Available online at www.sciencedirect.

com

Procedia CIRP 7 (2013) 616 – 621

Forty Sixth CIRP Conference on Manufacturing Systems 2013

Casting Defect Analysis using Design of Experiments (DoE) and

Computer Aided Casting Simulation Technique
Uday A. Dabade* and Rahul C. Bhedasgaonkar
Department of Mechanical Engineering, Walchand College of Engineering, Vishrambag, Sangli - 416 415, INDIA
* Corresponding author. Tel.: +91- 9820542138; fax: +91 0233 2300384; .E-mail address: [email protected]

Abstract

In this paper the design of experiments and computer assisted casting simulation techniques are combined to analyze the sand
related and methoding related defects in green sand casting. An attempt has been made to obtain the optimal settings of the
moulding sand and mould related process parameters of green sand casting process of the selected ductile iron cast component. The
green sand related process parameters considered are, moisture content, green compression strength, and permeability of moulding
sand and mould hardness (in horizontal direction). In first part of this paper Taguchi based L18 orthogonal array was used for the
experimental purpose and analysis was carried out using Minitab software for analysis of variance (ANOVA) and analysis of mean
(AOM) plot. ANOVA results indicate that the selected process parameters significantly affect the casting defects and rejection
percentage. In the second part, shrinkage porosity analysis is performed using casting simulation technique by introduction of a new
gating system designed, solid model developed for four cavities mould. Number of iterations using casting simulation software was
performed for mould filling and solidification analysis to reduce the level and intensities of shrinkage porosities in cast component.
With new gating and feeding system design reduction in shrinkage porosity (about 15%) and improvement in yield (about 5%) is
observed.

©2013
© 2013TheTheAuthors.
Authors. Published
Published by Elsevier
by Elsevier B.V.B.V.
Selection
Selectionand peer-review
and/or under under
peer-review responsibility of Professor
responsibility Pedro Filipe
of Professor do Carmo
Pedro Cunha
Filipe do Carmo Cunha

Keywords: Casting defects; Casting simulation and optimization; Green sand casting; Mould filling and solidification analysis; Taguchi method;

1. Introduction 2. Methodology

Foundry industries in developing countries suffer In this proposed method of casting defect analysis,
from poor quality and productivity due to involvement the DoE (Taguchi method) is used for analysis of sand
of number of process parameters in casting process. and mould related defects such as sand drop, bad mould,
Even in a completely controlled process, defects in blow holes, cuts and washes, etc. Whereas computer
casting are observed and hence casting process is also aided casting simulation technique is used for
known as process of uncertainty which challenges methoding, filling and solidification related defects such
explanation about the cause of casting defects [1]. as shrinkage porosity, hot tears, etc. Flow chart of
Casting defects analysis is the process of finding the root proposed method of casting defect analysis is shown in
cause of occurrence of defects in the rejection of casting Figure 1.
and taking necessary steps to reduce the defects and to
improve the casting yield. Techniques like cause-effect 2.1 DoE (Taguchi method) for casting defects analysis
diagrams, design of experiments (DoE), casting
simulation, if-then rules (expert systems) and artificial The literature review indicates that the Taguchi
neural networks (ANN) are used by various researchers method is the best option for design of experiments
for analysis of casting defects [1 2]. when number of process parameter are involved in the
In this paper, a new method of casting defects process. Taguchi approach is suitable in experimental
analysis is proposed which is combination of DoE design for designing and developing robust products or
(Taguchi method) and computer aided casting simulation processes irrespective of variation in process parameter
technique and discussed in the following sections. (within set limits) and or variation in environmental

2212-8271 © 2013 The Authors. Published by Elsevier B.V.

Selection and peer-review under responsibility of Professor Pedro Filipe do Carmo Cunha
doi:10.1016/j.procir.2013.06.042
 Uday A. Dabade and Rahul C. Bhedasgaonkar / Procedia CIRP 7 (2013) 616 – 621 617

conditions. The present research as associated with sand of variance (ANOVA) can be obtained to determine
casting process which involves various parameters at the statistical significance of the parameters. Means
different levels and affects the casting quality. plots can be plotted to determine the preferred levels
Considering these features of Taguchi method, it is used of parameters considered for experimentation.
to reduce the % of rejection due to sand and moulding Select optimum levels of control parameters,
related defects by setting the optimum values of the perform confirmation experiments and implement
process parameters of the green sand casting. The the process.
methodology used to achieve optimized process
parameters using DoE is as given below: 2.2 Casting simulation for casting defects analysis

Select any defect observed due to sand and mould. Simulation is the process of imitating a real
Set the phenomenon using a set of mathematical equations
adjusting the process parameters. implemented in a computer program. In casting
Select the most significant parameters influencing simulation the mould filling and solidification analysis is
defects in casting by cause effect diagram. done by using an algorithm or program based on finite
volume method, to identify the hot spots and hence
Rejection analysis defects like shrinkage porosities, hot tears, cracks, etc.
The simulation programs are based on finite element
analysis of 3D models of castings and involve
Identification of sophisticated functions for user interface, computation
defects to be analysed and display. The casting model (with feeders and gates)
has to be created using a solid modelling system and
Defects due to imported into the simulation program [3]. Ravi [4] has
Defects due to
sand / mould methoding / filling / suggested casting simulation and optimization
solidification methodology as shown in Figure 2.
Use DoE Use casting Part model, Material, Process
(Taguchi method) simulation technique STEP 1: Data Gathering parameters, Methods design,
Existing defects
Prepare cause- Parting line, Cores, Feeders,
Design and solid
effect diagram STEP 2: Methods Design Feedaids, Gating system, mold
modelling of gating
cavity layout
Identification of Model import, Mesh generation,
parameters, levels Mould filling and STEP 3: Simulation Material and process,
affecting defects solidification analysis, Computation, Visualisation
optimization Modify design, Simulation,
STEP 4: Optimization Check quality, Check yield,
DoE for Check cost
identification of Methods report, Analysis report,
optimized levels STEP 5: Project Closure Images slides, Compare results,
Archive project

Implementation Fig. 2 Casting simulation-optimization methodology [4]

N N 3. Experimental Work
Improvement in
casting and yield
Experiments were performed in a medium scale
Y ferrous foundry producing ductile iron cast components.
Accept and Implement The proposed method of casting defects analysis
involves two techniques, therefore according their
Fig. 1 Flow chart of analysis of casting defects using DoE and casting requirements; different castings with different defects
simulation techniques were identified by rejection analysis.
After carrying out rejection analysis it was found that
Select the parameters and their levels. Perform the in cast component named pipe turbocharger the
experiments (trial castings) as per DoE (Taguchi rejection was maximum up to 10 % due to sand related
method) and collect the data. defects like sand drop, bad mould, blow holes and cuts
Analyze the data using statistical tools. An analysis and washes, etc. Therefore this component was selected
618 Uday A. Dabade and Rahul C. Bhedasgaonkar / Procedia CIRP 7 (2013) 616 – 621

for defects analysis by DoE method. Whereas in another AOM plot in Figure 3 indicates that % rejection is
component named wheel hub severe problem minimum at first level of moisture content (A1), third
of shrinkage cavities inside the casting after level of green compression strength (B3), and second
solidification causing rejection of more than 25 %. level of mould hardness in horizontal direction (D2).
Therefore, front wheel hub was selected for defect-
shrinkage analysis by casting simulation technique.
Main Effects Plot for SN ratios
Data Means
3.1 DoE method for casting defects analysis
Moisture GCS Permeability Mould hardness
-13
Process parameters of green sand casting that
influences the identified defects in casting pipe
turbocharger with their levels are shown in table 1. -14

Mean of SN ratios
Table 1 Process parameters and their levels (range)
-15

Process parameters Level 1 Level 2 Level 3

-16
A : Moisture content
4.5 4.9 5.0 5.4 ------
(%)
B : Green compression -17
1151 1250 1251 1350 1351 1450
strength (gm/cm2)
4.7 5.2 1200 1300 1400 110 125 140 80 85 90
C : Permeability
105 117 118 132 133 145 Signal-to-noise: Smaller is better
(number)
D : Mould hardness
78 82 83 87 88 92
(number)
Fig. 3 Main effects plot for S/N ratio

Out of four selected process parameters, one 3.1.2 Confirmation experiments

parameter is at two levels and other three are at three
different levels. Therefore, L18 orthogonal array is Three confirmation experiments were performed at
selected for the experimentation [5]. Two parameters the optimized settings of the process parameters, results
interactions such as A*B and A*C was considered. The of which are shown in Table 3. Prior to the application
response variable was the % rejection of casting due to of Taguchi method rejection due to sand related defects
defects which is ratio of rejection due to considered for pipe turbocharger was 10% which is reduced to
process parameters to the quantity poured. As per L18 maximum up to 3.59 %.
orthogonal array 18 experiments were performed
randomly. Each experiment was performed thrice and Table 3 Results of confirmation experiments
average % of rejection in each experiment was
Rejection Average % of
considered as the response variable. Expt. No.
1 2 3 rejection
1 3.84 3.41 3.50 3.58
3.1.1 Analysis of experimental results
2 3.41 3.50 4.38 3.76
Analysis of experimental results was performed using 3 3.41 2.99 3.94 3.44
Minitab 16 software and ANOVA and AOM plots Total average of %
3.59
obtained are given in Table 2 and Figure 3 respectively. rejection
ANOVA in Table 2 indicates that moisture content,
green compression strength and mould hardness 3.2 Shrinkage analysis of wheel hub by casting
parameters significantly influence the % rejection at simulation technique
95% confidence level.
Casting simulation technique can be efficiently used
Table 2 ANOVA for % rejection at 95 % confidence limit
for analysis of casting defects related to methoding such
as shrinkage porosities by the mould filling and
Parameter DoF SS Variance F ratio P- value solidification analysis [8]. The casting simulation and
A 1 3.1920 31.920 16.73 0.005
B 2 20.6431 10.32 54.09 0.0001
optimization methodology is shown in Figure 2 and the
C 2 0.0594 0.0297 0.16 0.843 various steps involved in it are discussed below.
D 2 6.1523 3.076 16.12 0.003
A*B 2 0.3213 0.1606 0.95 0.438 3.2.1 Data collection
A*C 2 0.5741 0.2870 1.70 0.260
Error 42 1.0129 0.024
Total 53 31.9551 In this stage for selected casting the data regarding
S = 0.436840 R-Sq = 94.03% R-Sq(adj) = 89.85% the existing defects, process parameters, materials,
 Uday A. Dabade and Rahul C. Bhedasgaonkar / Procedia CIRP 7 (2013) 616 – 621 619

methods design is collected and analysed. Existing

image of wheel hub with defect is shown in Figure 4.
Input to the simulation program is the 3D solid CAD
model of the casting in STL format, so 3D solid CAD
model of hub was created as shown in Figure 5.

Fig. 6 Solid model of gating system of iteration I of Wheel hub (four

cavity mould)

3.2.3 Simulation and optimization

Fig. 4 Casting with defect-shrinkage cavities In this research paper, FEM based casting simulation
package Magmasoft (courtesy: TCML, Sangli, India) is
used for mould filling and solidification analysis. At the
initial stage, the gating and risering system for the four
cavities of the mould was designed and solid modeled as
shown in Figure 6. In this gating system, single separate
ingate is provided to each casting to fill the mould
cavity. Pouring is through central sprue and four ingates
are attached to the runner. The yield with this iteration
was 72 %.

Simulation for mould filling and solidification

analysis with this gating system reveals that the severe
shrinkage porosities of non acceptable range exist in
each casting of the four cavities mould, refer Figure 7(a).
Liquid to solid conversion plot shows that the height of
Fig. 5 Solid 3-D model of as-cast wheel hub
riser neck is insufficient, due to which feeders are not
properly feeding the casting during solidification though
3.2.2 Methods Design feeder sizes and locations are correct, refer Figure 7(b).
This results in occurrence of shrinkage porosities inside
In existing methoding, single central feeder was used the casting and hence causes rejection. Therefore, it is
to pour as well as feed the casting during solidification necessary to make suitable changes in gating system to
of four cavities mould. Feeder was not feeding the remove or reduce the level of shrinkage porosities.
casting properly which results in shrinkage porosities
inside the casting, refer Figure 4. Therefore, to minimize the shrinkage porosities
number of iterations for gating system design of wheel
In this stage a new gating system was designed (for hub was performed and optimum design for gating
four cavity mould) using theoretical formulae used for system was selected. In new optimum methoding,
gating and risering system design. Feeder was designed indirect chills in core were used in two cavities and
by modulus method and Caines method of feeder design provision for feeders was made into the core for
[6 7]. Two feeders with appropriate sizes were used to remaining two cavities to feed the casting during
feed the four cavities mould as shown in Figure 6. solidification (refer Figure 8).
620 Uday A. Dabade and Rahul C. Bhedasgaonkar / Procedia CIRP 7 (2013) 616 – 621

Fig. 8 Solid model of the gating system of optimum iteration

Finally, optimum iteration of gating system design for

wheel hub with four cavities mould was implemented
and it is found that rejection of casting due to shrinkage
porosities was reduced by 15 %.
(a) Shrinkage porosities

(b) Liquid-solid conversion plot

(a) Shrinkage porosities
Fig. 7(a b) Simulation results of iteration I of the gating system

Simulation results for shrinkage porosities refer

Figure 9(a); shows that with optimum iteration gating
system design the shrinkage porosity is considerably
reduced up to acceptable level for each casting of four
cavities mould as compared to that of iteration I stage
{compare Figure 9(a) and 7(a)}.

Liquid-solid conversion plot, refer Figure 9(b) shows

that, internal chills and feeder placed inside the core
works satisfactorily and results in sufficient feeding of
casting during solidification and hence reduction in level
of shrinkage porosities inside the casting. Though there
is small shrinkage porosity in the direction of ingate of
feeder and at thick cross-section of the casting which is
of acceptable level. With this methoding (optimum (b) Liquid-solid conversion plot
iteration) the yield is 70 % which is more than the Fig. 9(a b) Simulation results of optimum gating system design
existing methoding used in foundry. (optimum iteration)
 Uday A. Dabade and Rahul C. Bhedasgaonkar / Procedia CIRP 7 (2013) 616 – 621 621

4. Conclusions References

A new method of casting defect analysis is proposed [1] H.C. Pandit, Amit Sata, V. V. Mane, Uday A. Dabade, 2012,
Novel Web-
and studied which is a combination of design of
published in technical transactions of 60th Indian Foundry
experiments method (Taguchi method) and computer Congress, 2 4th March 2012, Bangalore, pp 535 544.
aided casting simulation technique for analysis of [2] Rahul Bhedasgaonkar, Uday A. Dabade, May 2012,
rejection of casting due to defects related to sand, Proceedings
moulding, methoding, filling and solidification in of 27th National Convention of Production Engineers and National
green sand casting. Seminar on Advancements in Manufacturing VISION 2020,
The optimized levels of selected process parameters organised by BIT, Mesra, Ranchi, India.
obtained by Taguchi method are: moisture content [3] B. Ravi, 2008, Casting Simulation and Optimization: Benefits,
(A): 4.7 %, green compression strength (B): 1400 Bottlenecks, Indian Foundry Journal, Special Issue.
gm/cm2, permeability number (C): 140 and mould [4] B. Ravi, 2010, Casting Simulation-Best Practices, Indian Foundry
hardness number (D): 85. Congress, Ahmadabad.
With Taguchi optimization method the % rejection of [5] Madhav S. Phadake, 1989, Quality Fngineering using Robust
castings due to sand related defects is reduced from Design, Prentice-Hall, Englewood Cliff, New Jersey.
10 % to a maximum upto 3.59 %.
[6] Richard Heine, Carl Loper, Philip Rosenthal, 1984, Principles of
Design of experiments method such as Taguchi
Metal Casting, Tata McGraw Hill Publications, New-Delhi.
method can be efficiently applied for deciding the
[7] P. N. Rao, 1998, Manufacturing Technology-Foundry, Forming
optimum settings of process parameters to have
minimum rejection due to defects for a new casting and Welding, Tata McGraw-Hill Publications, New-Delhi.
as well as for analysis of defects in existing casting. [8] Harshwardhan Pandit, Uday A. Dabade, 2011, -Need
For analysis of defect like shrinkage porosities itive age- A case study on Auto-
computer aided casting simulation technique is the NUiCONE, Nirma University, Ahmadabad.
most efficient and accurate method. The quality and [9] S. Guharaja, A. Noorul Haq and K. M. Karuppannan
yield of the casting can be efficiently improved by optimization of CO2
computer assisted casting simulation technique in International Journal of Advanced Manufacturing Technology
shortest possible time and without carrying out the (2009) 3, pp 41-50.
actual trials on foundry shop floor.