and Wintjens, P. (1998) Statistical evaluation of single sparks. Spectrochimica Acta - Part B Atomic Spectroscopy, 53(1), 49-62.

[23] Kuss, H.-M., et al. (2002) Comparison of spark OES methods for analysis of inclusions in iron base matters. Analytical and Bioanalytical Chemistry, 374(7-8), 1242-1249.

[24] Mohamed, W.T.Y. (2008) Improved LIBS limit of detection of Be, Mg, Si, Mn, Fe and Cu in aluminum alloy samples using a portable Echelle spectrometer with ICCD camera. Optics and Laser Technology, 40(1), 30-38.

[25] Nieuwenhuis, J.H., et al. (2004) Integrated Coulter counter based on 2-dimensional liquid aperture control. Sensors and Actuators, B: Chemical, 102(1), 44-50.

[26] Martin, J.-P., Hachey, R. and Painchaud, F. (1994) Online metal cleanliness determination in molten aluminum alloys using the LiMCA II analyzer. Light Metals (Warrendale, PA, United States), p. 915-20.


[27] Mei, L.I. and Guthrie, R.I.L. (2000) Numerical studies of the motion of spheroidal particles flowing with liquid metals through an electric sensing zone. Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science, 31(4), 855-866.

[28] Simard, A., et al. (2001) Capability study of the improved PREFIL-footprinter to measure liquid aluminum cleanliness and comparison with different techniques. Molten Aluminum Processing, International AFS Conference, 6th, Orlando, FL, United States, Nov. 11-13, 2001, p. 83-103.

[29] Li, M. and Guthrie, R.I.L. (2003) Molten Metal Inclusion Sensor Probes, Limca Research: United States.

[30] Pedneau, N. and Pekguleryuz, M.O. (1997) Equiaxed-grain size analysis in the mushy zone during solidification via an in-situ method based on the electrical sensing zone principle. Scripta Materialia, 37(7), 903-909.

[31] ABB (2007) The ABB Group - Automation and Power Technologies. [Webpage] [cited 31 Nov 2007]; ABB Group]. Available from: www.abb.com.

[32] ABB (2007) Analyze IT - LiMCA II Liquid Metal Cleanliness Analyzer. Retrieved 2007.

[33] Ludwig, R., Apelian, D. and Makarov, S. (2002) Systems for detecting measuring inclusions, United States. p. 19.

[34] Lemdiasov, R.A. and Ludwig, R. (2004) Mathematical Design of an Electromagnetic Separation Sensor in Molten Aluminum. IEEE Transactions on Magnetics,40(1 I), 37-42.

[35] Makarov, S., Ludwig, R. and Apelian, D. (2000) Identification of depth and size of subsurface defects by a multiple-voltage probe sensor: Analytical and neural network techniques. Journal of Nondestructive Evaluation, 19(2), 67-80.

[36] Brown, B.H. (2003) Electrical impedance tomography (EIT): A review. Journal of Medical Engineering and Technology, 27(3), 97-108.

[37] Schuessler, T.F. and Bates, J.H.T. (1998) Current patterns and electrode types for single-source electrical impedance tomography of the thorax. Annals of Biomedical Engineering, 26(2), 253-259.

[38] Dickin, F. and Wang, M. (1996) Electrical resistance tomography for process applications. Measurement Science and Technology, 7(3), 247-260.

[39] Jordana, J., Gasulla, M. and Pallas-Areny, R. (2001) Electrical resistance tomography to detect leaks from buried pipes. Measurement Science and Technology, 12(8), 1061-1068.

[40] Yi, M.J., et al. (2001) Three-dimensional imaging of subsurface structures using resistivity data. Geophysical Prospecting, 49(4), 483-497.


CHAPTER 6:

Cast House Productivity

Estimating the Production Capabilities of Casthouse Equipment Configuration Options

P. W. Baker

Hatch Associates Ltd., 5 Place Vile Marie, Montreal, H3B 2G2, Quebec, Canada

[email protected]

Keywords: productivity, cast house modelling, capacity analysis

Abstract

Operational casthouses must, above all else, be capable of accepting all primary metal produced by

the potlines so as not to limit the ability of the smelter to produce metal and to provide the most

stable operating regime for the reduction cells by allowing consistent metal pad depth. To be

effective, casthouse equipment configurations must accommodate not only the normal operational

state, but also both repair and catch-up scenarios.

Hatch has developed a simple first pass methodology for assessing the performance of equipment

configuration options for primary aluminium casthouses in this context.

Background

The operational focus of any modern aluminium smelter is the core reduction process itself. Process

engineers, technology providers and operators from across the plant must all contribute to the

optimised stable operation of the reduction line operating at the highest amperage and current

efficiency because this represents the fundamental source of revenue.

For its part, the casthouse must above all else be capable of accepting all of the metal that can be

produced by reduction. Bottlenecks in the casthouse result in the accumulation of metal in reduction

cells upsetting the regular tapping schedule and cell heat balance through excessive metal pad depth

fluctuation.

While the ability to accommodate the production needs of reduction is paramount, the casthouse

design must also represent the lowest investment cost for the performance level required.

Basis of Model

The following analysis demonstrates that generic expressions for the required utilisations of the key

components of the casting process line; the casting machines and the furnaces, can be derived as a

function of equipment numbers and furnace and caster throughput rates.

The expressions so obtained provide a simple first-pass comparison methodology for candidate

equipment configurations to produce the necessary product mix tonnages under the various

operational scenarios of steady-state, catch-up and repair.

This analysis is based on certain simplifying assumptions but provides a preliminary analysis

method before the decision is taken to conduct the more thorough and detailed Discrete Event

Model (DEM) simulation.

System Analysis

Simple Casting Line

Figure 1 below describes a simple steady state mass balance across a casting line comprised of a

furnace and a casting machine. The casting line processes NL ladles of metal each with a capacity of

CL tonnes throughout the calendar year.


Where:

PA = production, (t/y) RC = caster rate of production, (t/h) UC = caster utilisation rate, (dimensionless 0 -1) HA = scheduled operating hours per year, (h/y) RF = furnace throughput, (t/h) UF = furnace utilisation rate NL = number of ladles delivered, (y

-1)

CL = ladle capacity, (t)

Figure 1. Simple casting line.

Multi-component Casting Line

For the more complex multi-component casting line the relationship is similar and assumes that all

components are identical in their specification and operation.

Figure 2. Multi-component casting line.

Where there is more than a single furnace is arranged in parallel the combined furnace system

throughput is equivalent to:

AFFFA HURNP (1)

Where:

NF = number of furnaces in parallel

This assumes all furnaces are identical and have identical operating cycle times.

The equivalent expression for the caster system is;


ACCCA HURNP (2)

HA is the same for both caster and furnace systems because it represents the annual production time

scheduled for the entire production line depicted in Figure 2 above.

Furnace Throughput

The throughput of the furnace can be expressed as:

T

FF

F

CR (3)

Where:

CF = effective furnace capacity, (t)

FT = furnace cycle time, (h)

The Furnace cycle time may be expressed as:

FT = FP + FD (4)

Where:

FP = heat preparation time (5)

FD = metal delivery time (6)

FD is a function of caster system production rate and effective furnace capacity such that:

CC

FD

RN

CF (7)

Substituting FD from (7) into (4);

CC

FPT

RN

CFF (8)

and rearranging:

CC

FPCC

TRN

CFRNF (8a)

Substituting for FT in expression (3) we obtain;

FPCC

FCC

FCFRN

CRNR

(9)

System Utilisation

Substitute the above expression for RF into (1):

FPCC

FCC

AFFACFRN

CRNHUNP (10)

Rearranging for UF:


FF

FPCC

ACC

AF

CN

CFRN

HRN

PU (11)

The expression for caster utilisation is more simply obtained by rearranging (2);

ACC

AC

HRN

PU (12)

The above is identified as the first term of the expression for UF.

Application

The expressions above provide a simple means of comparing different casthouse equipment

configurations to compare each in respect to its ability to:

Produce product under steady state conditions;

Catch-up when a potroom metal inventory backlog exists (UC or UF = 1); and

Maintain parity when certain equipment is shutdown (furnace/caster N-1 operation).

Ratio parameters derived from the expressions above are also useful in indicating:

If the line is caster constrained (UC/UF >1) or furnace constrained (UC/UF <1); and

How well matched the equipment configuration is; UC/UF close to unity.

Shutdown Time and Scrap Production

The above analysis has not accounted for lost production time through planned or unplanned

shutdown. Additionally, it has not accounted for production hours lost casting scrap. These factors

may be accounted for by reducing accordingly, the scheduled available hours (HA) in the year.

Table 1 lists nominal values for purposes of illustration only.

Table 1. Shutdown & scrap loss.

Description % h/y

Scheduled time 100% 8760

Unplanned downtime 2.0% 170

Planned downtime 2.7% 240

Available time 8350

Scrap production time 4.0% 334

Capacity factor loss 5.0% 417.5

Total utilisable time 87% 7598.5

Scheduled Time

In this example the casting line is scheduled to produce this product 100% of the year, hence a

maximum of 8760h is scheduled. However this may not always be true such as where a casting line

produces more than a single product.

Unplanned Shutdown Time

Unplanned shutdown is time that the entire process line stops because of an equipment or

component failure.


Planned Shutdown Time

Planned shutdown is the time that the line is inactive because it is undergoing planned maintenance

or refurbishment

Available Time

Available time is the difference between the total scheduled time and the planned plus unplanned

shutdown time.

Scrap Production Time

Production hours lost casting scrap.

Capacity Utilization

Equipment, when running, does not always produce at 100% of rated capacity due to partial

functional loss and ramp up/down.

Total Utilizable Time

This is the time during which the line may be utilised at its rated capacity HA in the above

expressions. The Utilisation values derived above will be based on this time period.

Casthouse Configuration Analysis

Casthouse equipment is grouped into separate casting systems or casting lines each comprised of

one or more identical casters fed by one or more identical furnaces. The performance of the

individual casting systems is assessed using the above relationships in the form of a spreadsheet

model.

no launder connection

System 1

• A352

• 38 ktpa

• UC = 28%

• UF = 85%

System 2

• Purity

• 105 ktpa

• UC = 69%

• UF = 99%

System 3

• Purity

• 261 ktpa

• UC = 86%

• UF = 86%

no launder connection

System 1

• A352

• 38 ktpa

• UC = 28%

• UF = 85%

System 2

• Purity

• 105 ktpa

• UC = 69%

• UF = 99%

System 3

• Purity

• 261 ktpa

• UC = 86%

• UF = 86%

Figure 3. Example of casthouse systemization.

Several alternative equipment configurations would be assessed and ranked based on performance

and cost. Other outcomes such as operational safety are of course, important factors also.


The analysis method described above is applied to three operational states:

Normal – major equipment functional, producing desired product mix at average annual rates

Catch-Up – major equipment functional, casting Potroom backlog delivered at max rate to

casthouse (Ladle queue)

Repair – one or more major equipment items shutdown, casting product at normal metal

delivery rate

The performance or capability of the configuration is assessed for each operational state according

to the following criteria.

Normal - For desired PA determine required UC, UF. Performance measure is maximum

utilization, lower is better

Catch-Up – increase PA until either UC or UF reaches 100%. Performance measure is

throughput tonnage, higher is better.

Repair – reduce equipment numbers until no longer able to handle normal metal flow.

Performance measure is tolerance to equipment outage, higher is better.

Assumptions

There are several key assumptions required to permit the analysis above;

All like equipment within a casting system is identical in respect to both specification and

operational cycles (e.g. furnaces identical with other furnaces)

Furnaces may not receive metal from ladles when delivering metal to the casting process

Only one furnace supplies metal to the casting process at any given time.

Equipment is dedicated (not shared) during campaigns on individual products

It is assumed that there is a ladle queue at the casthouse which acts as a buffer to even-out

metal supply to filling furnaces. This is predicated by the assumption that metal delivery from

Potrooms is uniform.

No allowance is made for turnaround time of casting machines

Example Results and Outcomes

The results and outcomes of the application of this methodology to a practical casthouse example

are presented below. The values shown are nominal and provided only for purposes of illustration.

In this example various equipment configurations and product mixes are reviewed. It is desirable to

cast a proportion of the metal as value added A356 but under the proviso that the casthouse retains

its capability to cope with equipment outages and recover effectively from deficits. Capital cost and

safety of operation are also factors

Model Outcomes

The outcome of any analysis process is the selection of the preferred option.

Table 2 below lists three fundamental performance parameters derived from the overall analysis and

previously explained in this article.


Table 2. Typical outcomes of analysis.

Option Limiting Utilization Catch-Up Ratio Repair tolerance

1 99.0% 1.11 <1

2 88.5% 1.11 >1

3 90.0% 1.49 2

4 82.5% 1.49 >2

5 88.6% 1.39 2

Option 4 offers the best performance by virtue of its low limiting utilization under normal state

conditions while being most able to recover from a deficit and tolerate equipment outages. Options

1 and 2 are not viable.

In practice, the analysis is more complex and must take into account the capital investment cost

differentials and the complexity of the operation.

Summary

Hatch has developed a simple analytical approach for assessing the production performance

capability of casthouse equipment configuration options. Large primary aluminium casthouse

operations are amenable to analysis by this methodology.

The method is useful for Greenfield and Brownfield projects and can usefully indicate bottlenecks

and identify areas for improvement as well as rank potential configuration options.


CHAPTER 7:

Direct Chill & Continuous Casting

Downstream Considerations of Fusion Clad Casting

Robert Wagstaff1,a, Todd Bischoff1 and Dave Sinden1

1 Novelis, Solatens Technology Center 10221 E. Montgomery, Spokane, Washington, 99206, USA

Novelis Corp, 6060 Parkland Blvd, Cleveland, Ohio, 44124, USA a [email protected]

Keywords: direct chill casting, clad metals, rolling, fusion clad casting

Abstract

Fusion casting has been commercially available for over 3 years. Although there are many interesting points to casting, there are several downstream considerations such as scalping, preheat and rolling. Quality rolled clad sheet needs to have a consistent clad thickness. Appropriately varying the geometry of the clad layer during the transient portions of casting results in consistent clad thickness from head to butt. Controlling the edge wipe during rolling off does the same from edge to edge. The severity of edge wipe off can be predicted from relative flow stresses at rolling temperatures and guides the choice of alloy combinations.

Introduction

When Francis C. Frary [1] at Alcoa discovered that the rolling of strong alloys could be done, with the absence of surface cracking if the harder alloy was rolled as a sandwich between two softer alloy sheets, he set in motion a series of ideas and concepts which in retrospect have changed the aluminum industry for the best. Dix [1] applied Frary’s findings to the production of strong alloys for the fledgling aerospace industry and provided a corrosion resistant layer to these strong alloys and Mike Miller [2] discovered that two aluminum alloys could be joined if one or more of the two alloys had a layer of significantly lower melting point alloy in the area to be joined.

Aluminum clad alloys are commercially produced today by Frary’s process of hot roll-bonding a clad layer to a core alloy by rolling under significant loads at elevated temperatures. This conventional process has many manufacturing steps with ample opportunities to generate scrap. The clad layer must be produced via a separate route of casting, scalping, pre-heating, rolling and trimming to the necessary clad plate thickness and size. The clad and core surfaces must be clean when mated, possibly requiring a surface preparation step. As well, “tacking” (welding) or strapping the clad plate to the scalped core is required for pre-heating. The first rolling passes must be very light to reduce the slippage of the layers until they roll bond. There are limitations to the alloy combinations that can be roll bonded since some alloys form tenacious oxide films that are difficult to disrupt during the bonding process. This can result in “dirty” interfaces that degrade the useful strength of the sheet. In this latter regard, alloys high in Mg are particularly difficult to bond efficiently.

The commercial production of multi-alloy ingots have intrigued solidification and rolling experts for years as it was recognized that such a process would open doors to new products far beyond the classical selection of monolithic and conventional Al-clad products. In previous publications, a new process – “Novelis Fusion” – was introduced. This technology produces clad material directly by DC casting [3]. It was reported that this technology produced a clean, high-strength interface with few restrictions for the combination of alloys involved. It was also reported that Fusion sheet is produced by essentially the same hot and cold rolling route used for conventional monolithic alloys.


Process

Equipment Configuration

As reported in our U.S. Patent application [4] a Direct Chill (DC) mold and starting head, commonly used in the industry, serve as the primary heat removing, geometry-defining apparatus [4]. The molding apparatus also includes a secondary heat removal apparatus (SHRA) internal to the mold opening. Multiple coolant flows and liquid metal streams are controlled by a series of flow control and metal level sensors which maintain the necessary thermal, structural and mechanical boundaries during solidification (see Figure 1a).

Figure 1a. Figure 1b.

Casting

Solidification commences after liquid metal enters the cavity corresponding to the alloy with the highest liquidus temperature prior to the lowering of the starting head. See Figure 1a. After the starting head is lowered and before it exits the lower end opening of the mold, a second stream of metal enters the mold cavity and is raised in the cavity to contact the semisolid interface of the first alloy immediately below the secondary heat removal apparatus. Critical process parameters are modified to correct for the transient heat removal conditions typical to the start up and end portions of the cast.

Figure 1b, represents the apparatus set up and description for the production of a package where the clad layer has a higher liquidus temperature than the core material. In the Fusion process, alloy superheat and latent heat are removed at the SHRA, similar to any continuous or semi-continuous casting process. Care and attention is given to the alloy fraction solid and surface temperature immediately below the SHRA so that radiant effects of the core can re-heat the surface above the solidus if needed prior to contact with the core material.

Special care and attention needs to be placed on the equipment side of alloys with not only overlapping freezing ranges but also alloys with challenging coefficients of thermal contraction during solidifying and cooling of the ingot. All in all, the solidification portion of the process has developed and continues to develop into a technology package which is “cast house tough” and so at this point we then direct your attention to some of the subsequent down stream processing special considerations which are necessary to produce a successful sheet product.

Preparations for Scalping

In our patent application [4], we describe a methodology which we use to vary the geometry of the SHRA during the cast to compensate for the transient heat removal conditions at the start and end of the cast when compared to the more consistent heat removal conditions common to a steady state


process. The transient heat removal at the start often leads to a fatter ingot at the start, sometimes called “butt swell.” This technology effectively keeps the interface between the clad and core parallel to the centerline of the ingot. This assures that the clad thickness is consistent after scalping. Data supporting this claim is available [5].

Preheat

With some Fusion packages conventional heat to roll preheats are used. Other packages where discrete intermetallic precipitation, the dissolution of certain intermetallic phases or the transformation/modification of the cast intermetallic phase is desired, careful attention must be placed on understanding the liquidus-solidus reactions under Non-Equilibrium conditions so that the selected preheat conditions do not compromise the integrity of the clad or core alloy. Eg., localized melting or uncontrolled intermetallic precipitation. A technology to help Novelis produce cast structures with more equilibrium structures has been outlined in a related patent filing [6].

Rolling

Most alloy packages produced with the Fusion Process roll very similar to conventional monolithic sheet ingot. The virtual elimination of the multiple “Kiss” passes required to generate a friction bond between the core and the clad increases productivity and the elimination of clad layer slippage associated with the early passes of conventional clad products.

Although there is no slippage of the clad layer in Fusion casting, there is still the phenomenon of edge “wipe off” common to even unclad ingots where metal near the rolling surfaces extrudes laterally off the edges. Edge wipe off, and the associated mixed alloy scrap has presented its own challenges. From the technology point of view, accurate modelling of the SHRA and clad layer shape close to the edge of the sheet ingot to minimize the Mn content of the Si rich edge trim was an iterative process which generated success early in the process. Alloy segregation of this material at the hot mill always requires attention, and multiple alloy brazing packages creates its own set of challenges when it comes to proper scrap identification and segregation on its way back to the cast house.

Clad/core combinations with radically different work hardening coefficients and flow stress levels create their own set of challenges. In order to get a better understanding of the problem, we refer the interested reader to “Hot Working Guide” [7], a discussion on hot working where detailed hot working process maps are presented for various materials. In this discussion, the author uses a Plane Strain Compression (PSC) test at elevated temperatures with varied strain rates to give an indication of the flow stress compared to true strain indications. As an example, we present for discussion the challenges and solution of our Al-0.5Mg/Al-5.0Mg/Al-0.5Mg Fusion Package introduced as a high strength Anodizable Quality product. Note in Figure 2 that flow stress is reasonably linear to temperature with the Al-0.5Mg and Al-2.0Mg in the typical hot working temperatures (350-500˚C) and strain rates (0.01-0.1 ^-s) which in contrast to the commonly used Al-5.0Mg has a strong work hardening effect below 400˚C and which has an upper hot working temperature of 450˚C due to the presence of the Al5Mg8 under most heat to roll preheats.

Figure 3 shows the fraction of the rolling energy that goes into the clad layer as a function of clad percent. With an Al-5Mg, core it shows that both Al-0.5Mg and Al-2Mg absorb less than a proportional amount of rolling energy. This means that edge wipe off will be more pronounced for either of those than unclad Al-5Mg and Al-0.5Mg will have the most wipeoff. This analysis can be used to predict the severity of the wipe off before rolling. Based on this analysis, we are currently delivering a Al-2.5Mg/Al-5.0Mg/Al-2.5Mg Fusion Package since it has rolling advantages and still offers an acceptable anodizing response.


Figure 2. Figure 3.

Conclusion

The clad thickness can be kept constant from head to butt even considering butt swell. Fusion clad rolls better than roll-bonded clad because it cannot slip between the layers and the rolling is quicker because initial light passes are not necessary. The severity of edge wipe off can be predicted using relative flow stresses at constant strain of the two alloys at rolling temperatures.

References

[1] Graham, M.B., Pruitt, W. and Pruitt, B. H. (1990) R&D for Industry A Century of Technical Innovation at Alcoa, p 144, 145.

[2] Miller, M.A. (1952) Aluminous Brazing Product and Method of Brazing, US Patent 2,602,413 July 8, 1952.

[3] Sturgell, B.W. (2006) Breaking the Barriers: Unlocking New Possibilities for Aluminum, presentation at CRU 11th World Aluminum Conference, Montreal, June 2006.

[4] Anderson, M.D. et al. (2005) Method for Casting Composite Ingot, US Patent Filing 20050011630, January 20, 2005.

[5] Hudson, L.G. et al. (2007) Direct Chill Casting of Heat Exchanger Material Via the Novelis Fusion ™ Process, Light Metals, p 739.

[6] Wagstaff, R.B. and Fenton, W.J. (2007) Homogenization and Heat-Treatment of Cast Metals, US Patent Filing 20070102136, May 10, 2007.

[7] Prasad, Y.V.R.K. and Sasidhara, S. (1997) Hot Working Guide, A Compendium of Processing Maps, ASM International, 92-103.


Billet Sump Modification of Al-Billets for the Prevention of Starting Cracks

Marcel Rosefort a, Thomas Koehler and Hubert Koch

TRIMET ALUMINIUM AG, Aluminiumallee 1, 45356 Essen, Germany

a [email protected]

Keywords: DC-casting, aluminium, starting crack, billet sump

Abstract

Depending on the cast alloy, starting cracks are a common problem while casting aluminium extrusion billets. Hot crack formation is well investigated in theory as well as the formation of the sump shape using different casting technologies. Nevertheless, starting cracks are a problem in DC casting.

Because TRIMET ALUMINIUM AG Essen uses the spout and float technology for DC-casting of aluminium this paper presents an investigation into methods for the prevention of starting cracks in Al-billets using float and spout DC-casting and covers the research methods, the results of crack prevention and the results of implementing these methods. The main topics to be considered are the control of sump formation and the correlation between sump geometry and hot cracks in the billet.

Our experimental research covers DC-casting with varied casting parameters (stop-and-go casting, casting speed etc.), sump depth measurements, ultrasonic tests and metallographic examinations. We discuss crack prevention in the analysis and interpretation sections given consideration of the sump depth measurements. On the basis of these measurements the effect of the different casting technologies and parameters are shown.

Finally, the methods we describe have to be implemented in production. The essential points are the upscaling and the automation of the casting process, especially the start-up phase.

The results of this investigation provide excellent possibilities for crack prevention for DC-casting.

Introduction

Starting cracks make up a substantial part of the scrap rate during DC-casting and therefore generate substantial cost. Thus investigations on crack prevention were started based on the papers of W. Schneider [1] to [3].

In [1] and [2] W. Schneider describes some correlations between sump formation, sump depth and starting cracks. Due to the hot crack character of the starting cracks, Figure 1, varying and controlling the billet sump is an excellent possibility to avoid starting cracks. Amongst others, a new geometry like that pictured in Figure 2b was used. The corresponding sump formation showed no local maximum at the start-up phase, Figure 3. As a result no starting cracks could be found in any billet. So, lowering the residual stresses in the solidifying billet butt and maintaining a homogeneous temperature distribution support the prevention of starting cracks [4].

Thus an easy method was developed to influence sump formation in billets and especially in billet butts. Because TRIMET ALUMINIUM AG Essen uses the float and spout DC-casting technology for Al-billet production, our focus was to apply the examined technologies to the casting implements in Essen. The main objective was to eliminate starting cracks by influencing sump formation at the start-up phase of the casting.


Figure 1. Typical dendritic surface of a starting crack.

a) b)

Figure 2. a) Standard starting block geometry and b) new geometry compared by Schneider [1].

01020304050

60708090100

10 40 70 90 120

160

190

260

320

billet length [mm]

sump depth [mm]

a)

b)

Figure 3. Sump depth formation in billet butts by the use of the starting blocks a) and b), [1].

Experimental Research

The DC-casting technology used in these investigations is float and spout technology. The castings were carried out on an R&D casting machine with the following characteristics:

• teo casting furnaces (2.5t); • 7m casting length; • Fully automated casting process; • Stop-and-go-casting; • Pulsation water cooling; • Impeller cleaning (argon/nitrogen/chlorine); • CF-filter with pore burner.


The data for all castings are:

• Casting alloy: AlMgSi1Cu (EN AW –6013); • Mould diameter: 381mm; • Mould material: EN AW-6082; • Starting block material: EN AW-6082; • Starting block geometry: see Figure 2a.

Reference data based on typical production parameters, especially casting speed in the starting phase, were determined to enable an exact comparison. For this purpose starting cracks with lengths between 150mm and 300mm were generated systematically by lowering the grain refiner content, [3]. Shorter starting cracks are difficult to detect with ultrasonic tests, because of cold shuts or inclusions in the billet butt. If the cracks are longer they do not anneal and the billet will crack over the whole length. For the reference data the grain refiner content was varied between 0 and 5kg/t AlTi3B1. A content of 1.8kg/t AlTi3B1 yields reliable crack lengths of around 200mm to 300mm when using the standard casting technology and parameters. These starting cracks are easy to detect due to their dimensions, Figure 4. All further castings were carried out with this reference grain refiner content of 1.8kg/t AlTi3B1 for perfect comparability.

50mm50mm50mm

Figure 4. Slice of a reference billet with a starting crack.

The investigated test parameters of the casting technology are:

• Stop-and-go-casting in the start-up phase;

• Variations of the casting speed in the start-up phase.

All other casting parameters like temperature, water flow etc., were kept constant for all castings.

The stop-and-go-casting principle means that the table is stopped repeatedly for a short time during the start of casting, see Figure 5. The disadvantage of the stop-and-go technology is the tendency to float or mould overflow. If the stopping times are too long the melt level increases and the float or even the mould will overflow. To avoid this, it is possible to shorten the stopping times or to avoid a complete stop. Instead of a stopping phase, a very slow table speed is used, as shown in Figure 5. Three different variations of the stop-and-go technology were tried to optimise the start of casting without overflowing the float.

a) V1 Stop time: 5s (see Figure 5) Go time: 32.5s Stop-and-go start: 0mm Stop-and-go end: 400mm


b) V2 Stop time: 5s Go time: 180s Stop-and-go start: 0mm Stop-and-go end: 400mm

c) V3 Stop time: 5s Go time: 32.5s Stop-and-go start: 150mm Stop-and-go end: 350mm

Time [s]

Table Speed[%]

Time [s]

Table Speed[%]

Figure 5. Principle of the stop-and-go casting with the parameters used for the table speed at the start-up phase for the first casting tests. Left: The diagram shows the stop-and-go times

and the table speed in percent. Right: the start-up phase of a stop-and-go casting.

The intention of the stop-and-go technology is to remelt the top of the starting crack and to influence sump formation. First experiments with this technology show that it is very effective against starting cracks. At the same time, measurements show that sump depth decreases significantly. Thus it is obvious that the next step should be to test the effect of lowering casting speed at the starting phase without any stop, because a lower casting speed leads directly to a decreased sump depth.

In this second test run, casting speeds at the starting phase were varied, Figure 6. Slower casting speeds yield a decreased sump depth. A lower sump depth is synonymous with a lower temperature gradient, which leads to lower residual stresses. Four different variations were tested:

a) V4 Original casting speed;

b) V5 Slower casting speed at the start-up phase with the same average casting speed like the stop-and-go method V1 (-3mm/min in the start-up phase);

c) V6 Slower casting speed at the start-up phase (-6mm/min in the start-up phase);

d) V7 Slower casting speed at the start-up phase (first two steps –0mm/min, then –6mm/min).


Figure 6. Examples of different casting speeds in the start-up phase (V5 and V7).

For all casting experiments the following measurements were carried out:

• Sump depth measurements;

• Ultrasonic measurements of the billet butts;

• Metallographic analysis of billets slices;

• Metallographic analysis of polished specimen (microscopy);

• SEM analysis of specimen.

Results

Promising results have been found for the investigated modifications in the start-up phase. Particularly outstanding are the results of castings using stop-and-go technology (Version V1) during casting start. For all billets cast with the stop-and-go technology and with the reference grain refiner content (1.8kg/t AlTi3B1) no cracks could be detected with ultrasonic methods nor with metallographic specimens respectively.

To explore the efficiency of the stop-and-go method V1, further tests were carried out. The grain refiner content was gradually lowered below 1.8kg/t AlTi3B1. When using the stop-and-go-technology with a grain refiner content down to 1.4kg/t AlTi3B1 the billets are free from cracks. Using the standard aluminium starting blocks without the stop-and-go-casting this grain refiner content yields completely cracked billets.

All other variations (V2, V3, V5 – V7) were tested only with a grain refiner content of 1.8kg/t AlTi3B1. The crack prevention was only a little bit inferior to that produced using method V1. But there are clear differences in the practical casting process especially concerning the overflow of float and mould. Summarising these results is Figure 7.

All these results show excellent correlation with sump measurements. As expected, a low sump depth and consequent low residual stress works well in preventing starting cracks. This requirement is achieved perfectly when using the stop-and-go technology (V1). Less stop phases (V2) yield an increased sump depth and a later start of the stop-and-go phase (V3) yields a significant increased sump depth at the beginning of the start-up phase, Figure 8. The advantage of stop-and-go variations V2 and V3 is the easier casting start, because the risk of float overflow is minimized to an insignificant level.


0

0,2

0,4

0,6

0,8

1

1,2

0 0,2 0,4 0,6 0,8 1 1,2

Difficulties during start of casting

Efficiency against starting cracks

V4 (Standard)

V1 (stop-and-go)V2 (stop-and-go with fewer stop phases)

V3 (stop-and-go with later stops phase)V5 (slow casting)V7 (slow casting with first steps standard)

V6 (slowest casting)0

0,2

0,4

0,6

0,8

1

1,2

0 0,2 0,4 0,6 0,8 1 1,2

Difficulties during start of casting

Efficiency against starting cracks

V4 (Standard)

V1 (stop-and-go)V2 (stop-and-go with fewer stop phases)

V3 (stop-and-go with later stops phase)V5 (slow casting)V7 (slow casting with first steps standard)

V6 (slowest casting)

Figure 7. Schematic classification of the investigated technologies showing handling (float or mould overflow, float freeze-off) and starting crack suppression.

0

20

40

60

80

100

120

0 100 200 300 400 500 600 700 800 900

billet length [mm]

sump depth [mm]

Stop+Go V1Stop+Go V2Stop+Go V3

Figure 8. Sump depths in the start-up phase for the casting variations V1 – V3 (stop-and-go casting).

A reduced casting speed in the start-up phase without stop-and-go results in development of the sump depth. Following, the main points are visible in Figure 9:

• If the average casting speed is equal to the stop-and-go method V1 (like V5) the sump depth is much greater;

• For a low sump depth a significantly decreased casting speed is necessary; • The sump depth has a local maximum at the very start of the start-up phase.

The decreased casting speed increases the risk of float overflow clearly, while the increased sump depth at the first start-up phase raises the risk of starting cracks.

Figure 10 sums up the changes in sump depth for the most important methods. It is very clear that all tested methods are helpful for lowering sump depth and residual stresses respectively. The most effective method in this comparison is the stop-and-go method V1. It is exceedingly efficient, especially from the very start of the cast.


0

20

40

60

80

100

120

0 100 200 300 400 500 600 700 800 900

billet length [mm]

sump depth [mm]

Stop+Go V1slow casting speed V5slow casting speed V6slow casting speed V7

Figure 9. Sump depths in the start-up phase for the casting variations V1, V5 – V7 (stop-and-go casting).

0

20

40

60

80

100

120

0 100 200 300 400 500 600 700 800 900

billet length [mm]

sump depth [mm]

Standard V4Stop+Go V1slow casting speed V7Stop+Go V3

Figure 10. Sump depths in the start-up phase for the casting variations V1, V3, V4 and V7 (stop-and-go casting).

Further measurements show that there is a second effect caused by the stop-and-go method. It positively influences the shape of the sump. Due to table speed changes the sump has not only a lower depth, but the shape is also flattened in the middle of the billet, Figure 11. This is due to the different solidification progress at the billet surface and in the middle of the billet respectively while stopping the table. The stopping phase immediately influences the solidification speed at the billet surface, while solidification within the billet is decelerated and damped. In addition, this sump geometry can reduce the temperature gradient and the residual stresses in the area of starting crack formation.

Production parameter

Slow casting speed

Stop & Go

Production parameter

Slow casting speed

Stop & Go

Figure 11. Sump depth formation in billet butts by the use of the standard casting technology, the stop-and-go casting and a reduced casting speed respectively.


Conclusions

Compared to other publications, these investigations clearly show that sump depth and shape are very important parameters in the formation of starting cracks. Decreasing sump depth and a flatter sump geometry with a low temperature gradient in the middle of the billet helps to prevent starting cracks. Accordingly, it is recommended that influencing the sump can prevent starting cracks. Starting cracks can be prevented with each of the methods analysed for this paper, but there are significant differences in practice in the efficiency and feasibility of each method.

The best technology is the stop-and-go method with a multitude of stops. It results in a low sump depth and a flat sump geometry that means a significant decreased probability of starting cracks. But a higher casting speed with fewer and later stop phases is much simpler to handle. Thus a compromise between these methods must be made. This area of conflict, Figure 7, is best solved by the stop-and-go versions V1 or especially V3. The stop-and-go technology is relatively easy to implement. It only requires some changes in the controlling software and an adequate, quick hydraulic system for the casting table. Therefore TRIMET utilises the stop-and-go method at production plants. Provided that the starting conditions (especially the casting speed and the stop-and-go times) are properly adjusted, this technology can be easily applied in production at TRIMET ALUMINIUM AG Essen. Additionally, computer-controlled guidance gives perfect reproducibility and can be automated. If there is really no possibility to implement the stop-and-go method, decreasing the table speed in the start-up phase is a less efficient, alternative solution.

Summing up, it can be said that there are excellent and easy possibilities for preventing starting cracks, which are applicable for nearly all DC-casting technologies.

References

[1] Schneider, W. and Jensen, E.K. (1990) Investigation about Starting Cracks In DC Casting. Part I: Experimental results, The Minerals, Metals & Materials Society, TMS Warrendale PA, Light Metals, 931-936.

[2] Schneider, W. and Jensen, E.K. (1990), Investigation about Starting Cracks In DC Casting. Part II: Modelling results, The Minerals, Metals & Materials Society, TMS Warrendale PA, Light Metals, 937-943.

[3] Schneider, W. and Jensen, E.K. (1990) Untersuchungen zur Angußrissbildung an AlMgSi- Stranggussrundbolzen, DGM, 17-30

[4] Ji, M-S. (1990) Einfluss des Eingießvorganges auf Gefügeausbildung und Gussfehler beim Aluminium-Vertikalstrangguss, Dissertation RWTH Aachen.


New Billet Mould Casting Technology

Chris Emes1,a and Richard J. Collins2

1 Mechatherm International Ltd, Hampshire House, High St, Kingswinford, West Midlands, DY6 8AW, UK

2 R J Collins Inc., 5721 N. Forker Road, Spokane, WA, 99216, USA a [email protected]

Keywords: new billet, casting, technology, CBERJC

Abstract

The aluminum billet casting industry is continually looking for reliable, simple and efficient ways to produce high quality, aluminum billet stock for the extrusion and forging industry. Current demands for billet quality requires that the billet have a very thin inverse segregation of less than 250 microns, uniform cell size of less than 30 microns depending on the alloy, average grain size of less than 130 microns, dendrite arm spacing of less than 50 microns and is crack free. The characteristics of the mould technology needed to achieve the required metallurgical quality economically are described in this paper.

Introduction

R.J. Collins Incorporated and Mechatherm International Limited have joined forces to be able to supply the industry with such a system and provide another alternative for billet tooling to the billet casting industry. Features of the moulds and mould table designs include:

• An efficient water distribution box that ensures equal water flow to each mold;

• An efficient hot top metal delivery system that delivers clean metal to each mould and minimizes heat losses from the metal entry to the mould furthest away;

• A metal distribution system that minimizes the turbulence and velocity of metal entering the molds which can cause a metal flashing defect;

• A metal distribution system that minimizes the fill time from when the first mould receives metal to the last mould so that bleed outs, hot butt separation, or cold butt separation defects don't occur;

• A mould design that includes air and oil distribution so that casting is done on a cushion of air instead of contacting the mould wall that would increase the inverse segregation depth;

• A system to measure the graphite rings casting characteristics to ensure the cast defects are not related to the graphite ring condition;

• A mould system that is cost effective with replaceable alignment lugs and simple components designed to give maximum life while keeping operating costs low;

• An air distribution system that ensures the correct amount of air is delivered to each mold during the critical cast start phase of the cast to insure all billets go into air cast mode immediately;

• An air distribution system that during the run phase is automatically adjusted to a low run pressure that will give the very best billet surfaces with little to no bubbling from the mould openings;

• A simple oil distribution system that ensures the correct amount of oil is delivered to each mold that will prevent vertical drags and other types of casting defects;


• A starting head base with self centering starting heads and an alignment system with the mould table that ensures the correct alignment every time a cast is made;

• Comprehensive commissioning, technical service and support with operator training programs that ensures the very best overall recovery, productivity and return on investment.

Water Distribution Table Fabrication

This table must be constructed such that its design will ensure equal water flow to each mold. It is critical that each mould receives water at the correct flow level, pressure and temperature completely evenly around the mould to ensure the uniform solidification of the aluminum and resultant billet quality is achieved. Tables can be designed and manufactured from mild steel, treated and painted, aluminum or stainless steel. Water entry into the table should be symmetrical but can be dependent upon existing connections in the event the mould table is being designed and fitted to an existing machine. Generally, there are four main water connections onto the table each with stainless steel filter baskets.

The fabrication is then welded, stress relieved and precision CNC machined to close tolerances to fit the moulds to ensure reliability and safety. The water box is built to distribute the water evenly to each mould. The moulds are then installed into the bottom of the water box making the mould table assembly complete. The water distribution box comes with removal, gasket-sealed top cover to allow for easy maintenance, inspection and cleaning. Prior to shipment the water box complete with moulds will be shop tested to ensure accurate water flow to each mould and to prevent leakage.

Typically the entry water pressure to the mould water table will need to be about 0.5 – 0.75 bar and each mould can require a maximum of approximately 2.25 litres/cm of mould perimeter.

Molten Metal Hot Top Delivery System

An efficient hot top metal delivery system must deliver clean metal to each mould and minimize heat losses from the metal entry onto the table to the mould furthest away. Too high temperature drop across the table leads to cold defects at one end and hot defects at the other.

The patented hot top metal distribution launder system is manufactured from pre-cast refractory sections installed onto the top of the mould table and ensures that molten metal enters uniformly into each mould under a protective “Hot Top” cover with minimum turbulence or exposure to atmosphere. The refractory used has great insulation properties and its passage ways are narrow to keep the time from metal flowing onto the table from the supply launder to getting into the mould to a minimum. The unique trough designed in an “X” shape allows molten metal to be delivered into the each mould within the “X” section simultaneously this minimises casting defects by reducing temperature losses and metal flow fill time differential between the first and last moulds on the table. This system can also enable less central supply troughing (main metal supply launders) on the table, therefore enabling a greater density of strands and a more efficient pit casting a greater weight per drop.

The trough system can also include an automatic dam lift system that provides the capability of a totally hands free automated cast start and will minimise thermal gradient across the table even further. The system is pneumatically operated and raises all metal dams simultaneously on a common rig mechanism. We would recommend this system for larger tables (for example over 30 tonnes drop weight) where the metal travel distance is longer. It is important that the metal distribution system minimizes the turbulence and velocity of metal entering the moulds which can cause a metal flashing defect. The minimized fill time from when the first mould receives metal to the last mould ensures that bleed outs, hot butt separation, or cold butt separation defects don't occur.


Figure 1. Drawing of mould table refractory “X” type mould metal fill system. Metal Entry to moulds from single central main trough.

Mould Design with Air Cast System

It has long been proven that a mould incorporating a graphite ring through which lubricated compressed air flows at a controlled rate enables a better quality billet with a smooth surface and thin inverse segregation zone, less than 250 microns. The dual part mould design offers this same feature with the graphite ring in the main body of the mould but is easy to take apart for inspection and cleaning. Water jets can block easily and can be difficult to clean, sometimes machining out is required on a single part mould with drilled holes. The whole mould system is cost effective with replaceable alignment lugs and simple components designed to give maximum life while keeping operating costs low.

Figure 2. Two part mould diagram.

Air and Oil Distribution System

The mould design includes an air and oil distribution system so that casting is done on a cushion of air instead of contacting the mould wall. Contacting the mould wall would increase the inverse segregation depth. The graphite ring provides an air bearing that suspends the molten metal from contacting the mold wall while the water flow solidifies the billet upstream of the contact point creating the Air-Cast situation. This results in the very best billet surface quality and a very thin shell zone or inverse segregation.


Oil is distributed to each mould on the table via stainless steel pipe work and is delivered to the moulds via oil injectors that pulse the required amount of oil. The pipe work is fixed to the mould table and is not required to be disconnected for mould removal. The entire system uses the minimal amount of joints and connectors to ensure leak free flow and minimal maintenance. The oil distribution system that ensures the correct amount of oil is delivered to each mold that will prevent vertical drags and other types of casting defects.

Compressed air is distributed to each mould on the table via individual air lines. The air distribution system includes the Accu-Set Mass Flow Mould Calibration System that insures the correct amount of air is delivered to each mould during cast start. Then once each billet is in Air-Cast mode the air pressure is reduced to provide the very best billet surface quality, safety and billet recovery with little to no bubbling from the mould openings. The airflow is preset electronically to the required airflow required to get the logs into air cast mode. Then the air pressure is reduced to run condition so that there is minimal amount of bubbling and very good billet surfaces for the rest of the cast length.

Figure 3. Accu-Set Mass Flow Mould Calibration System.

Graphite Ring Testing

This is a system to measure the graphite rings casting characteristics to ensure the cast defects are not related to the graphite ring condition. It is very important that the graphite ring permeability is uniform all around its surface, or billet defects can result from differing air flow at different points around the ring. It is also important all graphite rings have similar characteristics. The rings eventually deteriorate from plugging of the pores of the graphite ring over many casts. The Accu-TestTM Casting Ring Testing System measures the total flow and uniformity of the graphite rings prior to them being put into a mold table. Then any time a mold is removed from service for any reason, it can be retested and compared to its previous and/or original condition. During testing, if the ring does not meet a certain uniformity and total flow criteria, it is recommended that the graphite ring be replaced. This prevents unnecessary casting of scrap and insures the graphite rings are being changed only when it is absolutely necessary, thereby reducing operating costs.

The starting head base assembly is made up of the starting head base plus the individual starting heads. It is designed to be located on the client’s platen and accurately mate each starting head to its corresponding mould.

The table is fabricated from mild steel plate which is then stress relieved and finished machined flat both top and bottom. The bottom is machined to be able to install and fasten to the client’s platen. The top is machined to locate with the self centering starting heads. All top surfaces of the table are sloped to shed water flow and metal spills. The fabrication is sandblasted and painted with approved safety coatings.


Figure 4. Starting head base table.

The starting heads are manufactured from mild steel and precision CNC machined to provide a snug fit with the mating mould. The starting heads are located to the table in such a way as they are free to move around ¼” (6mm) to allow self centering into the mould.

The starting head base will also include a unitised lift system to ensure proper alignment with the mould table prior to the starting heads engaging. The unitised lift system comprises two alignment pins that engage the starting base just prior to the starting heads engaging into the mould. Once the pair is mated together there are four bolts that connect through the table and thread into the four columns located at each corner.

Commissioning, Training, Technical Service and Support

The performance of the technology depends not only on its design and manufacture but also on commissioning, training and technical support. Training programs have been developed that cover all types of trouble shooting, general operating procedures and maintenance practices to give the most efficient operation, highest production and optimum quality.

Each training module can be coordinated with and supportive of the companies standard work procedures. This gives the very best opportunity for success as illustrated by the following:

• Classroom and “hands-on” shop floor training for casting pit operators, furnace batchers and mold-room personnel;

• Casting pit operator training that includes cast set-up, cast start and end of cast procedures, between cast maintenance and billet defect cause and corrective action;

• Mold room training that includes end of campaign maintenance, table storage, preparation for casting set-up and mold and table troubleshooting procedures;

• Furnace batching training that focuses on the “whys” of the key batching procedures that impact metal quality and casting pit recovery;

• Supervisor or casting co-coordinator training for ongoing support and reinforcement of the basic training. The principle is you only learn a percentage of what you are taught so repetitive training and follow-up is needed to insure success and continuous improvement.

In addition to this, a dedicated team has been set up to deliver spares and consumables as and when required together with on line commissioning and fast site presence of commissioning engineers to evaluate and solve any performance or quality issues.


Macrosegregation Mechanisms in Direct-Chill Casting of Aluminium Alloys

D.G. Eskin1,2, a and L. Katgerman2

1 Materials Innovation Institute, Mekelweg 2, 2628CD Delft, The Netherlands 2 Delft University of Technology, Dept. Materials Science and Engineering, Mekelweg 2, 2628CD

Delft, The Netherlands a [email protected]

Keywords: direct-chill casting, aluminium alloy, macrosegregation, floating crystals, shrinkage

Abstract

This review paper summarizes the results of recent studies on different mechanisms of macrosegregation upon direct-chill (DC) casting of aluminium alloys. In general, the main mechanisms of macrosegregation have been identified quite some time ago as thermo-solutal convection, free-moving crystals, shrinkage- and deformation-induced flow, and forced convection. Despite this general knowledge, the separation of the effects of these mechanisms on the overall macrosegregation pattern and the ratio of their contribution remained largely unexplored. With the advances in computer simulations and in experimental techniques it becomes possible to look at the impact of individual mechanisms in relation to the macroscopic parameters of the transition region of a DC cast billet and to the microscopic parameters of billet structure. Our systematic research helps in interpreting the apparently contradictory experimental macrosegregation profiles reported in literature. Paper is illustrated by own experimental and computer-simulation results.

Introduction

The current understanding of macrosegregation mechanisms can be formulated rather simply [1,2,3]: relative movement of liquid and solid phases during solidification. On phenomenological level, we can distinguish several types of such a relative movement that happens in the sump of a billet during direct-chill casting:

• thermo-solutal convection in the liquid sump caused by temperature and concentration gradients, and the penetration of this convective flow into the slurry and mushy zones of a billet (see Fig. 1);

• transport of solid grains within the slurry zone by gravity and buoyancy forces, convective or forced flows;

• melt flow in the mushy zone that feeds solidification shrinkage and thermal contraction during solidification;

• melt flow in the mushy zone caused by metallostatic pressure;

• melt flow in the mushy zone caused by deformation (thermal contraction) of the solid network;

• forced melt flow caused by pouring, gas evolution, stirring, vibration, cavitation, rotation etc., which penetrates into the slurry and mushy zones of a billet or changes the direction of convective flows.

We know that commercial alloys usually solidify as dendrites, forming overall equiaxed structure in a billet. In the slurry zone (above the coherency isotherm in the transition region) the equiaxed grains are free to move and can travel short or long distances, depending on their size and direction of melt flow (see Fig. 1). In the mushy zone, however, these dendrites form a continuous solid network and have a fixed position in the billet. It can be considered that they move only in the


direction of billet withdrawal and with the casting speed. Liquid flow within the mushy zone is limited to distances comparable to several grain sizes.

Convection-Driven Macrosegregation

One of the most recognized phenomena behind macrosegregation is thermo-solutal convection, or the melt flow driven by temperature and concentration gradients. These gradients exist in the liquid (or more correctly – fluid) part of a casting (billet) due to uneven cooling of the whole volume. In the case of a DC cast billet, the sides are cooled faster than the bottom and there is a noticeable temperature difference between the central part and the periphery of the billet sump. Temperature difference drives the difference in density for the liquid. As a result of this thermal convection, the cooler liquid sinks (we assume here that the liquid increases the density on cooling) at the periphery and creates the momentum that forces the liquid in the centre to rise. In the liquid part of the sump only thermal convection is active (if we neglect the solutal effects brought by washing out of liquid from the slurry zone). As soon as the liquidus isotherm is passed, the partitioning of alloying elements starts to produce the difference in composition and corresponding difference in density, causing the solutal convection.

The main reason why this flow may affect the distribution of alloying elements in the billet cross-section is the penetration of this flow into the slurry zone and washing out of the liquid with the composition already changed by the solidification process. The interaction between the liquid pool and the transition zone of the billet was noted as the main reason for convection-driven segregation by Tageev in 1949 [4]. In addition, the thermo-solutal flow may assist in transporting the solid phase within the slurry region and to the liquid pool (see Fig. 1).

Figure 1 shows that the penetration of the melt flow into the slurry zone occurs in the outer quarter of the billet cross-section. Thus the solute-enriched liquid from this part of the billet is mixed with the bulk liquid and the resultant mixture is brought to the centre of the billet. The result is centreline

Figure 1. A scheme showing the typical melt flow pattern in the transition region of the sump

of a DC cast round billet with liquidus, coherency and solidus isotherm separating liquid, slurry, mush and solid parts of the billet. Possible trajectories of free-floating crystals are

shown. Only half of the billet is shown with the centreline on the left.


positive segregation as shown in Fig. 2a by curve 1. At the billet periphery the melt flow is directed towards the surface (Fig. 1). Here the liquid of the nominal composition penetrates the mushy zone and dilutes the melt that is enriched there by solidification. Hence, a negative segregation at the billet periphery is facilitated (curve 1 in Fig. 2a). Generally, we can conclude that the natural thermo-solutal convection in DC cast billets of aluminium alloys enhances the normal (direct) macrosegregation.

a b

Figure 2. (a) Results of computer simulation of macrosegregation with only thermo-solutal convection (1); with thermo-solutal convection and shrinkage-induced flow (2); same as 2 but with permeability of the mush increased two times (3); same as 2 but with the Scheil model of solidification (4) and (b) experimental results on macrosegregation upon DC casting of a 200-

mm billet with forced centreline flow (indicated on the left bottom axis). Distance along diameter in (b) covers the entire billet diameter with 0 and 200 being at the surface. An Al–

4% Cu alloy is used in all cases.

A rather simple experiment clearly demonstrates the dependence of the macrosegregation pattern on the extent and direction of thermo-solutal convection. We used a screw pump placed in the centre of the liquid part of the billet sump, along the centreline of the billet. Depending on the direction of screw rotation it was possible to impose forced flow aligned or opposite to the natural convection (upward or downward, respectively). By changing the rotation speed it was possible to change the velocity of this forced flow. The results given in Figure 2b show that it is possible to enhance, suppress or reverse the macrosegregation pattern during DC casting. Strong upward flow extracts the enriched liquid from the centre of the billet and promotes negative centreline segregation; while the downward flow forces the enriched liquid to stay and solidify in the centre, effectively eliminating the macrosegregation or reversing it to positive. Structure examination did not reveal any significant changes in the structure and the amount and distribution of free-floating crystals. Hence, the observed effects can be attributed to the convection proper.

Real commercial alloys are always multicomponent. Different alloying elements can have opposite contributions to the density of the mixed melt and, as a result can affect the overall contribution of solutal convection to the macrosegregation [5]. Let us take as an example an Al–Cu–Mg alloy.

Two cases can be considered: the macrosegregation of Cu without (case 1) and with (case 2) taking into account the macrosegregation of Mg. The solutal buoyancy caused by the addition of a second alloying element, which is only present in Case 2, may make some difference in the final segregation pattern. It is expected that due to the negative contribution of Mg to the density in the liquid phase, which is opposite to the contribution by Cu, the extent of (positive) macrosegregation of Cu in Case 2 would be less than in Case 1. Computer-simulation results reported elsewhere confirm this estimation [5]. We can conclude that the addition of alloying elements, in this case –


Mg, will influence the final segregation pattern by its contribution to the solutal buoyancy. It is important to consider the contribution of every solute to the buoyancy force, as they may have opposite contribution to the overall flow.

Shrinkage-Driven Macrosegregation

Historical accounts, summarized elsewhere [2], show that the importance of shrinkage-driven flows for the formation of inverse segregation has been realized as early as in the 1930s. The inverse segregation is caused by the movement of the solute-rich liquid in the direction opposite to the movement of the solidification front. In the case of DC casting, that would be a melt flow directed from the centre to the periphery of a billet (ingot).

Solidification shrinkage is a result of density change during solidification and occurs throughout the entire solidification range. In the slurry zone, however, the solidification shrinkage is easily compensated by the melt flow. There is no pressure difference that may result in the additional flows and the solidification shrinkage does not play any significant role in the relative movement of solid and liquid in the slurry zone, the main influence being exerted by thermo-solutal convection. Deeper into the mushy zone when the permeability is limited and the feeding of the solid phase is restricted, the solidification shrinkage (assisted closer to the solidus by thermal contraction of the solid phase) causes the pressure difference over the solidifying layer of the mushy zone that creates the driving force for the so-called “shrinkage-driven” flow. The flow in the mushy zone, in spite of its small magnitude (velocities are around 10–4m/s or 6mm/min as compared to casting speeds of 100 to 200mm/min), involves highly enriched liquid, which determines its significance for the macrosegregation. It is important that this flow is directed perpendicular to the solidification front.

The horizontal component of shrinkage-induced flow velocity vector takes the solute away from the centre to the surface, though this solute transport physically occurs very slowly. Step by step, however, an overall solute transfer occurs from the centre of the billet to its surface. The depletion in the centre cannot be compensated, as there is no horizontal inflow of the solute from more enriched regions. At the surface, there is a pile-up of the solute as there is no outflow. Because the magnitude of the shrinkage-induced flow is dependant on the shrinkage ratio, one may conclude that the corresponding macrosegregation should depend on the dimensions of the mushy zone and the degree of shrinkage. There is also a clear correlation between the shape of the solidification front and the degree of shrinkage-induced segregation. The deeper the sump, the more solute is taken from the centre of the billet to the periphery [6]. Hence, there is a clear connection to the casting speed.

Computer simulations, using a model that includes solidification shrinkage, demonstrate the high potential of the shrinkage-driven flow in the formation of inverse segregation during DC casting [5, 6]. Figure 2a (curves 2–4) gives clear evidence of this. Note also the effect of permeability that is the function of structure. We can conclude that the shrinkage-induced flow can adequately explain the occurrence of negative centreline segregation.

Free-Floating Grains and Macrosegregation

Movement and sedimentation/growth of solid grains in the slurry zone is sometimes considered as the main mechanism of centreline segregation in DC cast billets and ingots [7,8]. The direct link between the inverse segregation and the presence of free-floating crystals was already rejected in 1940s as it contradicted the cases of inverse segregation in castings where no floating grains had been found. However, the applicability of this mechanism of macrosegregation to DC casting was supported by numerous experimental observations of a duplex structure, characterized by a mixture of fine-branched dendrites and coarse-cell dendrites, in the centre of billets and ingots, e.g. [9,10, 11,12]. The occurrence of such a duplex structure is taken as an evidence of solid-phase transport


(floating) within the transition region (see Fig. 1). Hence, the mechanism of negative centreline segregation by floating grains has been reinstated in 1980s.

The mainstream concept assumes that if the coarse-cell dendrites are solute-poor, fine dendrites (the last to solidify) are rich in solute [7,13].

If we assume that floating grains arrive from other part of the billet and settle in its centre, then they effectively bring there more solid phase than it should be at this point at time and space. As the primary solid in hypoeutectic aluminium alloys is always depleted of the solute, the accumulation of the floating grains has to result in negative segregation.

a b

Figure 3. Dependence of mass fraction of solid on the composition of solid aluminium at the liquid/solid interface during solidification of 2024 (a) and 7075 (b) alloys. Arrows and dots show the minimum concentrations found in different grain types found in

billets cast at 80 mm/min (CC–coarse-cell grain; FC–fine-cell grain; Fragm.–fragment).

We performed electron-probe microanalysis (EPMA) measurements on several coarse and fine cells in the duplex structures found in the centre of 195-mm billets from 2024 and 7075 alloys [14,15]. Both grain-refined (GR) and non-grain refined (NGR) alloys were tested. Line scans of Cu, Mg, and Zn concentrations in both NGR and GR DC cast samples clearly demonstrate that coarser cells are more solute-depleted as compared to the ‘regular’ finer cells. In the case of grain refining, the difference in minimum concentrations is less, probably due to a more efficient back diffusion in the finer structure. The areas occupied by fine-cell grains are close by composition to the average alloy composition or are enriched in solutes, especially in GR alloys cast at a low speed. From these measurements, the minimum Cu, Mg, and Zn concentrations in the centre of dendrite cells were tabulated. These values were then compared with the composition of the aluminium solid solution in equilibrium with the liquid during solidification and the corresponding volume fraction of solid was derived based on the Scheil solidification model using Thermocalc. The results shown in Figure 3 clearly demonstrate that coarse-cell grains, fine-cell grains and less developed dendrites (found in grain refined 7075 alloy) have been formed in different parts of the sump as depicted in Figure 1.

The obtained results and their analysis allow us to make the following conclusions. First, there is a considerable difference in the temperature and, therefore in the stage of solidification (or position in


the sump of a DC cast billet) where the grains of difference morphologies are formed. Second, the coarse-cell grains are formed in the upper part of the solidification range, possibly above the coherency isotherm, which can be rather low in grain refined alloys. These grains contribute to the negative centreline segregation. Third, fine-cell grains and fragments are formed in the lower part of the solidification range, with fragmentation occurring deeper in the mushy zone. Fragments may develop due to coarsening and remelting which occurs during recalescence at the late stages of solidification.

Therefore there are two mechanisms contributing to the negative centreline segregation: shrinkage-induced flow and free-floating grains. It is important to realize that the shrinkage-induced flow is a physical phenomenon that is always present in the transition region of a solidifying billet, whereas the transport of solute-lean solid with its accumulation in a certain part of a casting is a conditional incident, the occurrence of which depends on a number of factors such as the structure evolution, temperature regime, and the direction of strong flows.

Conclusions

Experimental studies together with computer simulations enable one to study the contributions of different mechanisms into the overall macrosegregation picture observed during DC casting of aluminium alloys. Computer simulation gives a unique opportunity to inspect the separate contributions of different mechanisms that are practically inseparable in experiment. On the other hand carefully designed experiments can shed a light on the mechanisms that are yet to be reliably modelled. Convection affects the segregation pattern in dependence on the flow direction. Natural convection in the sump of a DC cast billet enhances the positive centreline segregation. Shrinkage flow facilitates the negative centreline segregation. There is a direct correlation between the geometry of the billet sump (affected by the process parameters) and the degree of macrosegregation. The transport of solid grains in the sump of a DC cast billet contributes to the negative centreline inverse segregation. The overall macrosegregation pattern observed in real billets and ingots is a result of complex combination of different mechanisms, which is affected by the structure and process parameters.

Acknowledgements

The results reported in this paper have been obtained within the framework of research program of Materials innovation institute (www.M2i.nl), project 4.02134. Authors would like to thank Q. Du, R. Nadella, A.N. Turchin, and D. Ruvalcaba (M2i/TU Delft) for their valuable contribution to results and A. Ten Cate and W. Bounder (Corus R&D, IJmuiden) for support and fruitful discussions.

References

[1] Flemings, M.C. (2000) Our understanding of macrosegregation: past and present, ISIJ International 40, 833–841.

[2] Nadella, R., Eskin, D.G., Du, Q. & Katgerman. L. (2008) Macrosegregation in direct-chill casting of aluminium alloys, Progress in Materials Science 53, 421–480.

[3] Eskin, D.G. (2008) Physical Metallurgy of Direct-Chill Casting of Aluminum Alloys, CRC Press, 320 pp.

[4] Dobatkin, V.I. (1960) Ingots of Aluminium Alloys, Metallurgizdat, 175 pp. (citing Tageev, V.M. (1949) Doklady Akademii Nauk SSSR 67, 491–494).


[5] Du, Q., Eskin, D.G. & Katgerman. L. (2007) Modelling macrosegregation during DC casting of multi-component aluminum alloys, Metallurgical and Materials Transactions A 38A, 180–186.

[6] Eskin, D.G., Du, Q. & Katgerman, L. (2006) Relationship between shrinkage-induced macrosegregation and the sump profile upon direct-chill casting, Scripta Materialia 55, 715–718.

[7] Yu, H. & Granger, D.A. (1986) Macrosegregation in aluminum alloy ingot cast by the semicontinuous direct chill (DC) method, in E.A. Starke Jr. & T.H. Sanders Jr. (ed) Proceedings of International conference on Aluminum Alloys – Their Physical and Mechanical properties, Charlottesville, Virginia; USA, 17–29.

[8] Chu, M.G. & Jacoby, J.E. (1990) Macrosegregation characteristics of commercial size ingot cast by the direct chill method, in C.M. Bickert (ed) Light Metals 1990, TMS, 925–930.

[9] Håkonsen, A., Mortensen, D., Benum, S. & Vatne, H.E. (1999) A micro/macro model for the equiaxed grain size distribution in DC-cast aluminium ingots, in C.E. Eckert (ed) Light Metals 1999, TMS, 821–27.

[10] Suyitno, Eskin, D.G., Savran, V.I. & Katgerman, L. (2004) Effects of alloy composition and casting speed on structure formation and hot tearing during DC casting of Al–Cu alloys, Metallurgical and Materials Transactions A 35A, 3551–3561.

[11] Eskin, D.G., Savran, V.I. & Katgerman, L. (2005) Effects of melt temperature and casting speed on the structure and defect formation during direct-chill casting of an Al–Cu alloy, Metallurgical and Materials Transactions A, 2005, vol. 36A, pp. 1965–1976.

[12] Nadella, R., Eskin, D.G. & Katgerman, L. Effect of grain refinement on structure evolution, “floating” grains, and centreline macrosegregation in direct-chill cast AA2024 alloy billets, Metallurgical and Materials Transactions A 39, 450–461.

[13] Dorward, R.C. & Beerntsen, D.J. (1990) Effects of casting practice on macrosegregation and microstructure of 2024 alloy billet, in C.M. Bickert (ed) Light Metals 1990, TMS, 919–924.

[14] Eskin, D.G., Nadella, R. & Katgerman, L. (2008) Effect of different grain structures on centerline macrosegregation during direct-chill casting, Acta Materialia 56, 1358-1365.

[15] Eskin, D.G. & Katgerman L. (2008) Contribution of different grain structures to macrosegregation of aluminium-alloy billets, in J. Hirsch, B. Skrotzki, G. Gottstein (ed) Proceedings of 11th international conference on Aluminium Alloys: Their Physical and Mechanical Properties, Aachen, Germany, Wiley-VCH, 292–297.


Organic Coatings to Prevent Molten Metal Explosions

Alex W. Lowery1,a and Joe Roberts2

1 Wise Chem LLC, P.O. Box 97147, Pittsburgh, Pennsylvania, 15229, USA 2 Pyrotek Inc., 9503 E. Montgomery Avenue, Spokane Valley, Washington 99206 USA

a [email protected]

Keywords: molten metal explosions, organic coatings, wise chem, molten metal water explosions

Abstract

Over 60 years ago the first reported molten metal explosion from a bleed-out during direct chill casting in an aluminium mill was reported. Soon thereafter, testing was performed to determine the root cause of the explosion.

Upon determination of the root cause, an investigation to determine if any preventive measures could be instituted to prevent the explosions was conducted. Results found that a specific organic coating (e.g., Wise Chem E-212-F) prevented molten metal explosions, whereas some specific organic coatings initiated the explosions.

Fifteen years ago the U.S. Department of Energy in conjuncture with the Aluminum Association reinvestigated the root cause of molten metal explosions. Testing revealed that an initiation or trigger had to be present for a molten metal explosion to occur. Testing identified three additional coatings that could afford protection.

Introduction

Whenever two liquids with widely different temperatures come into contact, an explosion can result. This explosion is purely a physical phenomenon. With aluminium there is an additional concern. Because aluminium is a very reactive chemical element that has a strong chemical attraction for oxygen with which it is almost always attached with in nature. Just as aluminium requires a large amount of energy to break the aluminium-oxygen bonds and produce metallic aluminium in a reduction cell. That energy will be released if the aluminium is able to recombine with the oxygen from either water or air. The energy released of one half kilogram of aluminium fully reacts with oxygen is equivalent to detonating 1.4 kilograms of trinitrotoluene (TNT).

2Al+3H2O = Al2O3+H2 + Energy

There are three distinctly different types of explosions that can occur when molten aluminium comes in contact with water. The Aluminium Association has administered a molten metal incident reporting system for the past twenty years. In that system the different explosions are defined as Force 1, Force 2, and Force 3, which are characterized as follows:

Force 1 explosion, also referred to as “steam explosions” or “pops”, occur when molten metal traps water which quickly turns to steam. These explosions are characterized by metal thrown a short distance, usually up to about 4.5 meters and often less than 4.5 kilograms, with minimal or no property damage.

Force 2 explosions are violent steam explosions. As with Force 1 explosion, water is trapped and turns to steam instantaneously. But in this case, the water is trapped by the molten metal and pressure builds up to the point that considerably more metal is thrown a great distance of 4.5 – 6 meters, often to the roof of the plant. There maybe some accompanying equipment damage.

Force 3 explosions are the catastrophic events arising from reaction of molten metal with oxygen from water, air or both. They are characterized by considerable property damage and metal


dispersed more than 15 meters away. Often the metal has disappeared and what remains is a white powder – aluminium oxide. An example of a Force 3 explosion is shown below in Figure 1.

Figure 1. Force 3 explosion on August 20, 2007 at Binzhou Weiqiao Aluminum Company, in China. !ine workers were killed and 64 injured.

Research on Root Causes and Prevention

For more than 60 years, studies of molten aluminium water explosions have been conducted by company laboratories, at government laboratories and at independent laboratories to understand all aspects of molten aluminium water explosions. Some of the earliest tests were performed by G. Long of Aluminium Company of America (ALCOA). G. Long pioneered empirical experimental studies for studying aluminium water steam explosions. In his experiments, various quantities of molten aluminium were poured over coated or uncoated submerged surfaces. Elimination or occurrence of explosions was empirically inferred. Much of what Long’s research is still relevant today and forms the basis for the current prevention of molten metal explosions in casting pits.

Long determined that on certain surfaces such as rusted steel, gypsum, and lime promoted violent explosions. Other surfaces such as polished steel, aluminium and those with organic coatings displayed relative inertness to spontaneous explosions. Subsequent studies found that an organic coating Tarset Standard (TS) was the most practical organic coating at that time to prevent molten metal explosions. Unfortunately overtime Tarset Standard proved to not adhere well to wet concrete casting pit walls. Therefore, Wise Chem E-212-F was brought to market because of its attribute to adhere to wet concrete walls. Subsequent testing showed that Wise Chem E-212-F sufficiently protected against molten metal explosions.

Due to environmental regulations Tarset Standard production was discontinued by the manufacturer. The aluminium industry had one organic coating remaining that was thoroughly tested, Wise Chem E-212-F. The Aluminium Association in conjecture with aluminium companies and coating manufacturers sponsored testing at Alcoa Technical Center. The testing was not conducted to find a replacement for Wise Chem E-212-F, but to provide additional environmentally friendly coating(s) for the aluminium industry.

The standard explosion test was performed utilizing a 028 cubic meter steel open box whereas the prospective coatings were applied to the interior surface. The open topped steel container was filled


with approximately 15 centimetres of water. 40 centimetres above the open container was a clay graphite crucible with 23 kilograms of molten aluminium at 760° Celsius. Metal is released through an 8 centimetre diameter opening at the bottom of the crucible. Refer to Figure 2 below.

Figure 2. Standard explosion test set-up.

The testing had a pass fail criteria regarding occurrence of explosion or omission of an explosion. One particular coating initially passed because it prevented an explosion, but then failed when that particular coating combusted and produced a sizable flames after a few seconds after contact with molten aluminium.

In the past some molten metal-water explosions that were investigated centred around a root cause of a shock that before the explosion. The researchers added a swinging hammer that would provide force to the steel container upon impact to further investigate the role of an impact would have to an explosion. The shock impact test setup is illustrated in Figure 2.

Four coatings passed the standard explosion test: Wise Chem E-212-F, Wise Chem E-115, Multigard 955CP, Intertuff 132.

Larger explosions and some instances of multiple explosions were generated with the shock impact test was performed on bare substrate. Each of the three new coatings that passed the standard explosion set-up test passed the shock impact test to. Each coating was tested only once due to budget constraints. Past plant accidents have illustrate that the attributes of a protective coating can be overridden with a large enough shock during the casting process.


Figure 3. Shock impact test set-up.

Conclusion

Whenever two liquids with widely different temperatures come into contact, an explosion can result. This explosion is purely a physical phenomenon. Molten aluminium water explosions can produce large amounts of energy. For over 50 years the aluminium industries in association with government research institutes have studied the mechanisms of the molten aluminium water explosion.

Studies proved that certain organic coatings were found to prevent the explosions from occurring. Wise Chem E-212-F and Tarset Standard were successfully used throughout the aluminium industry. When Tarset Standard was removed from the marketplace, the aluminium industry began testing to find additional coatings to compliment Wise Chem E-212-F. Three new coatings successfully passed the testing.

References

[1] Epstein, S. G. (1991) Causes and Prevention of Molten Aluminium Water Explosion, Light Metals 1991.

[2] Long G. (1957) Explosions of Molten Aluminium and Water – Cause and Prevention, 107-112 Metal Progress.

[3] Taleyarkhan, R.P., Georgevich, V. and Nelson, L. (1997) Fundamental Experimentation and Theoretical Modelling for Prevention of Molten Metal Explosion in Casting Pits, Light Metal Age 1997.

[4] Leon, D.D., Richter, R.T. and Levendusky, T. L. (2001) Investigation of Coatings Which Prevent Molten Aluminum/Water Explosions, Light Metals 2001.

[5] Guidelines for Handling Molten Aluminium, Third Edition.


The Supply of Casthouse Equipment and Empowerment of Knowledge to Cast Aerospace Alloys the Almex Way

Shaun Hamer

Almex USA Inc., 6925 Aragon Circle, Buena Park, California, 90620, USA [email protected]

Keywords: casthouse, hard alloy, casting technology, billet tooling, degassing

Abstract

For well over a decade, Almex USA Inc. has been supplying hard alloy casting technology to the aluminum industry. Today, customers on five continents produce 2000 and 7000 series aluminum billets using this technology in sizes ranging from 75mm through to 1080mm in diameter. In order to produce such a range of sizes in hard to cast alloys and to aerospace standards, Almex has imparted complete casthouse process understanding and know-how, along with equipment supply to its customers. This paper provides an overview of Almex’s hard alloy casthouse technology and describes some of the critical parameters which must be controlled throughout the process route in order to produce defect free alloys for forging, extrusion and flat rolled applications. Explanation is also made of the microstructural requirements of the as cast product and how these influence the final product; along with the range of quality control solutions supplied by Almex to ensure these requirements are met on a continuous and repeatable basis by Almex’s technology users.

Introduction

Casting 2000 and 7000 series aluminium alloys in larger diameters and to aerospace specifications has been a well guarded market for a handful of industry players for several decades. Through recent industry mergers and acquisitions, the number of players has further reduced to a level where end users have just two or three options if they need to acquire hard alloy billet for further processing. The result of this is twofold; the remaining hard alloy producers benefit from generous premiums simply based on supply and demand economics. Secondly, downstream industries are prevented from being established because of high raw material costs and extended delivery lead times for inventory metal. On the face of it, it appears that aluminium producers can easily benefit from a growing high premium market by converting existing, underutilized capacity to production of 2000 and 7000 series. However, the lack of successful working examples of this proves otherwise. Typical limiting factors include scrap segregation, furnace contamination, equipment limitations, casting process complications and changing business models from a soft alloy, high volume mindset to a hard alloy, total quality vision.

A model which is gaining in interest is the philosophy of the small, dedicated hard alloy casthouse. With this strategy, the entire casthouse, business model and management structure is focused on the production of high strength aluminium alloys and dedicated to sales and market growth in that specific sector. Through growth of these small footprint, fast ROI facilities, Almex’s global vision is to help companies in both established and developing regions establish and develop a high strength aluminium alloy production path. This domestic supply of otherwise hard to source metal can create indigenous downstream industries and supply chains and build a closed loop production circle where downstream scrap is reprocessed rather than being sold offshore to overseas scrap merchants.

The Business Model

The hard alloy casthouse business model does not follow the same strategies as those for rolling ingot or commercial billet production. In the latter, production volume and cost of manufacture are the predominant considerations in order to remain competitive in these mature markets. Product


characteristics such as extrudability, repeatability, alloy conformity, quality and metal cleanliness will all gain a premium with educated customers, but the added cost of manufacture has to be very carefully controlled to maintain sufficient margin while focusing on this additional service to the client.

Conversely with hard alloy sales, total, repeatable metal quality is the customer’s benchmark for acceptance or rejection of the as cast material and without this a hard alloy casthouse will not survive. Accordingly, the business model for these facilities is a total quality model where production, consumable supply, manpower resources and operating practices are based on guaranteed quality of metal leaving the shipping dock.

It is rare that a company can implement these two very differing business scenarios under one roof without taking compromises which will affect both production paths. This is especially true if a casthouse is connected to a smelter facility. The primary function of the casthouse in this case is to keep up with the cell production and ensure the casthouse is not the bottleneck in the process. It is critical that the pot lines are well controlled for the production of metal with low silicon, low iron and low alkali metal content. However quality control of the chemistry of the hot metal transfer to the hard alloy line is extremely difficult to regulate and certify from heat to heat.

Accordingly, the only true path for senior management to take if considering a hard alloy casting capability is to make both the facility and the management structure a separate business entity to the soft alloy and / or primary production business. Almex’s Business Strategy Division, through over a decade of operating in these specific markets can assist company management in business planning, market assessment, investment planning and facility planning.

The Production Route

The Almex hard alloy billet production route is configured for the production of billet for both extrusion and forging applications. This concept provides strategic benefits for the casthouse owner in the following ways:

• The plant footprint can be minimized as production volume can be managed through lower drop tonnages;

• The casthouse can be based on attractive scrap buyback economics, permitting use of high scrap to prime ratios;

• All incoming materials can be approved prior to shipment and quality controlled upon receipt and rejected if necessary;

• The same casting line can cast extrusion and forging quality billet from 75mm to 1080mm in diameter with simple tooling changes;

• Downstream industries can be established relatively quickly through investment in extrusion presses or open / closed die forging presses, stimulating indigenous market growth;

• Hard alloy extrusion scrap ratios are typically 50%, generating a profitable domestic production cycle and complimenting environmental regulations;

• The as cast billet can be sold into the global marketplace;

• A growing number of applications exist in the aerospace, transportation, sporting good and other market sectors where high strength to weight ratio characteristics are required.

The production route is similar to any secondary plant operation with melting and holding, metal purification, filtration, hot metal transfer, direct chill semi continuous casting and homogenization. However, for hard alloy production, each step in this process requires far greater attention to detail and control than required for the production of conventional alloys.


Figure 1. Almex Hard Alloy casting line in India for production of 2000 and 7000 series aluminium alloy billets in the range from 152mm to 1080mm.

Furnace Control

Furnace charging begins with the correct specification of the charge to be loaded into the furnace. Primary aluminium with correct iron / silicon ratios is critical for 2000 and 7000 series alloy production in order to stay in specification in the final chemistry analysis. The primary metal must also have a chemistry map showing limited alkali metal content (sodium, lithium, calcium) and rare metal content. Although the alkali metals can be removed through metal purification processes, the alkali metal salts produced require significant additional processing to remove them from the melt prior to casting. Remaining salts captured in the as-cast product will lead to inclusion formation in the as cast product and ultrasonic rejection or undesirable properties in the final product. If these undesirable elements form a sufficient volume of the heat, then in-furnace systems may be required which add significant additional capital and operating cost to the process and also increase melt loss percentages.

The selection of hardeners, grain refiners and alloying elements used must be of correct purity and structure to prevent the introduction of inclusions and ensure total solution of the elements and prevent clusters, agglomerates and nucleation zones which will cause grain growth in the as cast metal. This is again an area which is considered to have potentially considerable cost saving by using low cost solutions; however, reject and scrap rates as a result of low quality additions are common and it is critical that additives with repeatable and high quality are used at all times.

Scrap is a valuable charge for production of 2000 and 7000 series alloys. As a rule, the clean, sorted scrap has reduced alkali metal content and, if well processed, contains limited salts which can be easily be removed in subsequent metal purification. Scrap charging as with all secondary remelt plants is an important technique to master to ensure correct chemistry, minimize melt losses and ensure casthouse safety. Several types of molten metal pumps and stirring devices introduced over the past few years provide rapid submersion for chips and light gauge scrap. However, operating costs for some of these devices are high, so care must be taken in the selection process.

Furnace selection, charging and production planning are also critical steps in optimizing production and ensuring alloy specification. Complications such as poor combustion characteristics, flame impingement locations, freeze planes, refractory sweating, burn off, oxidization and alloys with a wide range of melting points are all characteristics that must be considered in equipment selection and the melting and holding process must be mastered by the furnace operators in a hard alloy plant.


Metal Transfer

Metal transfer to the casting line is a further area where metal quality and conformity can be affected. Level pour of metal from furnace exit to the casting table is critical, where the metal is fed on the same plane by gravity to ensure a laminar flow. Any cascading or turbulence of metal will create a rupture in the protective aluminium oxide surface layer, resulting in increased hydrogen absorption into the metal and potential burn off of oxidisable elements such as magnesium.

Correct refractory preheating methods are also important to prevent damage of the refractory and ingress of hydrocarbons which can later be released back into the melt. Correct refractory surface preparation with non wetting agents is similarly important to maintain refractory life and prevent erosion of the joints which can cause carry through of solid inclusions into the as cast product. Heat loss through the casting line is also critical as the greater the degree of heat loss, the more initial temperature is required in the melting furnace. Accordingly, the line layout and correct launder materials and cross sections are critical as higher furnace temperatures result in higher operating costs from additional fuel consumption and increased hydrogen absorption into the metal.

Metal Purification

On line metal purification, grain refinement and filtration are critical in the production of 2000 and 7000 series alloys for critical end use applications. The function of the metal purification systems for these alloys are:

• Reduce hydrogen levels to the region of 0.08 to 0.10cc/100g Al;

• Reduce metal salt content to values of 2ppm or better;

• Remove alkali metals;

• Remove solid and liquid metallic and non metallic inclusions;

• Provide controlled grain refinement potency;

• Modulate the downstream temperature control.

To ensure optimum metal quality all of the metal purification devices must operate as a synchronous system, tuned to the specific alloy and casting rate. Grain refiner addition is the first step in the process where the specific grain refiner composition is alloy specific. The use of a high quality grain refiner rod is critical as the titanium performance as a modifier of grain growth during solidification can vary greatly, especially in lower cost rod. Poor quality grain refiner, incorrect addition rates and incomplete dissolution into the metal can also lead to low pit recoveries through hot and cold cracking of the billets, inclusions in the as cast product in the form or stringers, and other defects as well as non uniform and undesirable grain structure.

Degassing is a critical part of the metal purification process. In addition to lowering hydrogen levels in the metal, the degassing unit must provide alkali metal, alkali metal salt, solid and liquid inclusion removal, and be able to modulate the temperature of the liquid metal as it passes through the reaction vessel. The basic principle of any degassing unit is to remove hydrogen through contact with argon gas and the greater the contact and higher the contact time, then the greater the amount of hydrogen removal. Accordingly the vessel design, sizing and gas introduction methods are all critical factors in selection of a high quality system. However, even if a quality, well sized system has been installed, the use of a high quality of process gas is critical for optimum hydrogen removal. Unless high quality cryogenic inert gas is used then hydrogen removal efficiencies will be affected by the moisture contained in this process gas. Generally, the highest quality bulk gas supplied contains oxygen and moisture levels down to 2ppm each and this should always be specified as the gas requirement to the supply companies. The change from argon to lower cost nitrogen is always a debate from a financial perspective. However, especially when combining the


process gas with a halogen such as chlorine, the nitride formation generates further unacceptable impurities in the metal which need to be removed to avoid product rejection downstream so is not an acceptable alternative for high quality metal production. Further cost cutting by installing a site based nitrogen plant is acceptable, but this equipment must be maintained in optimum condition to ensure that the nitrogen cover gas over the surface of the metal used in the process is maintained with a total maximum of 6ppm impurities in the form of oxygen and moisture.

Utilization of a degassing system with in situ heating capability is extremely important. With controlled heating capability, the degassing unit can be used to compensate for initial temperature losses in the casting system until steady state casting conditions are reached. This capability negates the need to raise furnace temperatures to specific start temperatures, hence minimizing furnace fuel consumption and hydrogen entrainment into the metal. As the volume of metal in the degassing unit is comparatively small compared to the furnace charge, then the total energy consumption for this temperature rise is much smaller in comparison. With the metal in the degassing unit being controlled in an inert environment and the heat being supplied indirectly, then there is no additional hydrogen increase in the metal created by this raising of temperatures.

Finally, prevention of salt carry over into the metal, inclusion removal and design of the metal path which ensures that only the well processed, cleanest metal passes through to the casting machine are all characteristics of the design of a high quality degassing system. As aerospace quality metal must reach stringent ultrasonic cleanliness levels (USA MIL-2154 Class AA), the selection of the correct degassing system is critical.

By utilization of the correctly sized and designed degassing unit, the choice of filtration system becomes much simpler. Through reduction of inclusion content with up to 95% efficiency in the degassing unit, then final inclusion removal can be adequately achieved for most products by use of a ceramic foam filtration system. Only for exceptional critical end use products (for example, can stock and thin gauge foil) is it necessary to add more expensive filtration alternatives to provide the necessary metal cleanliness assurances. Systems such as deep bed filters and cartridge filters provide the necessary performance guarantees to assure ultra low inclusion levels, but these systems are higher priced and have greater operating costs and operating and maintenance complexity. Proper maintenance, operation and filter plate selection is however important for the optimal efficiency of CFF filter systems. Additionally, new generation filter plates are now available on the market from certain suppliers which are more impervious to the highly alloyed metals used in hard alloy production.

Casting Machine and Tooling

The casting machine control is at the heart of successful billet production. The upstream control is critical in imparting the correct chemistry and properties into the metal, but this is futile unless downstream quality billet can be produced to meet customer specifications with repeatedly high pit recovery and correct micro and macro structural properties. Several techniques are used in combination to meet these requirements which involve close and accurate control of the casting cylinder, cooling water conditions, metal feed, mold design and upstream system control in synchronicity. Typical issues in hard alloy casting and are tabulated below with typical process requirements to control the issues:

To achieve the necessary casting control, the casting cylinder and associated control systems require significantly greater precision than the equivalent systems used for soft alloy casting. Utilizing a double acting, self aligned cylinder the control sensitivity required at the start-up of a cast when low applied loads and most accurate motion control are required is possible. This configuration enables the necessary response from the cylinder to match the automation system’s demands and eliminates stiction which is a characteristic problem when controlling hydraulic cylinders under low load and low pressure regimes. The cylinder motion control via a closed loop system with a servo valve


controlling oil flow from the cylinder, an ultra sensitive flowmeter measuring the exact oil flow, and a dedicated PLC control loop providing the necessary integration of flow values to provide the servo valve with the error signal ensure the exact casting speeds are achieved throughout the length of a drop. For systems with a wide range of billet diameters and alloys, the casting speed control range can be from 170mm/min to just 13mm/min during steady state, so the control system and associated devices and actuators must have the capability of accurate control throughout and beyond this range.

Table 1. Typical casting challenges and solutions.

Casting Challenge Solution Excessive Hot Cracking Stop and Go start up procedure Excessive Cold Cracking Wiper technology and dry DC Casting In-Situ Inclusion Pick-Up Level Pour metal feed with no turbulence

Chromium and Zirconium Segregation Temperature and casting practice controls Ti-B Agglomeration Controlled inoculation practices Liquation Band Depth Very short mold length design and limited mold tang Macro Segregation Casting speed and metal feed control

Gas Porosity Correct degassing control Shrinkage Porosity Casting recipe control

Cooling water control is of greater importance when casting 2000 and 7000 series alloys. Flow rates, water pressures, water temperatures, water quality and water pH levels all have a significant influence on the successful casting of these alloys. Dedicated control systems are therefore required to maintain all of these parameters within limits and automatically control or alarm when out of tolerance values are reached. Without this control, issues such as increased liquation band, sub surface defects, cold cracking, hot cracking and bleed-out can all easily occur. Additionally, internal stresses within these billets can lead to cold cracking even after the termination of a cast. To minimize these internal stresses, water wiper technology and pit water level control is employed.

Stresses leading to both hot and cold cracking are also minimized by tooling design in conjunction with the automated recipe systems used to control the entire casting system. Tooling needs to be robust and configured so that the cooling water impingement angle and location, the wiper performance, and heat extraction characteristics of the starting heads are all carefully designed to optimize the heat transfer properties required for casting alloys with wide freezing ranges in excess of 100ºC. In spite of the closely controlled parameters of the tooling, it is critical that the tooling is designed for ease of operation and maintenance to maintain optimum performance from the tooling and ensure maximum casting safety at all times.

Automation and Process Control

In order to ensure casting success, repeatability and maximum pit recovery, casting recipes covering the processes detailed in this paper are critical. Today’s automation systems provide the capability of controlling and monitoring of field sensors, actuators and transducers and provide user friendly, intuitive and icon driven Human Machine Interfaces. However, for hard alloy success, all of the line parameters from furnace exit to casting machine must be controlled in this fashion and where direct human control and interface is required then strict procedures must be adopted to ensure total process repeatability. The following table identifies the critical parameters for successful casting of 2000 and 7000 series alloys, although this is by no means an exhaustive list:


Table 2. Casting parametric table.

Parameter Fixed Variable Recipe Controlled

Almex Specified Parameter

Almex Specific

Know-How Mold Design ü ü ü

Starting Head Design ü ü ü Wiper Location ü ü ü

Cooling water Flow & Pressure ü ü ü ü Water Temperature ü ü ü ü

Water Chemistry & Quality ü ü ü ü Casting Speed ü ü ü ü

Stop and Go Control ü ü ü ü Casting Speed Ramps at Start-Up ü ü ü ü

Metal Level ü ü ü Alloy Chemistry ü ü ü

Grain Refiner Addition ü ü ü ü Metal Temperature ü ü ü

Hydrogen levels in Metal ü ü ü Alaki Metal Content ü ü ü Alkali Salt Content ü ü ü

Conclusions

Low volume, small footprint casting lines for casting hard alloy to aerospace standards have been shown to be a viable opportunity for secondary aluminium producers. However to make these installations a qualified success, the management philosophy, equipment supply and operating practices all need to be aligned with the total quality and total process control philosophies. It is clear that tooling and equipment design, no matter how important, is only a part of the overall requirements of successful casting of hard alloy to aerospace forging and extrusion quality and understanding and attention to detail of all aspects of the process need to be understood, embraced and controlled with total repeatability.

With several complete installations on three continents, Almex have opened the process windows for companies to enter the business of producing high strength to weight aluminium alloys with complete casthouse equipment and know-how supply. By developing this hard alloy casting capacity in both developed and developing countries, it offers the possibility of indigenous growth of downstream industries which have previously been constrained by the price and supply lead times and reliability of overseas supply of these hard alloy billets.


Grain Refinement and Hot Tearing of Aluminium Alloys - How to Optimise and Minimise

Mark Easton1,a, David StJohn2 and Lisa Sweet3

CAST Cooperative Research Centre

1 Department of Materials Engineering, Monash University, Clayton 3800, Australia 2 Division of Materials, The University of Queensland, Brisbane, Australia, 4072

3 Rio Tinto Alcan, Edgars Road, Thomastown 3074, Australia a [email protected]

Keywords: grain refinement, hot tearing, casting, 6000 series alloys

Abstract

Grain refinement and hot tearing are important key factors affecting the quality of castings. There have been substantial advances in the understanding of both of these phenomena over the last two decades. The paper discusses strategies for obtaining the lowest cost grain refiner addition and provides an explanation for how the refinement of equiaxed grains leads to a reduction in hot tear susceptibility. However, it also provides a warning that adding more grain refiner may not be better for reducing hot tear susceptibility. Alloy factors affecting hot tearing are also discussed. Finally, a list of six key considerations is provided to help casthouse and foundry engineers when trying to optimise grain refinement and reduce hot tearing.

Introduction

Optimisation of grain refinement and reduction of hot tear susceptibility of alloys are key factors that need to be considered in the production of metal castings, whether they be continuous cast billet, ingot or shape castings. These are also intrinsically linked as grain refiner additions are often made to reduce the hot tear susceptibility of alloys. There has been considerable improvement in the understanding of the mechanisms of grain refinement and the manner in which grain refinement affects hot tearing. There has also been progress on other alloy factors that affect the hot tearing resistance of alloys. The purpose of this paper is to bring together all this information into a coherent framework so that grain refiner additions can be optimised for cost and for performance and key factors can be identified that may affect hot tear susceptibility.

Optimising Grain Refiner Additions

Reviews of grain refinement practice [1] and methods for reducing the cost of grain refiner additions have been discussed in detail elsewhere [2,3]. This section will provide an overview of the important considerations and developments over recent years.

The grain size of alloys, d, is controlled by the addition of potent nucleating particles (low nucleation undercooling ∆Tn), the number of those particles, v, the proportion that are active, f, the solute content of the alloy, Q, [4] and the cooling conditions such as cooling rate and superheat when casting into cold moulds [5,6]. A generic form of an equation for grain size of wrought alloys for a particular set of casting conditions is

QTb

fd n

v

∆⋅+=

3

1

(1)

where ( )∑ −=i

iiil kcmQ 1,0, , and ml is the liquidus gradient, co is the composition and k is the

partition coefficient for each element i in the alloy and b is a fitting factor.


The normal titanium boride grain refiners such as Al-5Ti-1B contain both TiB2 particles that act as nucleation sites and Al3Ti particles that dissolve supplying Ti solute to the alloy. Ti solute has a very large m(k-1) value (~240) compared with most other alloying elements (usually 2-6) [7,8], which means that very small additions dramatically increase the Q value especially in lean alloys, such as 1000 series alloys or the soft 6000 series alloys [9]. It has previously been shown that substantial savings to the cost of grain refiner can be achieved by the addition of small amounts of Ti to the alloy in the furnace (up to 0.02wt%) which can be used to reduce the addition of rod grain refiner [2,10]. It should be noted that in Al-Si foundry alloys, where Ti is routinely added, the addition of Ti has little effect on grain refinement because of the high Q value of the base alloys [11,12], although Ti has been shown to have a moderately beneficial effect in alloy 319 [13]. Additions of boride based refiners are much more effective as they contain nucleant particles [14].

TiB2 particles are the most common nucleant addition for grain refining purposes although TiC particles are used for some specialty products and other borides have been used to grain refine Al-Si foundry alloys. Whilst only a small percentage of TiB2 particles actually act as nucleating substrates [4,15], there are a number of things that can poison the grain refining effect. Firstly, if the Ti:B ratio is less than that required to form TiB2, i.e. a weight to weight ratio of 2 then grain refinement will not occur (Figure 1a) [9]. This is because the excess boron is able to combine with other elements coating the TiB2 surface and rendering the nucleants significantly less potent. In Al-Si foundry alloys, AlB2 particles have been shown to be more effective than TiB2 particles [16,17], however anecdotal evidence suggests that in practice Ti is required for these grain refiners to be effective [1]. Figure 1b) shows that a very fine grain size can be achieved in an Al-7Si-0.3Mg foundry alloy with the addition of B and no Ti, but the grain size is more variable than when the Ti:B ratio is above 2. This may also be related to the problem of excess boron, as AlB2 is a potent grain refiner but has a relatively low stability compared with TiB2 [18]. Hence sometimes boron additions work well, and other times it does not depending on which minor elements are present. TiB2 is a good combination of being a stable phase with relatively high potency (only ZrB2 is more stable which is probably why Zr poisons grain refinement [19]).

0

200

400

600

800

1000

1200

1400

1600

1800

2000

0 2 4 6 8

Ti:B ratio

Gra

in S

ize

(µm

)

1050

2014

3003

5083

6061

Ti:B ratio in TiB2

columnar equiaxed

0

200

400

600

800

0 2 4 6 8 10

Ti:B ratio

Gra

in S

ize

(µm

)

Ti:B ratio in TiB2

(a) (b)

Figure 1. Graphs showing changes in grain size with the Ti:B ratio. (a) wrought alloys [9] and (b) an Al-7Si-0.3Mg foundry alloy.

Another important consideration when trying to minimise the cost of grain refiner additions is whether scrap or returns add still potent nuclei into the alloy. This is of particular importance in remelting secondary alloys, but can be an issue in primary operations with moderate to large amounts of scrap. Opticast [20] has developed a technology that can reduce the addition of more


expensive grain refiner based on knowing the residual grain refinement efficiency of the melt. Further optimisation of grain refiner additions can be achieved by appropriate placement of the rod feed so that good dispersion of TiB2 particles is achieved in the launder before casting.

Minimisation of Hot Tearing

One of the key reasons for the addition of grain refiners is to reduce the hot tear susceptibility of alloys. However, it is not known how much is enough. In practice the aim is to keep the grain refiner addition to a minimum whilst retaining good hot tear resistance. It is clear that the most important thing is to generate equiaxed grains rather than columnar grains [21], but it is generally believed that more grain refiner is better although there is evidence that too much grain refiner can in fact lead to an increase in the hot tear susceptibility of alloys [22].

Hot tearing is a very complex area to study and whilst some effects are well documented, e.g. the lambda curve [23, 24], where the addition of an alloying element increases the hot tear susceptibility to a peak and then decreases again with an increase in the non-equilibrium solidification range, a good explanation for the beneficial effects of grain refinement on hot tearing has only recently been forthcoming. Grain refinement is known to change a number of important factors that affect hot tearing including the ability of the melt to feed the solidification shrinkage through the semi-solid zone, i.e. melt permeability [25], and the load development during solidification [26] and the capillary forces [27] that bind the almost fully solid dendritic network together. The most recent expression of this approach to hot tearing is known as the granular model, which considers the formation of liquid films around grains and grain clusters near the end of solidification [28].

Grain Refinement and Hot Tearing

To begin to understand how grain refinement affects hot tearing it is important to understand the hot tearing process. A schematic of many of the important features is given in Figure 2. As the alloy solidifies, applied strain develops due to differential thermal contraction. If this shrinkage is not fed by the liquid then a void may form in the region between where feeding begins to be constrained, fs,0, and the point at which the dendrites are bridged by solid bonds (which may occur before, during or after the formation of eutectic phases). The region most vulnerable to tearing is just before solid bonding occurs, as this is the most difficult area to feed and the structure is still being held together by capillary forces rather than solid bonds.

Flow to feed shrinkage and applied strain

Eutectic

fs=0

fs,co Solid bonding between dendrites

fs,0 Impingement of dendrites

Applied Strain from differential thermal contraction

Vulnerable region

Figure 2. Schematic of the important factors affecting hot tearing. Modified from [27].


To understand these factors more fully, the effect of the addition of grain refiner on the grain morphology is required. An unrefined alloy normally consists of very large equiaxed or columnar grains which are highly dendritic (Figure 3a). The addition of a small amount of grain refiner decreases the grain size, but the dendrite arm spacing (DAS) remains relatively constant and hence a dendritic equiaxed grain morphology is achieved. The addition of more grain refiner continues to decrease the grain size, whilst the DAS remains constant and the grain morphology changes from being equiaxed dendritic to being more rosette-like (Figure 3b). Adding even more grain refiner leads to a spherical grain morphology (Figure 3c). Hence, as well as changing the grain size, the grain morphology also changes.

(a) (b) (c)

Figure 3. Optical micrographs using polarised light of 6060 (a) without grain refiner additions, (b) grain refined with 0.01%Ti added by Al-3Ti-1B and (c) with a 0.02%Ti added

by Al-3Ti-1B with approximately 0.05%Ti added as solute. All samples were cooled at 1°C/s. After [29].

The grain size, DAS and grain morphology can all affect hot tearing [30]. One of the most accurate hot tear models for hot tearing was developed by Rappaz, Drezet and Gramaud (RDG) [31], and was modified by Grandfield and others [27] to include the case of equiaxed dendrites. The model provides a mathematical description of the environment illustrated Figure 2. Grain refinement effects can be incorporated into the RDG model in three ways [30]:

1. Changing the permeability length scale from the secondary dendrite arm spacing (DAS), in the case of columnar or equiaxed dendritic grains, to the grain size, in the case of globular or spherical equiaxed grains;

2. Changing the size of the permeable region, i.e. fs,0 and fs,co; and

3. Changing the capillary pressures by changing the liquid film thickness between grains. Smaller grain size implies thinner liquid films between grains at a given fraction solid and therefore greater capillary pressures to be overcome before a tear propagates.

Reducing the grain size increases the capillary pressure and hence reduces the hot tear susceptibility according to point 3. Reducing the grain size delays the point fs,0 which improves permeability and should reduce hot tear susceptibility, but it also appears to delay fs,co which would have the opposite affect so it is unclear whether grain refinement benefits point 2 or not. Reducing the permeability length scale reduces the permeability of the mushy zone and increases hot tear susceptibility. Hence, when the grain size is reduced whilst the grain morphology is still dendritic, it will have little effect on the permeability. However, reducing the grain size when the grain size is spherical will decrease the permeability and increase the hot tearing tendency. Hence, reducing the grain size has an effect on a number of competing factors that can result in an increase or decrease on hot tear susceptibility.


Figure 4. Predictions for the effect of the grain size on the hot tear susceptibility (HCCe) of alloy 6060, for globular grains (d) and for dendritic equiaxed grains (λ2) assuming dendrite

arm spacings of 50 and 100 µm. After [29].

Calculations performed on 6060 in which the permeability is controlled by either the DAS or the grain size are shown in Figure 4 [29]. It is seen that if reducing the grain size is the factor that controls the permeability then the hot tear susceptibility increases as the grain size decreases. However, when DAS controls permeability then the hot tear susceptibility decreases as the grain size decreases. When considering the grain morphology changes with grain refiner additions as shown in Figure 3, it suggests that whilst the grain morphology is dendritic reducing the grain size will decrease the hot tear susceptibility. However, if the grain morphology is spherical then a reduction in grain size may in fact increase the hot tear susceptibility of an alloy. This argument provides a theoretical understanding for the observation that the hot tear susceptibility increases once again when even higher additions of grain refiner are made [22]. The important practical point here is that too much grain refiner, as well as being costly, may also increase the prevalence of hot tearing during casting.

Alloys and Hot Tearing

Alloy composition also affects hot tearing. The well characterised lambda curve shows that there is a peak in the hot tear susceptibility which then decreases again with the addition of more alloying element. The height of the peak is also related to which element is the dominant alloying element with elements that have a lower eutectic temperature, leading to a larger freezing range, generally being more prone to hot tearing. For example, Al-Sn alloys are particularly hot tearing prone [32, 33] with a eutectic temperature of 228°C. However, this rule of thumb does not always hold as Al-Mg alloys are less susceptible to hot tearing than Al-Cu alloys [32], despite the eutectic temperatures being 450°C and 548°C respectively. Hence, there are other factors that affect the hot tear susceptibility of alloys other than freezing range but it is clear that small amounts of low melting point phases are detrimental to the hot tearing resistance of alloys.

Multi-component alloys also show peaks in hot tear susceptibility with alloy content, e.g. the Al-Si-Mg system (Figure 5). Once again this does not appear to present the full story, as generally 6005 is cast slower than 6082 which is cast slower than 6063 to reduce the prevalence of hot tearing [34], which is contrary to the observations in Figure 5. There has been some work recently on the role of minor elements on hot tear susceptibility of alloys [35, 36]. Surface active elements such as Na are also considered to have a detrimental effect on the hot tear susceptibility of alloys [36]. Considerable further work is required to understand the role of minor elements on hot tearing.

10

100

1000

10000

0 100 200 300 400 500 600d (µm)

HC

Ce

(s)

λ2=50µm

λ2=100µm

d


6063

6061 6082

Si (wt%)

Mg

(wt%

)

132

11

499

452

339

277

0.2

0.6

0.8

1.0

1.2

0.4

132395

6060 6005

67

266 502

528 549

236

149

240

28

232

61

0.2

0.4 0.6 0.8 1.0 1.2 1.4

50 100

400

200

500 300

Figure 5. A map of the hot tear susceptibility of ternary Al-Si-Mg alloys based on load at solidus measurements (numbers next to points) on the CAST hot tearing rig based at the

University of Queensland. Data is courtesy of Dr. Hao Wang.

Guidelines for Grain Refinement and Hot tearing

Due to better understanding of the mechanisms of grain refinement and hot tearing the following is a list of the key actions that can be made to industrial practice in order to reduce costs and improve casting quality:

1. If the alloy is lean in solute elements, e.g. 1000 series alloys or soft 6000 and 3000 series alloys, additions of titanium can be made to the alloy to reduce the amount of more expensive boride containing rod grain refiner that would normally be added;

2. For consistent grain refinement, the wt:wt Ti:B ratio needs to be greater than 2 especially in wrought alloys. This may be a problem if there is residual boron present in the melt. In foundry alloys, very good grain refinement can be achieved at lower Ti:B ratios but there is a risk that this approach may be less reliable than when the ratio is greater than 2;

3. Al-Si foundry alloys do not need elemental Ti additions for grain refinement. It is much better to use boron containing grain refiners, whether they be Al-B or Al-Ti-B containing to provide nucleant particles, whilst taking into consideration point 2;

4. The greatest improvement to hot tear resistance comes from the use of grain refiner to ensure the formation of equiaxed grains rather than columnar grains;

5. If grain refiner has been added to the point where the grain morphology is globular or spherical then adding more grain refiner may increase the hot tear susceptibility of the alloy and hence be detrimental to casting quality;

6. Small amounts of low melting point phases and surface active elements may lead to decreased hot tear resistance.


Conclusions

There has been considerable improvement in the understanding of grain refinement and hot tearing phenomena over the past 10-15 years that can be used to improve practices in the casthouse and foundry. These include the use of solute Ti to assist grain refinement in lean alloys, the understanding that solute Ti is generally unnecessary in foundry alloys, that too much grain refinement may be problematic to hot tearing and that minor element additions and impurity elements may significantly affect the hot tear performance of alloys. These findings have been collated into a number of guiding points that can be considered by casthouse and foundry engineers when optimising their process operating costs and casting quality.

Acknowledgments

The CAST Cooperative Research Centre was established under and is supported in part by the Australian Government’s Cooperative Research Centre’s Programme.

References

[1] Pearson, J. and Cooper, P. (1999). A Review of the Basics of Grain Refining. in Sixth Australian Asian Pacific Conference - Aluminium Cast House Technology, Sydney, Australia, 109-118.

[2] Easton, M.A., StJohn, D.H., Sweet, L. and Couper, M.J. (2003). Optimising the Cost of Grain Refinement by Separate TiB2 and Ti Additions to Wrought Aluminium Alloys. in 8th Australasian Conference on Aluminium Casthouse Technology, Brisbane, 153-165.

[3] Easton, M.A. and StJohn, D.H. (2007) Theoretical Advances Leading to Improvement in Commercial Grain Refinement Practices in Al Alloys, Materials Forum, 31: 57-63.

[4] Easton, M.A. and StJohn, D.H. (2005) An analysis of the relationship between grain size, solute content and the potency and content of nucleant particles, Metallurgical and Materials Transactions A, 36A(7): 1911-1920.

[5] Easton, M.A. and StJohn, D.H. (2008) Improved prediction of the grain size of aluminium alloys that includes the effect of cooling rate, Materials Science and Engineering A, 486(1-2): 8-13.

[6] StJohn, D.H., Easton, M.A. and Qian, M. (2008) Controlling the Semisolid Grain Size during Solidification, Solid State Phenomena, 141-143: 355-360.

[7] Johnsson, M. (1994) Influence of Si and Fe on the Grain Refinement of Aluminium, Zeitschrift für Metallkunde, 85: 781-785.

[8] Johnsson, M. and Bäckerud, L. (1996) The Influence of Composition on Equiaxed Crystal Growth Mechanisms and Grain Size in Al Alloys, Zeitschrift für Metallkunde, 87(3): 216-220.

[9] Easton, M.A. and StJohn, D.H. (2001). The Effect of Alloy Content on the Grain Refinement of Aluminium Alloys. in Light Metals 2001, 927-934.

[10] Easton, M.A., StJohn, D.H. and Sweet, E. (2004). Reducing the Cost of Grain Refiner Additions to DC Casting. in Light Metals 2004, Charlotte, NC, 827-831.

[11] Easton, M.A. and StJohn, D.H. (2001) A Model of Grain Refinement Incorporating the Alloy Constitution and the Potency of Nucleation Sites, Acta Materialia, 49(10): 1867-1878.


[12] Smith, C., Easton, M.A. Nie, J.F., Zhang, X. and Couper, M.J. (2004) The effect of Ti Content on the Mechanical Properties of an Al7Si0.35Mg Alloy, Materials Forum, 28: 1222-1228.

[13] Pasciak, K.J. and Sigworth, G.K. (2001) Role of alloy composition in grain refining of 319 alloy, AFS Transactions, 109: 567-577.

[14] Spittle, J.A. and Keeble, J.M. (1999) The Grain Refinement of Al7Si Alloys with Boron Containing Refiners. in Light Metals 1999, 673-678.

[15] Questedl, T.E. and Greer, A.L. (2004) The effect of the size distribution of inoculant particles on as-cast grain size in aluminium alloys, Acta Materialia, 52: 3859-3868.

[16] Sigworth, G. and Guzowski, M. (1985) Grain Refining of Hypoeutectic Al-Si Alloys, AFS Transactions, 172: 907-912.

[17] Tøndel, P. Halvosen, G. and Arnberg, L. (1993). Grain Refinement of Hypoeutectic Al-Si Foundry Alloys by Addition of Boron Containing Silicon Metal. in Light Metals 1993, 783-790.

[18] Johnsson, M. (1993), Chemical Communications No.5, in Department of Structural Chemistry, Arrhenius Laboratory. 1993, Stockholm University.

[19] Bunn, A.M. Schumacher, P. Kearns, M.A., Boothroyd, C.B. and Greer, A.L. (1999) Grain Refinement by Al-Ti-B Alloys in Aluminium Melts: A Study of the Mechanisms of Poisoning by Zirconium, Materials Science and Technology, 15: 1115-1123.

[20] Bäckerud, L. and Vainik, R. (2001). Method for optimised aluminum grain refinement. in Light Metals 2001, 951-954.

[21] McCartney, D.G. (1989) Grain Refining of Aluminium and its Alloys using Inoculants, International Materials Reviews, 34: 247-260.

[22] Warrington, D. and McCartney, D.G. (1991) Hot-Cracking in Aluminium Alloys 7050 and 7010 - A Comparative Study, Cast Metals, 3(4): 202-208.

[23] Campbell, J. and Clyne, T.W. (1991) Hot Tearing in Al-Cu Alloys, Cast Metals, 3(4): 224-226.

[24] Spittle, J.A. and Brown, S.G. (2005) Numerical modelling of permeability variation with composition in aluminium alloy systems and its relationship to hot tearing, Materials Science and Technology, 21(9): 1-7.

[25] Apelian, D., Flemings, M. and Mehrabian, R. (1974) Specific Permeability of Partially Solidified Dendritic Networks of Al-Si Alloys, Metallurgical Transactions, 5: 2533-2537.

[26] Eskin, D.G., Suyitno, Mooney, J.F. and Katgerman, L. (2004) Contraction of aluminum alloys during and after solidification, Metallurgical and Materials Transactions A, 35A(4): 1325-1335.

[27] Grandfield, J.F., Davidson, C.J. and Taylor, J.A. (2001). Application of a new hot tearing analysis to horizontal direct chill cast magnesium alloy AZ91. in Light Metals 2001, 911-917.

[28] Vernéde, S., Dantzig, J.A. and Rappaz, M. (2009) A mesoscale granular model for the mechanical behaviour of alloys during solidification, Acta Materialia, 57: 1554-1559.


[29] Easton, M.A., Grandfield, J.F., StJohn, D.H. and Rinderer, B. (2006) The effect of grain refinement and cooling rate on the hot tearing of wrought aluminium alloys, Materials Science Forum, 519-521: 1675-1680.

[30] Easton, M.A.,Wang, H., Grandfield, J.F., StJohn, D.H. and Sweet, E. (2004) An Analysis of the Effect of Grain Refinement on the Hot Tearing of Aluminium Alloys, Materials Forum, 28: 224-229.

[31] Rappaz, M., Drezet, .J.-M and Gremaud, M. (1999) A New Hot-Tearing Criterion, Metallurgical and Materials Transactions A, 30A: 449-455.

[32] Rossenberg, R.A., Flemings, M.C. and Taylor, H.F. (1960) Nonferrous binary alloys hot tearing, AFS Transactions, 68: 518-528.

[33] Instone, S. StJohn, D.H. and Grand, I. (2000) New Aparatus for Characterising Tensile Strength Development and Hot Cracking in the Mushy Zone, International Journal of Cast Metals Research, 12: 441-456.

[34] Schloz, J.D. (2001). New generation 6xxx alloys - casting considerations. in 1st Australasian-Pacific Alumininium Extrusion Conference, Sydney, Australia.

[35] Naguami, H., Suzuki, S., Okane, T. and Umeda, T. (2006) Effect of iron content on hot tearing of high-strength Al-Mg-Si alloy, Materials Transactions, 47(11): 2821-2827.

[36] Latter, D. (2001), Optimising pit recoveries on 6xxx extrusion billet, in 7th Australian Asian Pacific Conference Aluminium Casthouse Technology, P.R. Whiteley, Editor. 2001, The Minerals, Metals and Materials Society, Warrendale USA. p. 213-219.


DC Casting Using a TF Combo Bag: 3D Modeling of the Distribution of Molten Metal and Heat Transfer

Sylvain P. Tremblay1,a, André Arsenault2, Daniel Larouche2, Florin Ilinca3,

Jean-François Hétu3 and Jean-Philippe Dubé4

1 PYROTEK INC., 1623 Manic Street, Saguenay, QC, G7K 1G8, Canada 2 Laval University, Dept. of Mining, Metal & Mat. Eng., Adrien-Pouliot Bldg, Québec,

QC, G1K 7P4, Canada 3 Industrial Materials Institute, NRC; 75 de Mortagne; Boucherville, QC, J4B 6Y4, Canada

4 A.B.I., Aluminerie de Bécancour Inc., 5555, Pierre-Thibault street, Bécancour, Quebec, G0X 1B0, Canada

a [email protected]

Keywords: fluid flow, simulation, molten aluminium distribution, combo bag

Abstract

In the past 3 years, Pyrotek in association with Laval University and the Industrial Materials Institute of the National Research Council of Canada has developed a simulation package proposing a unique approach in the fluid flow calculation of molten aluminum distribution with a combo bag in the ingot head of sheet casting. This paper will summarize the difficulties encountered and the solutions that were adopted to render that simulation tool efficient and consistent.

Validation of the model using actual casting tests will be detailed as well as the major change in the simulation package in order to obtain decent calculation times. Some examples of simulations will be given to demonstrate that the initial goal of this project has been achieved by the development of a mathematical simulation tool to design and improve metal distributors used in the sheet casting production.

Introduction

In the middle of 70s, the automatic molten level control was introduced in vertical DC casting. The presence of the level probe was interfering with the channel bag/float used at the time to control the metal flow rate. A new bag was designed and used. It has been called “combo bag” because it captured oxides and distribute the molten aluminum around the mold. The combo bag is now the reference in DC sheet ingot casting.

Pyrotek has been a leader in designing and manufacturing combo bags since their introduction. Pyrotek has been very innovative in introducing distributor technologies such as TF combo bag [1] illustrated in Figure 1 ReMAD [2] shown in Figure 2. TF means Thermally Formed combo bag where the combo is shaped using a heated die instead of being sewn. This technology offers very consistent bag dimensions and rigidity compared to a sewn combo bag.

ReMAD stands for Reusable Molten Aluminum Distributor. It consists in a composite refractory frame coupled with a TF inner bag. The frame can be reused up to 50 times depending on casting conditions and maintenance while the inner bag is used one time.

These two technologies have been developed using our water model. However, even if water model is useful to see the flow coming out of the combo bag, it does not give any information on the thermal fluid flow distribution. To better understand the interaction between molten metal distribution from the combo bag and the ingot sump, the development of a mathematical simulation tool was initiated three years ago. The objectives of this research program were to find the real effects of combo bag geometry on the temperature distribution in a DC cast sheet ingot and to seek


for a performance criteria based on dimensional analysis on heat flow, sump profile and fluid velocity.

Figure 1. Typical sewn and TF combo bags. Figure 2. ReMAD.

Mathematical Simulation Tool – Part I

Pyrotek decided to use a quite innovative approach in trying to simulate the combo bag effect into the global casting set-up. It was therefore discarded to impose velocities at different positions in the bulk of the system or imposing inlet velocity boundary conditions directly on the liquid phase. This paper summarizes the first attempts to simulate the thermal and fluid flow patterns in DC casting coupling the flow in the entire set-up of the metal delivery system.

Numerical simulation [3] was conducted with the ProCASTTM software from ESI. This application allows fluid flow computation, heat transfer and solidification modeling. A dedicated functionality in ProCASTTM v.2006.1 is the continuous casting module, which allows calculations in the transient or steady state regimes. In the steady state regime with fluid flow, the model assumes that the solid phase is transported at the casting speed through a fixed domain. This strategy was adopted in the present contribution.

Figure 3 presents the geometry and the meshing of the metal delivery system and the upper portion of the ingot [4]. The ingot is not entirely shown but it is a parallelepiped having a cross section of 1346mm X 660mm and a length of 1350mm. Previous evaluations suggest that this was a sufficient length to obtain a sump profile independent of the ingot’s bottom heat extraction.

Control pin

Dip tube

Skim dam

Distributor bag

Spout sock

Impervious skirt

Metal inlet

Figure 3. Geometry and meshing of the model.

To reduce the numerical simulation effort, only the quart of the system was modeled because of its 4-fold symmetry. Aluminium alloy 5052 was considered and its composition is given in Table 1.


Table 1. AA5052 composition.

Mg 2,5 % Cr 0,25 % Mn 0.25 % Fe 0,1 % Al Balance

Material properties of AA5052 are illustrated in Figure 4. The properties have been evaluated using the rule of mixture, assuming equilibrium cooling condition.

0

0.2

0.4

0.6

0.8

1

550 570 590 610 630 650 670 690

Temperature (°C)

Frac

tion

solid

0

50

100

150

200

The

rmal

con

duct

ivity

(W/m

/K)

0

200

400

600

800

1000

1200

1400

100 200 300 400 500 600 700

Temperature (°C)

Ent

halp

y (k

J/kg

)2300

2350

2400

2450

2500

2550

2600

2650

2700

Den

sity

(kg/

m3 )

Figure 4. Physical properties of AA5052 used in the simulation.

Primary and secondary cooling conditions were modeled as in reference [5]. Table 2 presents the heat transfer boundary conditions used at different positions over the external surface of the ingot. A natural air cooling condition was given to all other surfaces. A film convective boundary condition (HTC = 500W/m2/K, T∞ = 25°C) was applied at the bottom of the ingot to simulate axial heat extraction in the steady state regime. Contact heat transfer coefficients between different volumes were constant and equal to 1000W/m2/K. Thermophysical properties of fiberglass fabric cloths, dip tube and control pin were assumed to be those of fused silica. For the skim dam, thermo-physical properties were taken from the N14 Pyrotek’s product data sheet, except for the specific heat, that was assumed to be equal to the specific heat of CaO-SiO2

Table 2. Heat transfer conditions applied at the external surface of the ingot.

Heat flux = h·(Tsurface – Tref) Distance from the surface of the pool

(mm)

Tsurface (K)

Heat transfer coefficient h

(W/m2/K)

Tref (K)

Primary cooling (h is constant) 0 – 87 * 500 373 Air gap (h is constant) 87-111 * 10 293

Water cooling (h is function of Tsurface)

111 and below

273 2000 283 373 5000 283 403 27000 283 573 8500 283 873 0 283

The velocity of metal at the bottom of the model was set to 5cm/min and a pressure inlet condition was prescribed at the top entry of the metal. Wall type conditions were given to the side of casting and a zero vertical velocity was prescribed at the top surface of the liquid pool. These conditions insured that the volumetric flow of metal at the entry was equal to the casting speed multiply by the cross section of the ingot.


The permeability of the fabric cloth was evaluated by calculating the pressure drop obtained when liquid aluminum flows perpendicularly through a grid made of rectangular holes, similar in size to the openings found in combo bags. Figure 5 presents the model used to calculate with ProCASTTM the pressure drops associated with metal flowing at different nominal velocities.

Sz

wz

Sx

wx

t

Figure 5. Geometrical model of the fabric cloth.

Pressure drops were averaged in the vicinity of the grid to calibrate the pressure drops associated to a thin filter material in ProCASTTM. The pressure drop is highly non-linear with velocity, which indicates that pressure drop at high velocities is more a consequence of the impacting effect on the screen than a consequence of friction effects. For the small range of fluid velocities generally obtained at the exits of the combo bag, it was acceptable to consider a linear dependency of pressure drop with velocity as required by the filter model in ProCAST.

The velocity magnitude contours in the two plane of symmetry are presented in Figure 6 and Figure 7 after a simulation time of 34 seconds. One can see that the velocity magnitudes in Figure 6 decrease rapidly outside the combo bag. In Figure 7, the impervious skirt redirects the flow normal to the XY plane. The recirculation loops are clearly visible and are similar in their general appearance with those obtained by Fortier et al. [6]. The pattern is more complex however in our simulation, mainly because of the complexity of the distribution system considered.

Figure 6. Velocity pattern in the symmetry plane YZ.

Figure 7. Velocity pattern in the symmetry plane XZ.

Figure 8 compares side by side a slice view of the velocity magnitudes in the XZ plane just below the bottom of the distributor bag and a snapshot of colorant dispersion in a water model obtained after injection of the colorant in the dip tube. Considering the high velocity magnitudes in the stream formed between the distributor bag and the impervious skirt and the colorant diffusion path in the water, it is clear that the preferential path of metal flow is in the space located between the distributor bag and the impervious skirt, the later being the true barrier redirecting the metal.


Figure 8. Comparison between the water model visualization and the calculated velocity magnitude in a section crossing the stream formed inside the impervious skirt of the

TF combo bag.

In conclusion, the first results of the mathematical simulation program on TF combo bag were in agreement with what could be seen from the previous water model. The two streams of liquid metal redirected by the impervious skirt were not just transient but prevail at steady state. These streams are the main source of metal and the question remaining was: at which level the temperature distribution in the sump is influenced by the momentum associated to these metal inlets?

Mathematical Simulation Tool – Part II – Model Validation

To validate the mathematical simulation and determine the sump profile, an experiment was conducted at Aluminerie de Bécancour Inc. on a casting table with five ingots in operation. The AA5052 was degassed using an Alpur D-5000 and filtered using a 40ppi reusable CFF. Two of the five ingots were assigned to our experiments. One ingot was equipped with a sewn combo bag, and another was equipped with a thermally formed combo bag. The sewn combo bag ingot was the farthest from the incoming metal source while the TF combo bag ingot was just beside the sewn combo bag one. The mold size was 1346mm x 660mm. Sumps profile were revealed by pouring pure zinc in the sump followed by cutting and etching. The results obtained on the ingot cast with a TF combo bag are shown in Figure 9 where the calculated sump profile is on the left and the zinc revealed sump of the cast ingot is on the right. Notice that the ingot casts with the sewn combo bag had a sump profile and a sump depth largely similar to those obtained with the TF combo bag, with only minor differences.

From the preceding sump’s depth comparison, it seems that the simulation results can be used to represent adequately the actual sump, at least as a good starting sump’s profile to be included in a more complex mathematical model. The discrepancy between the actual sump profile and the calculated one is expected to be in the order of 12%. The results also indicate that 1.5 meter (1800s x 51mm/min =1,530m) of cast ingot must be modeled to allows water cooling to extract the appropriate amount of heat at the sump profile level. The same shape is observed in both sumps i.e. going down and then up in the center axis.

Since computation times were too long to perform true steady-state simulations (one day of calculation per second of cast), we decided to migrate to another software driven by a cluster of computers. This software has been developed by the Industrial Materials Institute, a division of the National Research Council of Canada.


Figure 9. Comparison between calculated and measured sumps with the TF combo bag.

Mathematical Simulation Tool – Part III – New Simulation Package Results

The previous model was imported in the new platform, allowing much smaller computation time and giving us the opportunity to verify the influence of turbulence. This part presents new results and observations obtained with this platform. Computations were performed in parallel on a 64-node Beowulf cluster. Each node consists of an Intel motherboard with two Intel Xeon 3.4Ghz (FSB-800) and 2GB of RAM. Computational nodes are connected with Myrinet 2000 (PCIXD) cards and switches and MPICH-GM is used. The CFD model solves differential equations describing the conservation of mass, momentum and energy in order to evaluate the velocity, pressure and temperature fields. The flow is driven by the incompressible Reynolds Averaged Navier-Stokes equations.

The Isothermal Flow Model

The geometry of the model is basically the same as the one presented in part I. The computational domain was discretized using 4-node tetrahedral finite elements. The mesh for the isothermal flow case (no sump) had 141,971 nodes and 759,772 elements, whereas the one used for the thermal model had 148,161 nodes and 846,839 elements. Figure 10 illustrates the finite element discretization for the isothermal flow case.

Figure 10. Geometry and meshing of the isothermal flow model.


The inlet velocity was imposed on the cross section defining the entry plane of liquid metal. The casting speed is 50.8mm/min and the corresponding inlet velocity conserving the volumetric flow is 240mm/s. Non-slip boundary conditions were imposed on solid walls, whereas the liquid top surface was constrained to remain horizontal by imposing the vertical velocity to zero.

The fabric cloth considered had the same characteristic than the one described in Part 1. To determine the pressure drop through the fabric cloth a single cell unit was considered and the flow domain was meshed with 3D tetrahedral elements. The mesh was sufficiently refined in order to capture well the flow and contains 374,000 elements and 79,596 nodes. Several computations were carried out for velocities ranging from 0.002m/s to 0.256m/s. The results for the pressure drop are shown for four different fabrics in Figure 11.

Figure 11. Pressure drop for the four different fabric cloths.

A first computational configuration was used to validate the flow in the combo bag. The model does not consider the ingot and therefore it can be considered isothermal. Simulations were carried out for steady state conditions and an inlet velocity of 240mm/s. The flow is highly convective and therefore a time marching scheme was used to obtain the steady state solution. The Reynolds number based on the conditions in the inlet channel is Re = 56,800 indicating that the flow is turbulent. The observed velocity in the fabric cloth region is smaller than 300cm/min. At this flow rate, the laminar and turbulent flow solutions through the fabric cloth give similar pressure drop.

In order to assess the effect of the viscosity and of the turbulence model on the flow solution, simulations were carried out for the laminar flow conditions at viscosities equal to the nominal value ( 31.15 10 Pa s−⋅ ⋅ ), 10 times larger and 100 times larger and for the turbulent flow conditions using the κ-ε turbulence model. The solutions are compared in Figure 12 to Figure 15, which show the regions of the flow were the velocity is higher than 100cm/min. The colors are as from the velocity magnitude, with the red indicating velocities higher than 500cm/min.

The laminar flow solutions at the nominal viscosity and for the 10 times higher viscosity are comparable. The flow is highly convective and a strong jet develops from the region inside the impervious skirt. Flow of smaller intensity is observed through the fabric cloth at the bottom of the combo bag. The solution for a viscosity 100 times higher than the laminar viscosity seems to be over diffusive, as the jet of material from the impervious skirt is of much smaller intensity. This indicates that solving the laminar flow conditions with this value of the viscosity would be too diffusive. Finally, the solution was also obtained for the turbulent flow conditions using the κ-ε turbulence model. The turbulent flow solution also presents an important jet of material from inside the impervious skirt, but having smaller amplitude than observed for the laminar flow solution. Finally, it was found that the laminar flow solution for an effective viscosity 40 times the laminar viscosity gives results comparable to those for turbulent flow computation (see Figure 15 and Figure 16). This effective viscosity was considered in the work presented in the following section.


Figure 12. Laminar viscosity.

Figure 13. Laminar 10x (viscosity).

Figure 14. Laminar 100x (viscosity).

Figure 15. Turbulent, κ−εκ−εκ−εκ−ε model.

Figure 16. Effective viscosity laminar flow 40x (viscosity).

The Thermal Flow Model with Continuous Casting Conditions

The thermal flow computations were carried out starting with an initial temperature distribution obtained from a no flow computation (the velocity was imposed as from the ingot velocity) for the case that includes the natural convection term. It can be seen that the temperature varies very rapidly and the flow produces a more uniform temperature in the liquid pool above the solid ingot. The region of higher temperature decreases as the liquid metal with higher temperature has lower density and tends to flow towards the top free surface. The position of the liquid / solid interface does not change during the first six seconds of simulation (Figure 17). The temperature at later times is shown in Figure 18 which shows that the temperature changes little after the first 20 seconds of simulation. The higher temperature metal from the inlet flows towards the top surface and the temperature is more uniform at a level closer to the liquidus temperature of 649ºC in the liquid region below. In order to estimate the influence of the natural convection term on the flow and temperature distribution, a second simulation was carried out without the presence of the natural convection term. The temperature field for this case is shown in Figure 19. Significantly different temperature distributions are observed. In this case, the higher temperature material is pushed outside the bag laterally and towards the bottom as previously observed in the isothermal test case. The temperature distribution changes gradually with time and the high temperature region is larger than in the previous case.


Figure 17. Initial temperature distribution compared with the solution for t = 3s (left) and t = 6s (right) including the effect of natural convection.

Figure 18. Initial temperature distribution compared with the solution for t = 20s (left) and t = 60s (right) including the effect of natural convection.

Figure 19. Initial temperature distribution compared with the solution for t = 20s (left) and t = 60s (right) without the effect of natural convection.

The velocity distribution for the solution with natural convection at t=10s is illustrated in Figure 20 showing the velocity distribution in planes parallel to the longer side of the ingot. The flow is shown in the symmetry plane (Z = 0) and at 25mm from the symmetry plane. As can be seen, the effect of the flow from the combo bag is limited to a narrow region around the combo bag.

Figure 20. Velocity distribution in planes parallel to the XY plane.

Symmetry plane 25 mm from symmetry plane


The flow is downward in the region close to the liquid / solid interface and upward near the center of the domain were the temperature is higher. The results indicate that the flow outside the combo bag is driven by natural convection. The temperature in the liquid pool on top of the ingot is more uniform when considering the effect of natural convection and the material with higher temperature flows towards the top surface.

In conclusion, the present work allowed us to study the intensity of two important phenomena occurring in the liquid metal pool during DC casting. First, the turbulence intensity level was gauged against the forced convective flow area at the exit of the metal distributor. The shape and force of the liquid metal jets were used to determine an appropriate equivalent turbulent viscosity. Second, the impact of natural convection on the contour-plot of isotherms in the liquid metal pool has been illustrated. The drastic change in isotherm pattern when the natural convection model is activated indicates that natural convective flow is at least as important as the forced convective flow to drive the recirculation loops.

Despite the importance of natural convection in the development of the stationary temperature distribution below the combo bag, the forced convective flow produced by the combo bag still exerts an important influence in the upper portion of the pool, where solidification starts and defects are most likely to occur. Future works will then essentially focus on the quantification of the impact of combo bags on the heat transfer occurring at the periphery of the upper portion of the sump.

References

[1] Tremblay, S.P. and Ruel, M. (2003) The Manufacturing, Use and Plant Test Results of TF Combo Bags for DC Sheet Ingot Casting, Products, Applications, and Services Showcase, TMS (The Minerals, Metals & Materials Society), D.V. Neff, ed., 193-228.

[2] Tremblay, S.P. and Lapointe, M. (2002) The manufacturing, design and use of a new reusable molten metal distributor for sheet ingot casting, Light Metals 2002, 1634-1642.

[3] Larouche, D. and Tremblay, S.P. (2005) TF Combo bag – Mathematical simulation result and up-date, Presented at the 3rd International Melt Quality Workshop, Dubai, UAE, 14-16th November 2005.

[4] Arsenault, A., Larouche, D., Tremblay, S.P. and Dubé, J.P. (2008) DC cast thermal and fluid flow simulation using a semi-permeable model of TF combo bag, Light Metals 2008, D.H. DeYoung, ed., The Minerals, Metals & Materials Society, 781-785.

[5] Gruen, G.U., Buchholz, A. and Mortensen, D. (2000) 3-D modeling of fluid flow and heat transfert during the dc casting process – Influence of flow modeling approach, Light Metals 2000, 573-578.

[6] Fortier, M., Larouche, A., Chen, X.-G. and Caron, Y. (2005) The effect of process parameters on the metal distribution for DC sheet ingot casting, Light Meals 2005 , 1019-1024.

[7] Ilinca, F. and Hétu, J.-F. (2000) Finite Element Solution of Three-Dimensional Turbulent Flows Applied to Mold-Filling Problems, Int. J. for Numerical Methods in Fluids, vol. 34, 729-750.


CHAPTER 8:

Ingot Casting

Safety Enhancement in Ingot Casting at Tomago Aluminium

V. Nguyen1,a, J. F. Grandfield2, P. Rohan1 and B. Todd3

CAST Cooperative Research Centre (1, 2) 1 CSIRO, Clayton, Victoria, Australia.

2 Grandfield Technology Pty Ltd, Brunswick, Victoria, Australia. 3 Tomago Aluminium, Newcastle, New South Wales, Australia.

a [email protected]

Keywords: ingot casting, mould, boiling

Abstract

Tomago Aluminium experienced problems with the boiling of cooling water in ingot casting machines, where water splashed onto molten aluminium after mould filling. The boiling problem was particularly severe when new standard moulds were installed.

This paper describes the experiments and modelling employed to analyse and identify the cause of the problem and based on the results, modifications to mould geometry were suggested. Subsequent testing in lab and field trials showed that the boiling was suppressed. The first set of modified moulds in service at Tomago Aluminium showed mould life was improved as predicted. Other issues with the modified mould design, which arose in service, are also discussed.

Introduction

Chain conveyor ingot casting machines have been employed to produce millions of tonnes of aluminium ingots every year. In this process, ingot moulds moving along a conveyor are filled at one end and usually cooled in a water bath. The line speed is set such that the ingots are fully solidified when they reach the other end of an approximate 20m to 25m long conveyor, where they are knocked out and stacked into bundles.

At Tomago Aluminium (TAC), the standard moulds had problems with the boiling of cooling water, which resulted in water splashing onto molten aluminium after mould filling. The boiling problem was particularly severe when a set of new standard TAC moulds was installed. TAC approached CAST Cooperative Research Centre (CAST) to address this potentially hazardous situation.

In order to replicate the boiling behaviour and to understand the mechanism causing water ejection, casting trials with new and used moulds were conducted on the single ingot casting apparatus at CSIRO labs in Melbourne. Temperature data were collected at various positions on the walls of tested moulds to determine the hot spots.

In the next step, Alsim software [2] was employed to model and analyse the standard TAC mould design, and the results validated by experimental data. The established modelling procedure was then used to analyse alternative designs. A 19mm thick side wall mould proposed by TAC, a commercially proposed design and variants proposed by CAST/CSIRO project team as shown in Figure 1 were modelled. Based on the results predicted by modelling, a design by CAST/CSIRO, which showed the least tendency to cause boiling and had potential to enhance mould life and performance, was elected for further testing and evaluation.

Testing of the new CAST/CSIRO design showed that it had almost suppressed the boiling issue while its solidification performance was improved. The new mould design has been trialled and in service at TAC since January 2008 and confirmed the mould life improvement as predicted.


Figure 1. (a) Standard TAC mould (12mm thick wall); (b) 19mm thick side wall mould, proposed by TAC; (c) 14mm thick wall based on standard TAC design; (d) As design (c) with thickened centre ridge and wings.

Experimental Method

The measurements of temperature, air-gap and mould deformation have been carried out employing the single ingot test apparatus [1,2,3] which consists of a water tank, a tilting furnace, a feed launder and a steel frame to facilitate mould mounting. It is capable of reproducing various production conditions such as wet fill or dry fill followed by cooling water rising. Water was circulated by a pump through water nozzles positioned underneath the mould, which were controlled to simulate the effect of the mould movement in an ingot casting machine during production.

For each experiment, the test mould was preheated to approximately 100°C with the water level in the tank kept below the mould base. Just before the test started the water level was raised and then metal poured into the mould. The filling time was typically 12 seconds (slower than a typical 4 second fill time on a 20 tonne per hour machine). The amount of aluminium alloy charged into the tilting furnace was determined to suit the ingot-mould system to be measured. The air gap evolution and mould displacement were measured using Linear Variable Differential Transformer (LVDT) transducers mounted at various positions on the mould as shown in Figure 2a and 2b. To measure the air gap, a thin steel button attached to the LVDT core inside the mould was allowed to freeze in the ingot which then pulled away from the LVDT housing mounted in the outer mould wall, as illustrated in Figure 2c. Thermocouples were positioned at points of interest to measure mould and ingot temperatures shown in Figure 2d. A multi-channel data logger and a laptop computer were utilised to record all data simultaneously during tests, which were subsequently processed in Microsoft Excel. Underwater filming was also made to observe the film boiling on the underside of the mould.


Figure 2. (a) LVDTs mounted to measure air gap dimension. View is of the underside of the mould showing centre ridge; (b) Standard TAC mould fitted with thermocouples and LVDTs

shown suspended above water bath; (c) Detailed view of an LVDT connection to the mould for air gap measurement. The hole in the mould wall was sealed for waterproof operation; (d)

Thermocouple positions used in the tests, mounted at 3mm and 7mm from the hot face.

Modelling Method

Alsim, a fully coupled multi-physics 3D FEM solver developed by the Norwegian Institute for Energy, was used to predict ingot-mould air gap, complex ingot-mould deformation and stress, mould temperature and ingot solidification times. For each simulation, 3D solid models of the ingot-mould system analysed were imported into MSC-Patran, where the model meshes were generated and all boundary conditions were applied. A neutral file containing all information of the finite element model was then generated which was subsequently used as part of an input file for the Alsim solver. After the simulation was completed, the results were post-processed employing GLview Inova. A typical finite element model of an ingot-mould system for Alsim is shown in Figure 3.

All simulations were run with properties for pure aluminium. A starting temperature of 725°C for the molten metal was used in all cases. The mould material and mechanical properties were those used successfully before, typical for an ingot mould cast iron. Initially, the boiling curve (heat transfer coefficient versus external mould temperature) which was used to simulate previous ingot casting was employed. However this was found to predict the temperatures on the exterior surface of the standard TAC mould would not exceed 100-150°C and thus film boiling was not likely. The original curve, shown in Figure 4, had a critical transition temperature of 200°C and a gradual transition to film boiling. An inspection of the boiling literature indicated that a transition temperature of around 140°C and a sharp transition was likely. After some simulations with different boiling curves, followed with comparison with the experimental results, the curve Fb4 (Figure 4) was selected for use in all cases.


Figure 3. Typical mesh of an ingot mould FEA model.

0

2000

4000

6000

8000

10000

12000

0 50 100 150 200 250 300 350 400

Mould surface Temperature (C)

Hea

t Tra

nsf

er C

oef

icie

nt (W

/m2C

)

FB1

Fb2

Fb3

Fb4

Original curve

Figure 4. Various boiling curves investigated. Fb4 was the final curve used on all cases.

Results and Discussions

Boiling Behaviour of the Standard TAC Mould

Experimental rig tests with all standard TAC moulds showed evidence of violent boiling and associated spitting and splashing of water onto the splash guards and into the moulds. Water entered the mould creating the boiling by wetting the edge of the splash guard and wicking its way into the mould each time a spurt of water hit the splash guard. (Figure 5a). This occurred in the centre of the mould and under the wings. Large steam bubbles could sometimes be seen (Figure 5b). There was approximately a 20 second delay before boiling started, presumably till the external surface temperature increased above the boiling point. Violent boiling had not been observed on the ingot rig in the past with moulds from other plants.

The used standard TAC mould showed less violent boiling with just one wing showing water splash and spitting into the mould. Also, a distinct “singing” noise was heard when casting into the old moulds. This noise was also heard when casting into the new mould once rust had built up i.e. after 1-2 tests although to a reduced extent. In most cases, boiling at the mould wings was worse than at the centre ridge, which may indicate more hot spot areas at the wings than at the centre ridge.

Figure 6 shows typical mould temperatures measured at 3mm and 7mm from the hot face at the chosen positions shown in Figure 2d for a new standard TAC mould cast with normal water level.


Figure 5. (a) Boiling caused splashing of water into the mould; (b) Large bubbles formed under the wing at the cold end.

0

50

100

150

200

250

300

-50 0 50 100 150 200 250 300 350

Time (s)

Tem

per

ature

(°C

)

Wing 3mm wing 7mm

Centre Base 3 mm from hot face Centre Base 7 mm from hot face

End Base 7 mm from hot face centre wall at melt line 7 mm

Centre wall small side 7 mm Centre wall large side 3 mm

Centre wall large side 7 mm Centre wall small side 3 mm

140607

Figure 6. Mould temperatures measured at 3mm and 7mm from the hot face at the positions shown in Figure 2d for a new standard TAC mould cast with normal water level.

Predictions of Temperatures and Boiling Behaviour for the Six Mould Designs

Alsim predicted mould temperatures for the standard TAC moulds, shown in Figure 7, were found to go through a heating and cooling cycle typical of what was observed on the rig. This also occurs on other moulds of similar design and is due to the initial high heat transfer leading to fast mould heating and expansion followed by a reduced heat transfer when the air gap forms. The model predicted that the wing and the centre ridge would be hotter than the rest of the mould.

Whereas all the experimental data showed that the centre ridge was hotter than the wing (Figure 6), the model predicted the wing was the hottest part of the mould. The discrepancy indicated that the boiling curve Fb4 employed in the models required further fine tuning.


0

50

100

150

200

250

300

350

0 50 100 150 200 250 300 350 400

Time (s)

Tem

perature (C)

Centre baseWallwingEnd base

Figure 7. Predicted mould temperatures at various positions for the standard TAC mould.

Using the modelling procedure and material properties established with the standard TAC mould, alternative designs were modelled and analysed. The aim was to search for a design which eliminates hot spots and thus boiling, without decreasing mould life and performance. Five alternative designs were analysed and compared with the standard TAC mould. The maximum temperatures predicted for the six designs are tabulated in Table 1. It can be seen that the last two designs proposed by CAST/CSIRO substantially reduced the surface temperatures at hot spots to the extent that boiling hardly occurs, whereas the others still showed a high potential for boiling.

Table 1. Maximum predicted mould temperatures.

Mould design Figure Thickness at wing (mm)

Maximum predicted temperature at wing (°C)

Standard TAC Fig. 1a 12 315

A commercially proposed mould Not shown 14 237

19mm thick wall, thickened wing ridge, proposed by TAC Fig. 1b 15 290

Standard TAC with 14 mm wall (CAST/CSIRO) Fig. 1c 14 225

Standard TAC with thickened hot spots (CAST/CSIRO) Not shown 20 127

Standard TAC with 14mm wall and thickened hot spots

(CAST/CSIRO) Fig. 1d 22 125

Figure 8 shows the significant differences in the calculated temperature distribution on the outer surface of the six designs, captured at 100 seconds after mould filling when boiling severely occurs on the standard TAC moulds.


Figure 8. Temperature distribution on the outer surfaces predicted at 100 seconds for (a) standard TAC; (b) a commercial design; (c) 19mm thick wall, proposed by TAC; (d) standard TAC with 14mm thick walls; (e) 12mm thick with thickened centre ridge and wings proposed

by CAST/CSIRO; (f) as design (e) with 14mm thick walls.

Predictions of Maximal Principal Stresses in the Six Mould Designs

Stresses in the mould were predicted to increase as the mould was heated and then decrease as it cooled down. The interior was largely under compression (expansion of the hot interior is restrained by the cold exterior) and the exterior was under tension. The maximum principal stresses were predicted to occur on the mould exterior.

Figure 9 shows the maximum principal stresses on the six moulds. Four of the moulds were under the highest stresses, occurring in the regions at the edges of mould wings while both the CAST/CSIRO proposed designs, (e) and (f), did not have this high stress region. A more generous wing corner radius was predicted to reduce the stresses. The previously trialled 19mm thick wall mould, (c), was predicted to have the highest stresses both at the wing and the centre, where premature failures had been observed.

0

50

100

150

200

250

300

350

400

450

MP

a

Stress at centre ridge(MPa)Stress at wing

Figure 9. Maximum principal stresses predicted for the six designs (see Fig. 8 for legends).

(a) (b) (c)

(d) (e) (f)

(a) (b) (c) (d) (e) (f)


The Recommended Design

As stated before, the primary objective of the project was to eliminate the boiling issue associated with the standard TAC mould. The modelling results show that the two CAST/CSIRO designs namely (e) and (f), would achieve the above goal.

Additionally, the stress analyses showed that both designs did not only maintain, but potentially increase mould life above that of the standard TAC design. Modelling also predicted a reduced solidification time with either of the two CAST/CSIRO designs.

When installation and implementation were considered, the 12mm thick wall version of the two designs shown in Figure 8e, was more attractive since it could be provided as a bolt-on replacement without any required modification or adjustment to existing ingot casting machines.

Trials and Adoption

Testing of the new CAST/CSIRO design showed that it had almost suppressed the boiling issue. Depending on the water flow conditions it could be completely eliminated. Boiling in stagnant water could be suppressed with better water circulation. The new mould has reduced solidification time when compared with the standard TAC mould. The plant trial that took place in January 2008 showed that only a “handful” of moulds, in a set of approximately 270 moulds, named MK I, were observed to still exhibit very slightly boiling under one of the mould wings. The second set with a slightly thicker wing, named MK II, has totally suppressed the boiling. Very few ingot release issues were observed, the main reason for non release was subsequently identified as mould geometry tolerances, i.e. moulds not cast to the drawing. This issue has been rectified.

Field testing has confirmed the predicted improvement in mould life. A set of MK II moulds has, at the time of writing, cast over 150000 tonnes of aluminium without any sign of failure. It should be noted that failures used to occur at around 90000 tonnes for the old standard TAC moulds.

Conclusions

A new mould design, which successfully eliminates film boiling under normal operating conditions, has greatly enhanced the safety of the ingot casting process at TAC. Additionally, the new mould reduces solidification time and lasts much longer than the old standard mould.

Acknowledgements

The authors would like to acknowledge the permission to publish given by TAC and the contribution of the TAC team, including Michael Kilpatrick, Bob Todd, Leigh Thompson, Bruce Arnold, Dave Davies and all the people involved in ingot casting at TAC.

References

[1] Grandfield, J.F., Nguyen, T.T., Redden, G. and Taylor, J. (2001) Aspects of heat transfer during production of remelt ingots using chain casters, in Seventh Australian Asian Pacific Conference on Aluminium Cast House Technology Proceedings, P. Whiteley ed., TMS, 263-272.

[2] Grandfield, J., Mortensen, D., Fjær, H., Rohan, P., Nguyen, V., Sund, H. and Nguyen, T. (2006) Remelt ingot mould heat flow and deformation, in Light Metals 2006, TMS, 869-876.

[3] Grandfield, J., Nguyen, V., Rohan, P. and Nguyen, T. (2007) Ingot caster productivity improvement through examination of mould heat flow and deformation, in Ninth Aluminium Cast House Technology Conference, J. Grandfield, J. Taylor eds, CASTconsult, 147-152.


Technological Jump in Aluminium Ingot Production

Dr. Eng. Carmelo Maria Brocato

Continuus-Properzi S.p.A., Via Cosimo del Fante 10, Milan 20122, Italy [email protected]

Keywords: aluminium, ingot, production, casthouse

Abstract

Usual production of modern cast houses includes semi products and re-melt products. Billets, slabs and rod belong to the family of semi products, while T-bars, sows and ingots are incorporated in the category of re-melt products. Re-melt forms have been developed to be easily transported and easily processed in locations even far away from the cast house where these have been produced. Unlike sow and T-bars, ingots need to be stacked in bundles and securely strapped to allow safe and easier handling and transportation. Yet, the ingots can be processed one-by-one by small users or can be loaded into the melting furnace bundle-by-bundle.

The shape and weight of pure aluminium ingots of the old prior art were determined with two aims: maximization of the production rate and minimization of the production costs: for many years heavy ingots have been produced by pouring molten aluminium into a chain of open top moulds. The traditional complex shape of the ingots was intended to facilitate de-moulding operations and bundle piling.

For the above reasons, the most common ingot weight worldwide has ranged, until now, from 22.5 to 23.5kg.

Now a new technology launched by Continuus-Properzi offers bigger hourly production rates and handy, safe and sound ingots of 30 pounds (13.6kg) as well as more compact and stable bundles. These advantages are paralleled by low maintenance costs and a very high yield near to 90%.

With similar investment and production costs the new technology can give a 30% yearly production increase and winning ingot characteristics.

Preamble. From Liquid Metal to Transportation Means

Nowadays, anybody standing outside a modern smelter would see a continuous coming and going of trucks that transport, to the nearest port, containers full of both semi-finished products (rod, slabs, billets) and re-melt products (T-bars, sows, ingots) from the production site to the final destinations. The semis and the re-melt products are identified as “commodities” and have a reference price worldwide and standardized shapes and characteristics.

Now the question that could come on the table is: who/what is going to determine the shape and the packaging of the “commodities” produced in the smelters? The production equipment? The equipment used by the final users of these commodities? Both? In general, the grounds of the shape, weight and packaging of any aluminium commodities are:

a) easy, safe and cheap production;

b) easy, safe and cheap transportation;

c) easy, safe and cheap usage.

It would be very interesting to interview both the users and the producers of these commodities to better understand the reaction of the users in the case where the producers of such non ferrous commodities suggested new and better shapes and different weights.


Why are the “Light” Ingots so Heavy?

Talking about ingots, geometry and weight have varied in the past, but eventually the 50lb (22.7kg) ingot, with a design that varies just in small details from producer to producer, has piqued the worldwide preference and has become the most common ingot on the market, even though not the only one. Figure 1, by courtesy of Nalco, shows a typical 22.7kg ingot while Figure 2 shows a typical bundle of similar ingots.

Figure 1. Typical open top ingot (by courtesy of alco).

To understand the reasons for this weight, and consequently of this shape, we should go back to the pioneer systems used by the industry to produce ingots, and also consider the level of automation and the technology available at the end of the 50’s / beginning of the 60’s, when the standard weight of 50lb was consolidated.

The traditional system, based on a chain of open top mould, has some limitations due to the following principal factors:

a) the filling process is not continuous and must be interrupted between two subsequent moulds;

b) several studies have been done of filling star wheels, but still a difficulty remains to pour, in steady regime, a precise quantity of metal in each mould, again because the process is not a continuous one and is affected by start and stop sequence. Therefore the pouring flow in each mould goes from zero flow to a maximum flow and to zero flow again. Variation of speed involves turbulences and risk of dross formation.

In addition, some thirty years ago, at the beginning of the adventure of the industrial production of aluminium ingots, automatic stacking systems were not even in the fantasy of the most farsighted designers. Therefore, the size and weight of the ingots were such as to be compatible with the manual handling operations i.e. de-moulding and stacking (see Figure 3, taken from the cover page of “Aluminium Cast House Technology 2005”, showing the pioneer ingot casting machine at Comalco Bell Bay – Tasmania in 1968).

Figure 2. Typical bundle made of open top

cast ingots. Figure 3. Pioneer ingot casting line.


The above considerations have pushed the designers of these systems to find the best compromise between the weight of each ingot and the maximisation of the production output. Since the very beginning, this compromise has been satisfied by the weight of 22.7kg – 50lb, which has survived till the present day although some companies have turned their attention towards the production of lighter ingots of 10kg.

Weight of the Ingot and Ingots Production Rate

Looking at the present scenario, if we consider that the open top system has a production rate in the range of 25,000kg/hour it appears that this kind of equipment is able to cast some 1,100 ingots/hour, i.e. one ingot every 3.2 seconds. Indeed, the hourly output of a mould chain depends on the weight of the ingots and the length of the mould chain according to the equation:

ss tILW

P×

×=

Where : • P is the hourly output of the line

• W is the ingot mass

• L is the line length

• Is is the spacing between moulds

• ts is the solidification time

From this equation it appears that, once the length of the chain conveyor has been determined (i.e. the equipment cost), somehow, the greater the weight the higher is the output…. and 50lb is a nice round number. We do not see other technical reasons.

Despite some recent over-refinements of casting star wheels and the use of robotized skimming devices, dross-free ingots are still a commercial driving force. The complex shape of the ingot comes from the need to obtain a big surface of thermal exchange in order to reduce the length of the mould chain, which anyway requires about 1.3 to 1.4 meters in length per tonne/hour produced. Moreover, such geometry is necessary to facilitate the de-moulding and to “embed” the bundles layers to improve their stability. The 22.7kg open top Aluminium ingots are used worldwide and the technique to produce such commodities is well proven and well established but does not represent the latest state of the art, although it has become a consolidated tradition. However, even the most well-established traditions must be overcome when new process techniques are developed to offer a quality jump with lower or equal costs (equipment and transformation).

Now, if we consider that worldwide ingots are delivered packaged in bundles of 1 tonne approx., with dimensions meant to maximize the loading of containers and trucks, is it really necessary that the weight of each ingot is 22.7kg? And considering that the ingot users load their furnaces with one complete bundle after the other is it really necessary that the weight of each ingot is 22.7kg? Or is it possible that the same 1 tonne bundle could be composed of lighter ingots, for instance 30lb?

Last but not least, in the unlucky event of manual handling, lighter ingots – 30lb (13.6kg) are much handier and less harmful for a human being’s backbone. The new technology allowing higher production rate and lower ingot weight is now available.

Figure 4 (by courtesy of Vedani Carlo Metalli – Italy) shows ingot bundles produced with the revolutionary patented Track & Belt ingot casting machine, invented and developed by Mr. Giulio Properzi – Chairman of Properzi Company and author and owner of several patents in the casting and rolling area of the non ferrous industry.


These compact, stable and repeatable bundles have a weight in the range of 1,000kg and are composed, in this application, of subsequent of layers of 10kg ingots. Similar bundles are produced at Alba (Bahrain) and Dubal (UAE) using Properzi equipment, with growing success and OEE (Overall Equipment Efficiency).

Figure 4. View of Properzi ingot bundles before strapping.

Casting Speed and Production Rate of Properzi Track & Belt System

The ingots displayed in Figure 4 are produced at the rate of 2,000 ingots per hour, at the casting speed of 0.38m/s approximately. This casting speed in continuous process is given by the following equation:

kAP

castV ××∂

=

Where:

• P is the hourly output of the line

• Vcast is the casting speed, i.e. the speed of the liquid metal flowing into the casting machine, and the speed of the solidified cast bar. This speed is expressed in [m/s]

• ∂ is the specific weight of Aluminium (we assume 2,700kg/m3)

• A is the cross section area of the mould of the casting machine

• k makes the various units used homogeneous

Going through the numbers:

smkmmmkg

hkgVcast /381.0

5400/2700/20000

23 ≅××

=

To approach higher ingot production rates, we didn’t focus on the 50lb (22.7kg) ingots, but on lighter ingots 30lb (13.6kg), as we do believe that the future will belong to light ingots.

Keeping the same well-proven casting speed and ingot length, if we increase the weight of the ingot from 10kg to 13.6kg (30lb), the cross section area of the continuous mould of the casting machine must be increased from 5,400mm2 to 7,000mm2 approx.


The production rate is given by the following equation:

kAVcastP ×××∂=

The various units of the equation above have been already introduced, therefore we can go through the numbers:

hkgkmmsmmkgP /259007000/381.0/2700 23 ≅×××= *

(*) Remark: 7,000mm2 is the cross section area of the mould of the Track & Belt designed for producing 30lb ingots.

The above production rate surpasses the productivity of many open top systems and the bundles will still have a nominal weight of 1 tonne and will be much more compact and stable (as shown in Figure 4).

Today this quality/technological jump is offered by the ingots produced with the Properzi Track & Belt Ingot Caster (patented) which is described further on. We have already sold, installed and put into production several lines of this generation.

The Properzi Track & Belt Ingot Caster

Let’s make a brief presentation of the working principle of the Track & Belt system, derived from the experience gained in 60 years with more than 300 casting wheels in operation for Aluminium, Copper, Zinc and Lead. With reference to the Figure 5, the liquid metal coming from the furnace passes through the launders and tundish set, where the flow is controlled and adjusted and runs into the Track & Belt casting machine. In this machine a series of wheels (tracking wheel A and B) move the caterpillar formed by a plurality of Copper segments arranged in a chain. The steel belt is moved by wheel 2 and kept in tension by the tensioning wheel and wheel 1. The Copper segments and the steel belt during their motion determine a continuous closed mould of approximately 8m length that can be considered as a travelling heat exchanger.

This closed and continuous mould (see Figure 6) is formed on three sides by a copper block, while it is closed on the top by a continuous steel belt.

Figure 5. Front view of the Track & Belt.

Figure 6. Cross section of the continuous closed mould.

The continuous closed mould is cooled on the four sides by adjustable spray nozzles spraying water from the entry point of liquid Aluminium to the exit point where the cast bar leaves the caterpillar. In this ideal heat exchanger the molten Aluminium is cooled from the initial casting temperature of 690°C and is solidified in a form of a continuous cast bar, having a cross sectional area of 7000mm2


approx. (less the shrinkage) and trapezoidal shape. The cast bar passes through the bar cooler where it reaches a temperature of 480°C - 520°C (depending on the alloy being cast and the working conditions), before passing into the stamping machine and finally by the rotary shear, where the cast bar is cut into ingots of repeatable length (variation +/-0.5%). After being cut, the ingots are cooled in a dedicated continuous water cooling tunnel down to a temperature of some 50 - 70°C. A Cartesian stacker, designed by Properzi company, piles up the ingots in bundles ready for the downstream operations of weighing, strapping, labelling and eventually inkjet marking. Figure 7 shows a Track & Belt ingot casting machine in operation.

Data coming from equipment in operation, although producing 10kg ingots, have recorded, on yearly basis, OEE (Overall Equipment Efficiency) close to 90% (seven days per week and 330 days per year).

With such a line, considering an OEE of only 86% one can easily produce 170,000tpy of ingots each weighing of 30lb (13.6kg). In fact:

ytyddhhtP /000,17086.0/330/24/25 ≅×××=

The ingots produced with the Track & Belt system shown in Figure 4 are flat, smooth and without dangerous cracks in the top surface. The bundles are stable and three straps (steel or PET) are sufficient to secure each bundle even when long transportation is required. We might also add that ingots produced with the Track & Belt caster are engraved one by one with all the common traceability data (logo, cast nr, alloy type).

Cast Bar

Figure 7. Track & Belt in operation. Figure 8. Sketch of Properzi Bundle.

What is Heavier? One Kg of Lead or One Kg of Aluminium?

The question above is clearly a rhetorical wrong question. The right question would cover the volumes occupied by the same mass of Lead and Aluminium. But if we talk about weight and dimensions of ingot bundles, Mr. “Consumable Cost” and Mrs. “Logistic Cost” (chapters present in any company) could argue that my question is not 100% a rhetorical one. Indeed, more weight in the same or less volume surely brings an interesting saving at the end of the day.

We have taken the liberty of comparing the bundles made by Hillside Smelter (leaflet available on their web site) and the Properzi bundle in terms of dimensions and weight, reflected in the table below.


Open Top Ingots (22.7kg typical) (from Hillside Smelter data)

Properzi Ingots (13.6kg)

Ingot Dimensions: Length < 740mm < 740mm Height 118mm 62mm Width 170mm 120mm Weight 22.7kg (50lb) 13.6kg (30lb)

Bundle composition: 44 ingots 1,000kg – height 1,000 -1,160mm

Option 1 (70 ingots) 950kg; height ~805mm

Option 2 (82 ingots) 1,115kg; height~ 930mm

Option 3 (94 ingots) 1,278kg; height ~1,055mm

Option 4 (106 ingots) 1,441kg: height ~1,178mm

The Properzi ingots can be stacked either with the traditional robot stacker or with our Cartesian stacker, and different bundle height/weights can be easily obtained by operating the program of the stacker and downstream packaging machines. The bundle compactor is never required and…less equipment means less problems and less costs.

We want/need to ship bundles in the range of 1,000kg, therefore let’s compare the bundle composed of 44 ingots of 22.7kg each with the Properzi bundle composed of 82 ingots of 13.6kg each.

This bundle has a weight of 1,115kg and dimensions of 740mm x 740mm (base) and repeatable height of 930mm.

Is it possible to gain any saving in the strapping material?

Using the strapping pattern 2X1, sufficient even for long ocean transportation, we will use the following nominal quantity of strap (consumable material):

Strap 1: (740+930+740+930)mm = 3,340mm Strap 2: (740+806+740+806)mm =3,092mm Strap 3: (740+930+740+930)mm = 3,340mm Total ~ 9772mm The specific consumption is approx.: 8,770mm/ton.

Now, let’s take into consideration a bundle of traditional 50lb ingots and let’s make the same investigation considering the best case, i.e. bundle minimum height of 1,000mm and weight of 1,000kg:

Strap 1: (740+950+740+950)mm = 3,380mm Strap 2: (740+1.000+740+1.000)mm = 3,480mm Strap 3: (740+950+740+950)mm = 3,380mm Total ~ 10,240mm The specific strap consumption is: 10,240mm/ton.

We have not considered in either set of calculations the extra material (very minor amount) necessary for bending the strapping.

The saving in the specific strap consumption (mm/t) using Properzi technology is in the range of 15%-16%.


Further saving can be achieved if we take into account that 19mm PET straps can be used instead of the 32mm steel.

The compactness of a Properzi bundle also allows cost saving in storage and transportation, since three/four bundles can be easily piled up without any risk.

Simple ideas, creativity, and high technical skills are often the right “ingredients” for generating great inventions, and the intuition of applying the continuous casting method to the production of re-melt ingots is a vivid example of this concept.

Conclusions

1. Continuus-Properzi invented the casting system with casting wheel and belt some 60 years ago; almost all the worldwide Aluminium rod is produced using this system.

2. During the 90’s Properzi approached the ingot production industry with a modified casting machine having a wheel diameter of 4.2m.

3. At the beginning of the new millennium the casting machines for producing ingots have been completely changed and the casting wheel has been replaced by a plurality of Copper blocks in sequence as a caterpillar. This machine has been called Track & Belt Ingot Caster.

4. This equipment is available at a comparable or even more competitive price than other existing technologies (less equipment cost and saving in the consumables).

5. Five Track and Belt casting systems have been manufactured and sold for producing 10kg ingots at the rate of 2,000 ingots per hour.

6. The technological jump is addressed by the line Track & Belt type producing 25tph of 30lb ingots, meeting 1 tonne per each bundle or more, thus achieving savings in production, transportation and storage.

References

[1] Chain Conveyor Remelt Ingot casting Improvement (J. Grandfield, V. Nguyen, P. Rohan, M. Couper, K. Oswald – CAST Cooperative Research Centre (CAST)

[2] Ingot Line Productivity (Lyndon Meadows – o.d.t. Engineering Pty. Ltd.)

[3] Italian Patent no. MI2001A002556 and several international pending patents extension

[4] NALCO (National Aluminium Company – India) leaflet

[5] Aluminium Ingots - Hillside (BHP Billiton) leaflet


Aluminium Bahrain - benefits at ALBA from use of improved graphite rings for production of extrusion ingots

Mohamed Ali Kadhima, Hussain Al Ali Jalal Ghuloom, Garry Martin, Michael

Jacobs

Aluminium Bahrain, ALBA, Kingdom of Bahrain, Middle East. a [email protected]

Keywords: graphite ring, extrusion ingot, Airslip™, Alba, permeability

Abstract

Surface quality and metallurgical characteristics of extrusion ingot have a significant impact on the extrudability of the billet.

Both the type of casting mould and mould maintenance significantly influences billet surface quality. ALBA, like the majority of extrusion ingot producers, uses moulds with graphite technology, where a mix of oil and gas is injected through a graphite ring into the mould to improve billet surface quality.

ALBA has a policy of continual improvement in the areas of safety, product quality and productivity to enhance customer satisfaction. During 2008, an extensive campaign was conducted in ALBA’s casthouse to select the optimum type of casting rings for Airslip™ moulds supplied by Wagstaff Inc. The graphite casting rings evaluated must be suitable for the casting conditions and environment within the casthouse at ALBA.

This paper details the methodology and the criteria set for the selection of the optimum graphite casting ring type as well as the results achieved. The results of the work showed a preference for L type casting rings, which enhance the quality of the extrusion ingots and extend the service life of the graphite ring.

Introduction

Aluminium Bahrain, ALBA, in the Kingdom of Bahrain, Middle East, is one of the most modern and large aluminium smelters in the world. In its casthouse operation, which produces various ingot shapes, inclusive of extrusion ingot, there is a strong commitment to a continuous improvement approach in the operation including in the areas of safety, environment, quality and overall production efficiency, namely productivity and recovery.

For casting of extrusion ingot using Airslip™ technology the graphite ring is one of the most critical parts of the mould. Hence it was recognised at ALBA that there was an opportunity to further investigate the performance of graphite rings and to determine the differences between the various types, that is, C, K and L as supplied by Wagstaff Inc in terms of their influence on as-cast extrusion ingot surface quality and billet scrap and recovery.

The work reported in this paper outlines the methodology used in the study along with the results achieved and conclusions from the work done in ALBA’s Casthouse Number 3 operational environment.


Casthouse Description

Within ALBA there are three Casthouses.

Casthouse 1 is the original Casthouse, which has been in operation since the start up of Alba in 1971. This Casthouse is dedicated to production rolling slabs and T-ingot.

Casthouse 2 was commissioned in 1992, as part of the expansion of Pot Line 4. This Casthouse produces rolling slab, as well as standard and foundry alloy ingots.

Casthouse 3 was commissioned in 2005, as part of the expansion of Pot Line 5. This Casthouse uses state of the art equipment for production of high quality extrusion ingot. The sizes produced in Casthouse 3 are 152mm, 178mm, 203mm, 216mm, 229mm and 254mm diameters. The new Casthouse 3 has an open plan as shown in Figures 1 and 2. The equipment chosen facilitated the elimination of chlorine within the operation [1] and a safer work environment.

The three Casthouses have over 500 full time employees and produce a wide range of products as per Table 1.

Table 1. ALBA Ingot Product Mix

Product Metric Tonnes Per annum

Extrusion Ingot 350,000

Rolling Slab 155,000

Foundry Ingot 50,000

Standard Ingot 100,000

Liquid Metal Delivery 200,000

Total 855,000

Figure 1. ALBA’s Casthouse CH3. Figure 2. ALBA’s Casthouse CH3.


Function of the Graphite Ring in the Casting Mould

For extrusion ingot casting using Airslip™ technology, the graphite ring is one of the most critical parts of the mould and the overall casting process. The graphite ring influences the metal solidification in the mould and the resulting microstructure determines the properties of the as-cast ingot. When liquid metal enters the casting mould it touches the air film created inside the mould. The air film serves as a frictionless surface and a thermal insulator. Lubricant oil is also injected through the graphite ring to form a lubricant film and hence avoid direct contact of the billet surface to the graphite ring. The gas must balance the metallostatic head pressure. The permeability of the graphite is the primary factor for the even and uniform distribution of oil and air films and air pressure. If there is a significant variation in the permeability within the graphite ring, the flow characteristics will vary which in turn adversely affects the shell zone and surface characteristics of the as-cast extrusion ingot. Hence it is critical to have a uniform permeability across the graphite ring and to avoid variation in the process. Figures 3 and 4 show a cross-section of the casting mould and the graphite ring type used at ALBA. The graphite ring and its function are further described in references [3,4].

Figure 3. Cross-section casting ingot mould. Figure 4. Graphite ring.

Methodology and Results from Evaluation of the Graphite Ring at ALBA

Our study was carried out to investigate the difference between graphite rings types C, K and L, as supplied by Wagstaff Inc, in the work environment in ALBA’s casthouse 3 (CH3). The purpose of the work was to improve surface quality of the extrusion ingots and to reduce the drag mark on as-cast extrusion ingot.

The study was split into two phases. Phase I work was done in October 2007 and conducted on 203mm diameter ingot size, it used three types of graphite rings; C, K and L as supplied by Wagstaff Inc. The best result, in terms of surface quality, was achieved using an L type graphite ring. The relationship between the types of graphite ring used and surface quality of the as-cast ingot is strongly correlated. The alloys mainly cast during phase I work were AA6063 and AA6061.

In phase I of the work, a detailed study was conducted on all types of graphite rings available: C, K and L where the experiment was designed to quantify the difference in performance between the three types. The results are illustrated in Figure 5, which indicates the average performance in surface quality for the three ring types of 97% for L type, 90% for K type, and 74% for C type. Examining Figure 5 shows the sensitivity of the different types of the graphite rings to overall

graphite ring


process conditions and environment. While in ring types C and K the graphs shown are fluctuating, the L type ring is less affected. This in turn extends the life of the L type ring beyond the lives of the C and K type rings, which is obvious from the slope of the graphs.

It is important to note that these results were obtained in ALBA’s CH3 environment and results achieved in other casting centres may vary from the ones reported herein.

Figure 5. Comparison of C, K & L Graphite Rings % scrap performance with number of casts.

ote: The casts in the above chart are not consecutive.

As the above results suggested that ALBA shift to using L type rings, Phase II was conducted during 2008 on 216mm size ingots where all moulds on the station were of type L to quantify the effect with sufficient data to obtain a confident statistical conclusion.

The observations during the trial campaigns were very encouraging. During the few first drops slight drag marks were reported in the butt area (around up to 300mm). The idle air pressure was decreased from 3 bars to 2 bars in order to prevent the graphite ring from being dry at the beginning of the cast. This change indicates that the L type ring has better permeability compared with the C type ring that its norm was 3 bars during the idle stage.

During the campaign of L type no drag mark defects were reported as quality scrap, while for the latest C type campaign the drag mark scraped were 0.08% of the total metal cast. Table 2 below shows the percentage of surface defect by type for L type graphite rings compared with the historical data for C type graphite rings. The surface defects related to graphite ring type were analysed for the two campaigns of C and L type. The results are listed in Table 2 and presented in Figure 6.

A uniformity test was done for C and L types and descriptive statistics (Table 3) show that L type flow is more consistent than C type. Figures 7 and 8 show the uniformity test machine used in the study and the radar chart comparing flow output around the perimeter of each ring for C and L type with same input. From the radar chart in Figure 8 it is clear that the L type ring is capable of supplying higher flow rate compared to C type rings with more uniformity of the cast. This change indicates that L type rings have better permeability compared with C type that its norm was 3 bars during idle stage.


Table 2. Recoveries – comparison of C vs. L type rings

Comparison C-Type Campaign* L-Type Campaign No. of Casts 64 66

Total Cast Tonnage 3268 3172 Hot tear % 0.28% 0.10% Bleeding % 0.02% 0% Drag Mark % 0.08% 0%

Rough Surface % 0.06% 0% Total defects % 0.44% 0.10%

Not Casted Logs % 0.45% 0.10% Total defect loss (ton)/year 1,562 355

Opportunity lost of not casted/year 877 193 Net saleable difference (tonne)/year 1,702 using L type graphite

Figure 6. % defects versus type of defect and C and L type ring.

Table 3. Descriptive statistics of air outflow in graphite rings C and L types. C-Type L-Type

Mean 42 52 Standard Error 2 1 Median 42 53 Mode 54 45

Standard Deviation 8 6 Sample Variance 69 33

Range 28 19 Minimum 28 40 Maximum 56 59


Figure 7. Uniformity Air Flow test unit. Figure 8. Radar chart for air outflow.

Implementation of L type Graphite Rings

During the study and subsequent implementation of L type graphite rings, a set of criteria was developed to evaluate and quantify the differences in terms of billet surface quality. The quantitative scale that was used is listed below:

• 0: Full log scrap due to surface defect;

• 2: Continuous D/M. More than 2 lines;

• 4: Continuous D/M. 2 lines or less;

• 6: Non continuous Light D/M, 4 lines or more;

• 8: Non continuous Light D/M, less than 4 lines;

• 10: No marks.

The decision to change over to L type graphite rings was taken jointly between the technical service team and the production team in the casthouse operation. This was done following the comprehensive study of the available types of graphite rings as supplied by Wagstaff Inc. A contingency plan was also implemented to keep a minimum level of C type graphite rings as an emergency reserve while implementing the L type change over. The implementation period from November 2008, is still ongoing and as of March 2009 there have been no recorded issues.

Conclusion

The study and results at ALBA from use of three different types of graphite ring, as supplied by Wagstaff Inc, and under the operating conditions in ALBA's CH3 casthouse, have shown a preference for use of the L type graphite ring. The implementation of the L type graphite rings at ALBA has resulted in improvements in both the surface quality of as-cast AA6000 series extrusion ingot and cast recovery. This outcome has been achieved as a result of the detailed study done and with the team effort of all functions in the ALBA casthouse.


Acknowledgments

Thanks are conveyed to a Wagstaff representative during the commencement of the phase I campaign for his positive contribution. Thanks also go to the ALBA casthouse mould shop for their cooperation during the campaign. Also thanks to the casthouse operation team for their help achieving a successful assessment campaign.

References

[1] Ahmed, A. R. (2007) Elimination of Chlorine at ALBA, Proceedings of Aluminum Cast House Technology, 2007, Sydney, Australia, p. 25.

[2] Rahim Al-Ansari, T.A. (2009) Benefits to safety at ALBA from use of the Wagstaff AutofloTM system for casting of extrusion ingot, Light Metals 2009, TMS, p 675.

[3] Anon (2008), High-quality graphite casting rings can result in a superior billet, Aluminium Times, Sep. 2008, p 23.

[4] Grandfield J.F. and McGlade P.T. (1996) DC casting of aluminium: process behaviour and technology, Metals Forum, (1996),20, (Annual), p 29.

[5] Billet Casting operations manual, Wagstaff Inc.


Keywords Index

6000 Series Alloys 213

A

AirslipTM 251

Alba 251

Alloy Management 3

Aluminium Recycling 95

Aluminium Reverberatory Furnace 111

Aluminum 9, 37, 45, 53,61, 155, 179,

243

Aluminum Alloy 193

B

Batch Homogenizing 85

Billet Sump 179

Billet Tooling 205

Boiling 235

Boron Treatment 129

C

Capacity Analysis 165

Cast House 45, 53

Cast House Modelling 165

Casthouse 205, 243

Casthouse Safety 19

Casthouse Technology 27

Casting 187, 213

Casting Technology 205

CBERJC 187

Ceramic Foam 137

Charging 103

Clad Metal 175

Combo Bag 223

Continuous Homogenizing 85

Cyclone 147

D

DC Casting 179

Degassers 3

Degassing 205

Detection 155

Direct Chill Casting 175, 193

Dross 37, 45, 53, 61,77

Dry Hearth 103

E

Elaboration 71

EMP Pumps 111

Extrusion Ingot 251

F

Filtration 3, 137, 147

Floating Crystal 193

Fluid Flow 223

Furnace 3, 45, 77

Furnace Circulation 111

Furnace Design 103

Furnace Technology 95, 103

Fusion Clad Casting 175

G

Grain Refinement 213

Graphite Ring 251

Greenhouse Gas Emissions 27

H

Hard Alloy 205

Hot Tearing 213

I

Inclusion 155

Ingot 243

Ingot Casting 235

K

K-Mold 155

L

LAIS 155

Launder 119

Life Cycle Assessment (LCA) 27

LiMCA 155

M

Macrosegregation 193


Mechanical Pumps 111

Melt Quality 77

Melting Furnace 95

Metal Level Control 119

Metal Transfer 119

Modeling 37, 61

Molten Aluminium Distribution 223

Molten Metal Explosion 201

Molten Metal Pumps 111

Molten Metal Water Explosion 201

Mould 3, 235

N

New Billet 187

Nickel 129

O

Operation 77

Organic Coating 201

Oxidation 37, 61, 71

P

Packing 85

Permeability 251

Phosphine 137

Podfa 155

PPE 19

Prefil 155

Processing Technology 53

Production 243

Productivity 165

Protective Clothing 19

Q

Quality 71

R

Reference Casthouse Centre 9

Refractory 119

Rolling 175

S

Sawing 85

Secondary Smelting 95

SELEE 137

Shrinkage 193

Simulation 223

Starting Crack 179

Support 9

T

Technology 71, 187

Trace Elements 129

U

Ultrasonic Testing (UT) 85

Ultrasound 155

V

Vanadium 129

W

Wise Chem 201

Authors Index

A

Allen, L. 95

Arsenault, A. 223

Aubrey, L.S. 137

B

Baker, P.W. 165

Barresi, J. 77

Bischoff, T.F. 175

Brandt, M. 155

Brocato, C.M. 243

Butnariu, D. 71

Butnariu, I. 71, 71

C

Campbell, P. 111

Canullo, M.V. 119

Collins, M.J. 45, 53

Collins, R.J. 187

Courtenay, J.H. 147

D

Daroqui, F. 119

Dubé, J.P. 223

Dupuis, C. 77

E

Easton, M. 213

Emes, C. 187

Eskin, D.G. 193

F

Furu, T. 9

G

Ghuloom, H.A.A.J. 251

Girard, G. 77

Grandfield, J.F. 27, 129, 155,235

H

Hamer, S. 205

Herbert, J. 45, 53

Hétu, J.F. 223

Hipwood, P. 95

Houghton, B. 95

I

Ilinca, F. 223

J

Jacobs, M. 251

K

Kadhim, M.A. 251

Katgerman, L. 193

Koch, H. 179

Koehler, T. 179

Koltun, P. 27

L

Laje, R.A. 119

Larouche, D. 223

Lee, M. 19, 37

Liu, G.W. 61

Locatelli, J. 61

Lowery, A.W. 201

M

Martin, G. 251

Martín, M. 119

N

Newman, P. 103

Nguyen, V. 235

Niedermair, F. 85

O

Olson, R. 137

Ottaviani, J. 119

P

Peel, A. 45, 53

Pereira, G.G. 37

Poynton, S. 155


Prakash, M. 37

R

Reusch, F. 147

Rinderer, B. 37

Riverin, G. 77

Roberts, J. 201

Rohan, P. 37, 235

Rosefort, M. 179

Roth, D. 45, 53

S

Sinden, D. 175

Smith, D.D. 137

Steen, I.K. 9

StJohn, D.H. 213

Sweet, L. 213

T

Taylor, J.A. 37, 129

Tharumarajah, A. 27

Todd, B. 235

Tremblay, S.P. 223

W

Wagstaff, R.B. 175

Whiteley, P. 3

aluminium cast house technology xi : selected, peer reviewed papers from the international...

Documents