TECHNICAL PAPER

Indigenous Development and Industrial Application of MetalCasting Simulation Software

Bhallamudi Ravi1 • Savithri Sivaraman2 • Roschen Sasikumar2 •

Arjun Madhur Marwah3

Received: 29 June 2015 / Accepted: 8 September 2015 / Published online: 6 October 2015

� The Indian Institute of Metals - IIM 2015

Abstract Metal casting is a multi-physics process

involving many physical phenomena like fluid flow, phase

transformation, heat transfer, microstructure evolution,

defect formation and thermal stresses. Virtual casting based

on process modeling and computer simulation of the above

phenomena enables foundry engineers to reduce physical

trials, and optimize various process parameters to achieve

the desired quality and yield. A multi-disciplinary team of

researchers from IIT Bombay and CSIR-NIIST combined

their work in casting design and simulation, respectively,

and teamed up with 3D Foundry Tech to create an inte-

grated software package called AutoCAST FLOW?.

Foundry engineers are able to visualize mold filling, tem-

perature profiles during casting solidification and cooling

curves; predict related defects like air blow hole, cold shut,

shrinkage porosity and hard zones; and optimize the

feeding and gating design. Two industrial case studies

demonstrating its successful application for quality

improvement are presented. The software has been

implemented in several engineering institutes, raising the

awareness and interest among students as well as teachers

in metal casting field. These efforts are expected to lead to

indigenous capability in casting simulation technology

suitable for the local requirements.

Keywords Metal casting � Defects analysis �Methods design � Computer simulation

1 Introduction

Computer simulation of metal casting process, including

the flow, solidification and subsequent cooling of molten

metal in a mold cavity, provides the insight to analyze

internal defects, as well as the foresight to prevent them in

future. Since metal casting is a multi-physics process with

changing values of thermo-physical properties and

boundary conditions (such as interfacial heat transfer

coefficients), accurate solution of the relevant governing

equations is very difficult. Moreover, the intricate geome-

try of typical castings, and the presence of other bodies like

cores, insulating sleeves, chills, vents and mold coating,

makes it necessary to create a large number and variety of

mesh elements, increasing the computation time to several

hours for each iteration. In general, several iterations of

casting design (orientation, feeders, gating), pre-processing

(meshing, boundary conditions), solution of the governing

equations using an appropriate scheme, and post-process-

ing (interpreting the results) are required to analyze and

solve casting defect issues while achieving the maximum

possible yield. Several researchers have successfully

demonstrated the application of casting simulation for

defect prediction (Table 1), yet foundry engineers in India

are unable to use such high-end tools. In addition, the high

cost of simulation per casting project makes them

& Bhallamudi Ravi

b.ravi@iitb.ac.in

Savithri Sivaraman

sivakumarsavi@gmail.com

Roschen Sasikumar

roschen.csir@gmail.com

Arjun Madhur Marwah

arjunmadhurm2401@gmail.com

1 Mechanical Engineering, IIT Bombay, Mumbai, India

2 Computational Modelling Group, CSIR-NIIST, Trivandrum,

India

3 3D Foundry Tech Pvt. Ltd., Mumbai, India

123

Trans Indian Inst Met (2015) 68(6):1227–1233

DOI 10.1007/s12666-015-0710-x

unaffordable to small and medium foundries, which form

80 % of the Indian foundry industry. Thus a strong need

was felt to develop an easy to use integrated software

package for casting design and simulation that is accessible

and affordable for SMEs.

2 Casting Process Modelling

The detailed modelling was presented by Savithri, et al.

[19] in an earlier publication, and summarized here. Molten

metal poured into sprue cup flows through runners and

Table 1 Casting defect prediction by simulation

Researcher (s) Year Method Alloy Process Prediction

Lewis & Ravichandran [1] 2000 FEM Al SC, GDC Shrinkage

Bounds et al. [2] 2000 FVM SS IC Shrinkage, misrun

Seetharamu et al. [3] 2001 FEM MS CC Shrinkage, distortion

Ou and Beckermann [4] 2003 FDM SS SC Shrinkage

Hamilton et al. [5] 2003 FEM Al IC Microporosity

Huat et al. [6] 2005 FDM SS IC Shrinkage

Kim et al. [7] 2006 FEM SS SC Shrinkage

Brooks et al. [8] 2007 FDM MS SC Burn-on, penetration

Ravi and Joshi [9] 2007 VEM MS SC Shrinkage

Kermanpur et al. [10] 2008 FVM CI SC Micro shrinkage

Hai-dong et al. [11] 2010 FDM Al HPDC Gas porosity

Zhanli et al. [12] 2010 FEM SS SC Shrinkage

Sun et al. [13] 2011 FDM DI SC Shrinkage, cold shut, hot crack

Sturm and Busch [14] 2011 FDM, FVM CI SC Shrinkage

Stetina, et al. [15] 2012 FDM MS CC Shrinkage

Reilly et al. [16] 2013 FDM, FVM – – Entrapment bubble

Sistaninia et al. [17] 2013 FEM Al CHC Hot tear

Tavakoli [18] 2014 FDM MS SC Shrinkage porosity

Fig. 1 Visualization of mold filling with temperature drop

1228 Trans Indian Inst Met (2015) 68(6):1227–1233

123

gates, and enters the casting cavity. This is governed by

conservation equations of mass and momentum, mainly,

incompressible Navier–Stokes equations. Appropriate ini-

tial and boundary conditions are required to obtain the

solution to these equations. In addition, tracking the

evolving free surfaces and satisfying boundary conditions

at the free surfaces becomes part of the solution procedure

which makes mold filling simulation a computationally

expensive task, limiting its practical applications. In the

present work, a novel approach for quick simulation of

metal flow during mold filling has been developed. This is

based on a set of rules governing the mass and momentum

conservation. The flow sequence is simulated based on the

computation of directional weights to velocity components,

as a function of inertia, head pressure and gravity. The

energy conservation equation also needs to be solved to

obtain casting temperatures, considering the heat transfer

between the molten metal and the surroundings, and the

release of latent heat during solidification. The variation of

solid fraction between the liquidus and solidus tempera-

tures for a given alloy is determined by assuming either a

linear relationship, or by using the Scheil’s equation.

For mold filling simulation, the initial conditions include

the user-specified value of pouring temperature of liquid

metal, fill time and the temperature of the mold. The tem-

perature distribution at the end of mold filling simulation

serves as the initial temperature distribution for solidification

simulation. The boundary conditions must be satisfied at

both external boundaries (exterior mold wall and top surface

of the riser) and the internal boundaries (metal-mold, mold-

insulator and chill-metal). The heat flux at the external sur-

faces is approximated using Newton’s law of cooling. The

internal boundaries are represented by the respective inter-

facial heat transfer coefficient values, which depend on the

gap between metal-mold interface, which in turn can vary

with respect to time and temperature. Hence a provision has

been made for handling specified temperature, flux or con-

vective heat transfer coefficient, which are either constant,

time-dependent or temperature-dependent at other material

interfaces like the metal-chill, metal-core etc. Thus, any one

of the following five types of boundary conditions can be

handled: (a) specific temperature, (b) specified heat transfer

coefficient to the ambient, (c) specified flux which can be a

constant or can have time or temperature dependent values,

(d) specified gap heat transfer coefficient which can be a

constant or can have time or temperature dependent values,

and (e) perfect contact.

The governing energy equation along with appropriate

initial and boundary conditions are solved using finite

volume method on structured grid. Explicit time integra-

tion scheme has been employed for transient simulation.

The internal time-step for solver computations is auto-

matically selected based on stability criteria.

3 Software Inputs and Outputs

The physics based Virtual Casting solver developed by

CSIR-NIIST team was integrated with the casting design

software AutoCAST developed at IIT-B, giving birth to

AutoCAST-X1 FLOW? (Fig. 1). The design module takes

3D model of an as-cast part as the initial input, and allows

the user to create the feeding and gating systems. The

structured mesh of mold along with relevant material

properties (density, specific heat, thermal conductivity), as

well as material properties and initial conditions (taken

from a database) form the intermediate output. Based on

these, FLOW? computes the values at every voxel (volume

cell) as a function of time. For mold voxels, the main

output is the temperature values from the instant of pouring

to solidification and beyond. For casting voxels, in addition

to temperature, the results include solid fraction, air frac-

tion, solidification time, and cooling rate. Major results and

analysis tools are briefly described next.

3.1 Flow Sequence

The initial entry of metal into the mold can be simulated to

get an understanding of possible problems of turbulence,

and area of mold being impacted. The flow sequence along

with volume filled, melt front velocity, and volume solid-

ified, aid in gating design and balancing.

3.2 Air Fraction

Air fraction displays the amount of air present in the cavity

at any point of time during filling. The user can identify

Fig. 2 Casting with shrinkage defect

Trans Indian Inst Met (2015) 68(6):1227–1233 1229

123

Fig. 3 Original methoding and simulation (temperature and shrink-

age results)

Fig. 4 Revised methoding and simulation(temperature and shrinkage

results)

1230 Trans Indian Inst Met (2015) 68(6):1227–1233

123

regions having isolated air pockets susceptible to air

entrapment. This enables correct placement of vents.

3.3 Liquid Fraction

This shows the residual liquid in the casting at any point of

time during solidification. Regions that solidify last and

may lead to shrinkage porosity (macro and micro) can be

identified.

3.4 Temperature Gradients

Large difference in temperature gradients along with high

temperatures causes hot tears, which can be located and

analyzed.

3.5 Cooling Curves

Virtual thermocouples can be placed inside the part as well

as mold to produce cooling curves, which can be compared

with experimental values to calibrate the software, as well

as analyze the microstructure.

4 Industrial Case Studies

4.1 Steel Coupler - Shrinkage Porosity

The casting exhibited corner shrinkage defect close to a

boss (Fig. 2). To analyze the cause of the defect, the

methods design was exactly replicated and simulated. The

casting had four side feeders and two top feeders.

Exothermic sleeves and covers were used to increase

effectiveness of feeding. The gating system included two

runners and two gates connecting to side feeders. Figure 3

shows the temperature profile at the end of solidification

and the computed locations of shrinkage porosity, match-

ing the actual observations. The colour scale corresponds to

the range from room temperature (dark blue) to solidus

temperature (white–yellow).

To prevent future occurrence of the shrinkage defect, the

methods design of the casting was revised. Now five top

feeders and one side feeder (close to the defect location)

were designed and modelled (Fig. 4). The revised design

was simulated and analyzed using various tools. This

included: temperature profile at the end of solidification,

and shrinkage porosity (Fig. 4), as well as additional tools:

liquid fraction during solidification, feed metal paths, and

cooling rate (Fig. 5). It was seen that as the solidification

progressed, liquid metal was present only in the feeders

(not in the casting); the peak temperatures were also in the

feeders. The feed paths and cooling curve in the thermo-

couples placed at the suspected location and the side feeder Fig. 5 Additional results: liquid fraction, feed paths, and cooling rate

Trans Indian Inst Met (2015) 68(6):1227–1233 1231

123

(not included here for the sake of brevity) also showed that

the feeder was able to supply feed metal and prevent the

occurrence of shrinkage defect. The casting was found to

be free of shrinkage defect. The revised methods design

was communicated to the foundry and they were happy that

the casting quality could be improved without affecting

yield.

4.2 Cast Iron Connector—Blow Hole

This casting suffered from frequent rejections due to blow

holes at the upper surfaces (Fig. 6). Initial methoding of the

casting is shown in Fig. 7. The casting had two hot feeders

and one top vent. The methods design was exactly modeled

and simulated to visualize the motion of air inside the mold

cavity, since blow holes are primarily caused by air

entrapped in solidifying metal (Fig. 8). The simulation

Fig. 6 Casting with blow hole

Fig. 7 Model of the methods design

Fig. 8 Molding filling simulation showing: top- metal progress,

middle- air fraction, and bottom- air blow hole formation

1232 Trans Indian Inst Met (2015) 68(6):1227–1233

123

clearly showed the accumulation of air in the defect loca-

tion during mold filling process. To prevent the blow hole

defect, the methods design was modified by adding flow-

offs (not shown here to protect the interest of the manu-

facturer). The mold filling time was also increased by

20 %. The results of air fraction obtained from mold filling

simulation of the revised methods design showed a good

casting without blow hole defects.

5 Conclusion

The case studies clearly demonstrate the ease and useful-

ness of casting process simulation, providing an insight to

isolate the causes of internal defects, based on which the

methods design and process parameters can be modified.

The in-built design and modelling facility makes this task

also easy for practicing foundry engineers. Further, the

modified design can be simulated again to verify and assure

casting quality, thus avoiding foundry trials that consume

time and production resources. Indian foundries, especially

small and medium units can utilize such tools for com-

puter-aided design and simulation to enhance their quality,

productivity and value-addition. Further, engineering and

polytechnic institutes can integrate computer simulation

with experimental casting, allowing students to compare

the results, so that they can gain a better understanding and

interest in this field.

References

1. Lewis R W, and Ravichandran K, Int J Numer Methods Eng 47(2000).

2. Bounds S, Moran G, Pericleous K, Cross M, and Croft T N,

Metall Mater Trans 31 (2000).

3. Seetharamu K, ParaGasam R, Qadir G A, Zainal Z A, Prasad B S,

and Sundararajan T, Sadhana 26 (2001).

4. Ou S, and Beckermann C, in Simulation of Dimensional Changes

in Steel Casting. Chicago, Proceedings of the 57th SFSA Tech-

nical and Operating Conference (2003).

5. Hamilton R W, See D, Butler S and Lee P D, Mater Sci Eng

A343 (2003).

6. Huat N C, Hassan S, Subudin S, and Ariffin S, Defect Analysis of

Investment Casting Using Computer Simulation. Johor, Interna-

tional Conference on Information and Communication Technol-

ogy (2005).

7. Kim K H, Hwang J H, Oh J S, Lee D H, Kim I H, and Yoon Y C,

in Prediction of Shrinkage Defect in Steel Casting For Marine

Engine Cylinder Cover by Numerical Analysis. Harrogate, Pro-

ceedings of 67th World Foundry Congress (2006).

8. Brooks B E, Beckermann C, and Richards V L, Int J Cast Metals

Res 20 (2007).

9. Ravi B, and Joshi D, Indian Foundry J 53 (2007).

10. Kermanpur A, Mahmoudi S, and Hajipour A, J Mater Process

Technol 206 (2008).

11. Hai-dong Z, Yan-fei B, Xiao-xian O, and Pu-yun D, Trans

Nonferrous Metall Soc 20 (2010).

12. Zhanli G, Young-Hoon Y, Shuya K and Mook Y J, Simulation of

Shrinkage Defect Formation in EXhaust Manifold. Korea, The

8th Pacific Rim International Conference on Modeling of Casting

and Solidification (2010).

13. Sun Y, Luo J, Mi GF, and Lin X, Mater Design 32 (2011).

14. Sturm J and Busch G, in Cast Iron - A Predictable Material.

Hangzhou, Proceedings of 69th World Foundry Congress (2011).

15. Stetina J, Klimes L, Mauder T and Kavicka F, Mater Technol 46(2012).

16. Reilly C, Green N R, and Jolly M R, Appl Math Model 37 (2013).17. Sistaninia M, Drezet J M, Phillion A B, and Rappaz M, J Miner

Metals Mater Soc 65 (2013).

18. Tavakoli R, Int J Adv Manuf Technol 74 (2014).

19. Savithri S, Roschen S, Jacob E, Chourasia V, Sivakumar Y V M,

and Lanka B, Indian Foundry J 60 (2014).

Trans Indian Inst Met (2015) 68(6):1227–1233 1233

123