1 INTRODUCTION

In the context of global competition, complementary experimental and numerical approaches allow to develop high performance cutting processes. However, understanding of the interface phenomena (Fig. 1) occurring at the tool-chip contact (secondary shear zone) and at the tool-workpiece contact (rubbing zone) have to be improved [1].

Cutting tool

Chip

Primary shear zone

Workmaterial Vc

Vc Vchip ~

h

f

Rubbing zone

hVc.f

Vc : Cutting speedf : feed

Fig. 1. Illustration of the various strategic zones in cutting.

The Coulomb model, with a constant coefficient usually used to simulate the friction, is not relevant because it doesn’t take into account the temperature and pressure influences. In the case of steel

machining, usual cutting conditions lead to severe tribological conditions: high velocities (60-600 m.min-1), high temperatures (up to 1000°C), high pressures (up to 2 GPa) [2-3].

The pin-on-disc system is the most widely known set-up for tribological investigations. However, it is unable to simulate temperature and pressure found during cutting [4]. Moreover, during a cutting operation, the workmaterial is never more in contact with the tool. Based on these statements, many scientists [5] have work on open tribo-system. They proposed configurations in which a pin is placed just after a cutting tool during the machining of a tube’s flat face in order to rub on a refreshed surface (Fig. 2a and 2b). Friction conditions are well reproduced but the set-up is very difficult to perform, expensive and the duration limited. As a consequence, a new configuration of frictional tests has to be designed. The principle proposed by Hedenqvist et al. [6] is a promising configuration (Fig. 2c). This paper present a new experimental set-up based on his system in order to investigate the friction coefficient

ABSTRACT: This paper aims to identify a friction model able to describe the friction coefficient at the interface between the tool, the chip and workpiece during the dry cutting of a AISI316L austenitic stainless steel with TiN coated carbide tools. A new tribometer has been designed in order to reach relevant values of pressures, temperatures and sliding velocities. This set-up is based on a modified pin-on-ring system. Additionally a numerical model simulating the frictional test has been associated in order to identify local phenomena around the spherical pin, from the standard macroscopic data provided by the experimental system. It has been shown that the friction coefficient is mainly dependant on the sliding velocity, whereas the pressure has a secondary importance. Finally a new friction model has been identified based on this local sliding velocity.

Key words: Friction; tribology; numerical modelling; cutting; coating

Development of a friction modelling method in dry cutting of AISI 316L austenitic stainless steels

C. Bonnet1, F.Valiorgue1,2, J. Rech1, J.M Bergheau1, P.Gilles2, C.Claudin1 1Ecole Nationale d’Ingénieurs de Saint Etienne (ENISE), Laboratoire de Tribologie et Dynamique des Systèmes (LTDS), UMR CNRS 5513, 58 rue Jean Parot 42023 Saint-Étienne, France e-mail: cedric.bonnet@yahoo.fr

frederic.valiorgue@enise.fr joel.rech@enise.fr

2AREVA NP, 92084 Paris La Défense, France

occurring at tool-chip-workpiece interfaces during the high speed dry cutting of a AISI 316L stainless steel with TiN coated carbide tools.

(a) (b)

(c) Fig. 2. Various technologies of tribometer (a) Olsson’s

tribometer [5], (b) Zemzemi’s tribometer [5], (c) Hedenqvist’s tribometer [6].

2 EXPERIMENTAL SET-UP

2.1 Material description

A cylindrical bar, made of AISI 316L, is fixed onto a lathe chuck, as illustrated in Figure 3.

Fn

Ft

Heat Fluxmeas.

1-Workpiece : 316L

3- Pin :Carbide +

TiN coating

Forcesmeas.

2-Cutting tool refreshingthe surface

6-Dynamometer

4-Pneumatic jack

5-Pin holder 7-Thermistor

Fig. 3. Tribometer designed for cutting applications.

A pin made of cemented carbide with a TiN coating and having a spherical geometry Ø 9 mm is pressed onto the cylindrical surface by means of a jack. Thanks to the helical trajectory of pin, the surface contact is continuously regenerated. The pin is maintained by an instrumented pin-holder able to provide data about the instantaneous heat flux entering into the pin [7]. It is fixed onto a dynamometer, in order to provide the apparent normal and tangential force (macroscopic forces). After each friction test, a cutting tool refreshes the

surface and a belt finishing operation is performed in order to obtain a very low initial surface roughness.

2.2 Testing conditions

Testing conditions have to be chosen in accordance with the frictional conditions estimated along the tool/chip/workpiece interfaces. For instance, the machining of a AISI 316L with a TiN coated carbide tool in dry cutting conditions is around Vc~120 m.min-1. Based on Figure 1, it is of evidence that the macroscopic sliding speed at the tool-workpiece interface is almost equal to the cutting speed ~ 120 m.min-1. On the chip-tool interface, the average sliding velocity depends on the compression ratio β, which is around 2. So at this interface, the macroscopic velocity is 2 times lower than the cutting speed � 60 m.min-1. As a consequence, the characterization of the frictional properties at the tool/chip/workpiece interface needs performing friction tests in the following conditions: sliding velocity: 60-120 m.min-1, pressure: 1-2 GPa [2].

2.3 Experimental results

The sliding velocity was the single modified parameter in this work. A normal force equal to about 1000 N has been applied onto pins, which lead to an average pressure of 1.8 GPa.

The first output data provided by this set-up is the ratio between the tangential and the normal force. This coefficient can be defined as an apparent

friction coefficientn

tapp F

F=µ . (1)

The evolution of µapp versus sliding velocity is plotted in Figure 4. This confirms that it is not relevant to consider the friction coefficient as independent from sliding speed. Over 120 m.min-1, a modification of the frictional behavior is observed.

Workmaterial : AISI 316 L steel Pin : Carbide + TiN coating

Sphere diameter : 9 mm

Fig. 4. Apparent friction coefficient.

The evolution of heat flux transmitted to the pin φpin versus sliding velocity is plotted in Figure 5. It appears that heat flow increases with the velocity, which is coherent with the fact that more energy is produced and dissipated. The frictional behaviour may be modified over a certain value (saturation).

Workmaterial : AISI 316 L steel Pin : Carbide + TiN coating

Sphere diameter : 9 mm

Fig. 5. Heat flux measured into the pin.

3 NUMERICAL POST-TRAITEMENT

A finite element model, simulating the same experimental conditions of plowing and friction, has been developed. Macroscopic experimental data µapp

and φpin are used to fit the numerical model for 3 sliding velocities: 60-90-120 m.min-1.

3.1 Description of the model Vs

Vs

Fig. 6 Geometry of the numerical friction model.

The 3D model (Fig.6) is based on the work of Zemzemi et Al. [5]. Considering the high strain and strain rate, an explicit formulation has been chosen. A thermo-mechanical model has been programmed with ABAQUS explicit.

A Johnson-Cook flow stress model has been used to model the AISI 316L steel. This model is dependent on strain pε , strain-rate pε& and temperature T so as

to observe softening phenomenon:

0

0 0

( ) . 1 . ln . 1m

pneq p

F

T TA B C

T T

εσ ε

ε

− = + + − −

&

&

(2)

The Johnson-Cook parameters used have been chosen thanks to the works of Umbrello et al. [8]. Moreover our model considers the hypothesis from Bowden et al. [9] with: µplast describes the plastic deformation part of µapp and µadh corresponds to the adhesive friction coefficient: app adh plastµ µ µ= + . (3)

The model uses thermo-dependent properties for the WC/Co substrate [6] and the AISI 316L workpiece: Table1. AISI 316L workpiece properties Parameter Temperature Value Specific heat (J.kg-1.°C-1)

20 450 300 545 500 570 800 625 1100 670

Thermal Conductivity (W.m-1.°C-1)

20 14 300 18 500 21 800 24 1100 29

Density (kg.m-3) 20 8000 300 7890 500 7800 800 7660 1100 7510

3.2 Thermal boundaries conditions

The model includes the heat generation induced by the friction in addition to the heat dissipated by the plastic deformation (plowing) of the workmaterial. As shown by Figure 7, the frictional heat flux φfriction is dissipated at the interface in a narrow layer, whereas the plastic deformation heat flux φplast is dissipated only in the workpiece in a large volume.

Fric tion heat fluxtransm itted to part 1Plastic deformation

heat fluxTransm itted to part 1

Fric tion heat fluxtransm itted to part 2

Plastic deformation heat flux

D iffused in part 2 Fig. 7. Heat flux sources during friction tests.

3.3 Fitting method

For each testing condition simulated, the indentation depth h, the adhesive friction coefficient µadh and the heat partition coefficient α have to be determined. In this aim, an iterative method, presented in Figure 8 has been developed. The experiments has provided

the intensity of the normal force Fn, an average value of the apparent friction coefficient µapp and an average value of the heat flux transmitted to the pin φpin. So the key is to adapt h, µadh and α in order to fit the experimental (deviation lower than 5%).

Fig. 8. Optimization methodology.

3.4 Numerical results : Pressure and Temperature

The average contact pressure is around 1.8 GPa for each velocities. This result was expected since the pressure can only be modified by the viscosity component of the flow stress model, in a small range: 60 to 120 m.min-1. The contact temperature on is about 500°C for the test performed at 90 m.min-1.

3.5 Friction model depending on local sliding velocity

The numerical model can also be used to analyze the local sliding velocity at the pin/workpiece interface. It is of evidence that adhesion limit the sliding velocity along the contact. Figure 9 shows an example of local sliding velocity Vls.

Fig. 9. Local sliding velocity at the surface of the workpiece.

Equation (4) presents the model linking the adhesive friction coefficient and the local sliding velocity in the range 35-90 m.min-1, corresponding to macroscopic friction velocities 60-120 m.min-1.

(mod ) max. adh el lsµ r V µ= + ; with µmax=0.39 ; r=2.10-3 (4)

4 CONCLUSIONS

The new tribometer proposed, enables to reach temperature, pressure and velocities relevant with cutting application. The experimentation, made between AISI 316L and WC/Co with TiN coating pin in dry conditions, has shown that the friction coefficient is very sensitive to the sliding velocity. A numerical friction model enables to extract local informations, difficult to obtain experimentally: average contact temperature, average contact pressure, local sliding velocity. A friction law depending on local sliding velocity is proposed so as to control the frictional phenomena at the tool/chip/workpiece interface in cutting FEM model.

ACKNOWLEDGEMENTS

More information will be communicated later. Authors would like to express their gratitude to the AREVA NP and UGITECH Company.

REFERENCES

1. T. Özel, The influence of friction models on finite element simulations of machining, International Journal of Machine Tools and Manufacture 46 (2006) 518-530.

2. E.M. Trent, Metal Cutting, Butterworth Heinemann, 1991, ISBN 0-7506-1068-9.

3. D. Buryta, R. Sowerby, I. Yellowley, Stress distributions on the rake face during orthogonal machining, International Journal of Machine Tools and Manufacture 34 (5) (1994) 721-739.

4. W. Grzesik, An integrated approach to evaluating the tribo-contact for coated cutting inserts, Wear 240 (2000) 9-18.

5. F.Zemzemi, J.Rech, W.Ben Salem, P.Kapsa, A.Dogui, Development of a friction model for the tool-chip-workpiece interface during dry machining of AISI4142 steel with TiN coated carbide cutting tools, International Journal for Machining and Machinability of Materials Vol.2, Is.3-4 (2007) 361-367.

6. P. Hedenquist, M. Olsson,’Sliding wear testing of coated cutting tool materials‘, Tribology International 23 (3) (1991) 143–150.

7. A.Kusiak, J.L.Battaglia, J.Rech, Tool coatings influence on the heat transfer in the tool during machining, Surface and Coatings Technology 195 (2005) 29-40

8. D. Umbrello, R. M’Saoubi, J.C. Outeiro, The influence of Johnson-Cook material constants on finite element simulation of machining of AISI 316L steel, International Journal of Machine tools and Manufacture 47 (2006) 462-470.

9. F.P. Bowden, D. Tabor, Friction and lubrication of solids, Oxford University Press, 1951.

Investigation of friction in warm forging of AA6082

B. Buchner, A. Weber, B. Buchmayr

Chair of Metal Forming, University of Leoben – Franz-Josef-Strasse 18, 8700 Leoben, AustriaURL: www.metalforming.at e-mail: metalforming@unileoben.ac.at

ABSTRACT: This paper presents an experimental investigation of friction in hot forging of AA6082, whichis a standard forging alloy in automotive engineering, mechanical engineering and in naval architecture, byemploying a modified ring-on-disc test. The experiments were performed with three different commercialgraphite-based lubricants and at various loads, sliding velocities and specimen surface conditions. In addition,some tests were performed without lubrication.

KEYWORDS: Friction, Warm Forging, Aluminium

1 INTRODUCTION

Forged parts made of aluminium play a significantrole as components of light weight structures in theautomotive and aerospace industry. The main param-eters influencing forging processes are the flow curvesof the specimen material, the heat transfer at the con-tact area and the friction in the die-workpiece inter-face [1, 2]. The exact knowledge of the latter is paidmuch attention to as it affects power requirements,material flow, die filling, tool life and workpiece qual-ity. In order to understand the tribological processesand interactions in the tool-workpiece interface, theinfluences of the load collective and the surface condi-tions on friction were investigated systematically. Theexperiments were performed with a facility based onthe ring-on-disc test introduced by Schey [3].

2 EXPERIMENTAL WORK

2.1 Testing device

Figure 1 shows the Rotational Forging Tribometer.The circular motion is supplied from the bottom sideby means of a servo motor and a bevel gear system.The compression force is applied by a hydraulic cylin-der from the top side. The specimen (workpiece)is mounted on a rotary disc and transmits the fric-tional torque to the pivot-mounted ring-shaped tool.The tool is supported by a load cell via a lever armwhich enables an accurate measurement of the fric-tional torque. The acquisition of the compression

force is realised by another load cell, the rotationalspeed of the specimen is calculated from the speed ofthe servo motor. The specimen is brought to forgingtemperature by an inductive heating system, whereasthe tool is heated by a heating sleeve. The lubricantis sprayed onto the tool by an automatic applicationsystem that allows a reproducible dosing. The testingdevice is controlled by a programmable logic controlunit (PLC), the data acquisition is realised by a mea-surement amplifier that is connected to a commercialpersonal computer.The tribometer has a maximum upsetting force of ap-proximately 100 kN and provides a maximum torqueof more then 800 Nm on the rotary disc. The rota-tional speed can be up to 1.7 s−1. The sliding dis-tance is not constrained by the testing device. Themaximum temperature of the specimen is 1200°C,the maximum temperature of the tool is 450°C. Thetemperatures can be kept constant in an interval of±2.5°C.

Figure 1: Rotational Forging Tribometer.

1

2.2 Toolkit and specimen geometry

Figure 2 shows the toolkit used in the investigations.The ring-shaped workpiece is placed in the cavity ofa container and compressed by an annular tool withthe same inner and outer diameter as the specimen.Sliding on the bottom face of the workpiece is pre-vented by preparing the bottom of the cavity with ra-dial ridges, and the container is split in order to alloweasy removal of the tested specimen. The tempera-tures of die and workpiece are measured via thermocouples. Tool and specimen are made of hot worktool steel 1.2344 (hardened to 55 HRC) and AA6082,respectively.

Figure 2: Toolkit used in the investigation.

2.3 Experimental details

The experiments were performed with three differentcommercial graphite-based lubricants (see Table 1)and at various loads (20–150 MPa), sliding veloci-ties (10–100 mm/s) and specimen surface conditions(turned, sand blasted). In addition, some tests wereperformed without lubrication. The changing param-eters of the test series are summarised in Table 2.On the one hand, the sliding velocities were varied atthe same interface conditions (series 1–4), and on theother hand, the interface conditions were changed atconstant sliding speeds (series 5–9). Tool and work-piece temperature were set to 250°C and 450°C, re-spectively, and the relative displacement was 70 mm.For each parameter set and when pick-up occurred,the tool was prepared with grit 800 sandpaper and3 µm polishing suspension. Within a parameter set,

seven tests were performed with exception of series9, where just two experiments were carried out perload level due to heavy wear at some stages.The experiments were carried out in the followingway: First, the tool was brought to operating tem-perature. When the testing sequence was started bythe PLC, the specimen was heated, and the final tem-perature was kept constant for 180 s in order to allowtemperature equalisation. Then, the lubricant was ap-plied by the automatic spraying device (series 1–8).After lubricating, the execution of the test itself wasstarted. In order to avoid an interaction of differentinfluences, the specimens were compressed in a firststep and thereafter the rotation began.

Table 1: Lubricant test matrix.lubricant type rec. dilution ratio

A dispersion of graphite in water 1:15B dispersion of graphite in water 1: 7C emulsion of a dispersion of 2: 1

graphite in water and mineral oil

Table 2: Parameters of the test series.series no. lubricant surface cond. velocity, mm/s

1 B sand-blasted 102 B sand-blasted 403 B sand-blasted 704 B sand-blasted 1005 B, 80 %1 sand-blasted 406 B, 80 %1 turned 407 A sand-blasted 408 C sand-blasted 409 without sand-blasted 40

1 In series 5–6, 20 % less lubricant than in series 1–4 was used.

2.4 Evaluation of the tests

The evaluation of the experiments was performed inthree steps (see Figure 3):

1. First, the stationary region of the experimentwas determined from the velocity curve: Con-stant conditions were assumed in the intervalwhere the actual velocity at time stepi wasequal to or greater then the reference velocity.

2. In the stationary region, the mean values of nor-mal pressureσn, friction stressτ f and frictioncoefficientµ were calculated by the following

2

equations:

σn =1

f · (te− tb)

te

∑i = tb

σn,i , (1)

τ f =1

f · (te− tb)

te

∑i = tb

τ f ,i , (2)

µ =τ f

σn. (3)

Herein,tb andte indicate the begin and the endof the stationary region andf is the samplingrate of data acquisition.σn,i and τ f ,i are thenormal pressure and friction stress at time in-crementi, respectively.

3. For easier illustration, theσn-, τ f - and µ-values of each parameter set were averaged tothe mean valuesσn,m, τ f ,m and µm and thestandard deviations were calculated. In orderto avoid falsified results, apparent outliers werenot considered in this calculation.

Figure 3: Evaluation of the measurements.

3 RESULTS

3.1 Results of the tests without lubricant

The friction stresses and friction coefficients obtainedfrom the experiments without lubrication are pre-sented in Figure 4. Even at the lowest load level,the asymptotic behavior of the friction stress is wellpronounced; at normal loads higher than 65 MPa, amore or less steady state is reached. When regard-ing the friction coefficient, a decrease from an initialvalue ofµ = 1.83 (!) to a final value ofµ = 0.34 isobserved. As a friction coefficient higher than 1 can

only be explained by the adhesion theory, extensivewelding in the tool-workpiece interface has to be as-sumed in these cases. This assumption was verifiedby a visual inspection of the surfaces and by a com-parison of the lifting forces of the friction facility afterthe tests.The shear yield stress of AA6082 at an interface tem-perature of 350°C was determined to bek = 45–50 MPa. From the performed experiments, shearingof the specimen (and therefore sticking friction) canbe presumed at normal pressures equal to or greaterthan 65 MPa. Micrographs confirmed that (most of)the relative motion was done by shearing in subsur-face layers of the specimens at all pressure levels.

Figure 4: Results of the tests without lubrication.

3.2 Results of the tests with lubricant

Figure 5 presents the friction stress at different inter-face conditions. For comparison, the region of thevelocity dependent curves is indicated hatched. Ex-cept of series 5, all series have a maximum value at anormal stress of 130 MPa and have similar or lowervalues at contact loads of 150 MPa. However, thehigh friction stress of series 5 at a normal pressureof 150 MPa was due to material extruded in the gapbetween tool and container and did not indicate an-other tendency. In contrast to the other curves, series 7shows a sharp rise fromσn = 100 to σn = 130 MPa.When comparing the sand-blasted with the turned sur-face, the curve of the untreated specimen rises signifi-cantly steeper than that of the sand-blasted workpiece.The lowest friction stresses are obtained with lubri-cantC (series 8), whereas the highest friction stressesare caused by the conditions present in series 7 (lu-bricantA). Figure 6 shows the friction coefficients independence of the surface conditions.

3

In general, the friction coefficient increases sig-nificantly at low normal pressures and stays approx-imately constant or shows an light decrease at ele-vated contact stresses. LubricantA has a well pro-nounced local maximum atσn = 130 MPa, and thefriction coefficient rises at high normal pressured inseries 5. However, the behavior of series 6 is contrary:Here, the friction coefficient rises from the beginning,reaches a maximum at a normal pressure of 130 MPa,and decreases slightly atσn = 150 MPa.

Figure 5: Friction stresses in the tests with lubrication. Theregion of the velocity dependent curves is indicated hatched.

Figure 6: Friction coefficients in the tests with lubrication. Theregion of the velocity dependent curves is indicated hatched.

It has to be stated that all lubricants investigated re-duced friction significantly compared to the dry fric-tion condition. In fact, lubricantA (series 7) thatshowed the poorest lubricating effect reduced the fric-tion stress about 80 %, and lubricantC (series 8) re-duced friction about more than 90 %. Moreover, alllubricants prevented the tool from wear. Micrographs

confirmed that shearing was restricted to the graphitelayer and the asperities; the bulk material of the spec-imen was not affected.

4 CONCLUSIONS

The present analysis aimed in the characterisation ofthe load collective and the interface conditions onfriction. From this point of view, especially two con-clusions can be made:

1. As in solid lubricated interfaces (nearly) allshearing is done in the lubricant layer, the lu-bricant itself and the surface conditions of thefriction partners are the dominating parametersof the tribological system.

2. When regarding the load collective, the effectof normal pressure was in the observed rangemore significant than the influence of the slid-ing velocity.

The results were compared to the results of a previousstudy [4] employing a pin-on-disc test for lubricantevaluation, and good qualitative agreement was foundin terms of friction coefficient and friction evolutionduring the tests.

ACKNOWLEDGEMENT

The authors want to thank the federal province of Styria(”Zukunftsfonds Steiermark”, Project 19) for financing theproject and Fuchs Schmiermittel GmbH and Acheson Industriesfor the provision of the lubricants.

REFERENCES

[1] P. Groche and U. Weiss. Numerical identification of forgingparameters. InModelling of Metal Forming Processes: Pro-ceedings of the Euromech 233 Colloquium, pages 237–244,Sophia Antipolis (France), 1988.

[2] R. G. Snape, S. E. Clift, A. N. Bramley, and A. N.McGilvray. Forging modelling – sensitivity to input param-eters using FEA. In12th National Conference on Manu-facturing Research – Advances in Manufacturing Technol-ogy X, pages 51–55, Bath (UK), 1996.

[3] J. A. Schey.Tribology in Metalforming. Friction, Lubrica-tion and Wear. American Society for Metals, Ohio (USA),1983.

[4] B. Buchner, G. Maderthoner, and B. Buchmayr. Character-isation of different lubricants concerning the friction coeffi-cient in forging of AA2618.Journal of Materials Process-ing Technology, 198(1–3):41–47, 2008.

4

1 INTRODUCTION

In conventional manufacturing operation significant contact phenomena take place between tool and workpiece. Due to this contact and the resulting surface interactions, friction is generated. Friction knowledge is very important because it affects many aspects of manufacturing processes, such as the required force, the die wear and in some cases the process feasibility. To investigate and to determine the friction value several tests, such as ring test [1] and strip test [2], and the Pin-on-Disk [3] have been introduced. In recent years many Authors have proposed new methods to evaluate friction, for example the modified LDH [4], the Twist Compression Test [5]. Among these tests Pin-on-Disk (PoD) allows to directly measure the compression (F) and friction (T) forces and to estimate the friction coefficient (µ=T/F) using one single couple of surfaces in contact. In previous works the Authors studied the influence of contact pressure, sliding velocity and material roughness on friction coefficient by using a self-designed Pin-on-Disk equipment [6, 7, 8]. In these

works, the H13 steel and different die coatings were tested on AISI 5115 and on Series 1 Aluminium painted sheets. The idea was to identify the dependence of friction on the process parameters and to implement it in FE codes. The aim of the present work is to study the influence of contact pressure, sliding velocity and temperature on friction while forming different materials, namely deep drawing steel FeP04 and AZ31 Magnesium alloy sheets. Since this latter is characterized by a high formability in warm forming [9], the temperature influence was studied too.

2 EXPERIMENTAL EQUIPMENT

The experimental device (Fig. 1 left) is a large scale Pin-on-Disk tribometer designed by the Authors [7]. The friction coefficient is evaluated by pushing into contact two components called Pin and Plate. The Pin (a cylinder with a diameter of 12 mm and a corner radius of 2 mm) stands for the die (or more in general the tool) and is mounted on a carriage which can move orthogonally with respect to the plate surface. The Plate (flat) represents the workpiece and it is mounted on a disk that rotates so

ABSTRACT: In conventional sheet forming processes, such as stamping or drawing, significant contact phenomena take place between workpiece and die surfaces. Especially, relative motion and normal loads generate friction which influences some aspects of processes such as material flow, tools wear and life and total force needed to complete the process. In the current paper an experimental test campaign has been carried out using a large scale pin-on-disk device designed and realized by the Authors to investigate the influence of pressure, sliding velocity and temperature. The purpose is to test the developed device and to find which and how these parameters mostly affect friction. The pin-on-disk test consists of two specimens, a pin and a plate representing respectively die and workpiece, which are compressed by means of a known force and then moved one over the other. Compression and friction forces are sampled during the tests and the friction coefficient is estimated as the ratio of these two forces. The tested materials are H13 die steel on FeP04 and AZ31 sheets.

Keywords: sheet forming, friction, contact problems, temperature, pressure, velocity.

Process parameters influence on friction coefficient in sheet forming operations

E. Ceretti1, A. Fiorentino1, C. Giardini2 1 University of Brescia, Dept. of Mechanical and Industrial Engineering - Via Branze 38, 25123 Brescia, Italy URL: www.ing.unibs.it/tecmec e-mail: elisabetta.ceretti@unibs.it; antonio.fiorentino@unibs.it 2University of Bergamo, Dept. of Design & Technologies - Viale Marconi 5, 20044 Dalmine (BG), Italy e-mail: claudio.giardini@unibg.it

determining a relative movement between the two surfaces whose parallelism is guaranteed by the accuracy of the pin realization and of its holding system. The two bodies are kept in contact by means of a hydraulic cylinder acting on the carriage. The use of a pressure transducer and load cells allows to measure the normal (F) and the friction (T) forces acting between pin and plate. The dynamic friction coefficient can be estimated by using the Coulomb relationship (µ=T/F). The test data are collected by an acquisition system sampling and processing the transducer signals. To carry out tests at different working temperatures an additional device for heating the plate was used. This heating system (Fig. 1 right) consists of a base placed between the disk and the plate and containing a heating rod. A feedback system based on thermocouples positioned one under the plate and one in the rod, keeps the plate at the desired constant temperature (up to 310°C).

3 EXPERIMENTAL TESTS

Tests were conduced using H13 steel pins on AZ31 magnesium alloy and on FeP04 deep drawing steel plates. The experimental campaign was designed as a general full factorial experiment with multiple factors and multiple levels. Each factor combination was performed with 3 replications. The considered factors were: velocity and contact pressure for both FeP04 and AZ31. Temperature influence was studied on AZ31 due to its higher formability at warm temperature. Since the forces exchanged between punch, die and blank holder with the material under deformation are strictly related with its yield strength in stamping operations, the contact pressures were chosen as given percentage of the yield stress itself. In such a way, it was also possible to consider the yield stress reduction due to the temperature influence for AZ31 alloy. All tests were performed in dry conditions and

with the same sliding length. The tests summary is reported in Table 1. Table 1 – Design of the experiments.

Plate FeP04 deep drawing steel sheet (t=2 mm)Pin H13 steel Velocity 21 – 42 mm/s Pressure 7.5 – 15 – 27% σ0 Temperature 20°C σ0 170 MPa Lubrication Dry Plate AZ31 magnesium alloy sheet (t=1 mm) Pin H13 steel Velocity 21 – 42 mm/s Pressure 7.5 – 15% σ0 Temperature 200 – 250 – 300 °C σ0 (T) 127 – 86 – 63 MPa Lubrication Dry

The working procedure adopted for the tests was: • specimens surfaces cleaning; • machine set-up and compression load (F)

application; • temperature assessment (if necessary); • plate movement and data acquisition; • load removal. Each repetition was performed with a new pin and plate couple.

4 EXPERIMENTAL RESULTS AND ANALYSIS

The results were statistically analyzed and compared to identify which of the tested parameters significantly affects the µ behaviour and how it does. Single and multiple interactions were considered too.

4.1 FeP04 results

The typical friction coefficient behaviour during FeP04 tests is shown in Fig. 2. It is characterized by an initial transient (A) followed by the steady state (B). The friction coefficient for each test was estimated as the average value in the steady state interval (B).

MOTOR

CARRIAGE CASE

LC

HYDRAULICCYLINDER

PT

PIN PLATE DISK

F T

HEATING ROD

AND THERMOCOUPLE

THERMOCOUPLEPLATE

Fig. 1. PoD (left) and heating device (right).

Table 2 reports the experimental friction value results in terms of mean and standard deviation, estimated considering three repetitions for each test parameters combination.

FeP04 vs H13 | p = 7.5% 0 - v = 42 mm/s

0

0.02

0.04

0.06

0.08

0 0.5 1 time [s]

µFeP04 vs H13 | p = 7.5% σ0 - v = 42mm/s

A B

Fig. 2. Example of detected friction profile for FeP04.

Table 2 – FeP04 experimental results summary.

Plate FeP04 Pin H13 µ Velocity

[mm/s] Pressure ratio

[% σ0] mean σ 7.5 % 0.080 0.011 15 % 0.125 0.018 21 27 % 0.143 0.037 7.5 % 0.082 0.017 15 % 0.094 0.009 42 27 % 0.140 0.019

Pareto Chart of the Standardized Effects (α=0.05)

v

p*v

p

Standardized Effect

Fo limitFo

Fig. 3. Pareto Chart for FeP04 results.

pp

vv

4221

0.14

0.12

0.10

0.08

27.015.07.5

0.14

0.12

0.10

0.08

p

27.0

7.515.0

v2142

Interaction Plot (data means) for µ

[mm/s]

[% ]σ 0

% % %

[mm/s] [mm/s]

Fig. 4. FeP04 parameters interaction effects.

Fig. 3 compares the Fisher reference values with the relative F0 value estimated for single and multiple parameter interactions. The comparison shows that pressure strongly influences friction. In particular Fig. 4 shows the parameters interaction effects where it is possible to see how an increase of

pressure causes an increase of friction for both the sliding velocities. Considering sliding velocity Fig. 4 seems to demonstrate a dependence of friction on velocity, but this is not confirmed by the statistical test (see Fig. 3).

4.2 AZ31 results

A typical friction profile for AZ31 tests is shown in Fig. 5: after an initial transient (A) the curve grows as the two surfaces slide (B). The slope in B interval increases as the temperature increases. Also for AZ31 sheets, the friction coefficient was estimated as the mean value in the B interval. The results are reported in Table 3 and Fig. 6 where the parameters effect on friction is shown.

FeP04 vs H13 | p = 7.5% ∠0 - v = 42 mm/s

00.10.20.30.4

0 0.5 1 1.5 2 2.5 3

µ

AZ31 vs H13 | T = 300°C - p = 15% σ0 - v = 21mm/s

time [s]

A B

Fig. 5. Example of friction profile for AZ31.

Table 3 – AZ31 experimental results summary.

Plate AZ31 Pin H13 µ Temp.

[°C] Velocity[mm/s]

Pressure ratio [% σ0] mean σ 7.5 % 0.205 0.086 21 15 % 0.179 0.055 7.5 % 0.216 0.090

200 42 15 % 0.167 0.098

7.5 % 0.314 0.095 21 15 % 0.320 0.044 7.5 % 0.265 0.069

250 42 15 % 0.213 0.031

7.5 % 0.630 0.051 21 15 % 0.274 0.013 7.5 % 0.449 0.018

300 42 15 % 0.366 0.054

In particular, Fig. 6 compares the estimated Fisher values for the AZ31 results. It is possible to observe that temperature and contact pressure affect friction in the tested ranges, while velocity does not. In particular, the most affecting parameter is temperature which causes a friction raise, while an increasing contact pressure causes a friction reduction as shown in Fig. 7. An explanation of the temperature influence can be given as follows: as the temperature increases, the material surface (at peaks roughness level) is more deformed. This means that, for the same pressure ratio, the real contact area and, consequently, the

force needed to move the pin (friction force) increase. This is in agreement with the experience where hot forming causes higher friction forces. Observing the contact surfaces after the tests, it was found a deposition of AZ31 on H13. In this situation the material slides on itself so varying the friction conditions. This fact can explain the friction growth highlighted in B interval of Fig. 5. This implies the need of lubricant in AZ31 warm forming to keep the integrity of the die and workpiece surfaces.

Pareto Chart of the Standardized Effects (α=0.05)

T*vp*v

vT*v*p

T*ppT

Standardized Effect

Fo limitFo

Fig. 6. Pareto Chart for AZ31 results.

TT

0.50

0.35

0.20

pp

vv

4221

15,07,5

0.50

0.35

0.20

300250200

0.50

0.35

0.20

T

300

200250

p7.5

15.0

v2142

Interaction Plot (data means) for µ

% %

°C °C °C

[°C]

[mm/s]

[% ]σ0

mm/s mm/s Fig. 7. AZ31 parameters interaction effects.

p [% ]p [% ]

µ

18.515.011.57.5

0,55

0,50

0,45

0,40

0,35

0,30

0,25

0,20

T200300

σ 0

% % % %

[°C]

Fig. 8. Results of the tests at 200°C and 300°C and sliding

speed equal to 21 mm/s.

To better understand the contact pressure influence, additional tests were performed. Namely 11.5% and 18.5% of the yield stress at the temperature of 200°C and 300°C and a sliding velocity of 21 mm/s were considered. The result comparisons (Fig. 8) show a

common trend between the two curves with a more evident friction minimum at the temperature of 300°C and at a pressure of 15% of the yield stress.

4.3 Considerations on data scattering

Comparing the standard deviations reported in Table 2 and Table 3, it is possible to observe that the tests conduced on AZ31 sheets are affected by an higher scattering. This scattering decreases as the temperature increases and should be correlated with a problem in temperature control system.

5 CONCLUSIONS AND FUTURE STUDIES

The present study allowed to identify the friction most affecting parameters while forming AZ31 and FeP04 sheets. Furthermore, it was identified how these parameters influence friction. Further studies will be conducted to understand the dependence of friction on pressure, temperature and lubrication especially for AZ31 sheets.

REFERENCES

1. H. Sofuoglu, J. Rasty, On the measurement of friction coefficient utilizing the ring compression test, Tribology International, 32 (1999) 327-335.

2. R. Combarieu, , P. Montmitonnet, Effect of additives on friction during plane strain compression of aluminium stip test, Wear, 257 (2004) 1071-1080.

3. E. Yoon, H. Kong, O. Kwon, J. Oh, Evaluation of frictional characteristics for a pin-on-disk apparatus with different dynamic parameters, Wear, 203-204 (1997) 341-349.

4. H. Cho, T. Altan, Determination of flow stress and interface friction at elevated temperatures by inverse analysis technique, J. of Mat. Proc. Tech., 170 (2005) 64-70.

5. Y.C. Lam, S. Khoddam, P.F. Thomson, Inverse computational method for constitutive parameters obtained from torsion, plane-strain and axisymmetric compression test, J. of Mat. Proc. Tech., 83 (1998) 62-71.

6. F. Klocke, G. Messner, C. Giardini, E. Ceretti, FE-simulation of micro-tribological contacts in cold forming: experimental validation, In: 2° Int. Conf. on Tribology in Manuf. Proc, ICTMP, Nyborg – Denmark (2004), 395-402.

7. E. Ceretti, A. Fiorentino, A. Attanasio, C. Giardini, Influence of die material and roughness on friction coefficient in cold forming, In: Proceedings of the 8th AITeM conference, CET, Montecatini Terme (PT) – Italy (2007) 115-116.

8. E. Ceretti, C. Contri, C. Giardini, Study on micro-tribological contacts in cold forming: simulations and experimental validation, In: Proceeding of 7th A.I.Te.M Conference, CET, Lecce – Italy (2005)117-118.

9. F.K. Chen, T.B. Huang, Formability of stamping magnesium-alloy AZ31 sheets, J. of Mater. Proc. Tech., 142 (2003) 643-647.

1 INTRODUCTION

The choice of an efficient lubricant is still a critical point in the development of a metal forming process. A bad lubricant, or a good lubricant applied according to bad conditions, may lead to prematurely tool wear, and consequently have a dramatic influence on production. The aim of the present work is to improve the comprehension of wear phenomenon during hot steel forging with sprayed lubricants [1]. The study uses a specific tribological testing device: the Warm Hot Upsetting Sliding Test (WHUST). The WHUST simulates hot forging contact conditions between specimens heated up to 1100°C and contactor heated up to 200°C. Friction tests are performed so that WHUST contact pressure, sliding velocity, contactor sliding velocity, contactor and specimen temperatures are similar to industrial ones. Before performing the friction tests, lubricant is sprayed on contactor surface. Then the contactor slides against the specimen with a constant penetration, leaving a residual deformed track on its surface. Main WHUST results are tangential and normal loads on

contactor, surface roughness and chemical compositions on both specimen and contactor surfaces. A methodology is proposed to analyse these results and provide “wear markers” which characterized the contactor wear at microscopic and macroscopic scales. The methodology is the applied to the hot forging of a steel tripod using water based graphite lubricants. The effect on friction of lubricant film thickness, and solid lubricant particle size are analysed.

2 WHUST PRINCIPLE

During the test, the contactor penetrates the specimen and moves along specimen surface with a constant penetration and sliding velocity. The WHUST parameters are the specimen and contactor temperatures, the contactor velocity, the contactor geometry and the contactor penetration within the specimen (figure 1). Those test parameters are adjusted so that the test temperatures, the contact pressure at contactor/specimen interface, the specimen plastic strain and the sliding velocity are closed to those of the studied process. Contactors are machined from actual industrial tools

ABSTRACT: The aim of the present work is to improved comprehension of wear phenomenon during hot steel forging with sprayed lubricants. The study reported in this paper uses a specific friction test, called the Warm Hot Upsetting Sliding Test (WHUST), which reproduces hot forging contact conditions. Wear markers –such as friction force/normal load ratio or sliding distance before the first scratch– are proposed to characterize the contactor wear at microscopic and macroscopic scales. Friction tests are performed on 1100°C heated specimens to characterize the influence of graphite based lubricant film thickness and particle sizes on friction and wear in the flash zone of a nitrided steel die.

Key words: Hot forging; steel; friction test, wear, lubricant, graphite

Effects of lubricant and lubrication parameters on friction during hot steel forging

E. Daouben1,2, A. Dubois1, M. Dubar1, L. Dubar1, R. Deltombe1, N.G Truong Dinh2, L. Lazzarotto3 1Laboratoire d’Automatique, de Mécanique et d’informatique Industrielles et Humaines, UMR 8530, University of Valenciennes, F-59313 Valenciennes Cedex 9, France URL: www.univ-valenciennes.fr/LAMIH e-mail: andre.dubois@univ-valenciennes.fr 2CONDAT Lubrifiants –Avenue Frédéric Mistral, F-38670 Chasses sur Rhônes, France 3CETIM – Etablissement de Saint Etienne – 7 rue de la presse – BP 802, F-42952 Saint Etienne Cedex 9, France

and specimens are parts of actual industrial work pieces. That specificity enables the WHUST to simulate physical and chemical contact conditions encountered in the studied forming process: the lubricant, surface roughness, chemical properties involved in the test are the same as those of the process [2].

Fig. 1. Overview of the W.H.U.S.T

3 WHUST RESULTS

WHUST direct results are normal and tangential forces on contactor, specimen and contactor surface profiles. From these results we can define macroscopic or microscopic “wear markers”.

3.1 Macroscopic markers

Three parameters are defined: the lubrication index Li, the contactor roughness Ra and the critical length Lc [3].

3.1.a Lubrication index Li The lubrication index corresponds to the ratio of the tangential and normal forces: Li = Ft/Fn. This index provides quick and reliable information on the evolution of friction. Correlation between lubrication index Li and the Coulomb’s coefficient of friction can be found in [4].

3.1.b Contactor roughness Ra The mean surface roughness Ra is measured on contactor contact surfaces before and after each test. It provides information on the occurrences of adhesive and/or abrasive wear.

3.1.c Critical length Lc Plastic strains generate a residual track on the specimen surface during the tests. Scratches can be observed on this track when the lubricant film breaks down. The critical length Lc corresponds to the dimensionless distance measured along the track, from the beginning of the contact to the point where the first scratch is observed. Lc equals zero when

scratches appears as soon as the contactor comes in contact with the specimen; Lc equals 1 when no scratches are observed. This marker characterizes the ability of lubricant to delay the first metal-to-metal contact between tool and work piece.

3.2 Microscopic marker

Microscopic markers are observed on contactor contact surfaces. They derive from SEM-EDS analyses.

3.2.a SEM analysis SEM pictures are taken on contactor surface at a magnification of 200. These pictures show the presence or absence of material deposits on contactor surface (presence of adhesive wear area, oxide scale…). It allows quantifying the size of these deposits.

3.2.b EDS analysis The SEM observations are completed by EDS analyses in order to determine the chemical composition of the material deposits. First, a mapping is taken before the friction test in order to examine lubricant structure. Then, two mappings are realised after the test, one to determine the presence and nature of oxide scale [5] and the other to quantify the residual lubricant on contactor surface.

4 APPLICATION

4.1 Forming process: hot forging of a tripod

The hot forging of a tripod (work piece with three symmetry planes at 120°) is studied in the present work. The process involves two forming sequences: a preform followed by the finishing. Both are carried out using a mechanical press with a maximum velocity of 1m/s [6]. The present study focuses on the lubricant behaviour in the flash zone, at the end of the preform sequence. Dies are made of nitrided tool steel and the tripod is in medium alloyed steel. Dies and work piece temperatures respectively equal 200°C and 1100°C when they come in contact. A graphite suspension in water lubricant is sprayed on the dies before forming to reduce friction and limit wear [7]. Measurements operated on industrial dies show that the graphite film thickness varied from 10 to 40 µm, depending on the angle between the lubricant spray nozzle and the surface of the tool. A finite element simulation of the process shows the contact pressures equalled

to 190 MPa and sliding velocities equalled to 60 mm/s in the flash zone [2]. Each testing condition is performed three times, with three “new” contactors.

4.2 Effect of lubricant film thickness

Four film thicknesses are simulated on the WHUST: • 0 µm: dry lubrication, • 10 µm and 40 µm, which correspond to the

minimum and maximum thicknesses measured on the industrial tools,

• 30 µm which is the mean film thickness measured on the flash zone of the dies.

Table one sums up the main results for the macroscopic wear markers. Contactor surfaces before and after the friction tests are presented on figures 2 and 3. Bonding elements used in the formulation of the lubricant have a great influence on the structure of the graphite layer on tool surface. In the present study, for thin lubricant thickness, the graphite layer is almost only composed of powder, without particular structure (figure 2b). The layer structure evolves to a “honeycomb” structure when the film thickness equal 30µm (figure 2c), and to a stratified layer for thicker films (figure 2d). WHUST performed without lubricant lead to strong material deposit on contactor surface. The lubrication index is then very high and scratches appears on the work piece surface after a short sliding distance (Lc = 0.5). SEM-EDS analysis shows that the material deposit is oxide scale coming from specimen surface. Large oxide scales particles are observed on contactor surface when performing test with a 10µm lubricant film thickness (figure 3b). The lubrication index Li is 28% lower than in dry contact condition, and the critical length Lc increase from 0.5 to 0.87. The graphite layer reduces friction and delay the first direct metal-to-metal contact, but is not bonded sufficiently to the surface to avoid scratch occurrence. Friction tests operated on thicker lubricant films present less material deposit on contactor surface (figures 3c-d). SEM-EDS show that the surface is mostly cover with residual graphite. No scratch is observed on the specimen surfaces and the lubrication index decreases with the increase of film thickness. As a conclusion, the film thickness in the flash zone must by greater than or equal to 30µm to limit tool wear and reduce friction. Nonetheless, it should be

notice that the aim of the flash being to slow down material flow in order to guaranty a good die filling, a strong reduction of friction is not always wanted. So, for the studied process, a film thickness equal to 30µm is the better compromise between surface defect protection and friction control.

Fig. 2. SEM pictures of contactor surface before friction test

a) dry friction, b) 10 µm, c) 30 µm, d) 40 µm

Fig. 3. SEM pictures of contactor surface after friction tests a) dry friction, b) 10 µm, c) 30 µm, d) 40 µm

Table 1. Evolution of Li and Lc macroscopic markers in function of lubricant film thickness Lubricant film thickness

0 µm 10 µm 30 µm 40 µm

Lubrication index Li 0.81±0.07 0.58±0.04 0.39±0.03 0.26±0.02

Critical length Lc 0.50±0.04 0.87±0.04 1.00±0.00 1.00±0.00

4.3 Effect of lubricant particle size

Common particle sizes in graphite lubricants range from 2 to 50 µm [8]. To study the lubricant grading effect, two lubricants are tested, one with medium particle sizes (25-30 µm), another with large particle sizes (50µm). Both lubricants have the same composition (same bonding agent), and are sprayed on the contactor so that their thickness equals 30 µm. Table 2 exhibits the change of tribological behaviour when increasing the particle sizes of the lubricant: the lubrication index increases from 0.39 to 0.48 and the critical length becomes lower than 1. Figure 4 presents the surface of the contactor just before the test. Blue areas are for carbon and red for iron. Lubricant with medium particle sizes has a more homogeneous covering on tool surface. Zones where almost no graphite is present are observed on contactor surfaces sprayed with the large particle size lubricant. As a consequence, the whole contactor surface is not protected, direct metal-to-metal contact occurs between specimens and contactors and scratches appear on specimen surface. Scratches lead to an increase of the friction force. As a conclusion of these tests, when large particle sizes are used, the lubricant film layer has to be thicker so that the bonding agent can react to form a homogeneous graphite layer.

5 CONCLUSIONS

An original methodology focused on the analysis of wear phenomena in hot forging of steel has been presented. A prototype testing device called WHUST reproduces industrial hot forging contact conditions. WHUST results are analysed using microscopic and macroscopic wear markers. Tests operated with different lubricant film thicknesses and different lubricant particle sizes have shown that the efficiency of the lubricant film is linked to structure of the graphite layer more than the lubricant thickness. These results are very promising and will help to understand wear mechanisms in a very near future. The next steps of this study will be the understanding of the behaviour of graphite free lubricant for hot forging and the testing of industrial dies at the end their lifecycle.

Table 2. Evolution of Li and Lc macroscopic markers in function of lubricant film thickness Lubricant particle size 25-30 µm 50 µm

Lubrication index Li 0.39 ±0.03 0.48 ±0.02

Critical length Lc 1.00 ±0.00 0.90 ±0.04

Fig. 4. SEM-EDS pictures of contactor surface before friction

tests a) EDS, medium grading, b) EDS, large grading

ACKNOWLEDGEMENTS

The present research work has been supported by the European Community, the Délégation Régionale à la Recherche et à la Technologie, the Ministère de l’Education Nationale, de la Recherche et de la technologie, The Région Nord-Pas de Calais, the Centre National de la Recherche Scientifique, CONDAT SA and CETIM, ASCOFORGE and Forges de Courcelles. The authors gratefully acknowledge the support of these institutions and industrial partners.

REFERENCES

1. O. Barrau, C. Boher, R.Gras, F. Rezaï-Aria, ‘Analysis of the friction and wear behaviour of hot forging’, Wear 255 (2003) 1444-1454.

2. R. Deltombe, N. Morgado, M.Dubar, A.Dubois and L. Dubar, Hot forming of aluminium: a new methodology to identify friction parameters, 7th ESAFORM, Norway – Trondheim (2004) 253-256.

3. E. Daouben, L. Dubar, M. Dubar, R. Deltombe, A. Dubois, N. Truong-Dinh, L. Lazzarotto, ‘Friction and wear in hot forging of steels’, Proceedings of the Int. Conf. ESAFORM 2007, Zaragoza, Spain (2007)

4. L. Lazzarotto, L. Dubar, A. Dubois, P. Ravassard and J. Oudin, ‘Identification of Coulomb’s friction coefficient in real contact conditions applied to a wire drawing process’, Wear, 211 (1997) 54-63

5. B. Picqué, P.O. Bouchard, P. Montmitonnet and M. Picard, ‘Mechanichal of iron oxide scale: Experimental and numerical study’, Wear, 260 (2005) 231-242.

6. S.Sheljaskow, Tool lubricating systems in warm forging. J. Mater. Process. Technol., 113 (2001) 16-21.

7. T.Iwama and Y. Morimoto, Die life and lubrication in warm forging. Journal of Materials Processing Technology 71 (1997) 43-48.

8. R. F. Deacon and J.F. Goodman, ‘Spreading Behavior of water based graphite Lubricants on Hot Die Surfaces’ , CIRP 55/1 (2006) 299-302.

1 INTRODUCTION

In hot rolling of steels, the rolled material in fact

consists of thin layers of a ceramic, the complex iron

oxide layers (20 – 50 µm thick), on hot metal. The

difference in hardness and ductility of these two

materials [1] often leads to oxide cracks of various

kinds, through-thickness or interfacial (oxide

spalling) [2-5]. The present paper focuses on

through-thickness cracks.

In a previous work [6-7], fracture occurring just

before bite entry had been studied both

experimentally (hot bending test) and theoretically

(FEM). Cracks open wherever superficial tensile

stresses occur, then they may open wide in the bite

due to strip elongation. If the pressure is high

enough, "micro-extrusion" of fresh hot metal takes

place through the open cracks. The interface

becomes wavy as in Figure 1 ("rolled-in scale"), a

defect which may become visible after pickling and

remain even after cold rolling. A numerical

parametric study resulted in a damage risk chart,

where thicker oxides (e.g. due to high temperatures)

were found to be most detrimental, unless sufficient

plasticity allows them to keep up with part of the

elongation of the strip. In terms of material data, the

oxide-to-steel hardness ratio and the temperature-

dependent and strain-rate-dependent oxide fracture

stress were found most important.

Figure 1: oxide cracks and wavy oxide – metal interface.

Above: top view [2] shows an array of cracks normal to rolling

direction (RD). Below: cross section (this work).

The present work aims at extending these notions to

cracks opening in the roll-strip contact (bite). There

are few data available on the morphology, nature

and origins of this particular category.

ABSTRACT: The behaviour of oxide scales in the finishing Hot Strip Mill is simulated by the hot Plane Strain Compression Test (PSCT). Compared with the ideal case of homogeneous plastic co-deformation of the oxide layer and the underlying metal, different types of defects are described: delamination at the interface or within the oxide layer; interfacial plastic instability due to the jump of the mechanical properties; perpendicular, through-thickness cracks where the axial strain parallel to the interface dominates, followed by micro-extrusion of metal between the fragments; oblique cracks followed by sliding along the lips, where shear dominates. The Finite Element Method (FEM) is used to bring elements of interpretation, as to which conditions determine each mechanism. Conclusions for the behaviour in hot rolling are sketched.

Key words: Oxide scales, hot rolling, Plane Strain Compression Test, PSCT, Finite Element Modelling

Behaviour of oxide scales in hot steel strip rolling

C. Grenier 1,2, P.-O. Bouchard

1, P. Montmitonnet

1,*, M. Picard

2

1 Ecole des Mines de Paris - ParisTech, UMR CNRS 7635, BP 207, 06904 Sophia-Antipolis Cedex, France URL: www.cemef.ensmp.fr e-mail: Claire.Grenier@ensmp.fr ; Pierre-Olivier.Bouchard@ensmp.fr ; Pierre.Montmitonnet@ensmp.fr

2 ArcelorMittal, R&D Centre, BP 30320, 57214 Maizières-les-Metz Cedex, France e-mail: Michel.Picard@arcelormittal.com

x

z

x

z y

x

y

2 EXPERIMENTAL

2.1 PSCT set-up and procedure

Figure 2: PSCT test rig (left) and procedure (right).

PSCT consists in upsetting a strip between two flat

dies (figure 2 left). Here, the oxidation of steel strip

samples is made in situ, by allowing temporarily an

oxidative atmosphere in the protective glass vessel

(figure 2 right). The oxidation temperature is 900°C

in all tests, i.e. close to entry temperature in the

finishing Hot Strip Mill (HSM). The oxidation time

is varied to give oxide thickness between 10 µm and

100 µm. The system is then brought up or down to

the mechanical test temperature. After temperature

equilibration, the test is performed in a fraction of a

second; in the tests reported, no lubricant was used.

Then the sample is allowed to cool freely in N2.

2.2 Materials

The steel strip is an ultra-low carbon, DWI deep-

drawing steel, with 0.015%C, 2.29% Mn, 0.23% Si

(atomic concentrations). The coupons are 5 mm

thick, 50 mm wide and 62 mm long.

The die material is yttria-toughened zirconia; dies

are 70 mm long and 12 mm wide. Rough dies (Ra =

3.6 µm, isotropic) are compared with smooth dies

(Ra = 0.42 µm, isotropic); on occasion, grooved dies

are used to simulate damaged (“banded”) rolls; the

grooves are in the punch width direction x,

equivalent to the rolling direction (RD).

2.3 Experimental plan

Oxide thickness: 10µm, 25 µm, 50 µm, 75 µm.

PSCT Temperature: 800°C, 900°C and 1050°C

Strain: ε = 0.2 ; 0.4 ; 0.6 ; 0.8.

Strain rate: 0.1 s-1, 1 s

-1 and 10 s

-1.

3 EXPERIMENTAL RESULTS

3.1 General observations: cracks and interface

a b Figure 3: Top view: vertical, normal cracks on the side of a

PSCT indentation. (a): enlarged view of the framed area in (b);

the white bar is 1 mm.

Smooth die, 4.0=ε , 1.1 −= sε& , T = 900°C, 50 µm oxide.

Figure 4: SEM picture of a cross-section in the flow direction

(x = punch width direction ↔ RD). Same conditions. Metal is

white, oxide is light grey.

Figure 3 shows a typical aspect in top view, an array

of cracks perpendicular to the major flow direction

x, very similar to cracks opening before the bite [6-

7]. Figure 4 shows they run through the oxide

thickness, with a uniform density. Their origin here

is not the oxide bending ahead of the bite as in [6-7]

but probably the flow of the underlying metal

putting the oxide in tension. It might also be due to

punching by die roughness peaks; this will be

discussed using numerical simulation in paragraph 4.

Spalling (figure 4) may be due to sample polishing

before microscopic observation; but slight interface

waviness suggests spalling or crushing during the

tests, followed by micro-extrusion, as in [6-7].

This observation answers one of the questions

behind this work: crack array formation may go on

in the contact - PSCT is known to be a good

simulator of strip rolling. The subsequent evolution

seems very similar to pre-bite cracks.

In the case of pre-bite cracks, the density could be

related to oxide thickness, oxide fracture stress and

interface shear stress [6-8]. We expect numerical

simulation to help determine the parametric

dependencies for the present, in-bite opening cracks.

It has been found occasionally that cracks may not

always be normal to the surfaces. Figure 5 shows a

case where oblique cracks, followed by rotation of

fragments, have occurred near the edge of a die

(note that rotation may have been facilitated by the

presence of a thick lubricant film in this particular

400

500

600

700

800

900

1000

1100

0 500 1000 1500time (s)

tem

pera

ture

(°C

)

Oxi

datio

n(N

2+O

2+H

2O)

Def

orm

atio

n

N2 N2

400

500

600

700

800

900

1000

1100

0 500 1000 1500time (s)

tem

pera

ture

(°C

)

Oxi

datio

n(N

2+O

2+H

2O)

Def

orm

atio

n

N2 N2

x

zy

x

zy

x

zy

x

y

x

y

x

y

case). The same pattern has been found by [9] on hot

rolled strips. This proves the relevance of this

phenomenon, tentatively attributed to the presence

of significant shear stress (die edge / friction) which

induces the rotation of principal axes.

Figure 5: oblique cracks (top) near the PSCT die edge,

(bottom) on a rolled strip [9].

3.2 Roughness transfer

Figure 6: comparison of the oxide surface state after PSCT

with rough / smooth dies. Die width 12 mm. 4.0=ε , 1.1 −= sε& ,

T = 900°C, oxide thickness 50 µm.

Figure 7: cross-section of the samples shown in figure 5,

showing the smooth metal (white) – oxide (grey) interface,

whatever the die roughness.

Another possible defect is interface waviness due to

tool roughness printing, in particular when rolls are

severely worn. Figure 6 shows that the oxide

surface, to a large extent, takes the roll roughness,

which suggests a certain degree of plasticity of the

oxide, but the oxide – metal interface remains

smooth (figure 7). This may not be a general

conclusion however, certainly dependent (i) on the

ratio of the tool roughness to the oxide thickness,

and (ii) of the temperature-dependent mechanical

properties of the oxide (toughness, oxide-to-metal

yield stress ratio). Here again, numerical simulation

can contribute in the study of these parameters.

3.3 Interface

3.3.a Delamination / spalling

This is another failure mechanism whereby oxide

fragments may be spalled off the sample surface; in

rolling, such fragments may then be embedded

inside the strip surface by the contact pressure.

Figure 8: two examples of delamination. Left: interfacial

delamination on the flank of a groove. Right: delamination

within the oxide layer.

In figure 8 (left), bending of the sample in the flank

of a grooved die (representing the “roll banding”

defect, i.e. peeling of an orthoradial strip of roll

oxide) has resulted in a normal, through-thickness

crack which has bifurcated along the interface; in

such situations, the formation and embedding of a

fragment becomes highly probable. The die groove

is oriented in the die width direction, equivalent to

the rolling direction. Such a crack would thus be

longitudinal, contrary to those shown above. Figure

8 (right) shows delamination within the oxide. Lines

of pores have occasionally been found in oxide

layers, parallel to the interface; they may be the

origin of such defects. Yet delamination might also

have taken place at the interface, with subsequent re-

oxidation during cooling in imperfectly pure N2.

Here again, fragmentation and embedding of the

spalled layer in the next rolling stand is inevitable.

3.3.b Interface instability In a series of tests devoted to varying strain rate, a

sinusoidal interface waviness has been observed at

the lowest strain-rate, decreasing and disappearing

as ε& increases (figure 9).

Figure 9: Interface waviness without cracking. Strain rate

increases from top to bottom (0.1, 1 and 10 s-1).

There is no evidence of any crack / micro-extrusion

phenomenon, hence the interpretation by plastic co-

deformation instability (see [10] e.g.). It is not sure

that this can occur in strip rolling, since it has been

y

z x

z

found here only in a low strain rate range; yet it

might occur at higher speeds under different

conditions (temperature…).

4 THEORETICAL INTERPRETATION

4.1 Model description

The Forge2005® FEM software has been used for

this study. Its description can be found in [6,7],

together with the added through-thickness crack-

opening algorithm (based on a critical fracture

stress). Only such cracks are addressed here,

although a delamination capability has also been

added (Cohesive Zone Modelling approach).

4.2 An application to interpretation of crack origin

Figure 10 shows a crack opening simulation in the

PSCT at 1000°C and 1 s-1, with a critical fracture

stress of 200 MPa for the oxide. The oxide and the

metal yield stresses are respectively given by:

159.0223.0)(.00299.0 ).3(..exp.69)( εεσ &KTMPa = (1)

09.022.0 ).3(..)(

3340exp.5.8)( εεσ &

=

KTMPa (2)

The die-oxide friction factor is 08.0=m . The oxide-

metal interface is assumed perfectly adherent.

Figure 10: PSCT FEM modelling and crack formation.

Under these conditions, cracks form either under die

edges (figure 10, left) due to the stress singularity, or

at asperity tops (figure 10, insert right), but never on

flat parts of the die. Looking in more details, with

the roughness peak height chosen (3 µm, to be

compared with oxide thickness 50 µm), fracture

does not occur because of the stress singularity at

peak apex, since full penetration of the asperity into

the plastic oxide occurs before crack opens. The

critical tension is in fact reached later on, when the

flow of the underlying metal shears the interface and

puts the oxide in tension. In another simulation with

a periodic roughness covering the whole die width

(Ra = 0.3 µm), cracks appear periodically, but the

wavelength is much larger than the roughness

wavelength. Study of the influence of parameters

and comparison with experiments are in progress.

The effect of friction and shear stress on oblique

crack formation (as in figure 5) is also under study.

5 CONCLUSIONS

Experimental results present a number of

deformation and fracture phenomena which occur in

oxidized metal / tool contact; their relevance towards

interface defects in hot strip rolling has been

commented. A first example of application of multi-

body FEM to the analysis of the origins of these

oxide and interface defects has been presented.

ACKNOWLEDGEMENT

The authors thank ArcelorMittal (Arcelor Research S.A.) for

financial support and authorization to publish the results.

REFERENCES

1. G. Vagnard and J. Manenc, Etude de la plasticité du

protoxyde de fer et de l’oxyde cuivreux. Mem. Et. Sci.

Rev. Met. LXI 11 (1964) 768-776 (in French).

2. Y.H. Li and C.M. Sellars, ‘Modelling deformation of

oxide scales and their effects on interfacial heat transfer

and friction during hot steel rolling’, Proceedings of the

2nd Conf. Modelling of Metal rolling processes, London

(1996) 192-201

3. M. Krzyzanowski and J.H. Beynon, 'Oxide behaviour in

hot rolling', in: Metal forming science and practice, ed,

J.G. Lenard, Elsevier, Amsterdam (2002) 259-295

4. M.F. Frolish, M. Krzyzanowski, W.M. Rainforth and

J.H. Beynon, 'Oxide scale behaviour on aluminium and

steel under hot working conditions', J. Mater. Process.

Technol. 177, 1-3 (2006) 36-40

5. M. Schütze, 'Mechanical properties of oxide scales',

Oxid. Met. 44, 1-2 (1995) 29-60

6. B. Picqué, Experimental study and numerical simulation

of oxide scales mechanical behaviour in hot rolling, PhD

Thesis, Ecole des Mines de Paris (2004)

7. B. Picqué, P.-O. Bouchard, P. Montmitonnet and M.

Picard, 'Mechanical behaviour of iron oxide scale:

experimental and numerical study', Wear 260 (2006) 231-

242

8. D.C. Agrawal and R. Raj, 'Measurement of the ultimate

shear strength of a metal-ceramic interface', Acta Met.

37, 4 (1989) 1265-1270

9. F. Platteau, G. Lannoo and D. Espinosa, 'Control of strip

surface quality during hot rolling', Internal Report, CRM

(2007) (personal communication).

10. S.L. Semiatin and H.R. Piehler, 'Formability of sandwich

sheet materials in Plane Strain Compression and rolling',

Met. Trans. A 10A (1979) 97-107

Modelling of friction with respect to size effects

Z. Hu1, F. Vollertsen1

1BIAS Bremer Institut fuer Angewandte Strahltechnik – Klagenfurter Str. 2, D-28359 Bremen, Germany URL: www.bias.de e-mail: hu@bias.de; vollertsen@bias.de

ABSTRACT: In this paper a size dependent FEM-simulation for sheet metal forming was realized applying the size dependent friction functions acquired from strip drawing tests [1]. The software ABAQUS 6.6.3 was used to simulate the strip drawing tests and deep drawing process in different process dimensions. A 2-dimensional model was built for strip drawing test and a 3-dimensional model was built for deep drawing process. Both processes were experimentally carried out in different dimensions. A constant friction coefficient as well as a friction function, which shows a dependence of friction coefficient on the contact pressure, was applied in the simulation. The simulated punch force vs. punch travel curves were then compared with the experimental curves. A discussion about the difference between these curves with consideration of tribological size effects will be given.

Key words: Scaling, Tribological size effects, Friction

1 INTRODUCTION

When downscaling the size of the work piece to micro forming, not all parameters can be changed according to the rule of similarity, e.g. grain size. This causes the so called size effects, i.e. the occurrence of unexpected results concerning the forming force of the forming limit [1]. Since deep drawing is essentially affected by the friction between the work piece and tools [2, 3], which is also affected by the size effects [4, 5], the tribological size effects in sheet metal forming were investigated in our former work [6, 7] and the size-dependent friction functions were acquired. The aim of this work is to realize size dependent FEM-simulation for deep drawing applying the acquired friction functions.

2 TRIBOLOGICAL SIZE EFFECTS

2.1 Scaled experiments

Compared to usual deep drawing there is no

tangential force Ft at flange area in strip drawing, see figure 1.

Fig. 1. a) Strip drawing test b) Deep drawing

This difference makes it easier to find the relation between the punch force and the friction coefficient. Thus the friction function method was developed only for the strip drawing, which can be used later to identify the tribological size effects in deep drawing. The effective friction function method yields a friction coefficient from deep drawing (according figure 1b), having some disadvantages concerning the dependence on the contact pressure. Both processes were carried out with 5 different punch diameters in this investigation. They are 50, 20, 10, 5 and 1 mm. According to the theory of similarity, all process parameters are kept constant while almost all geometrical parameters of tools and work

pieces are scaled by the same scaling factor. For example the thickness of the work piece has the same ratio to punch diameter in each experiment [8]. Al99.5 is used as work piece material in every process dimension. The properties of the material are also affected by the size effects [9, 10]. For the determination of the friction function and the friction coefficient in this work the flow stress is required. Thus, the flow curves of Al99.5 in each thickness were acquired through tensile tests and show clearly difference from each other, see figure 2. In order to avoid the influence of the surface quality on the friction, all tools as well as blanks used in this work have nearly the same roughness respectively.

Fig. 2. Flow stress of Al99.5 in different thicknesses

2.2 Friction function

The punch force was measured in strip drawing tests and was used later in the calculation model described in [6]. For each punch diameter, a friction function is determined. A graphic display of these friction functions is shown in figure 3.

Fig. 3. Friction functions from strip drawing tests

A difference with a factor of about 1.5 between the friction functions with punch diameter of 1 and 50

mm is described, which illustrates a tribological size effect within this investigation. The friction functions can be expressed mathematically through an exponential form:

)exp()exp( 53421 CPCCPCC ⋅−⋅+⋅−⋅+=μ (1)

Where µ = friction coefficient, P = contact pressure, C1-C5 = coefficients (They are listed in table 1). Table 1. Coefficients of friction functions Punch diameter [mm]

C1 C2 C3 C4 C5

50 0.000 0.159 0.130 0.871 0.007 20 0.000 0.180 0.130 0.571 0.011 10 0.125 12.00 0.449 7.759 1.346 5 0.116 0.319 0.032 3.571 2.348 1 0.000 0.188 0.180 0.687 0.010

2.3 Effective friction coefficient

According to theory of Storoschew [11] the friction coefficient in deep drawing can be evaluated from the measured maximum punch force for each process dimension [12], see figure 4.

Fig. 4. Effective friction coefficients from deep drawing

In this theory there is only one friction coefficient. Thus the friction coefficient calculated using this method is call effective friction coefficient. By these effective friction coefficients the tribological size effects can also be shown: with decreasing the process dimension the friction increases.

3 FEM-SIMULATION

3.1 Scaled strip drawing tests

In this work the software ABAQUS 6.6.3 was used to simulate both forming processes. A 2-dimensional

model was created for the strip drawing test, in which the tools were defined as analytical rigid line and the blank was defined as deformable object. The 4-node bilinear plane stress element CPS4R was used to mesh the blank. Within the thickness of the blank there are four elements. For simulation of the experiments with blanks in different thicknesses, the flow stress curves of Al99.5 in different thicknesses were applied respectively in the simulation model. Using the normalization described in [12] the simulated punch force vs. travel curves for strip drawing with punch diameters of 1 and 50 mm shown in figure 5 were obtained.

Fig. 5. Comparison of simulated and experimental punch force

vs. punch travel from strip drawing tests

For changes of the process dimension the simulation program can not take into account the change of the size automatically. Thus the result of simulation for micro forming can not differ from that of macro forming using the conventional simulation method. Applying the friction function obtained above into the simulation, the simulated curves (the curve FEM-1 for the punch diameter of 1 mm, the curve FEM-50 for punch diameter of 50 mm) show the same trend as the experimental curves (the curve EXP-1 for the punch diameter of 1 mm, the curve EXP-50 for the punch diameter of 50 mm). The normalised punch force for punch diameter of 50 mm is lower than that for punch diameter of 1 mm. All curves have the maximum point at the normalised punch travel of about 0.2. A FEM-simulation with consideration of tribological size effects was realised. The maximum point of the simulated curve FEM-1 is about 11% lower than that of the experimental curve EXP-1. While the maximum point of the curve FEM-50 is about 8% lower than that of EXP-50. The

reason for this might be the assumption in the calculation model: The normal pressure at the radius is uniform [1]. However, the simulation shows two local contact zones, see figure 6. The pressure on these two zones is much higher than the uniform distributed pressure assumed in the calculation model. Since the friction coefficient depends on the normal contact pressure, if the pressure on the contact surface is not right, the incorrect friction coefficient will be used. Thus the simulated punch force differs from the experimental one.

Fig. 6. The local contact zones at radius of the die

3.2 Scaled deep drawing

Similar to the strip drawing a 3-dimensional model was created for deep drawing. The effective friction coefficients from deep drawing as well as the friction functions from strip drawing using the same punch diameter were applied in the FEM-simulation for each dimension. The comparison of the simulated and experimental punch force vs. punch travel for deep drawing with punch diameter of 1 mm is shown in figure 7.

Fig. 7. Comparison of simulated and experimental punch force

vs. punch travel from micro deep drawing

The maximum punch force of the simulated curve using the friction function is about 5% higher than that using the effective friction coefficient. Both of them show a good agreement with the experimental curve from beginning till the maximum punch force. Then both simulated curves begin to decrease while

the experimental one shows nearly no trend to decrease. This result confirms our statement in [12], that the thickness of lubricant can result in an increase in punch force, because in micro deep drawing the sum of blank thickness and lubricant thickness might be more than the drawing clearance. This effect can not be detected in macro deep drawing and was also not implemented in the FEM-simulation. For punch diameter of 50 mm, the maximum punch force of the simulated curve using friction function is about 400 N lower than the experimental curve, while the one using effective friction coefficient shows a difference of about 1000 N, see figure 8.

Fig. 8. Comparison of simulated and experimental punch force

vs. punch travel from macro deep drawing

This means the theory of Storoschew is not suited to calculate the friction coefficient precisely from measured punch force in different dimensions. Oppositely the friction functions from strip drawing tests show a good scalability. However, in accordance to the results in strip drawing, the simulated punch force vs. punch travel curve does not agree with the experimental curve perfectly. The reason for that might be the distribution of contact pressure, since the local contact zones exist also in deep drawing. Thus in our future work the calculation model will be enhanced in order to take into account the distribution of contact pressure.

4 CONCLUSIONS

• Tribological size effects were observed in both scaled strip drawing tests and deep drawing.

• The friction functions from strip drawing tests can be integrated into FEM-simulation e.g. ABAQUS. This enables to simulate sheet metal

forming processes with consideration of tribological size effects.

• The distribution of contact pressure will be taken into account in our future work.

ACKNOWLEDGEMENTS

The work reported in this paper is funded by the Deutsche Forschungsgemeinschaft (DFG) within the project “Modelling of tribological size-effects in deep drawing” (DFG project no. Vo 530/6). The authors would like thank the DFG for their beneficial support. Moreover the authors would like thank the institute of Metal Forming and Casting (UTG) in Munich in Germany for carry out the tensile test for the Al99.5 in thicknesses of 0.02, 0.1, and 0.2 mm.

REFERENCES

1. F. Vollertsen, Z. Hu, Tribological Size Effects in Sheet Metal Forming Measured by a Strip Drawing Test, Annals of CIRP 2006, vol. 55/1, 291-294.

2. D.D. Olssen, N. Bay, Prediction of Limits of Lubrication in Strip Reduction Testing, Annals of CIRP 2004, 53/1, 231-234.

3. P. Becker, H.J. Jeon, C. C. Chang, A.N. Bramley, A Geometric Approach to Modelling Friction in Metal Forming, Annals of CIRP 2003, 52/1, 209-212.

4. F. Vollertsen, Z. Hu, H. Schulze Niehoff, C. Theiler, State of the art in micro forming and investigations into micro deep drawing, Journal of Materials Processing Technology, 2004, 151, 70-79.

5. N. Tiesler, U. Engel, Microforming – Effects of Miniaturisation, Proceedings of the 8th International Conference on Metal Forming, Eds. Pietrzyk, M., Kusiak, J., ec al., Kraków, 2000, 355-360

6. Z. Hu, H. Schulze Niehoff, F. Vollertsen, Determination of the Friction Coefficients in Deep Drawing, Process scaling, eds.: Vollertsen F., Hollmann, F., BIAS-Verlag, ISBN 3-933762-14-6, Strahltechnik 24, 2003, 27-34

7. Z. Hu, F. Vollertsen, Scaled Friction Test Integrated in Deep Drawing, ICTMP2004, Denmark, Ed. Niels Bay, ISBN: 87-91035-12-0, 561-568

8. Z. Hu, F. Vollertsen, Tribological Size Effects in Sheet Metal Forming, ICTMP2007, Yokohama, Japan, Ed. Akira Azushima, ISBN 978-4-9903785-0-9, 163-168

9. A. B. Richelson, E. van der Giessen, Size Effects in Sheet Drawing, 9th International Conference on Sheet Metal 2001, eds. Duflou, J.R., Geiger, M., et al., 263-270

10. A. Messner, Kaltmassivumformung Metallischer Kleinstteile –Werkstoffverhalten, Wirkflaechen -reibung, Prozessauslegung-, Eds. Geiger, M., Feldmann, K., ISBN 3-87525-100-8, 1998

11. M.W. Storoschew, E. A. Popow, Grundlagen der Umformtechnik, VEB Verlag Technik Berlin, 1968

12. Z. Hu, H. Schulze Niehoff, F. Vollertsen, Tribological size effects in deep drawing, ICNFT2007, eds. F. Vollertsen, S. Yuan, ISBN 978-3-933762-22-1, 573-582

1 INTRODUCTION

Discrete element method has been widely developed for rheological study of true discrete materials like sand, powder and granular materials [1]. In the past ten years, their fields of applications have been extended to heterogeneous materials like concrete, biological materials or foams. Their ability to simulate multi body behavior is used for problems where:

- A great number of dissociated elements must be taken into account,

- A great number of default is encountered The main recent fields of applications are multi-fractures problems, where detached elements must be taken into account (wear in tribology [2], avalanches in geophysic, milling, grinding …). In material forming or cutting, the contact zone between the tool and the working piece is often very difficult to analyze because:

- The affected area has little dimension, - High mechanical, rheological, thermal

gradient are involved, - Physical phenomena are highly dynamic.

To analyze the contact behavior Godet developed the third body concept [3]. This included a description of the formation and movement of fragmented particles in the interface region. To study the behavior of the third body inside and outside the contact, Berthier proposed [4] the tribological circuit which allows the study of mass transfers inside the contact. Based on this tribological circuit and coupling Discrete Element models (DEM) to experimental, but simplified, wear studies, Fillot et al [5] proposed a set of equations that allows a qualitative modeling of wear as a mass balance in the contact area. Iordanoff et al. [6] showed how abrasion process can be studied as a particular and controlled wear process.

This paper first presents the Discrete Element Methods developed in the special case of material forming. Then, three examples are given to illustrate how this numerical tool can be used to study local properties in material forming: investigation of Sub Surface Damage in abrasion process, welding joint characterization in Friction Stir welding and thermal investigation of the tool-chip contact during cutting operation.

Discrete Element method, a tool to investigate contacts in material forming I. Iordanoff1, D. Richard2, S. Tcherniaieff1

1ARTS ET METIER PARITECH, LAMEFIP – Esplanade des ARTS et Metiers, Talence, FRANCE URL: www.lamef.bordeaux.ensam.fr e-mail: ivan.iordanoff@lamef.bordeaux.ensam.fr; Serge.tcherniaeff@bordeaux.ensam.fr 2INSA de LYON, LAMCOS – 18-20 Rue des Sciences, 69621 Villeurbanne Cedex, FRANCE URL: lamcos.insa-lyon.fr e-mail:david.richard@insa-lyon.fr;

ABSTRACT: This paper is devoted to the description of a numerical tool that allows the local study of contacts in material forming: Discrete Element Method (DEM). Discrete Element Methods allow the study of local properties (cohesion, thermal generation, fractures) on process behavior. This tool is used as molecular dynamics but allows the simulation of much representative volumes. Discrete element method is applied as a tool to understand/propose/confirm, physical scenario involved in the contact zone. It is shown in this paper how such numerical simulations can be used as a complementary tool for forming processes study. Examples are given on thermal study in cutting process, subsurface damages analysis in abrasion process and welding joint characterization in Friction Stir welding. Key words: Discrete Element Method, Abrasion process, Friction Stir welding, Tool Machining.

2 NUMERICAL TOOL

2.1 General description

The method described is based on smooth particle dynamic Method. The material is considered as a set of discrete particles that moves under a force field. The forces are contact forces to simulate multi body interactions and joint forces to simulate a solid made of discrete elements. Particle movements are calculated using an explicit algorithm to integrate the dynamic Newton law. A thermal model can be couple with the mechanical one. Both Conduction through contact and thermal energy source due to mechanical dissipation are taken into account. The two next paragraphs briefly described the mechanical and thermal part. More details can be found in references [2] and [7].

2.2 Mechanical Part

The particle interaction laws define the micro mechanical properties of the media. They are divided into contact forces and joint forces. Every particle is subjected to contact force. This force acts only when two particles geometrically interact. Two adjacent particles belonging to the same solid are linked by a solid joint that acts under traction and shear state.

δnc

δs

δntContact ForceLink Force

δnc

δs

δntContact ForceLink Force

Fig. 1. Contact and Link force calculation.

Contact force Contact force is divided into three parts: repulsion, adhesion (both are energy conservative) and energy dissipation. Geometrical interaction δnc allows the force calculation (Fig.1). • Repulsion is represented by a linear spring,

whose stiffness is K. Repulsive force Fr is: ncr KF δ*= (1)

• Adhesion has been simplified to a constant γ: γ=aF (2)

• The energy dissipation in the contact is due to the damping force written as:

ncijd MKF δα &*.2= (3)

where α is the damping coefficient (<1), ncδ& the impact speed, and Mij the equivalent mass of the contact. The sum of the interaction forces is:

ncijncdarcontact MKKFFFF δαγδ &..2. −+−=++=

Solid joint law The solid link is created between two adjacent spheres belonging to the same solid. The normal force due to this link is calculated only under traction state (Fig.1) and is equal to :

ntnt KF δ*= (6) The shear force is calculated thanks to the shear distance calculated between two points at the contact sphere periphery.

ss KF δ*= Particle detachment In the present study, it is assumed that a link breaks only under traction state: when the link is under traction and when the link force reaches a yield value Fy, the link is broken. The law used in this paper is characteristic of a brittle material. Numerical integration A Verlet velocity algorithm is chosen to integrate the velocities of the sphere centers, the rotational velocities, the new positions of sphere centers and the new position of the solid joint action points over a small time step dt while accelerations have been calculated thanks to Newton laws and the force laws described above.

2.3 Thermal part

The thermal process has been divided into two parts: - the heat generation is directly linked to the dissipative term of the interaction law and leads to a temperature increase dTi for the sphere i as:

0.5* . /( )i d nc i pdT F dt c Vδ ρ= &i i

where ρi, cpi and Vi are the density, thermal capacity and volume of the sphere i. - the diffusion is based on the work of Vargas et al. [10] and leads to a temperature increase dTij as:

( / ). (ij i i ij i jj contacts

dT dt V a T Tα−

= − )∑

where ai is the diffusivity of the sphere i, aij and Ti, Tj respectively the contact radius and the temperatures of the spheres i and j. Convection effects within the granular medium have

been neglected due to the strong density of the domain and the higher influence of diffusion.

3 APPLICATIONS

3.1 SSD Study in abrasion process.

Upper wall

Lower wall

DegradableFirst body

Imposed velocity

Periodical boundaries

x

y

z

Abrasive particles

Upper wall

Lower wall

DegradableFirst body

Imposed velocity

Periodical boundaries

x

y

z

Abrasive particles

Fig.2 Simulated domain

3.1.a Simulated domain The studied material is silica. According to fig. 2, the upper first body (Silica piece to be surfaced) is constituted by spheres linked together by elastic solid joints. This first body is linked to the upper wall. A normal load is applied to the upper wall which is free to move along axis z. The lower first body (tool) is simply defined as a lower rigid wall made of adjacent spheres. A constant velocity along x is applied to the lower wall. The abrasive particles are plates composed by 4 adjacent spheres linked by an elastic non breakable solid link. The effect of abrasive size and abrasive quantity through the contact are studied. Abrasive size is the same than silica particles (*1) or twice silica particles (*2). Table 1 summarizes the studied cases.

Nb of Abra. plates abrasive size Case ‘DiscreteBig’ 6 *2

Case ‘LayerOfLittle’ 39 *1 Case ‘DiscreteLittle’ 19 *1 Case ‘L.ayerOfbig’ 24 *2

Table 1 :Abrasive layer definition.

3.1.b Results Calculations are carried out till a certain amount of silica particle is removed from the silica volume, by the effect of broken links. The fixed amount is 150 particles. Broken links through the volume are plotted in terms of the distance from the surface. Curves fig. 3 show the results. The thickness is divided by particle mean radius.

5 10 15 20 25 30

10-3

10-2

10-1

100

0 5 10 15 20 25

Nb/

unit

ofV

olum

e

Dimensionless Thickness

1 : DiscreteBig2 : DiscreteLittle3 : LayerOfLittle4 : LayerOfBig

3

2 41

5 10 15 20 25 30

10-3

10-2

10-1

100

0 5 10 15 20 25

Nb/

unit

ofV

olum

e

Dimensionless Thickness

1 : DiscreteBig2 : DiscreteLittle3 : LayerOfLittle4 : LayerOfBig

3

2 41

Fig.3 :Number of broken joint through the thickness

The number of broken joints decreases exponentially with the depth. It greatly depends on abrasive geometrical properties and quantity. A little number of big abrasive particles creates more residual cracks in the material. These results are qualitatively in accordance with experimental results [9]. This first simple study demonstrates how discrete element simulations have the ability to simulate the formation of a great number of cracks and the link between sub surface damage and abrasive properties.

3.2 FSW joint characterization

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 115

20

25

30

35

40

45

50

50%0% 100%

0

+10

-10

VhighVlow

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 115

20

25

30

35

40

45

50

50%0% 100%

0

+10

-10

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 115

20

25

30

35

40

45

50

50%0% 100%50%0% 100%

0

+10

-10

0

+10

-10

VhighVlow

Fig.4 : quality of mixture in an FSW joint.

Simulating friction Stir welding requires the simulation of material flow, material mixture, tool-material contact and thermo mechanical coupling. D.E.M. could be in that case an interesting tool, because it has been widely used to simulate mixture of true granular flow. A first simulation is carried out with a tool that passes through two bodies initially in contact (two colours in fig.4). The bodies are simulated by an assembly of cohesive spherical particles. The particles of the two bodies are able to mix thanks to the rotating tool that pass through the interface. Fig.4 shows the result, in terms of mixture quality for two cases. For one simulation the translation velocity has the same order of magnitude than the rotating velocity. For the other one the translation velocity is largely less than rotating velocity. It is found that the quality of mixture is

much better when rotating velocity is high compared to translation velocity

3.3 Thermal Tool-Chip Contact investigation

Per

iodi

cal b

ound

arie

s

CHIP

CUTTING TOOL SURFACEVelocity

PressureAIR/CHIP SURFACE

Fig. 5. Tool-chip contact domain in DEM

The chip is modelled as a set of cohesive spheres with Steel properties. The upper wall represents the contact between the chip and the ambient air (for convection process) and the lower wall is the cutting tool surface (for diffusion process) sliding on the chip. During its generation, the chip faces different pressure fields from high levels (penetration of the tool in the material) to low levels (evacuation of it). The rheology of the chip greatly varies for each level of constraint, leading to different velocity accommodation processes and heat generation localisation [7]. The pressure effects on temperature increase within the chip and the tool are studied, assuming that all the mechanical dissipative power (from Eq. 3) is converted into heat.

0 10 20 30 40 50 60 70 80

Chi

p th

ickn

ess

Tool

Low pressure

Intermediate pressure

High pressure

Temperature increase °K

Fig. 6. Temperature profiles in the chip and part of the tool thickness for 3 different applied pressures.

Simulations have been carried out for a sliding velocity of 0.58m/s and three pressures (10, 20 and 200Mpa). It shows that the air plays an important insulating role: the maximum temperature is at the air-chip interface. The temperature increase of the tool is very small compared to the chip, the diffusion allowing an efficient evacuation of the heat, especially during the high pressure process. These first results are in accordance with classical thermal study of the tool-chip interface.

4 CONCLUSION

The discrete Element model has been presented. Three examples showed the capability of such simulations to analyse locally thermo mechanical properties of contacts between the tool and the material. The studied cases are very simplified but allow finding classical results from the literatures. Generally, to find such results with continuous models, both rheological laws and numerical methods reach high level of complexity. Here, the micro-mechanical model is simple and the behaviour is complex (multi-cracks, high deformation levels, thermo-mechanical interaction). D.E.M is a promising tool for the material forming field, where materials are submitted to very high stress, strain and thermal gradients and where rheological characterisation is still a true challenge. The main default of such method is the limitation in terms of the number of simulated particles (104 to 106) due to calculation duration. Coupling discrete model in the contact areas to continuous model elsewhere should be a solution for future and more realistic studies. REFERENCES 1. GDR MIDI, On Dense Granular Flows, Eur. Phys. J. E

14, 341-365, 2004 2. Fillot, Iordanoff, Berthier, “Modelling third body flows

with a discrete element method-a tool for understanding wear with adhesive particles”, Tribology International, Volume 40, Issue 6, June 2007, Pages 973-981

3. Godet, M., 1984, “The Third Body Approach: a Mechanical View of Wear”, Wear, 100, pp. 437-452.

4. Berthier, Y., 1995, “Maurice Godet’s Third Body”, 22nd Leeds-Lyon Symposium on Tribology, Tribology series, 31, pp. 21-30.

5. N.Fillot, I. Iordanoff, Y. Berthier : « Wear modelling and the Third Body concept », Wear, Elsevier, 2007, Vol.262 n°7-8, pp. 949-957.

6. Iordanoff, I., Charles, J.L., 2007, “Discrete Element Method: An Helpful Tool for Abrasion Process Study”, proceedings of the I MECH E Part B, Journal of Engineering Manufacture, Vol 221, Vol 6, pp 1007-1019.

7. Richard, D., Iordanoff, I., Renouf, M., Berthier, Y., 2008, “public ASME 2008”, to be published in ASME Journal of tribology

8 Verlet, L., 1967, Computer “Experiments” on Classical Fluids. I. Thermodynamical Properties of Lennard-Jones Molecules, Phys. Rev., 159, pp. 98-103

9 T. Sutatwala, L. Wong, P. Miler, M. D. Feity, J. Menapace, R. Steele, P. Davis and D. Walmer, "Sub-surface mechanical damage distribution during grinding of fused silica", J. Non Crys. Sol. 352 (2006) 5601-5617

10 Vargas, W. L., McCarthy, J. J., 2001, “Heat Conduction in Granular Materials,” AIChE Journal, 47, pp. 1052-10

Simulation of interface temperature and control of lubrication in the study of friction and wear in cold rolling

K. Louaisil1, M. Dubar1*, R. Deltombe1, A. Dubois1, L. Dubar1

1Laboratoire d’Automatique, de Mécanique et d’informatique Industrielles et Humaines, UMR 8530 Université de Valenciennes et du Hainaut Cambrésis, F-59313 Valenciennes Cedex 9, France URL: www.univ-valenciennes.fr/LAMIH/ *e-mail: mirentxu.dubar@univ-valenciennes.fr

ABSTRACT: A prototype testing device called upsetting rolling test (URT) was developed a few years ago and was recently optimised in our laboratory to simulate mechanical and thermal contact conditions in cold rolling. Moreover a new experimental protocol has been designed to reproduce the industrial quasi-boundary lubrication regime. Numerous experiments have been then carried out to study the influence of contact temperature and forward slip on friction, iron fines residues and microscopic surface aspects. A great influence of temperature on friction and wear has been put forward. An increase of the Coulomb friction coefficient associated with a decrease of the iron fines quantities has been shown when temperature increases which seems to indicate more adhesive wear. Key words: Cold rolling; Experimental simulation; Lubrication; Interface temperature; Friction; Iron fines 1. INTRODUCTION In cold rolling, contact at roll-strip interface, which is the main location of slipping, chattering or important wear rate phenomena, has to be controlled. Thus a simulation test, the Upsetting Rolling Test (URT) [1], was developed in our laboratory LAMIH to study the contact for the industrial Sendzimir’s 20-high, a few years ago. This reversible mill is used for reduction of low carbon steel strips. Lubrication is made by an emulsion of oil in water. All industrial contact conditions have to be reproduced on URT tool. Mechanical, thermal and lubrication contact conditions are first presented and specified. Then the principle and the methodology of the URT and its recent optimisations are described. Finally many tests have been performed in different cases of contact temperature and relative speed. Results such as mean Coulomb friction coefficient, iron fines pollution and observations of strip surface aspects are reported and analysed.

2. INDUSTRIAL CONTACT CONDITIONS 2.1 Mechanical contact conditions Operators drive the mill with force parameters: the roll separating force, the rear and front tensions and the torque. Output parameters are exit speed, Vs, peripheral roll speed, �R and reduction rate. Motion parameters are generally defined by the forward slip, Sfwd = (Vs – �R)/�R. The strip is driven forward thanks to the front tension and the frictional forces between rolls and strip [1]. 2.2 Thermal contact conditions Temperature has a decisive effect on lubrication, especially on the efficiency of additives [2]. The interface temperature depends on a lot of factors: contact partners properties as well as rolling parameters and convective cooling by the emulsion water are influent. First, heat created by plastic strain and friction are respectively favoured by reduction rate and rolling speed [3]. Secondly, a good efficiency of adequate lubricant additives limits friction and heat generation. Finally, heat

transfer across the roll-strip interface has also a decisive impact. From a general point of view the contact can be thermally defined by a mean contact thermal resistance which depends on strip and roll properties, such as roughness and conductivity, contact pressure and lubricant behaviour [4,5]. Nevertheless, it is quite important to note that there are important temperature gradients across the interface [4]. It present peaks on asperity locations where conditions, such as local plastic strain and friction, are more severe and where a temperature of 300°C could be locally reached [6]. Therefore, considering all these factors, it is quite difficult to determine the mean contact temperature corresponding to the studied process from literature. Indeed the thermal studies for cold rolling were generally carried out on tandem mill with notably the use of bigger work rolls than on a Sendzimir one [3,4,7]. According to the value of the convective coefficient used to define the cooling by emulsion, the mean interface temperature is evaluated between 100°C and 160°C [7]. As a conclusion, the mean contact temperature cannot be defined accurately for process analysed herein: a temperature of 120°C will be first tested.

2.3 Lubrication contact conditions

Fig. 1. SEM observations (a) pickled strip (b) industrial rolled strip after two passes (c) industrial rolled strip after four passes

(d) experimental rolled strip after two passes on URT

The contact ratio, R = Ar /Aa, where Ar and Aa are respectively the real and the apparent contact areas,

is representative of the lubrication regime. In cold rolling process, most of the computations carried out at the end of the roll bite estimate this ratio at more than 0,8 [8]: strip-roll interface is essentially governed by contact between asperities. Industrial strips have been observed by SEM after pickling (Fig. 1.a), after two passes (Fig 1.b) and in its final state after four passes (Fig. 1.c). Once the second pass is performed (Fig 1.b), a scaly surface is pointed out: the quasi-boundary lubrication regime concerning the studied process is confirmed.

3. UPSETTING ROLLING TEST (URT) 3.1 Objectives The upsetting rolling test (Fig. 2.) was designed by R. Deltombe & al. [1] and has been recently optimised. The mechanical running of the machine is detailed in [1]. Its aim is to reproduce contact characteristics in order to study their influence on the following parameters: • friction: a mean Coulomb friction coefficient is identified [1] • iron fines production [1] • surface aspects thanks to SEM observations

Fig. 2. Upsetting rolling test designed by authors’ laboratory LAMIH [1]

3.2 Principle: reproduction of industrial contact conditions

3.2.a Materials and geometry Industrial roll and strip specimens are used in order to reproduce industrial geometry roll bite and real material contact conditions.

3.2.b Stresses and plastic strain It is first important to note that the test is driven in a different, but equivalent, way than on industrial mill. The test parameters are sliding forward and

reduction rate (normal and tangential forces are measured as output parameters). A methodology was developed by Deltombe & al. [1]: thanks to industrial and experimental FEM models, the test parameters enabling the experimental reproduction of industrial contact stresses and industrial plastic strain are computed.

3.2.c Thermal conditions A convective heating system has been recently designed to control interface temperature. This system is associated with a convective cooling system to protect sensors and thus to avoid drift of measurements or even their degradation.

3.2.d Lubrication conditions A new methodology of lubrication application has been designed to solution this problem and reproduce industrial lubrication regime.

4. SIMULATION OF LUBRICATION 4.1 Lubrication by oil-in-water emulsion in cold rolling process The Wilson theory, called dynamic concentration theory [9] corresponds to the process studied herein. Because of oil high viscosity, this theory supposes that quasi no water or a tiny amount of water enters the roll bite [10]: water is mainly used as coolant fluid. The thickness of oil penetrating the contact is directly influenced by rolling speeds, oil viscosity and concentration, materials, contact pressure and roll bite geometry [8]. In section 2, lubrication was defined as almost boundary. Generally the minimal limit of a mixed regime thickness is considered equal to:

hlim = 0,35.Ra [6]

(1)

where Ra = mean strip roughness. 4.2 Methodology: computation of required oil quantity to apply No water is considered entering the roll bite: test will be made with neat oil. Speeds not being reproducible, lubricant feeding cannot be simulated on URT: the solution is to apply the good amount of neat oil on URT specimen beforehand. The used lubricant thickness for this study is the most critical

value we could meet in cold rolling, i.e. hlim, defined in 4.1. The mass corresponding to this thickness is computed according to the following procedure. Thanks to a 3D profilometer, the first step is to calculate the mean roughness, Ra, of strip studied herein and to deduce (1) the corresponding thickness to apply on it. Then, from a 5 mm² specimen surface previously analysed by the profilometer, computer software enables to calculate which volume of oil is necessary to fill in the valleys in order to obtain the required lubricant thickness computed in first step. Finally, knowing oil density the mass is deduced. As a validation of these recent optimisations, experimental surface aspect shows a scaly surface as for the industrial one after two passes (Fig. 1.b, 1.d).

5. TESTS AND RESULTS 5.1 Test objectives and conditions The aim is to analyse the influence of rolling parameters such as the pass number, the interface temperature and the forward slip on friction and wear (iron fines creation and surface aspects). All the test configurations are indicated for each pass on table 1. The quantity of lubricant applied on specimen strip is the mass corresponding to the thickness hlim (Eq. 1). Each configuration of test has been reproduced twenty five times. Table 1. URT test configurations for each pass

Configuration number 1 2 3 4

Forward slip 2% 2% 7% 7%

Interface temperature 40°C 120°C 40°C 120°C

5.2 Results and discussions • Friction: The coefficient values corresponding to each configuration are represented on Fig. 3.a: o Coulomb friction coefficient is higher in pass

two than in pass one whatever the test configuration. This can be explained by the increase of the contact ratio with the pass number. As an evidence, nearly no hole (or lubricant “traps”) remains once pass 2 has been performed (Fig 1.b)

o Coulomb friction coefficient increases with temperature. For a quasi-boundary lubrication

regime, the contact behaviour is governed by contact between asperities where no lubricant fluid film can remain [2]. To prevent real contacts between roll and strip at asperities locations, additives react with surfaces to form a protective film. The friction reducer additives are the polar ones and they get desorbed from a critical temperature estimated at about 100°C [2].

o No clear influence of forward slip on friction coefficient is observed

• Iron fines residues: After pass one, iron fines are measured only on roll. After pass two, they are measured both on URT specimen and roll. The quantities of iron fines residues collected for each configuration test are represented on Fig. 3.b: o The quantity of iron fines measured decreases

with temperature contrary to friction coefficient. As previously observed (Fig. 3.a), a higher contact temperature causes a higher friction coefficient which favours the wrenching of asperities: a high amount of iron fine residues should be found [1] contrary to what has been observed and measured (Fig. 3.b). As a consequence, the high temperature conditions imply that more wrenched particles must adhere (directly or indirectly) on strip or roll and cannot be collected.

o No clear influence of forward slip on the quantity of iron fine residues is observed

a) b)a) b)

Fig. 3. Results for each test configuration (a) Mean computed Coulomb coefficient (b) Total iron fine residues collected on roll and strip 6. CONCLUSIONS AND PERSPECTIVES To better understand cold rolling contact behaviour a new methodology was created and recently optimised to simulate mechanical, thermal and lubrication contact conditions. It has enabled to point out that:

o friction increases with temperature because of the desorption of polar additives

o quantity of iron fines residues decreases with temperature. It means that adhesive wear and transfer layer formation is favoured by a high temperature.

o friction increases with the number of the pass because of the decrease of the contact ratio

Despite these interesting conclusions, important questions remain: What are the nature and the real impact of the transfer layer? What is the pertinence of the current industrial measurement of iron fines pollution? Finally, what are the behaviour and the role of the lubricant additives?

ACKNOWLEDGEMENTS The present research work has been supported by Myriad, a CORUS group company, the CNRS, the European Community, the Association Nationale de la Recherche Technique, the Conseil Régional du Nord Pas de Calais. The authors gratefully acknowledge the support of these institutions. REFERENCES [1] R. Deltombe, M. Dubar, A. Dubois, L. Dubar, A new methodology to analyse iron fines during steel cold rolling processes, Wear 254 (2003) 211-221 [2] P. Montmitonnet, Tribologie du laminage à froid de tôles, Revue de la Métallurgie Paris, N°2 (2001), 125-130 [3] S. Matysiak, S. Konieczny, A. Yevtushenko, Distribution of friction heat during cold rolling of metals by using composite rolls, Numerical Heat Transfer, Part A, 34: 719-729 (1998) [4] A.K. Tieu, P.B. Kosasih, A. Godbole, A thermal analysis of strip-rolling in mixed film lubrication with O/W emulsions, Tribology International 39 (2006) 1591-1600 [5] A. Boutonnet, Etude de la résistance thermique de contact à l’interface de solides déformables en frottement : application aux procédés de forgeage, Thèse INSA de Lyon, 1998 [6] J. Molimard, Etude expérimentale du régime de lubrification en film mince – Applications aux fluides de laminage, Thèse INSA de Lyon (1999) [7] O. U. Khan, A. Jamal, G. M. Arshed, A. F. M. Arif, S. M. Zubair, Thermal analysis of a cold rolling process – A numerical approach, Numerical Heat Transfer, Part A, 46: 613-632 (2004) [8] N. Letalleur, Influence de la géométrie des aspérités dans un contact hydrodynamique ultramince. Effets locaux et comportements moyen, Thèse INSA de Lyon (2000) [9] A.D. Bugg, A. Shirizly, J.G. Lenard, The use of emulsions during cold rolling of steel strips,, The second European rolling conference rolling , VASTERAS Suède, (2000) [10] K. Dick, J. G. Lenard, The effect of roll roughness and lubricant viscosity on the loads on the mill during cold rolling of steel strips, Materials Processing Technology 168 (2005) 16-24

1 INTRODUCTION

In the blanking industry, the punch wear has a significant impact on production. In particular, important wear causes a wrong geometry of the cut pieces, which can lead to a rejection of products. Among other defects, the apparition of burr is the most incapacitating in the use of the blanking piece. So, in order to overcome this problem, industrials regrind their tools after a number of press storks. But this operation has a negative consequence on costs. In fact, the production stopping for regrinding punches creates substantial losses for the industry. Thus, it is important to control aspects of the blanking operation leading to the appearance of burr. To do this, it is necessary to have in advance a reliable way to measure the quantity of burr. In this paper, we present an original method for the quantification of the volume of burr. Moreover, a first approximation of the burr amount with the level of wear and the cutting materiel is also presented.

2 BLANKING BURR

The burr occurs on the cutting edge of pieces (fig.1), this presence is one of the most serious limitations of the blanking process. Indeed, beyond a certain burr

amount, the cutting pieces can be unusable.

Fig. 1.Geometry of the cutting edge.

HR : Roll-over depth. HS : Sheared edge. HF : Fracture depth. Hb : Burr height. φ : Fracture angle. Given the important issues related to the appearance of blanking burr, a lot of work has been undertaken to explain the appearance. Most of this work was devoted to the study of the influence of blanking conditions (clearance, radius, etc...) (Fig. 2), on the appearance of burr.

ABSTRACT: Blanking burr and punch wear are two phenomena closely linked. This leads to consider burr as the best criterion for regrinding or renewing tools. Thus, in order to control the quality and cost in blanking process, it is necessary to measure with the best accuracy the amount of burr and wear punch. To do this, a method for measuring the volume of burr was developed. This method is based on the use of 3D topographical images. On the other hand, punch wear was calculated by combination of a mathematical model to the geometric shape of punch. This model gives the geometry variations of the punch, depending on the number of press storks. At the end, the aim of this study is to identify a relationship between burr and punch wear

Key words: Blanking, Burr amount, Punch wear, Cut edge profile, Worn geometry

Metrology of the burr amount - correlation with blanking operation parameters (blanked material – wear of the punch)

H. Makich1, L. Carpentier1, G. Monteil1, X. Roizard1, J. Chambert2, P. Picart2

1LMS – ENSMM, 26 Chemin de l’Epitaphe 25000 Besançon, France URL: www.lms.ens2m.fr e-mail:hamid.makich@ens2m.fr; Luc.Carpentier@ens2m.fr;

guy.monteil@ens2m.fr; xavier.roizard@ens2m.fr 2Institut FEMTO-ST – DMARC, 24 Chemin de l’Epitaphe 25000 Besançon, France URL: www.femto-st.fr e-mail: jchamber@univ-fcomte.fr; philippe.picart@univ-fcomte.fr

Fig. 2.Geometrical parameters characteristic of blanking.

The main conclusions of these studies show that the size of burr depends mainly on the blanking clearance, the wear condition of the tool edge and properties of the cutting materials and the tool materiel. According to the studies carried out in [1] and [2], it appears that the wear promotes the appearance of burr and that the increase in the blanking clearance results in an increase in the fracture angle of the shear zone and the height of the burr. This is confirmed also in [3] and [4]. With regards to the properties of blanking materials, Gréban in [5] says that the composition of cutting alloys has a strong influence on the quantity of burrs that is created during the blanking operation. However, the works are more on the theoretical study of the conditions of burr appearance, than on an experimental metrology on the burr amount.

3 METROLOGY OF BURR

The possibility to estimate the burr size on the blank pieces remains of great importance in the blanking industry, because the quality of the products is determined with the evaluation of the level of acceptable burr on the parts. It is possible to study, in theory, the emergence of blanking burrs. However, if we wish to verify the theoretical predictions, it is necessary to be able to accurately and reliably measure the quantity of burrs. To summarize, we can return to what might be a definition of the burr height, the main criterion mentioned here. According to [6], because the sheet can have a residual macroscopic deformation during the cutting operation, the height of burr is defined as "the difference between the highest point of burr and the surface of the sheet metal immediately adjacent to the burr". This definition seems satisfactory and will be the basis for development of our work.

3.1 Metrology

We can see that in literature, there are many ways of measuring burr, but fundamentally based on a destructive test of blanking pieces and therefore of metrology in a few positions of the cutting edge. Other methods that have been proposed for metrology burr heights were the non-contact or tactile optical profilometry [7], or the vision metrology [8] of the cutting edge, which is aimed at pieces with straight faces. The latter allows then access to the profile heights of burr over a length of cut board of 4mm. Another example is a measurement using the shadow of burr or a more traditional measure, the mechanical profilometry.

3.2 Crushed burr

In the case of progressive blanking, complex geometry of the parts is produced by several press strokes, which has the effect of crushing burrs. This makes it more difficult to measure the height of the burr by traditional means.

Fig. 3. Crushed burr.

This has prompted us to seek, in this work, a new method more able to give us a precise quantification of the crushed burr. To do this, we sought to define a reference plan around the cutting form, because the adjacent sheet is distorted.

3.3 Establishing a means of measurement

In our approach to the burr quantification, we sought to meet a need, which is to confront the theoretical results with experimental robust measurements.

3.3.a Images Acquisition The device used is a microscope Infinitefocus ®

Punch

Die

Sheet metal

/ 2J

pr

mr

/ 2J

α

β

Blank-holder

pD

mD

e

Alicona (Fig.4). It is based on an optical microscope coupled to a camera. Unlike a confocal microscope, which uses a monochromatic sensor, the device uses a sensor of contrast in color. On the whole height, the microscope collects images which are then analysed by the software. The latter analyses each pixel and compares it to its neighbours to see if it is focused or not. Then, it reconstructs a three-dimensional image from this focus. From this image, surface-, volume-, surface state- and topography dimensional measurements are possible. In accordance with selected lens, the image size ranges from 2.1x1.6mm ² to 103x83 µm² and resolution in height from 444nm to 20nm. If the sample or area of interest is larger than the size of capture of the selected lens, it is possible to achieve a multi-field acquisition through motorized tables. We only need to identify the entire area to be analysed, and by interconnection between image fields acquired with recoveries, we reconstructed a joined-fie image.

Fig. 4. The optical device measurement: InfiniteFocus ®

3.3.b Measurement The measurement method is developed within the LMS specifically for quantifying the burr volume. The originality of this method lies in the transformation of the blanked shape, (here a circular shape). The main stages of this transformation are: • First, a straightening of the image around the

blanked shape (hole) is achieved (Fig. 5). This is done in order to eliminate any imperfections in the image rectitude. The straightening of the image (outside the zone of burr) is made with the method of least squares.

Fig. 5 . Profiles on the surface of the sheet before and after

straightening .

• Secondly, an image processing is performed in

order to extract the profiles that we wish vertically from the contour of the blanked shape (Fig. 6).

Fig. 6. Example of obtained profile.

• Thirdly, the profiles are extracted in multiples so as to build a rectangular matrix; each profile being the average of 100 adjacent profiles. Thereafter, the image is adjusted, outside the zone of the blunder, with a linear polynomial of degree 2, which assembles all the profile (Fig. 7).

Fig. 7. Representation of the profiles assembling.

• Finally, the volume is determined only in the areas of the burr presence on the image.

Profile beginning

Sheet metal

Blanked Shape

Profile

Profile after straightening

Profile before straightening

Fig. 8. Calculations on the defined area

4 RESULTS AND CONCLUSIONS

Figure 9 shows the evolution of the burr volume according to the number of press stroke. We observe an increase of a burr volume with the increasing number of press stroke.

0,00E+00

2,00E+05

4,00E+05

6,00E+05

8,00E+05

1,00E+06

1,20E+06

0 50000 100000 150000 200000 250000 300000

Number of press strokes

Burr volume (µm3)

Fig. 9. Burr volume.

Other results (Fig. 10) can show several very interesting phenomena. Indeed, a quick initial analysis of the curve allows noting that the kinetics of appearance of burr, as it was highlighted by Gréban [5], depends directly on the blanking material.

0,00E+00

2,00E+05

4,00E+05

6,00E+05

8,00E+05

1,00E+06

1,20E+06

1,40E+06

1,60E+06

0 100000 200000 300000 400000 500000 600000 700000 800000

Number of press stroke

Burr volume (µm3)

Fig. 10. Burr volume for different materials.

Because of the non-homogeneity of the burrs on the blanked contour, the method seems the most appropriate, because it is obtained from a large

number of profiles, which has the advantage of allowing a complete scan of the blanking edge.

REFERENCES

1. N. Hatanaka, K. Yamaguchi, N. Takakura, T. Iizuka « Simulation of sheared edge formation process in blanking of sheet metals » Journal of Materials Processing Technology, 140 (2003) 628-634.

2. C. Husson, C. Poizat, L. Daridon, S. Ahzi « Travail des métaux en feuilles - Simulation numérique 2D du découpage d’alliages de cuivre » 7ème Colloque National en Calcul des Structures, 2005.

3. A. Touache « Contribution à la caractérisation et à la modélisation de l’influence de la vitesse et de la température sur le comportement en découpage de tôles minces. » Thèse à l’Université de FRANCHE-COMTÉ, 14 Décembre 2006.

4. Z. Tekiner, M. Nalbant, H. Gürün « An experimental study for the effect of different clearances on burr, smooth-sheared and blanking force on aluminium sheet metal » Materials & Design, 27 (2006) 1134-1138.

5. F. Gréban, G. Monteil, X. Roizard « Influence of the structure of blanked materials upon the blanking quality of copper alloys » Journal of Materials Processing Technology, Vol 186 N°1-3, (2007), pp 27-32.

6. M. Grünbaum, J. Breitling, Influence of high cutting speeds on the quality of blanked parts. ERC report N°5-96-19, 1996.

7. F. Gréban « Découpabilité du cuivre et des alliages cuivreux. » Thèse à l’Université de FRANCHE-COMTÉ, 21 février 2006.

8. E. Levy « Mise en oeuvre des systèmes de vision pour le contrôle des pièces découpées ».Document interne Statimage. 2000.

Mat 1

Mat 1 Mat 1

Mat 2 Mat 3

1 INTRODUCTION

Frequently, the life of the tools used in sheet metal

forming operations is determined by a phenomenon

known as galling. Galling is described in the ASTM

G40 Standard [1] as “a form of surface damage

arising between sliding solids, distinguished by

macroscopic, usually localized, roughening and

creation of protrusions above the original surface; it

often includes plastic flow or material transfer or

both”. In spite of the significant number of

publications on this phenomenon, some attention is

still devoted towards its effective occurrence [2]

and, more importantly, towards the development of

tests able to reproduce galling in laboratory [2-10].

One important reference regarding these tests is the

ASTM G98 method [3], according to which a

threshold pressure for galling is calculated based on

the visual inspection of the presence of galling at the

surfaces of a button and/or a block that were pressed

against and manually rotated with respect to each

other at increasing loads. Despite the popularity of

the ASTM G98 test method, many works criticize

the applicability of such threshold pressure value for

design purposes and suggest modifications to the

original standard method. These changes intend to

overcome some of the method limitations, such as

the heterogeneity of contact pressure distribution [4]

or the fact that the velocity is zero at the centre of

the rotating button [2,4]. Other questionings on the

ASTM G98 test method include the non

consideration of the statistical nature of galling

[4,5]; the absence of constant speed during manual

rotation [6] and even cost issues associated with the

necessity of a large number of specimens or the

availability of an equipment capable of applying

large normal loads [7].

Some of the alternatives for laboratory testing of

galling include a modification in the geometry of the

specimens, such as the shape of initial contact,

which is considered as a line [8] or a point [7,9],

rather than an area (ASTM standard). In these tests,

the analysis of the occurrence of galling may remain

based on the visual inspection of contact surfaces [8]

or, as in the case of two crossed cylinders that slide

against each other, may be based on the friction

coefficient calculated during the experiments [7,11].

ABSTRACT: Frequently, the life of the tools used in sheet metal forming operations is determined by a

phenomenon known as galling, which originates from the adhesion of the sheet to the forming tool surface.

The application of coating architectures composed by single or multiple layers of Physical Vapor Deposition

(PVD) films, such as TiN, TiCN, CrN, TiCNAl, may significantly reduce the chemical interaction in the

contact, up to the point that no significant adhesion may be observed for an extended number of forming

operations. Usually, the evaluation of the behavior of different thin film architectures is conducted using

tribometers that may or may not reproduce the conditions found in industrial practice. This work presents a

tribological analysis of coated and uncoated surfaces of tools used in industrial sheet metal forming

operations and discusses the capability of laboratory tests in reproducing the situations found in practice.

Key words: Galling, PVD coatings, Laboratory tests, Field tests

Laboratory and field analysis on the tribological behavior of coated and

uncoated forming tools

M.A.R.S. Mendes1, R.M. Souza

1, P.K. Vencovsky

2, Y. Berthier

3

1Surface Phenomena Laboratory, Department of Mechanical Engineering, Polytechnic School of the

University of São Paulo – Av. Prof. Mello Moraes, 2231, 05508-900 São Paulo, SP, Brazil URL: www.poli.usp.br e-mail: roberto.souza@poli.usp.br; marcorsm@usp.br 2Bodycote Brasimet – Av. Nações Unidas, 21476, 04795-912 São Paulo, SP, Brazil

URL: www.brasimet.com.br e-mail:Paulo.vencovsky@bodycote.com 3INSA de Lyon, LaMCoS – 20 avenue Einstein, 69621 Villeurbanne, France

URL: lamcos.insa-lyon.fr e-mail: yves.berthier@insa-lyon.fr

In spite of the predominance, galling is not the only

phenomenon observed in forming operations and,

during the past decades, more sophisticated tests

have been developed [10,12-15] aiming a better

laboratory reproduction of the conditions found in

practice. Most of these tests are conducted in

specially designed rigs and involve the actual

contact of metal strips against tool materials.

Many of the laboratory studies mentioned above

[2,9,13] have also explored the point that the

evolution of galling at a given surface presents a

correspondence with variations in roughness [16].

The intensity and continuity of the search for a

laboratory test that accurately reproduces the

tribological conditions in forming operations are

understandable, not only based on the possibility of

knowing these operations in detail, but also based on

the ability that such test would have to provide hints

towards materials development. In particular, such

test would provide a means of keeping track of the

applicability in forming operations of constantly

emerging developments of Physical Vapor

Deposition (PVD) coating architectures.

In this work, eight sheet metal operations, conducted

with coated and uncoated tools, were analysed in

terms of wear phenomena observed at tool surface.

Results were then discussed in the light of the

laboratory tests mentioned in the previous

paragraphs.

2 FIELD TESTS AND ANALYSIS

Table 1 presents a list of the industrial sheet metal

operations studied in this work. All operations were

part of the actual production line of a bearing

industry. Operations 1 to 6 were conducted as steps

of the manufacturing process of cups with diameter

of 30 mm and depth of 25 mm. Similarly, operations

7 and 8 were part of the manufacturing procedure of

cups with diameter of 16 mm and depth of 12 mm.

As indicated in the table, two types of tool steel were

selected for the tools (or substrates), AISI H13 and

AISI M2, which presented Rockwell C hardness of

52 (approx. 6.0 GPa) and 61 (approx. 7.5 GPa),

respectively. The TiCN and TiCNAl coatings were

deposited in a commercial cathodic arc evaporation

deposition chamber and, according to the

manufacturer, presented hardness on the order of 29

GPa and 35 GPa, respectively. All operations listed

in table 1 were lubricated with oil and conducted on

0.8 mm thick 16MnCr6 steel sheets. In table 1, the

column labelled as “Production” indicates the

number of parts manufactured by each tool, whose

life limiting factor was based on the quality of the

formed part. The end life of all uncoated operations

(1-4) was determined by galling and the end life of

all coated tools was not determined by galling, but

by other wear phenomena that resulted from the

contact between the punch and the die (e.g. wear on

the sides of the punches due to slight centre

misalignment).

Table1. Overall conditions in sheet metal forming operations.

Operations were lubricated with oil and conducted on 8.0 mm

thick 16MnCr6 steel sheets

Operation

number

Substrate Coating Type of

Operation

Production

(x103 parts)

1 AISI H13 None Ironing 32

2 AISI M2 None Blanking 175

3 AISI M2 None Bending 175

4 AISI M2 None Coining 94

5 AISI H13 TiCN Ironing 1000

6 AISI M2 TiCN Coining 1000

7 AISI M2 TiCNAl Ironing 1200

8 AISI M2 TiCNAl Coining 1200

After removal from the press, tool surfaces were

cleaned and observed with optical and scanning

electron microscopy (SEM). Surfaces were also

analysed in terms of roughness.

Figure 1 presents a detail of the surface of the tool

used in operation 2. Two distinct regions may be

observed in figure 1, one that presents the original

surface, and an external ring where the original

grinding lines are no longer visible. Additional

analyses of the external ring revealed features

typical of galling, such as surface protrusions

(figure 2) and the presence of additional peaks and

valleys in the roughness profile (figure 3).

Fig. 1. Optical microscopy of the tool used in the blanking of

8.0 mm thick 16MnCr6 steel sheets (operation 2 in table 1)

Fig. 2. SEM analysis of the tool used in the blanking of 8.0 mm

thick 16MnCr6 steel sheets (operation 2 in table 1). Arrows

coincide with the radial direction of the cylindrical tool: (a)

Secondary electron image and (b) backscattered electron image

Fig. 3. Roughness analysis of the tool used in the blanking of

8.0 mm thick 16MnCr6 steel sheets (operation 2 in table 1): (a)

Unworn surface and (b) worn external ring.

Operations 4 and 6 are representative of the

differences observed with coated and uncoated tools,

since the same tool geometry and overall operation

conditions were present in both cases. In the tool

used in operation 4, it was once again possible to

observe that worn surfaces present additional peaks

and valleys in the roughness profile, which were not

clearly noticeable at the surface of tool used in

operation 6.

Differences between coated and uncoated surfaces

were also observed in the SEM analysis. Although in

a scale different from that shown in figure 2, galling

features were also observed at the surface of the

uncoated tool used in operation 4. These features

were not distinguishable at the TiCN coated surface

(figure 4), which presented several scratches in the

radial direction, but where it was not possible to

identify the presence of material transferred from the

sheet to the coated tool surface, in spite of the larger

number of parts produced with this tool.

3 DISCUSSION

As a general trend, coated tools, independent of the

coating material, outperformed uncoated tools in all

industrial operations analysed in this work. The

better performance of the coated tools may be

attributed to their ability to minimize, or postpone,

the occurrence of galling; an ability that was verified

as an absence of particles adhered to the tool surface

and the absence of additional peaks and valleys in

the roughness profile.

Fig. 4. SEM analysis of the TiCN coated tool used in the

coining of 8.0 mm thick 16MnCr6 steel sheets (operation 6

in table 1).

It is not appropriate to make a direct comparison of

the results obtained in this work with those of the

tests presented in the literature, since most of the

elements of the tribossystem were different,

including materials, geometries, loads and

lubrication condition. However, some considerations

are still possible and, to this end, the laboratory tests

were intentionally grouped into three categories,

which are the tests based on the ASTM G98

standard; tests with initial line or point contact and

tests that involve actual sliding of strips against the

surface of tools.

It is possible to suppose that the use of the tool

materials of this work in tests based on the ASTM

G98 standard would provide lower threshold galling

pressures for the uncoated tools than the coated

ones. However, in addition to the questionings on

the concepts of ASTM G98 standard [2,4-8],

literature [4,9] data also indicates that the

preparation of these tests must be careful in order to

provide good alignment between the contacting

surfaces, even if some of the alternatives to the

original standard are selected.

The simplicity and quickness of tests with line or

point contact may be advantageous. However, taking

the example of the crossed cylinder test, it is

possible to state that the use of the outputs of this

test to study forming operations is based on the

assumptions that: (i) galling is associated with an

increase in friction coefficient and (ii) galling is the

predominant mechanism in forming operations. The

first of these assumptions may seem intuitively

correct and a direct correlation between friction and

galling has been observed in some cases [14]. On the

50 µm a b

a b mm mm

µm

µm

Unworn Worn ring

other hand, it has not been observed in others [2].

The second assumption is also plausible, but the

literature also presents examples where wear was

predominantly dictated by ploughing, rather than by

galling [14]. In terms of the comparisons provided

by the crossed cylinder test, Podgornik [7] observed

that uncoated tools performed better than tools

coated with titanium nitride (TiN). Based on the

results of the industrial operations analysed in this

work, it is difficult to imagine that this result would

occur in industry. Therefore, in addition to the

validity of the assumptions presented above, it is

possible to suppose that some of the conditions

imposed during the line or point contact tests are

unable to entirely reproduce the conditions found in

practice. One of the possible explanations for these

differences is the nominal contact pressure, which,

in these tests, tends to be larger than in the industrial

conditions [9]. A reduction in pressure would be

possible in the crossed cylinder test, but, in this case,

the sliding distance would probably have to be

significantly increased until detectable wear would

be noticed. As a result, repetition of the regions in

contact would possibly occur during each test cycle;

a condition that is frequently not observed in

forming operations [12].

Finally, it is possible to expect that tests that involve

actual sliding of strips against the surface of tools

would provide results in good agreement with those

observed in industrial operations. However, these

tests frequently demand specially designed

equipment and may require large and complex

shaped specimens.

4 CONCLUSIONS

The literature presents a large number of laboratory

tests dedicated to the study of the wear phenomenon

known as galling and/or the tribology in forming

operations as a whole. A qualitative comparison of

some of these tests with results obtained directly in

industrial operations indicate that, in spite of the

large number, most of the options are still not

entirely satisfactory. Apparently, none of them is

able to provide a simple means to analyse factors

such as contact pressure, lubrication, continuous or

intermittent contact, refreshment of contact area, in

order to not only analyse galling, but also, if

necessary, conduct the test with different levels of

predominance of galling over other phenomena.

ACKNOWLEDGEMENTS

The authors acknowledge two Brazilian founding agencies for

financial support: the São Paulo State Research Foundation -

FAPESP through projects number 2006/02006-2 and

2003/10157-2 and The National Council for Scientific and

Technological Development (CNPq) through project

550235/03-5.

REFERENCES

1. ASTM G40, Standard Terminology Relating to Wear

and Erosion, ASTM International, West Conshohocken

(2005).

2. K.G. Budinski, M.K. Budinski and M.S. Kohler, A

galling-resistant substitute for silicon nickel, Wear 255

(2003) 489-97.

3. ASTM G98, Standard Test Method for Galling

Resistance of Materials, ASTM International, West

Conshohocken (2002).

4. S.R. Hummel, Development of a galling resistance test

method with a uniform stress distribution, Tribol. Int. 41

(2008) 175-80.

5. S.R. Hummel and B. Partlow, Comparison of threshold

galling results from two testing methods, Tribol. Int. 37

(2004) 291-5.

6. K. Gurumoorthy, M. Kamaraj, K. Prasad Rao and S.

Venugopal, Development and use of combined wear

testing equipment for evaluating galling and high stress

sliding wear behaviour, Mater. Des. 28 (2007) 987-92.

7. B. Podgornik, S. Hogmark and J. Pezdirnik, Comparison

between different test methods for evaluation of galling

properties of surface engineered tool surfaces, Wear 257

(2004) 843-51.

8. S.R. Hummel, New test method and apparatus for

measuring galling resistance, Tribol. Int. 34 (2001)

593-7.

9. P.A. Swanson, L.K. Ives, E.P. Whitenton and M.B.

Peterson, A study of the galling of two steels using two

test methods, Wear 122 (1988) 207-23.

10. L.M. Bernick, R.R. Hilsen and C.L. Wandrei,

Development of quantitative sheet galling test, Wear 48

(1978) 323-46.

11. M. Hanson, N. Stavlid, E. Coronel and S. Hogmark, On

adhesion and metal transfer in sliding contact between

TiN and austenitic stainless steel, Wear in press.

12. E. Schedin, Galling mechanisms in sheet forming

operations, Wear 179 (1994) 123-8.

13. F. Clarysse, W. Lauwerens and M. Vermeulen,

Tribological properties of PVD tool coatings in forming

operations of steel sheet, Wear 264 (2008) 400-4.

14. C. Boher, D. Attaf, L. Penazzi and C. Levaillant, Wear

behaviour on the radius portion of a die in deep-drawing:

Identification, localisation and evolution of the surface

damage, Wear 259 (2005) 1097-108.

15. M. Hirasaka and H. Nishimura, Effects of the surface

micro-geometry of steel sheets on galling behavior, J.

Mater. Process. Technol. 47 (1994) 153-66.

16. J.L. Andreasen, N. Bay and L. de Chiffre, Quantification

of galling in sheet metal forming by surface topography

characterization, Int. J. Mach. Tool Manu., 38 (1998)

503-10.

1 INTRODUCTION

Sulfur-containing organic molecules have been used for a long time as extreme pressure additives in lubricating oils. Several papers report on comparisons of the tribological efficiency of organo-sulfur additives as a function of number of sulfur atoms, alkyl- or aryl-chains [1]. In tribotests, involving ferrous materials and S-containing additives, the formation of the FeSx compound has been detected in superficial layers by different surface analysis techniques [2, 3]. FeS, also known as troilite, has a layered hexagonal structure, low shear strength and high melting point (1100 °C), so it is an effective solid lubricant like graphite and MoS2 [4]. HSAB theory may also shed light on those species which react to form these layers [5]. The molecule may be thermally decomposed or oxidized in the lubricant during rolling process or reacts on the metallic surface. The details of the reaction path are nevertheless poorly known, and so is the effect of the environment (dissolved oxygen, temperature, …) on preferred reaction paths and reactivity and will be analyzed in the present work. Quantum models have been applied to study the reactivity of some sulfur molecules on various metallic surfaces [6]. Most of the papers are focused

on copper, palladium and nickel for catalytic purpose with simple sulfur molecules. Reactions act on the first layers of metal mainly for gas phase reactions [6,7]. The sulfur may be randomly distributed on copper surface or organized as mobile sulfur metal clusters Cu3S3 [8]. The organisation of metal sulfur molecules may depend on the reaction path of the sulfur during its reaction with metal surface. We consider here the reaction of a di-tertio-dodecyl-pentasulfur molecule (C12H25 S5 C12H25, DTDP) with a bcc iron [001] surface.

2 SIMULATION

The simulations have been done with a semi-empirical model MOPAC using the functional PM5. The results were confirmed by simulations with other semi-empirical functionals AM1 and PM3 and a DFT functional B88-LYP. The MOPAC software is coupled to the graphical user interface CAChe. Full geometry optimization of a standalone molecule is searched first to define the equilibrium conformation of the molecule. This conformation is searched from various initial non-equilibrated state. A molecular dynamics simulation allows us to follow the conformational change around the equilibrium conformation. The stable conformation

ABSTRACT: sulfur-containing molecules are used as extreme pressure additive for low-alloy steel surfaces.These molecules react with ferrous surfaces and form FeSx compounds which strength the surface and reduce scuffing. The stable conformation and the chemical stability and reactivity of an alkyl sulfur molecule (DTDP) with oxygen and an iron surface are here predicted by DFT model. The effect of the conformation of the molecule respective to the surface acts on the surface reactivity.

Key words: polysulfur, iron, reactivity, semi-empirical functional,

DFT – Modeling of the Reaction of a Polysulfur Extreme-Pressure Lubricant Additive on Iron Surface

B. Monasse, P. Montmitonnet

Ecole des Mines de Paris PARISTECH, CEMEF, UMR 7635, BP 207, 06904 Sophia-Antipolis Cedex, France URL: www-cemef.cma.fr e-mail: bernard.monasse@ensmp.fr e-mail: pierre.montmitonnet@ensmp.fr

corresponds to the minimum energy state. The reactivity of this molecule at the contact is then predicted with various approaches of this molecule near the metal surface.

3 RESULTS

3.1 Molecular conformation of DTDP molecule

The DTDP molecule is symmetrical from the central sulfur atom, which may act of its equilibrium conformation. A fully extended molecule is created with CAChe software and relaxed with various semi-empirical and DFT functionals at 0K (fig. 1).

a b c Fig. 1. Equilibrium conformation of DTDP molecule along the

three main orientation directions The molecule forms a loop with parallel alkyl chains in extended conformation. The loop results from gauche conformations of S-S bonds. The parallel chains results from maximal van der Waals and electrostatic interactions of aliphatic segments. The conformation change of the molecule results only from the S-S bond rotation. The aliphatic chains are almost normal to the mean plane containing sulfur atoms. Complementary simulations with a variable number of sulfur atoms show that this loop conformation appears from three sulfur atoms up. Molecular dynamics simulations at 300K of a DTDP molecule show out a fluctuation of orientation and distance of almost all-trans aliphatic sequence around the equilibrium conformation. They come from fluctuations of dihedral angle S-S-S around the equilibrium conformation. The loop corresponds to

the equilibrium conformation of the molecule. The conformational energy decreases by 57.3 kJ.mol-1 from the fully extended molecule to the stable conformation.

3.2 Stability of bonds and oxidation

The homolytic scission of S-S bonds and their oxidation are two main hypotheses proposed to explain the reactivity of sulfur based molecules. We have calculated the intrinsic stability of the various bonds (S-S, C-S, C-C) inside the molecule and their reactivity with O2 molecules. The energy of each bond is estimated by the following energy balance deduced from calculation of the energy of the molecule and of the molecular fragments resulting from the bond leakage:

Ebond = Efragment1 + Efragment2 - Emolecule (1)

The atoms are numbered from one methyl carbon and sulfurs range from S13 to S17, and each bond is defined by the linked atoms (fig. 1). The bond energy depends on the nature and localization of bonds inside the molecule, half molecule is represented thanks to molecular symmetry and most of C-C bonds are not considered for their high bond energy (table 1).

Table 1: bond energy inside DTDP molecule bond Energy kJ.mol-1

C4-C5 268 C11-C12 246 C12-S13 206 S13-S14 215 S14-S15 164

The sulfur and C-S bonds are significantly weaker than carbon bonds and the two bonds linked to the central sulfur are the weakest leading to the most probable position for an homolytic rupture. The following mechanism is considered for a thermo-oxidative degradation (fig. 3) :

C12H25S5C12H25 + O2 → R-S-O. + R’-S-O (2)

and leads to the following reaction energy (table 2)

Table 2: energy of oxidation reaction on various sulfur bonds

reaction Energy (kJ.mol-1) C12-O S13-O - 83 S13-O S14-O - 171 S14-O S15-O -337

C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11 C12

C29 C28 C27 C26 C25 C24 C23 C22 C21 C20 C19 C18

S13 S14 S16 S17 S15

The high oxidation energy on bond S14-S15 may suspect a thermal oxidation of the molecule.

Fig. 2. excited state for oxidative reaction on S14 and S15

atoms The bond S14-S15 is the weakest bond for oxidation and for homolytic scission. This part of the molecule is a clear candidate to react with a metallic surface.

3.3 Reaction of DTDP molecule with iron surface

A direct reaction on an iron surface is also analyzed. A layer of Fe atoms is created and locked to represent a surface of bcc α-iron, with the same crystallographic parameters. The iron functional is available inside PM5 basis-set and is able to predict reaction with DTDP sulfur based molecule. A DTDP molecule is then located near the surface with various orientations of sulfur atoms. Figure 3 presents one initial scheme.

Fig. 3. An initial conformation of a DTDP molecule near the iron surface

The relaxation of this molecule on the iron surface predicts the reaction of two non consecutive sulfur atoms with iron atoms (fig. 4).

Fig. 4. Conformation of a DTDP molecule after reaction with

the iron surface (tilt/iron plane ~ 60°) This type of reaction leads to isolated sulfur-iron bonds distributed on the surface which are weakened inside the DTDP molecule (fig. 6). Other simulations with the same molecule but with a different tilt-angle with the surface lead a different reaction path (fig. 5).

Fig. 5. Conformation of a DTDP molecule after reaction with

the iron surface (side and bottom view) (tilt/plane ~ 80°) All the sulfur atoms react with an iron atom, or more, which leads to a cluster of Fe-S bonds on iron surface. These bonds are strong and weaken the intramolecular bonds, measured by the reducing their covalent order (fig. 7).

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 5 10 15 20 25 30 Fig. 6. Covalent order of intramolecular bonds inside DTDP

molecule (from fig. 4)

ordre liaison

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 5 10 15 20 25 30

liaison

Fig. 7. Covalent order of intramolecular bonds inside DTDP

molecule (from fig. 5)

The two (fig. 6) and four S-S bonds orders (fig. 7), respectively, are almost reduced to none after reaction on iron surface in favor to Fe-S bonds. The iron surface is highly reactive to sulfur atoms. The resulting sulfur groups can be directly organized in cluster and randomly distributed as a function on the approach path of the DTDP molecule onto the iron surface.

4 CONCLUSIONS

The conformation and the stability of the bonds of a sulfur based molecule are mainly dependent on sulfur bonds. These ones are much more reactive and oxidized than carbonyl bonds. Sulfur atoms preferentially react with iron surface and lead to randomly distribution of isolated Fe-S bonds and on small clusters of Fe-S as an effect of contact path of the molecule on the surface. These results, in one hand, should be completed with similar analyses on other crystallographic planes of iron and of iron oxide surfaces and, in other hand, by prediction of the reaction rate of oxidation or bond rupture by calculation of the excited state energy.

REFERENCES

1. K.D. Allum and J.F. Ford, The influence of chemical structure on the load carrying properties of certain organo-sulfur compounds, J. Instit. Petrol., 497 (1965) 145-169.

2. I. M. Petrushina, E. Christensen, R. S. Bergqvist, P. B. Møller, N. J. Bjerrum, J. Høj, G. Kann and I. Chorkendorff, On the chemical nature of boundary lubrication of stainless steel by chlorine- and sulfur-containing EP-additives, Wear, 246 (2000) 98-105.

3. G. Dauchot, R. Combarieu, P. Montmitonnet, M. Repoux, G. Dessalces and F. Delamare, Tribochemical reactions in cold-rolling : a ToF-SIMS study of the chemisorption of the lubricant additives on the sheet, Rev. Met. Paris, CIT-Science et Génie des Matériaux, 2 (2001) 159-168.

4. D.M. Zhuang, Y.R. Liu, J.J. Liu, X.D. Fang, M.X. Guang and Y. Cui, Microstructure and tribological properties of sulphide coating produced by ion sulfuration, Wear, 225 (1999) 799-805.

5. R.G. Pearson, Chemical hardness, Wiley-VCH, New-York (1997).

6. H. Toulhoat, P. Raybaud, S. Kaztelan, G. Kresse, J. Hafner, Transition metals to sulfur binding energies relationship to catalytic activities in HDS: back to Sabatier with first principle calculations, Catal. Today, 50 (1999) 629-636.

7. D.R. Alfonso, First-principles studies of the √7 x √7 19.1° structure of sulfur on the Pd(111) surface, Surface Science, 601 (2007) 4899-4909.

8. G.B.D. Rousseau, A. Mulligan, N. Bovet, M. Adam, V. Dhanak and M. Kadodwala, A structural study of disordered sulfur overlayers on Cu(111), Surface Science, 600 (2006) 873-903.

bond

bond order S17-C18 C12-S13 S15-S16 S13-S14 S14-S15 S16-S17

S17-C18 S13-S14 S14-S15 S16-S17

bond

bond order

1 INTRODUCTION

Massive hot forging of any aluminium alloy can be

performed with the help of several types of metal-

forming machines, namely hydraulic press,

mechanical press, screw press etc. Applying one of

these machines allows production of either typical

forgings or near net shape forgings. On the other

hand, the quality of a forging made of an Al-alloy

depends on the composition of a lubricant

significantly. The choice of the lubricant for hot

forging is major task, especially in case of

aluminium alloys deformation. Moreover, inaccurate

description of friction for numerical simulation of a

forging technology may be one of the reasons why

the results of simulation and experiments are not in a

good agreement.

The tribological properties of the lubricant can be

determined with the help of the ring-compression

test which is the most simple and widely used

method for the quantitative estimation of the contact

friction during bulk metal forging. It was developed

by Kunogi [1], Male & Cocroft [2] and further

modified by Burgdorf [3] etc.

This method allows to define a friction factor’s

value for the definite temperature only. But the

numerical analysis of forging technology requires

knowing the effect of temperature or/and strain rate

on the value friction factor.

In connection with it as well as hot forging of Al-

alloys there is a lack of systematized data

concerning the research on this effect for wide range

of temperatures of forging and strain rates. For

instance, in [4] it is only presented the results of

effect of temperature on friction factor during

deformation of Al-5Si and Al-4Mg alloys. These

data were obtained for temperature range of 20-

500°C but it is unknown what the value of strain rate

in those tests was. Then in [5] the properties of two

different lubricants were investigated during

deformation of Al-Mn, Al-Mg, Al-Cu-Mg and Al-

Cu-Mg-Fe-Ni alloys within temperature range of

200-470°C. In this case the hydraulic press was used

for deformation. It provided the initial strain rate of

0.14 s–1

. But the question is whether the effect of

temperature will be significant if the screw press is

chosen for deformation? It implies that the

temperature range and the materials under study

should be the same as in [5].

So, the aim of the present paper is linked to the

investigation of effect of temperature on friction

during the deformation of Al-alloys on the screw

press and farther generalization of the obtained data

which can be applied in industry in order to optimize

the new or/and being technological processes of

metal forming. In the present paper, the ring test and

a numerical simulation were applied to estimate the

tribological properties of lubricants.

ABSTRACT: The paper is linked to the investigation of lubricants for massive hot forging. It was observed

temperature range of 200-470°C and screw press was used for tests. The research on friction has been done

for Al-Mg and Al-Cu-Mg-Fe-Ni aluminium alloys. In this connection, physical and numerical investigation of

friction are performed. On the basis of the ring upsetting technique, the effect of temperature on the friction

factor value has been investigated. The regressions for the relationship between friction factor and

temperature for all lubricants under study have been obtained. Moreover, the sets of calibration curves were

drawn. Each set of calibration curves corresponds to the definite type of aluminium alloy as well as definite

conditions of deformation. Some practical recommendations were given.

Key words: interfacial friction, friction factor, aluminium alloy, non-ferrous alloy, isothermal hot forging, lubricant, massive hot forming

Research on Friction during Hot Deformation of Al-Alloys at High

Strain Rate

P. Petrov1, M. Petrov

1, E. Vasileva

1, A. Dubinchin

1

1Moscow State Technical University “MAMI” – B.Semenovskaya str. 38, 107023 Moscow Russia

URL: www.mami.ru e-mail: petrov_p@mail.ru; petrovpa@online.ru

2 EXPERIMENTAL PROCEDURE

Two lubricants were chosen for the investigation as

in [5]. These are the lubricant based on industrial oil

(IO+G) and based on synthetic oil (SO+G). Both

lubricants contain colloidal graphite particles as

lubricant’s components and the size of particles is

less than 15 µm. The chemical composition of alloys

under study is given in Table 1 [5]. The bold type in

Table 1 indicates the amount of the basic impurities,

which the investigated alloys contain.

The sizes of the ring samples were as follows: inner

diameter = 20 mm; outer diameter = 40 mm; height

= 14 mm. The ring samples were heated to

temperatures of 200, 350, 430, 450, and 470°C in an

electric furnace. Deformation of the heated samples

was carried out on flat dies that were warmed up to

temperature of 120°C. Samples were compressed

with lubrication. Die velocity was constant at V≈400

mm/s (screw press with nominal load = 10 MN and

maximum load = 16 MN), which corresponded to an

initial strain rate of 28.6 s–1

. This value of strain rate

belongs to the strain rate interval (100-10

2 s

-1) within

that the massive hot forging is usually carried out.

Table 1. Chemical composition of alloys [5]

Element Percentage, %

A95456 A92618

Al base base

Cu 0.04 2.12

Si 0.16 0.20

Mn 0.63 0.03

Mg 6.80 1.56

Ti 0.1 0.05

Zn 0.2 0.06

Fe 0.22 1.0

Ni - 0.80

Cr - -

The values of height hexp

and inner diameter dexp

were determined after compression of the ring

samples. The inner diameter was measured in three

locations along the height of the rings. Finally, the

value of the inner diameter was determined as

dexp

=(dtop+dmid+dbot)/3, where dtop, dmid and dbot are

respectively inner diameter at the top, middle and

bottom along the height of the ring, accordingly.

3 NUMERICAL SIMULATION OF RING

UPSETTING

A numerical simulation of the ring test was carried

out by means of the FE system QFORM-2D

(QuantorForm Ltd., Russia). The aims of the

simulation were to determine the values of the

friction factor based on the obtained experimental

data for the lubricants under study, and further to

construct calibration curves. Levanov’s friction

model [6, 7] was implemented by the QFORM

simulation software.

Levanov’s friction model can be considered as a

combination of Coulomb’s friction law and constant

friction law. It gives almost the same results as

Coulomb’s friction law for low value of contact

pressure σn. In case of high contact pressure,

Levanov’s friction model and constant friction law

allow us to obtain approximately the same values of

friction shear stress.

Fig. 1. Scheme of calculation

Figure 1 illustrates the scheme of numerical

calculation performed. Here dexp

and dfem

are the

inner diameter of the ring sample obtained

experimentally and by FEM, respectively.

The variable parameter in the simulation was the

friction factor. The simulation was carried out for

the same values of temperature as the experiments

had been done.

Stress–strain curves of alloys under study were taken

from the handbook [8]. These curves were obtained

under the conditions given in Table 2.

Table 2. Identification of flow stress-strain curves

Alloy type T, °C ε ε& , s-1

A92618 200-470 0-3 0-3

A95456 150-450 0-0.8 0.01-100

To perform the simulation, we assumed that the

contact friction was constant at the defined

temperature of deformation within the investigated

range.

4 RESULTS

4.1 Effect of temperature, strain rate and

lubricant’s composition

According to scheme (see Fig.1), one parameter

should be compared. That is the inner diameter of

the deformed ring specimen.

Comparing the volumes of a ring before and after

the FE simulation showed that the loss in volume

was approximately equal to 0.5%. As a result, it is

useless to control the outer diameter before and after

each simulation trial. That is why the inner diameter

is only controlled.

The determined values of the friction factor kn are

given in Table 3. Figures 2 and 3 show the

relationships between the friction factors and the

temperatures for the lubricants under study.

Moreover, in Table 3 the value of friction factor

taken from [5] are given. These data correspond to

the upsetting of rings made from the same Al-alloys

with the help of hydraulic press. The deformation

process was isothermal in those tests. It means that

the initial temperature of a specimen and tools was

equal to each other. The initial strain rate was

0.14 s-1

.

Table 3. Friction actors kn values

Friction factor kn

Strain rate, s-1

T, °°°°C

0.14 s–1

28.6 s–1

0.14 s–1

28.6 s–1

A95456 A95456 A92618 A92618

IO+G

200 0.171 0.41 0.25 0.21

350 0.153 0.287 0.236 0.202

430 0.155 0.245 0.146 0.179

450 0.118 0.216 0.15 0.17

470 0.11 0.21 0.138 0.143

SO+G

200 0.245 0.45 0.19 0.34

350 0.206 0.396 0.223 0.32

430 0.15 0.37 0.143 0.26

450 0.12 0.31 0.114 0.24

470 0.143 0.285 0.133 0.235

Figures 2 and 3 illustrate the effect of two

parameters on friction factor, i.e. temperature and

initial strain rate. It can be seen that the more

temperature of deformation, the less friction factor

value is. It is valid for both investigated alloys.

On the other hand, using two different presses in

terms of their design allowed to investigate the

effect of strain rate on friction factor value.

Unfortunately, we could not perform this

comparison within the whole range of investigated

temperatures. The reason for that is linked to the fact

that the conditions of deformation in each cases

differed from each other. In case of using of the

screw press for deformation the isothermal condition

was approximately observed at the temperature of

200°C.

Taking it into consideration it can be concluded that

at temperature of 200°C the effect of strain rate is

more significant for Al-Mg alloy. Moreover, it is

independent from the lubricant’s composition.

Increasing the strain rate from 0.14 to 28.6 s-1

gives

rise to increase in friction factor in over two times. It

is valid for both lubricants.

Fig. 2. Deformation of A92618 alloy: solid and dash lines –

regression; points - experiment

Fig. 3. Deformation of A95456 alloy: solid and dash lines –

regression; points - experiment

On the other hand, for Al-Cu-Mg-Fe-Ni alloy the

effect of strain rate on friction factor value is not so

obvious. For the lubricant based on synthetic oil

(SO+G) the influence of initial strain rate on kn is

almost the same. The increase in friction factor value

is over 1.5 times. For IO+G lubricant the effect of

strain rate is opposite to it. The difference between

the values of friction factor for both initial values of

strain rate is inessential.

4.2 Regression for friction factor vs temperature

The temperature dependence of the friction factor

can be described using the following equation: 2

o2o1on TATAAk ×+×+= , (1)

where Ao, A1 and A2 = coefficients, and Tо = the

temperature of the deformed material.

The values of the coefficients in equation (1) are

given in Table 4. Table 4. Coefficients of temperature dependence for kn

Strain rate, s-1

Coefficients

0.14 s–1

28.6 s–1

0.14 s–1

28.6 s–1

A95456 A95456 A92618 A92618

IO+G

Ao, °C 0.16 0.387 0.086 0.10

A1, 1/°C 0.00037 0.00037 0.0013 0.00086

A2, 1/(°C)2 -1.01×10

-6 -1.65×10

-6 -2.55×10

-6 -1.6×10

-6

SO+G

Ao, °C 0.155 0.32 0.145 0.224

A1, 1/°C 0.00088 0.00113 0.00085 0.00101

A2, 1/(°C)2 -2.05×10

-6 -2.53×10

-6 -2.03×10

-6 -2.13×10

-6

4.3 Calibration curves

The numerical simulation also allowed us to

construct calibration curves. The results of FEM are

shown in Figures 4 and 5, in terms of reduction in

internal diameter of ring specimens as functions of

the reduction in their height.

Fig. 4. Calibration curves - A92618 alloy

The obtained curves correspond to an initial

temperature of the deformed material of Tо = 450°C.

This temperature is the optimal temperature of

massive hot forging of both aluminum alloys under

study.

These plots allow to estimate the friction property of

a new lubricant which is using for deformation of

either A92618 alloy or A95456 alloy. Temperature

of material’s specimen/tools as well as initial strain

rate in the test should be the same as mentioned in

Section 2.

5 CONCLUSIONS

In summary, the following conclusions can be drawn

from the results presented here:

• The obtained temperature equations for the

friction factor can be used for FE simulation of

the massive hot forging of A95456 and A92618

alloys at strain rate of order 10-2

and 101 s

-1.

• The calculated calibration curves for Tо = 450°C

can be applied to estimate the friction factor of a

lubricant for use in the massive hot forging of

investigated alloys.

Fig. 5. Calibration curves - A95456 alloy

REFERENCES

1. M. Kunogi, On Plastic Deformation of Hollow Cylinders

Under Axial Compressive Loading. Rep. Sci. Res. Inst.

(Tokyo) 2 (1954) 63-92.

2. A.T. Male, M.G. Cocroft, Method for the Determination

of the Coefficient of Friction of Metals Under

Conditions of Bulk Plastic Deformation. J.Instit.Metals.

93 (1964-65) 38-46.

3. M. Burgdorf, Investigation of Friction Values for Metal

Forming Processes by the Ring Compression Method.

Industrie-Anzierger. 89 (1967) 799

4. K.P.Rao, K.Sivaram, A review of ring-compression

testing and applicability of the calibration curves,

J.Mat.Proc.Technol. 37 (1993) 295-318.

5. P.Petrov, Generalized approach to the choice of lubricant

for hot isothermal forging of aluminium alloys,

Computer Methods in Materials Science 7(2) (2007)

106-111.

6. A.N. Levanov, V.L. Kolmogorov, S.P. Burkin, B.R.

Kartak, U.V. Ashpur and U.I. Spasskiy, Contact Friction

in Metal Forging, Metallurgia, Moscow (1976).

7. P.Petrov, M.Petrov, Experimental and numerical

investigation of friction in hot isothermal deformation of

aluminium alloy AA3003, The 7th International

ESAFORM Conference on Material Forming, Norway,

Romania, Cluj-Napoca, 27-29 April, 2005, p.511-514.

8. P.G. Mikliev, Mechanical properties of Light Alloys in

Temperatures and Strain Rates of its Forging,

Metallurgia, Moscow (1976).

1 INTRODUCTION

The process of aluminium die casting presents a complex phenomenon of degradation, where erosion, soldering, corrosion, thermal fatigue usually occur jointly. All of these phenomena are major sources of limitation to the die castings service life. The application of hard PVD coatings may result efficient against these degradation mechanisms when applied on particular parts of the die: e.g. the inlet and the pins. The development of physical vapour deposition (PVD) has provided engineers and systems designers with the ability to tailor the surface properties of a range of mechanical devices to suit a growing range of applications. Advanced PVD coatings are designed to withstand severe mechanical and thermal stress conditions. Generally, the main requirements expected by advanced coatings are high hardness and compression strength, high wear resistance, high

mechanical and thermal fatigue resistance and low friction coefficient. These parameters may be attained by properly functionalising tools surfaces with innovative targeted thin films. Furthermore, the adhesion strength at the interface between coating and substrate results of major importance to guarantee a long endurance of the tool lifetime and surface properties. In the recent years, single layer, multilayer and gradient micro-structures have been developed and now, with testing of different coating chemistry and stoichiometry, representing the cutting edge solutions in coating technology. The present paper reports the results of a research on the effect of PVD films applied on a heat treated hot work tool steel substrate. The aim of this study was to determine the behaviour exhibited by each coatings in relation with the modulated chemical composition. To this purpose a set of coated specimens with chemical composition variations based on the CrAlSiN system were produced

ABSTRACT: Wear and failure of die casting dies involve a complex interaction between various mechanisms. The most important wear and failure modes are summarized as follows: (i) the so-called washout damages on working die surfaces are attributed to erosion, corrosion and soldering; (ii) thermal fatigue is the most important failure mode in die casting. The sector of surface thin PVD coatings is constantly enhancing in order to meet the increasing demand for improved performances of tooling. Advanced PVD coatings are designed to withstand severe mechanical and thermal stress conditions. A CrAlSiN nanostructured coating system was deposited on the base material, modulating the chemical composition so as to increase either the chromium or the aluminium-silicon content. Then, another set of specimens was subjected to a cyclic immersion program in a molten aluminium bath. The washout signs were detected and monitored in function of the increasing number of cycles.

Key words: PVD ceramic coatings, aluminium die casting, washout, soldering, modulated PVD composition

Performance enhancements of die casting tools trough PVD nanocoatings

M. Rosso1, D. Ugues1, E. Torres1, M. Perucca2, P. Kapranos3 1Politecnico di Torino – Cso Duca degli Abruzzi, 24, 10129 Turin (Italy) URL: www.polito.it e-mail: mario.rosso@polito.it; daniele.ugues@polito.it; eloy.torres@polito.it 2Clean NT Lab, Environment Park - Via Livorno, 58/60, 10144 Turin (Italy) URL: http://www.cleanntlab.com/ e-mail: massimo.perucca@envipark.com 3The University of Sheffield - Mappin Street, Sheffield, S1 3JD, United Kingdom URL: www.shef.ac.uk/materials e-mail: p.kapranos@shef.ac.uk

through an arc cathodic equipment. The so produced specimens were subjected to a program of cyclic immersions in molten aluminium alloy bath. The assessment of damages on the cycled specimens was performed through optical and SEM microscopy of both surfaces and transverse sections, so as to analyse the presence, if any, of soldering pits.

2 EXPERIMENTAL PART

2.1 Specimens fabrication

A parent block was cut from an annealed AISI H11 hot rolled bar, produced through vacuum melting and remelting processes. From this test coupon a set of cubic specimens were fabricated. These specimens were then vacuum heat treated according to the following parameters: austenitizing at 1,000°C – quenching with 5 bar nitrogen flow - first tempering at 550°C – second and third tempering at 595°C to about 45.5 HRC, which is rather typical for hot work tool steels. Ceramic coatings were deposited through the Physical Vapor Deposition (PVD) process provided by the PL-55 prototype unit installed at the Clean NT Lab. The unit is equipped with the innovative Lateral Arc Rotating Cathodes (LARC®) system. Two rotating cylindrical cathodes allow enlarged target surface and continuous surface refreshing during evaporation of the metallic constituents characterizing the ceramic coating. The CrAlSiN coating systems were deposited on the all specimens steel base material AISI H11 with different modulations of the chemical composition were developed so as to increase either the chromium or the aluminium-silicon content. The obtained coatings were roughly 3 μm thick. They were characterized by a multilayered structure covering about 50% of the entire coating depth, followed by a complementary massive monolayer of defined chemical composition. Six chemical composition modulations of the external layer were prepared according to the variations of the cathode arc current and of the nitrogen flow. The coating thickness was evaluated through the ball crater technique. A Rockwell indenter was used to study the adhesion properties of the deposited films by the indentation method.

Fracture cross-section of the coated AISI H11 samples was prepared to the scanning electron microscope (SEM) morphology investigation. X-ray diffraction (XRD) analysis was performed using a Siemens D5000 X-ray diffractometer. Monochromatic Cu-Kα radiation, produced at an acceleration voltage of 40 keV and 40 mA current, was used an initial incidence angle of 10°. Table 1. Deposition parameters used for the fabrication of the six different CrAlSiN coating systems

Cathodic arc current AlSi/Cr [A]

N2 [sccm] 110 / 70 100 / 90 80 / 100 60 / 110

120 A1 D1

150 B2 C2

190 A3 D3

The coatings hardness was determined using a Mitutoyo MVK-G1 microharness tester at 100, 50, 25, 10 g loads. Later the hardness of the different coatings was extrapolated from the evaluation of microhardness at increasing weights and applying a modified form of the work-of-indentation model.

( )XSf

SC

ββHHHH

011

+=

−− (1)

Where Hc is the composite hardness, that is the hardness due to the effects of the film and the substrate; Hs is the substrate hardness; Hf is the film hardness; β is the relative indentation depth (RID), defined as the ratio of the maximum indenter penetration depth to the coating thickness; β0 and X are systems parameters. Tribological properties of the coatings deposited on AISI H11 substrates were evaluated by means of low frequency (5 Hz) reciprocating sliding wear test (at 5 N normal load for 500 m sliding distance) with a 10 mm diameter SAE 52100 hardened steel ball, at a temperature 25 ± 2 °C and 50 ± 5 % relative humidity and amplitude 10mm (according to the ASTM G133 standard). The wear scar was observed through electronic microscopy (SEM) and EDS spot analysis was used to confirm whether the coating had failed or if there was simply transfer as counterface material from the steel ball. The specimens were then subjected to cyclic immersions in a molten aluminium alloy so as to simulate the environmental conditions that occur at the surface of a high pressure die casting. The

aluminium alloy was AlSi8Cu3Fe, commonly used for die casting. The cooling bath is constituted of a diluted solution (1:80) of a commercial die lubricant. The duration of a typical cycle is of 30s with an actual immersion time of 4s for both the aluminium alloy and cooling baths. All the specimens tested were periodically analyzed to detect the formation of surface defects by SEM inspection.

3 RESULTS AND DISCUSSION

The reflected light micrograph of the ball crater (figure 1) clearly shows the multilayer structure of the coatings and led to a thickness measurement of 2.875±0.074 μm, and clearly shows the multilayered structure of the coatings.

Fig. 1. Reflected light micrograph of a ball crater on A1

The appearance of Rockwell C indentation on representative coated systems is reported in figure 2. The adhesion was evaluated on the base of an observation of the aspects of the cracks and of the presence of coating detachments around the indentation. The adhesion evaluated through this test can be considered good since no detachments of the coating along the edge of the indentation could be revealed. The hardness evaluations for the different coatings as a function of increasing applied weights and, as a consequence, of increasing indentation depth. The A coatings presented the highest film hardness (46 GPa), the D coatings the lowest (38 GPa), whereas the B and C coatings exhibited an intermediate hardness value (43 GPa).

Fig. 2. Adhesion Test: appearance of Rockwell indentations on

specimen A3

The coatings deposited tested in reciprocating sliding in term of friction coefficient exhibited a similar trend, where the friction coefficient increased gradually from 0.55 to 0.65. SEM micrographs of the wear scars for coatings A1, A3, B2 and C2 (figure 3) shows ploughings marks along the direction of sliding, instead for coatings D1 and D3 (figure 4) demonstrate ploughings marks along the direction of sliding plus signs of delamination areas.

Fig. 3. SEM micrographic wear scars specimen A1

Fig. 4. SEM micrographic wear scars specimen D1

To verify the absence of coating in the critical zones, a deepening was performed using an EDS (Energy Dispersive X-ray Spectrometer) probe (figure 4). By this method, the coating spoliation could be confirmed since contents were detected in the critical points. figure 5 and figure 6 shows one of the critical points and details of the EDS spectrum in the two significant zones.

Fig. 3. Details of the SEM micrographic specimen D1

Fig. 4. EDS spectrum graphs by 2 different zones

As for the washout resistance, a first ranking of the coated specimens performances can be drawn just after the first step of immersion in molten aluminium. Actually, specimens D1 and D3 (those with the lowest AlSi/Cr ratio) presented the formation of a thick soldered layer just after 5000 cycles. On the contrary the observation of the other specimens did not revealed at all the presence of soldering points (specimens A1 and B2) or revealed very few soldering points (specimens A3 and C2)

after the first 5000 cycles. The cyclic immersion test in the molten aluminium alloy clearly demonstrated that the ceramic coatings deposited on the steel surface may play a positive role in terms of the reduction of the soldering effect. In the best performing systems, A1 and B2, the soldering didn’t appear up to ca. 7500 cycles

4 CONCLUSIONS

On the base of the complete coating quality assessments (thickness, adhesion, hardness, tribological properties), the A coatings were (systems with high content aluminum-silicon) considered the most promising candidate for the final applications on protection of aluminium die casting tools. The coatings where a high aluminium-silicon composition with a concurrent medium to low nitrogen content was realized resulted to be the most effective against soldering. On the contrary, the enrichment in chromium of coating composition was found to be not so effective in preventing aluminium soldering.

REFERENCES

1. Chowdhury, A., Cameron, D., Hashmi, M. “Adhesion of carbon nitride thin films on tool steel” Surface And Coatings Technology, 116-119, (1999).p. 46

2. , D., Holler, F., Mitterer, C. “Hard coatings produced by PACVD applied to aluminium die casting” Surface And Coatings Technology, 116-119, (1999) p.530.

3. Holler, F., Ustel, F., Mitterer, C., Heim, D., Proc. “Thermal cycling and oxidation behaviour of hard coatings in aluminium die casting” 5th Int. Conf. on Tooling, Leoben, Inst. Für Metallkunde und Werkstoffprüfung, Leoben, Austria, (1999) p.357.

4. Joshi, V., Srivastava, A., R. Shivpuri, R. “Investigating tribochemical behavior of nitrided die casting surface” Proc. 6th Int. Conf. on Tooling, Karlstad, Karlstad University, Karlstad, Sweden, (2002) p.809.

5. Mitterer, C., Holler, F., Ustel, F., Heim, D. “Application of hard coatings in aluminium die casting — soldering, erosion and thermal fatigue behaviour” Surface And Coatings Technology 125, (2000) p.233.

6. J.R. Tuck, A.M. Korsunsky, D.G. Bhatc, S.J. Bull, “Indentation hardness evaluation of cathodic arc deposited thin hard coatings” Surface and Coatings Technology 139 (2001).p. 63-74.

7. Ugues, D., Rosso, M., Albertinazzi, M., Raimondi, F., Silipigni, A. (2004a) “Heat treatments and innovative surface treatments for high performing dies” Proc. 2nd Int. Conf. High Tech Die Casting, Brescia, Italy, p. 155.

1 INTRODUCTION

The upsetting of billets with conical punches of known surface roughness and appropriate angles is proposed as a new friction test for the parameter identification of workpiece-tool interface behaviour during bulk forming processes. The experimental and numerical analysis of this forming operation showed a plastic flow of the material layer beneath the workpiece surface that led to an “unpredictable” deformed shape of the specimen with the classical friction laws. So, this simple process is simulated with the finite element method and very fine meshes in order to predict numerically the deformed shape of the cylinders near the workpiece-tool interface with the actual geometry of the tool roughness. In order to model the behaviour of the actual interface with a macroscopic law, an improvement of the plastic wave friction model is developed with tilted boundary conditions linked to the displacements of the nodes sliding on the surface tool.

2 THE PLASTIC WAVE FRICTION MODEL

2.1 Summary of the theory

For rough contact, Challen and Oxley [1] had

proposed a friction model using the plastic deformation of the material workpiece in local asperities of the rigid tool surface. The model is based upon an ideal 2D geometry of the tool surface. For the sake of simplifying the theory, the actual roughness is supposed to be equivalent to triangular waves, height and wavelengths of which equal the mean height R and the mean wavelength AR of the actual roughness profile respectively, see figure 1.

Fig. 1. Triangular asperity

These profile parameters are defined by the standard ISO 4287-1997 and can be measured with profilometers. With this simple geometry, the flow of the workpiece material under the tool asperity is analysed with the slip line theory. Assuming a perfectly plastic material, the value of the friction stress τf is given by the balance of the external tractions, see equation (1). τf is a function of two characteristic angles Φ and η of the plastic wave, σy the yield stress of the billets material, α the asperity

ABSTRACT: In the case of a rough contact without friction between axisymmetric billets and punches, experimental evidences as well as very fine numerical modelling show that the workpiece material flows as a lubricant layer in the vicinity of the contact surface. As no existing friction laws used for bulk forming processes can take into account this typical surface behaviour, an improvement of the so-called “plastic wave” friction law is proposed in order to describe more accurately the plastic flow in the workpiece-tool interface. Key words: Bulk Forming, Rough Friction, Plastic Wave Model

An improved « plastic wave » friction model for rough contact in axisymmetric modeling of bulk forming processes

E. Vidal-Sallé1, S. Boutabba2, Y. Cui1, J.C. Boyer 1

1Département de Génie Mécanique Conception, Institut National des Sciences Appliquées, 20 Av A. Einstein, 69621 Villeurbanne Cedex, France URL: www.insa-lyon.fr e-mail: Emmanuelle.Vidal-Salle@insa-lyon.fr; Jean-Claude.Boyer@insa-lyon.fr 2Département de Génie Mécanique, Université de Guelma, Guelma, Algérie e-mail: boutabba_s_lpg@yahoo.fr;

Roughness Profile

R/2

-R/2 Height =R

Ra=R/4 α

Wawelength = AR

angle, and m0 the constant friction coefficient between the tool asperity and the billet surface :

( )

( )

yf

110

0

1 2 242 3

m

2 1 m

sin coscos

cos sinsin

−−

σ π τ = + + Φ − η α + α + Φ α

α α + Φ = η = −

(1)

With this theoretical approach, the proposed friction law for rough contact looks like a mix of the Coulomb and Tresca laws, see figure 2 where the abscissa axis is the ratio of the normal stress to the tensile yield stress of the workpiece material, and the ordinate axis, the ratio of the friction stress to the shear yield stress:

0.05

0.1

0.15

0.2

0.25

0.3

0 1 2 3 4 5 6 7 8 Normalised normal stress

Angle = 0.5° Angle = 1°

Angle = 2°

Angle = 3°

Angle = 4°

Angle = 5°

Normalised friction stress

Fig. 2. Plastic wave friction law

For normal stress lower than 1.5 the tensile yield stress of the workpiece material, the friction stress obeys to the Coulomb law with a friction coefficient sensitive to the angle α of the tool asperity. For high normal stress, the friction stress is quite constant as proposed by the Tresca model with a constant coefficient also sensitive to the asperity angle α. In order to use this interface constitutive law, three parameters have to be identified: the yield stress of the workpiece material in the vicinity of the contact area, the tool asperity angle α, and the constant friction coefficient m0 between the tool asperity and the plastic wave surface. The identification of the first parameter needs torsion tests of the material workpiece as the plastic strain could reach values in the range 100% to 500%. The asperity angle α can be deduced from the roughness profile if the actual mesoscopic geometry of the tool surface is not too far from the assumed triangular profile used for the slip line theory discussion. The main unknown parameter is the constant coefficient m0 which, at present, is still “measured” with the so-called ring-compression test proposed first by Kunogi [2] for

comparison of cold forging lubricants for rolling where two opposite plastic flows are separated by a neutral zone under high contact pressure. Other authors like Petersen et al. [3], proposed alternative ring geometries in order to modify this test for the low contact pressure range. Anyway, the sliding velocity distribution and the contact pressure distribution at the tool-ring interface are non-uniform, see Vidal-Sallé et al. [4], and the influence of strain hardening is not negligible, so a simple test has to be designed for a consistent identification of the constant friction parameter m0. As a new attempt, a compression test with conical punch geometries and low constant friction parameter m0 is proposed in order to check the predictions of the plastic wave model on simple axisymmetric cases.

2.2 Actual and theoretical roughness

For axisymmetric geometry, the punches are machined by turning and the plastic flow of the workpiece is purely radial so the 2D model of the plastic wave model is pertinent. Nowadays, machining is using inserts of known circular geometry. The turned punch surfaces were analysed with a profilometer and the roughness profiles of the surfaces are similar to the record of figure 3:

Fig. 3. Actual profile of a turned surface

For this conical punch, the asperity wave length of the circular arc is close to 360 µm and its asperity height close to 38 µm but one issue to address is how to transform this circular arc geometry in a triangular profile in order to evaluate the asperity angle α. The simplest solution is a triangular asperity with the same height and wave length but the volume of the plastic wave trapped by the tool roughness will be slightly underestimated. If the mass balance must be fulfilled, the triangular asperity must have an area equal to the circular arc. This second solution leads to a height of the triangular asperity greater than the actual one if the wave length is held constant and so to a sharper asperity angle. Numerical tests were conducted without local friction for five different roughness profiles in the range of the machined punches in order to compare displacements, stress and plastic strain under a conical punch modelled with its actual

circular arcs on the first hand, triangular asperities with equivalent height and triangular asperities with equivalent area on the second hand. From these comparisons, the equivalent height triangle asperity has been found to be the better solution with distributions of displacements and of von Mises stress very close to the distributions with the actual circular arc profile. So, for further numerical and theoretical investigations, the equivalent height triangle asperity will be used instead of the actual circular arc.

3 EXPERIMENTAL DATA

3.1 Billets

The billets were cylindrical with an initial diameter of 60 mm and an initial length of 80 mm. The material used was an aluminum alloy with a low yield stress identified with a tensile test. The stress-strain curve was measured only up to 10% and then extrapolated with a logarithmic function for larger plastic strain, see figure 4:

Fig. 4. Stress-strain curve of the billet material

3.2 Upsetting with conical punches

The classical friction test for bulk forming processes are mainly based on compression that induces heterogeneous strain hardening in the billets with rough punches and friction. In order to reduce the heterogeneity of the yield stress in the billets, the upsetting with conical punch is tested with particular values of the cone angles set equal to the asperity angle α of the triangular roughness and quasi perfect sliding. The billets are cylindrical but their ends are machined as concave cones for matching with the upper and lower punches. Different upsetting tests were conducted for roughness ranging from 3° to

12° with different heights and wave length asperities, see table 1. Experiments were conducted only with rough dies.

Fig. 5. Experimental set up

Table1. Roughness parameters of the punches

3.3 Experimental results

Two different lubrication conditions were applied for the upsetting with conical punches, the first one was dry friction and the second one “perfect sliding” with a bisulphide-molybdenum grease. The deformed specimens for three different roughnesses are presented on figure 6. Dry friction induced bulging while perfect sliding led to “diabolo” like shape. This macroscopic difference can be used to identify friction parameter but it is not worthwhile to notice that the displacements of the equator of the billets depend also on other parameters than friction.

Fig. 6. Deformed specimens with and without friction

The variations of the upper and lower diameters of the deformed billets are more representative of the workpiece-tool interface behaviour and the displacements of the upper and lower billet edges will be the main variables used for the identification of the constant friction parameter m0 of the plastic wave friction model.

Cone Angle R(µm) AR(µm) θ(°) 3° 2.913 120.681 2.76 5° 13.842 238.970 6.61 12° 38.310 358.978 12.04 3° 9.164 240.131 4.36 7° 20.985 359.060 6.67

4 NUMERICAL MODELLING

4.1 Upsetting with conical punches

The problem symmetries allow considering only one half of one punch and one quarter of the billet.

Fig. 7. Plastic strain distribution in the billet for perfect sliding

The punch material is considered as elastic and the billet material elastic-plastic with the strain hardening defined by the constitutive law given on figure 5. In the area of contact with the punch, the mesh of the billet is very fine with twelve elements in front of each tool asperity of a 7° cone angle. The explicit algorithm of the Abaqus software with an appropriate mass scaling is used for solving this huge number of equations. In spite of this choice, only small billets with 6 mm diameter and 8 mm length were analysed in a reasonable time with a dual core PC. Nevertheless, the results were qualitatively attractive as a mesoscopic effect of the roughness was observed. Some kind of Venturi effect is observed in a 0.2 mm thick layer of the billet material with very large plastic strains (greater than 200%). This plastic flow feeds a flash at the edge of the billet, see figure 7. Accurate observations of the deformed specimens presented on figure 6 confirm this fact with a 0.7 mm large flange on both edges of the billets.

4.2 Improved plastic wave model

The plastic flow in the vicinity of the punch surface was found to be sensitive to the local slope including the roughness profile of the tool. As it is not realistic to define the tooling geometry of any bulk forming process with any roughness profile, a first improvement would be the definition of tilted boundary conditions coupled with the displacements of nodes sliding on the surface tool. This solution has been implemented with the plastic wave friction

model and flat punches. In figure 8 both billets are given with the same scale. Discrepancies found can be explained by the differences on the geometric description. With the rough tool, the mesh is fine enough to take into account all the asperities of the interface. It is not the case for the left map where the meshing is coarse.

Fig. 8. Upsetting with flat punches - Plastic strain distribution

The prediction of the displacements of the upper edge with the improved plastic wave model is very close to the values of the punch defined with its actual roughness, but the deformed shape is not in good agreement as the work hardening is found very different in the two billets.

5 CONCLUSIONS

The numerical analysis of an axisymmetric rough contact with the finite element method and an explicit integration scheme coupled to a fine mesh taking into account the roughness geometry of the tools gives results similar to the experiment. The actual plastic flow of the billet material in the close vicinity of the tool has to be introduced in the friction laws. Further modifications of the interface constitutive equations are necessary for improving the numerical predictions of the billet-tool contact.

ACKNOWLEDGEMENTS

The experimental part of this work has been completed under the supervision of Pr. A. Torrance at the Department of Mechanical Engineering and Manufacturing of Trinity College Dublin with an Erasmus Mundus Masters grant from the E.C. REFERENCES

1. J.M. Challen, and P.L.B. Oxley, An explanation of the different regimes of friction and wear using asperity deformation models, Wear 53 ,(1979) 229-243

2. M. Kunogi, J. Sci. Res. Inst. .2 ,(1954), 63 3. S.B. Petersen, P.A.F. Martins and N. Bay, Friction in

bulk metal forming, J. of Mat. Proc. Tech. 66, (1997), 186-194

4. E. Vidal-Sallé, L. Baillet , J.-C Boyer., Friction law and parameter identification, 2nd ESAFORM Conference on Material Forming, Guimarães (1999), 603-606

1 INTRODUCTION

The cold forging processes, including extrusion,

drawing, upsetting, and heading, have been widely

used to realize the net-shape manufacture in the

transportation industry. The friction has a large

effect on the cold forging, because it not only

changes the forming force but also determines the

surface quality of the formed part and dies life. The

problem is especially critical in cold forging of steel

which induces high contact pressure and new surface

generation. So the billets are generally coated with

zinc phosphate and soap and lubricated with oil to

reduce the friction during forging [1].

For reduction of the cost and time of manufacture,

almost all of manufacturers and research institutes

use FE simulation to predict defects in the forged

part and optimize the die shape. However, it is still

difficult to choose the friction model and assign the

friction coefficient. Therefore, several methods were

invented to evaluate the friction condition during

cold forging, such as ring compression, forward

extrusion, upsetting-sliding test, spike test, double

cup extrusion and T-shape compression. Studies

illustrated that the contact surfaces in ring

compression and spike test are shorn surfaces, which

haven’t been coated with phosphate/soap layer, so

friction coefficient can not be adequately evaluated

by using them. Forward extrusion and double cup

extrusion [2] can induce the large contact pressure

and new surface generation. Meanwhile, the load of

forward extrusion and cup height ratio of double cup

extrusion are sensitive to friction. In upsetting-

sliding test [3] friction coefficient is calculated

starting from the tangential and normal forces of

ABSTRACT: The measurement of friction in the industrial metal working operations is a complex problem because the friction test must impose at the tool/metal interface conditions similar to those in the industrial operations. So in the European Network VIF (Virtual Intelligent Forging), a workshop (WP3) was held to evaluate friction condition in cold forging by using numerical simulation and experimental methods. It was chosen an industrial cold extrusion operation, in which a low carbon steel bar, covered with phosphate layer and soap, was drawn, cropped and then formed by extrusion. In order to know the friction condition in this industrial extrusion operation, four kinds of friction tests, forward extrusion, double-cup extrusion, upsetting-sliding test and T-shape compression were carried out in four labs, IPU, DIMEG, LAMIH and CEMEF, respectively. In a first preliminary step we simulate the drawing and extrusion operations in order to estimate the contact conditions in extrusion along the container and the die surfaces. Then by numerical simulation we estimate the contact conditions in the friction tests and define the parameters of the tests which insure the better similarity with the industrial operation. In the next step experiments will be performed in order to compare the results of these various friction tests.

Key words: Friction test, Extrusion, Cold Forging, Low Carbon Steel.

Measurement of friction in a cold extrusion operation: Study by

numerical simulation of four friction tests

Q. Zhang1, M. Arentoft

2, S. Bruschi

3, L. Dubar

4, E. Felder

1

1CEMEF, UMR CNRS/Ecole des Mines de Paris 7635, B.P. 207, F 06904 Sophia Antipolis Cedex, France URL: http://www.cemef.cma.fr/ e-mail: qi.zhang@ensmp.fr; Eric.Felder@ensmp.fr 2 IPU, Technical University of Denmark, Produktionstorvet, Bygning 425 DK-2800 Kgs. Lyngby, Denmark URL: http://www.ipu.dk e-mail: ma@ipu.dk 3 DIMEG, University of Padova, Via Venezia, 1-35131 Padova, Italy URL: http://www.dimeg.unipd.it/ e-mail: bruschis@ing.unitn.it 4LAMIH, Université de Valenciennes Le Mont Houy F59313 Valenciennes Cedex 9, France URL: http://www.univ-valenciennes.fr/LAMIH/ e-mail: laurent.dubar@univ-valenciennes.fr

indenter. T-shape compression, a new friction test,

was developed to evaluate the friction condition by

using the formed part shape and compression load.

So four suitable methods, forward extrusion, double-

cup extrusion, upsetting-sliding test and T-shape

compression (see Fig. 1), can be chosen to determine

friction coefficient in cold forging. However, each of

them has special features and they haven’t been

compared. In this paper, we investigated their ability

by FE simulation and evaluated the friction

condition of an industrial cold extrusion process.

(a) Forward extrusion (b) Double-cup extrusion

(c) Upsetting-sliding test (d) T-shape compression

Fig. 1 Friction tests for cold forging

2 MATERIAL PROPERTY AND FRICTION

LAW USED IN SIMULATION

The material of specimen used in the simulation is

low carbon AISI 1010 steel in drawn state. The

relationship between strain and stress has been

obtained by uniaxial compression test (Fig.1). For

accurate simulation, several points on the

strain/stress curve were chosen to input into the FE

code, and we assume that the stress is constant, for

strain higher than 0.7.

There are two important laws, Coulomb and

(Tresca) shear stress friction law, which are applied

widely to calculate the tangential stress between

specimen and tool in metal forming processes.

Coulomb friction law was used in this work. It is

because Coulomb friction law can be used in the

high pressure condition, if the friction coefficient is

small.

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.80

100

200

300

400

500

600

700

Tru

e st

ress

(M

Pa)

True strain

Fig. 1 The strain/stress curve of AISI 1010 by uniaxial

compression test (drawn bar with 7.15 mm diameter)

3 INDUSTRY COLD EXTRUSION PROCESS

The whole manufacturing system includes drawing,

cropping and extrusion processes. Cropping process

can induce non-uniform shorn surface of billet and

small deformation of billet, but it is not an important

factor on friction. Therefore only drawing and

extrusion processes were considered. That is AISI

1010 steel bar with 7.5 mm diameter was drawn to

7.15 mm diameter and then extruded to 4.43 mm

(extrusion ratio λ =2.6) in a die with a complex

shape. The axisymmetrical drawing and extrusion

process were simulated by using FORGE2005®.

According to the simulation results, the maximum

contact pressure on the extrusion die can reach 2500

MPa, whereas it is near 850 MPa along the container.

The equivalent strain in the extruded workpiece is 2.

4 FRICTION TESTS FOR COLD FORGING

4.1 Forward extrusion

The operation is performed by a conical die with

29.6 deg semi-angle and the extrusion ratio λ

equals to 1.96. The loading curve of forward

extrusion is sensitive to the friction coefficient along

the die and along the container. So the force can

decrease with decreasing the length of workpiece in

the container, whereas the die region has been filled

with the metal. However, usually the friction

coefficients on container and die are considered as

equal despite the fact that some factors along

container and die are different such as material,

roughness, pressure and lubricant film thickness. For

deeply investigating the friction on the extrusion,

two different friction coefficients are assigned to

contact pairs of die/workpiece and container/

workpiece (See Fig 1 (a)),

Fig. 3 shows the effect of friction coefficients of die

and container on the extrusion loading curves. It is

seen that the extrusion load changes by assigning

different friction coefficient on die and container in

simulation. After the extrusion experiment, the

friction coefficient of container can be defined by

the slope of the loading curve and that of die can be

defined by the maximum load.

0 2 4 6 8 10 12 14 16 18 200

5

10

15

20

25

30

350.07 0.05Die Containerµ µ= =

0.02 0.05Die Containerµ µ= =0.02 0Die Containerµ µ= =

Load

(K

N)

Stroke (mm)

Fig. 3 Effect of friction coefficients of die and container on the

extrusion loading curve

4.2 Double-cup extrusion

Double cup extrusion test includes four parts: upper

punch, lower punch, container and specimen, as

shown in Fig. 1 (b). In the test, the upper punch

moves down with the press ram while the lower

punch and container are stationary [2]. After

extrusion, a part with two cups was formed. The

height of upper cup H1 is larger than that of lower

cup H2, because the container has relative velocity

with respect to the upper punch, then the friction

force from container promotes the metal flow into

the upper cup. So the ratio of the upper cup height to

the low cup height (H1/H2) was defined as cup

height ratio, which can be used to determine the

friction coefficient.

For the studied configuration, the ratio of the

container diameter to the punch diameter is 1.66. As

in [2], we verified that the friction at the punch/billet

interface has no significant influence on H1/H2.

Two kinds of material were chosen to simulate

effect of strain hardening on cup height ratio, one is

AISI 1010 (see Fig.2) and another is AISI 1010

without strain hardening. Simulation results

illustrate the material strain hardening reduces

significantly the cup height ratio (see Fig.4). It is

because the difference in strain between both metal

flows increases with the cup height ratio and so the

strain hardening promotes its reduction. Therefore,

in order to determine the friction coefficient by

double cup extrusion, the material property,

especially the strain/stress relationship, should be

well-known and carefully assigned to FE model.

0 1 2 3 4 5 60

1

2

3

4

5

6

With strain hardening

H1/

H2

Stroke (mm)

Without strain hardening

Fig.4 Effect of strain hardening on cup height ratio ( 0.05µ = )

4.3 Upsetting-sliding test

The basic methodology of upsetting-sliding test is

similar to the sliding indentation test. However it is

special to simulate the real contact conditions of the

drawing and extrusion processes [3]. As shown in

Fig.1 (c), this test includes two main components:

indenter and specimen. A cylindrical indenter plays

a role of drawing or extrusion die, so the indenter is

made with the same material and surface roughness

than the die. The specimen with coating was fixed in

the special device. Before test, the position of

indenter is adjusted to induce the penetration

depth p of the indenter in the specimen. Then the

indenter moves up with a constant tangential speed

and scratches the surface of specimen. The forces in

both normal ( nF ) and tangential (Fτ ) directions are

measured with load sensors and recorded in a

computer to evaluate the friction condition between

indenter and specimen [3].

In simulation, the radius of indenter was 5 mm.

Numerical simulation demonstrates that for a given

penetration depth p , the force ratio / nF Fτ increases

with the friction coefficient µ , but the contact

pressure and the equivalent strain do not change

significantly. Fig. 5 shows that the force ratio and

the equivalent strain increase with increasing the

penetration depth. The penetration depth of indenter

in experiment can be decided depending on the

equivalent strain of extrusion parts. Because

equivalent strain in industrial extrusion part is 2, the

penetration depth of indenter should be chosen as

p =0.5 mm (see (Fig. 5).

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.80.05

0.10

0.15

0.20

0.25

0.30

0.35

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

/ nF Fτ

Equ

ival

ent s

trai

n

Tan

gent

ial f

orce

/ N

orm

al fo

rce

Penetration depth (mm)

Equivalent strain

Fig. 5 Effects of the penetration depth of indenter on the force

ratio and the equivalent strain (µ=0.02, R=5mm)

4.4 T-shape compression

T-shape compression, a new method to determine

friction coefficient in cold forging, is developed. The

die in this test, is machined as a flat-topped die with

a V-shape groove (total angle 20 deg, 7 mm depth

and 1 mm entry radius) (see Fig. 1 (d)). With the flat

upper punch moving down some metal of specimen

will be extruded into the groove, others will be

compressed between punch and die. The sectional

shape of formed part in this test looks like ‘T’,

suggesting us the name: T-shape compression test.

The friction force generated between interface of

billet and V-shape wall can directly change load and

the length H of the extrude part.

The stroke and load curves with different friction

coefficient are shown in Fig. 6 (a). Results illustrate

the load increases when increasing friction

coefficient and the large stroke can improve the

sensitivity of load to friction coefficient.

0 1 2 3 4 50

20

40

60

80 0.10µ =

0.15µ =

0.05µ =

Load

(K

N)

Stroke (mm)

0 20 40 60 800.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0.15µ =

0.10µ= 0.05µ=

Hei

ght o

f ext

rude

d pa

rt (

mm

)

Load (KN)

(a) Load (b) Height of extrusion part

Fig. 6 Effects of friction coefficient on load and height of

extrusion part

Fig. 6 (b) shows effect of friction coefficient on the

height of extruded part. It is seen that the height of

extruded part increases with decrease in the friction

coefficient. Therefore, we can determine the friction

condition in the T-shape compression by using

compression load the deformed part shape.

5 DISCUSSION AND CONCLUSION

According to simulation of four kinds of friction

tests, testing conditions of them are shown in table 1.

Table 1 Testing conditions for cold forging

Types of

testing Maximum

Pressure

(MPa)

Maximum

equivalent

strain

New surface

generation

Forward

extrusion 1800 (die)

≈500 (cont.) 1.6 (die)

0 (cont.)

40 % (die)

0 (cont.)

Double-cup

extrusion 800 (cont.) 1.9 (cont.) 44 % (cont.)

Upsetting-

sliding test 1900 2.6 6.1 %

T-shape

compression 1900 3 50 %

*Industry

extrusion

2500 (die)

850 (cont.)

2 (die)

0.15 (cont.)

61 %

0 (cont.)

Concerning the main features of these tests, the

friction conditions of die and container in forward

extrusion may be different, because of different

contact conditions. We can expect higher friction on

the die in industry extrusion because contact

pressure, equivalent strain and surface generation

obtain highest values in die region. For double-cup

extrusion material strain hardening has a large effect

on the cup height ratio and determination of friction

coefficient. In the upsetting-sliding test, the

penetration depth of indenter can be defined by the

equivalent strain of extruded part. In the T-shape

compression, the friction coefficient can be

determined by both compression load and shape of

formed part. But according to table 1 we can find it

is not possible to insure similar contact conditions

for each test and industry extrusion. So experiments

will demonstrate which contact parameters have

significant influence on friction.

ACKNOWLEDGEMENTS

The authors want to thank the VIF (Virtual

Intelligent Forging - CA) project for financing the

research of friction condition in forging.

REFERENCES

1. N. Bay, The state of the art in cold forging lubrication. J.

Mater. Process. Technol, 46 (1994) 19-40

2. T. Schrader, M. Shirgaokar, T. Altan. A critical

evaluation of the double cup extrusion test for selection

of cold forging lubricants, J. Mater. Process. Technol,

189 (2007) 36-44

3. L. Lazzarotto, L. Dubar, A. Dubois, et. al A selection

methodology for lubricating oils in cold metal forming

processes, Wear, 215 (1998) 1-9