See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/299544864

Effect of molding parameters on chemically

bonded sand mold properties
Article in Journal of Manufacturing Processes April 2016
DOI: 10.1016/j.jmapro.2016.03.007

CITATIONS

READS

80

2 authors:
Himanshu Khandelwal

B. Ravi

Indian Institute of Technology Bombay

Indian Institute of Technology Bombay

11 PUBLICATIONS 18 CITATIONS

105 PUBLICATIONS 825 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.

SEE PROFILE

Available from: Himanshu Khandelwal

Retrieved on: 05 October 2016

Journal of Manufacturing Processes 22 (2016) 127133

Contents lists available at ScienceDirect

Journal of Manufacturing Processes

journal homepage: www.elsevier.com/locate/manpro

Effect of molding parameters on chemically bonded sand

mold properties
Himanshu Khandelwal , B. Ravi
Indian Institute of Technology Bombay, Mumbai 400076, India

a r t i c l e

i n f o

Article history:
Received 31 December 2015
Received in revised form 9 March 2016
Accepted 9 March 2016
Keywords:
Sand-casting
No-bake binder
Chemically bonded mold
Mechanical property
Shrinkage
Optimization

a b s t r a c t
Chemically bonded sand molds and cores have better mechanical properties and produce more dimensionally accurate castings, compared to green sand molds, and hence are being increasingly preferred for
near net shape metal parts. While it is well known that the properties and quality of a mold depend on
its material composition and molding process, this has not been investigated well for chemically bonded
molds. In this work, the effect of sand grain size, binder percentage and curing time (each varied in three
levels) on the mechanical properties (compression strength, shear strength and core hardness) as well as
dimensional changes (shrinkage) of no-bake chemically-bonded molds and cores was studied through
lab experiments. Their mechanical properties were found to increase with an increase in binder content and curing time; and decrease with increasing grain neness number. The shrinkage was found to
increase with an increase in all three process parameters. The results are explained in terms of bonding
and curing phenomena of binders, and evaporation of solvent. These are also supported by SEM analysis
of crosslinked resin bridges between sand grains. The ndings were incorporated into a multi-objective
optimization model to obtain the desired combination of mold properties, which is solved as a linear
programming problem. The model and its results were successfully veried through experiments. This
work paves the way for automatic optimization of molding parameters of chemically-bonded sand molds
and cores, to achieve the desired quality.
2016 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.

1. Introduction
Near-net shape metal parts can be manufactured by many
routes; yet metal casting is still the most preferred process due
to its ability to produce intricate shaped parts in a variety of metals
and range of order quantity at a low cost compared to either subtractive or additive manufacturing [1]. Among various metal casting
processes, green sand casting is the most widely used one worldwide, due to the availability of raw materials (sand and bentonite),
established process equipment, and ease of recycling of molding
materials [24]. This process is however, unable to meet the evertightening requirements of original equipment manufacturers in
terms of internal soundness, dimensional accuracy and surface
nish, and has to be followed up with considerable machining
[5,6]. These limitations are reduced by some of the newer molding processes, which use various chemical binders instead of clay
or bentonite [7,8]. The binder systems are classied according to
the curing mechanism and the type of chemicals used: heat cured,

Corresponding author. Tel.: +91 9167481947.

E-mail address: khandelwal [email protected] (H. Khandelwal).

cold box, and no-bake [9,10]. The quality of the mold, in terms of
strength, hardness and dimensional stability plays a signicant role
in achieving dimensionally accurate castings with good mechanical properties [11,12]. The quality of the mold depends on the type
of sand used (silica or zircon; grain size and shape), binder composition (proportion of different chemicals), type of molding process
(manual, machine), and the curing time.
Several researchers have investigated the effect of molding composition on the dimensional stability and mechanical properties of
sand molds [9,11,1316]. Werling [14] conducted a set of experiments with different molding mixtures to investigate their effect
on core strength, shrinkage, and sand owability. Dimensional
changes in chemically bonded mold have been investigated by
Showman et al. [15], who reported continuous shrinkage of mold
over the curing time, and ascribed it to shortening of the resin
bridges. Khandelwal and Ravi [9] also observed the shrinkage in
cores prepared using alkyd urethane binder system, and linked
it to the evaporation of the solvent. Core shrinkage and hardness
were found to be affected by the amount of resin used and not
by the amount of catalyst. Lowe and Showman [16] used various
binder systems with different type of sands and process conditions.
Their research showed that sand grain size, shape, distribution and

http://dx.doi.org/10.1016/j.jmapro.2016.03.007
1526-6125/ 2016 The Society of Manufacturing Engineers. Published by Elsevier Ltd. All rights reserved.

128

H. Khandelwal, B. Ravi / Journal of Manufacturing Processes 22 (2016) 127133

density also affect dimensional changes of sand molds. Cores with

ne sand grain size were found to have more shrinkage and less
hardness for the same percentage of binder.
A few researchers have explored different techniques for optimizing the composition of sand molds to achieve the desired
quality, though most of the work was conned to green sand
molding [2,1720]. The molding composition was optimized for
quality improvement of iron casting by Kumar et al. [17], who
used Taguchi method to minimize the defects (shifts, blow holes,
sand drop, warpage, etc.) in housing covers. Saikaew et al. and
Kumar et al. [2,17] used mixed experimental design and propagation of error analysis to optimize green sand molding composition.
Ganesan et al. [18] introduced the use of Normal Boundary Intersection method for optimization of green sand molds. The technique
was used for obtaining the local maximum solution of the dened
problem. Kumaravadivel and Natarajan [19] used response surface method to optimize molding composition (recycled molding
sand, bentonite, and water) of green sand casting. They developed
a non-linear regression model to optimize permeability, compression strength and shear strength of chromite-based resin bonded
system.
In general, there appears to be very little work investigating the
effect of material composition and process parameters on the properties of no-bake chemically-bonded sand molds. Further, there
appears to be no report of any systematic approach to optimize
the above parameters to achieve the desired combination of mold
properties. This has been taken up in the present work. The design
of experiments was carried out, followed by the analysis of results,
understanding the inuence of individual parameters on mold
properties, and nally the formulation and validation of an optimization model. The overall methodology is illustrated in Fig. 1,
and described in the following sections.
2. Experimental work
The experimental work was taken up to investigate the effect
of mold composition (sand grain neness and binder content)
and curing time on mold properties: compression strength, shear
strength, core hardness, and core shrinkage.
The sand core samples were prepared using oil urethane nobake binder system. This three part binder system comprises alkyd
resin, metallic catalyst, and polymeric MDI (methyl di-isocyanate).
Their proportions were selected based on the weight of sand.
Double screened silica sand was used for preparing the molds

Step 1 - Deciding the Objecves

Invesgate the eect of molding parameters on mold properes

Opmizaon of mold composion to obtain the desired range of properes
Step 2 - Determine Process Variable and their level

Parameters selected: Binder percentage, Sand Grain Size, Curing Time

All parameters varied in three levels: Low, Medium, High

Table 1
Experiment parameters and their levels.
Designation

Parameter (unit)

Level 1

Level 2

Level 3

G
B
T

Sand grain size (GFN)

Binder (wt% of sand)
Curing time (h)

40
1.6
1

60
2
2

80
2.4
4

and cores. Three key molding parameters were selected based

on a preliminary study, which included sand grain size, binder
composition, and curing time. The grain neness number (G) of
silica sand used for these experiments was varied in three levels:
40, 60, and 80, which are common in industrial use. The standard
deviation of three test samples of sand with GFN 40, 60, 80 are 2.45,
2.17 and 1.78. The percentage of alkyd resin binder (B) was varied
in three levels: 1.6, 2 and 2.4, by weight percentage of sand. The
curing time (T) was also varied in three levels: 1 h, 2 h and 4 h. All
parameters and their levels are listed in Table 1.
Taguchi design of experiments (DOEs) methodology was
employed to determine the optimum number of experiments and
to study the inuence of various parameters on mold quality. The
appropriate orthogonal array selection is based on the experimental resolution required, which is estimated by degree of freedom
(DOF). Two DOF for each factors (number of levels-1), suggests
minimum 6 DOF required. An L9 orthogonal array was selected,
wherein the experimental design consists of nine experimental
runs, and various factors were assigned in three columns of the
array, as shown in Table 2.
Test samples were produced by oil urethane chemical binder
system in line with AFS (American Foundry Society) standard
guidelines. Alkyd resin percentages were selected as shown in
Table 1. The amount of catalyst and cross-linking agent was taken
as 10% and 20% of the amount of weight of alkyd resin. The samples were prepared using the assigned experimental array shown
in Table 2. The chemicals mentioned were mixed well with the
sand using a sand mixer. Cylindrical samples of 5 cm diameter and
10 cm length were produced using a gang compression core box,
for compression and shear strength measurements. Rectangular
blocks of size 15 cm 15 cm 5 cm (for hardness measurement)
and cylindrical samples of 20 mm diameter and 100 mm length
(for shrinkage measurement) were also prepared. All the test
samples were produced in similar laboratory conditions of temperature and humidity. All four types of test samples are shown in
Fig. 2.
The compression strength and shear strength were measured
using a digital universal strength testing machine (Versatile Equipments Pvt. Ltd, Kolhapur), shown in Fig. 3. This is a pneumatic
machine which can exert a continuous direct load up to 400 kg;
it can measure compression strength up to 80 kg/cm2 and shear
strength up to 64 kg/cm2 . The machine has a loading frame with
provision to accommodate various samples, attachments and calibration kit. The core hardness is measured using a scratch tester

Step 3 - Experimental Plan

Taguchi based design of experiments

Full factorial design - L9 orthogonal array
Step 4 - Analysis of Experimental Results

Table 2
Experimental (L9) orthogonal array.
Experiment no.

Invesgate the eect of process parameters on mold properes (Compression strength,

Shear Strength, Core Hardness, Core Shrinkage) using main eect plots
Step 5 - Molding Sand Composion Opmizaon

Construct the opmizaon model as a linear programming problem

Solve for dierent weighng factors of parameters using Opmizer Solver
Step 6 - Verifying the Results

Compare the results obtainded from opmizaon model with experiments.

Fig. 1. Process ow framework used for optimization.

1
2
3
4
5
6
7
8
9

Sand (G)

Binder (B)

Time (T)

Level

GFN

Level

Wt. % of sand

Level

Hour(s)

1
1
1
2
2
2
3
3
3

40
60
60
60
60
60
80
80
80

1
2
3
1
2
3
1
2
3

1.6
2
2.4
1.6
2
2.4
1.6
2
2.4

1
2
3
2
3
1
3
1
2

1
2
4
2
4
1
4
4
2

H. Khandelwal, B. Ravi / Journal of Manufacturing Processes 22 (2016) 127133

129

Fig. 2. Samples prepared for compression, shear, hardness, and shrinkage tests (left
to right).

shown in Fig. 4(a). For obtaining the hardness reading, the sharp
edge radius tipped plough of tester is pressed and pulled (up to
25 mm) against the surface of the sand mold. The shrinkage (dimensional change) of the sand cores was measured using a dial indicator
measurement, using a setup and protocol developed during a previous investigation [9]. The xture for core shrinkage is shown in
Fig. 4(b). The produced core specimens were placed in the xture
and the reading from the dial indicator was recorded in predened
time interval.
The Taguchi approach predicts the effect of individual process
parameters by the main effect plot. ANOVA analysis was performed
to observe the signicant effect of parameters. The Taguchi analysis will give the optimum setting among the selected levels, which
can be treated as local optimum values. Therefore, additionally
an optimization model was developed in for global optimization
of all four responses (mold compression strength, shear strength,
core shrinkage and core hardness). The optimization function was
designed to maximize compression strength, shear strength and
core hardness; and minimize the core shrinkage. The optimization
problem was constructed as a linear programming problem and
solved to predict the optimum mold properties. Further the model
was veried by comparing the solver result with conrmation
experiments.
3. Results and discussion
The quality attributes of sand molds and cores measured in this
work included compression strength, shear strength, hardness and
shrinkage. These properties were measured for all nine experimental conditions shown in Table 2 and are reported in Table 3. Their
main effects are plotted in Fig. 5 to gain a better understanding. The
compression strength was found to decrease with ner sand and
increase with binder content as well as curing time [Fig. 5(a)]. A
similar trend was observed in shear strength and hardness measurement, as shown in Fig. 5(b) and (c). The inuence of sand
grain size on core shrinkage was found to be the reverse of that
on mechanical properties [Fig. 5(d)]. The higher slop of the plots
shows the signicant effect of the individual parameters. Although
all the parameters have signicant inuence, the curing time was

Fig. 4. Hardness measurement with scratch tester (left) and shrinkage measurement with a dial indicator setup (right).
Table 3
Mold properties for different experiment trials.
Trial no.

Compression
strength CS
(kg/cm2 )

Shear strength
SS (kg/cm2 )

Core hardness
CH (Number)

Core shrinkage
SH (microns)

1
2
3
4
5
6
7
8
9

4.08
7.56
16.16
5.12
10.31
5.21
5.30
4.20
7.21

9.10
13.10
17.20
12.40
13.64
9.64
11.61
9.40
11.20

33
57
83
42
64
46
40
37
51

15.2
33.0
45.5
30.1
63.5
53.2
52.1
57.1
80.6

found to be the most dominant parameter for mechanical properties. The sand grain neness was found to be the most dominant
parameter for core shrinkage.
The phenomena can be explained in terms of the bonding and
curing mechanisms of chemically bonded mold. When the binder is
mixed with the silica sand, the mixing process allows a thin coating
of binders on individual sand particles. When the surfaces of two
binder coated sand particles come in close proximity, then they
crosslink with each other, and form a resin bridge. During curing,
these resin bridges shorten due to the evaporation of the solvent
present in binder, and result in hardening of the bridges. The curing generally takes about 24 h. To get a better insight in the curing
mechanism, the SEM analysis of a sample were carried out. The
SEM images of samples (after 24 h curing) taken at two different
locations are shown in Fig. 6. The images show the resign bridges
between two irregular shaped sand grains. The evidence of solvent
evaporation and shortening of resin bridges was obtained in an
earlier set of experiments [9], in which weight reduction of bare
chemical mixture with varying binder content was investigated.
The inuence of molding composition on mold properties can
be explained in term of curing phenomenon. The effect of molding parameters on compression strength, shear strength and core
hardness can be clearly observed from Fig. 5(a)(c). An increase in

Fig. 3. Sand strength testing machine (left) for compression (middle) and shear (right).

130

H. Khandelwal, B. Ravi / Journal of Manufacturing Processes 22 (2016) 127133

Fig. 5. Main Effect Plot of mold properties.

grain neness number of sand leads to more sand particles being

coated with the same amount of binder. This provides less strength
to the crosslink bond between two sand particles, resulting in lower
mechanical properties. The observation is in agreement with previous research [16]. As resin binder percentage increases, the same
number of sand particle are coated with more amount of binder,
which results in an increase in mechanical properties. Further, it
was observed that compression strength, shear strength and core
hardness increase with curing time. This occurs because curing
allows hardening of resin bridges between sand grains, providing
greater mechanical strength to the mold or core.
The effect of molding parameters on core shrinkage can be
clearly observed from Fig. 5(d). The inuence of sand grain size
on core shrinkage was found to be the reverse of that on mechanical properties [Fig. 5(d)]. The shrinkage phenomena of the core
is the result of shortening of the resin bridge due to evaporation of solvent present in binders. This was also shown in earlier
research [9,21]. Core shrinkage increases with ner sand, because
more number of sand particles lead to more resin bridges, and these
shorten during curing, resulting in more shrinkage. The shrinkage
also increases with binder content and curing time. More amount
of binder leads to more volumetric reduction of resin bridges during curing, thereby leading to increased shrinkage. Similarly, as
curing time increases, more solvent evaporates, leading to more
volumetric reduction of resin bridges and thereby more shrinkage.
The experimental results were used to obtain empirical relationships between individual responses (compression strength, shear
strength, core hardness and core shrinkage) based on predictor

variables (sand grain size, binder content and curing time), through
general regression analysis performed using Minitab V17 statistical software. Variance analyses (ANOVA) were performed to
investigate the signicance of the parameters. The regression
model can determine how a response property changes when a
predictor variable changes. The equations were developed by least
square tting, which is suitable for tting data to a linear model. The
software uses the data obtained from the experiments (Table 2) to
develop these regression equations, which take the following form.

Response = constant + coffecient1 predictor1 + + coefcient

predictorn
An empirical relationship of compression strength (in kg/cm2 )
of mold or core, based on sand grain size, binder content, and curing
time was obtained as follows:
CS = 3.67 0.09G + 5.87B + 2.02T

(1)

here CS is compression strength, which acts as a response variable;

G, B and T are sand grain neness number, percentage of binder content, and curing time, respectively, which act as predictor variables.
The predictor variables can be either continuous or categorical. The
predictive model is based on the sequential sum of squares analysis. It was found that linear terms are sufcient for the compression
strength model (p < 0.05). The sequential F test in ANOVA indicates statistically signicant contribution of sand grain size, binder

H. Khandelwal, B. Ravi / Journal of Manufacturing Processes 22 (2016) 127133

131

hardness and 94% for core shrinkage proves the stability of model
t. These two models were also found to be statistically signicant.
CH = 0.83 0.38G + 27.09B + 7.64T

(3)

SH = 77.38 + 0.80G + 34.13B + 3.80T

(4)

The above empirical relations can be used to estimate the properties of alkyd no-bake bonded molds and cores, for a given set of
values of sand grain size, binder content and curing time.
4. Optimization model and validation
To determine the optimal values of the above parameters,
an optimization model has been formulated. For this purpose,
an objective function was designed to maximize compression
strength, shear strength and hardness, and to minimize shrinkage.
The values of sand grain size, binder content and curing time are
varied within the limits described earlier. The objective function
and the constraints are mathematically represented as follows.
Objective function:
Maximize
f (X) = CS + SS + CH SH

(5)

Subject to

Fig. 6. SEM micrograph shows irregular sand grains bonded with alkyd resin.

content, and curing time in determining mold shrinkage. The stability of the model t was determined by R2 values, which indicate the
total deviation of the response variable from the predictive model.
The adjusted R2 is a measure of the amount of variation around
the mean. In this work, the value of the determination coefcient
R2 = 92%, represents only 8% of the total variation. The high value of
the adjusted R2 (adj. R2 = 87%) also conrms a high signicance of
the predictor variable. Hence the predictive model of compression
strength has only linear terms as shown in Eq. (1). The higher coefcient of binder content in the regression model exhibits a more
signicant effect as compared to sand grain size and curing time.
The empirical relation between shear strength (in kg/cm2 ) as
response; and sand grain size, binder content and curing time as
predictor is given in the following equation:
SS = 7.91 0.06G + 2.05B + 1.50T

(2)

here SS is shear strength acting as the response variable. According

to the sequential sum of squares analysis (p < 0.05), the effect of
predictor variable was found to be signicant. From the analysis
of variance, the determination coefcient (R2 = 84%) was found to
be high and adjusted determination coefcient (adj. R2 = 75%) was
also found to be high. This indicates a good statistical signicance
of the model.
Similar equations were developed for core hardness (CH) and
shrinkage (SH), given in Eqs. (3) and (4). The R2 value of 95% for core

CS = f (G, B, T )

(6)

SS = f (G, B, T )

(7)

CH = f (G, B, T )

(8)

SH = f (G, B, T )

(9)

CH 100

(10)

SH 150

(11)

40 G 80

(12)

1.6 B 2.4

(13)

T 4

(14)

CS, SS, CH, SH and T are non negative

(15)

The constraints numbered 6, 7, 8 and 9 are represented by

regression equations 1, 2, 3 and 4, respectively. Constraint 10
implies that core hardness is measured on a scale of 0 to 100.
The maximum permissible shrinkage is taken as 150 microns (constraint 11). The last three constraints (12, 13 and 14) prescribe the
limits of grain neness number, binder content and curing time.
Since the objective function as well as the constraints are linear,
we can consider this problem as a linear programming problem.
The optimization problem is modelled using AMPL (A Mathematical Programming Language), an algebraic modelling language. The
Gurobi Optimizer was employed to solve the problem. The optimal solution for the above problem is given by: CS = 14.81 kg/cm2 ,
SS = 16.44 kg/cm2 , CH = 81.40 and SH = 51.75 m, at G = 40 GFN,
B = 2.4% and T = 4 h. In this instance, equal importance was given
to all the responses and hence their coefcients in the objective
function were kept as one. Since all four regression equations
cumulatively give a negative coefcient for GFN, the optimum solution for GFN takes the lower bound. In the case of binder and time,
the model gives a cumulative positive coefcient, so their solution
remains in the upper bound. The analytical solution of the model is
the same as the solution obtained from the optimizer. This approach
however, can work only for simple models. For problems with nonequality constraints, the model cannot be solved analytically.
In practice, the four responses may have different levels of
importance. For example, the casting of a light weight precision part
may not require high values of mechanical properties of mold, but

132

H. Khandelwal, B. Ravi / Journal of Manufacturing Processes 22 (2016) 127133

Table 4
Optimum values of process variables and responses.

5. Conclusion

Parameters/
responses

Case 1
(1,1,1,1)

Case 2
(1/2,1/2,1/2,1)

Case 3
(1/3,1/3,1/3,1)

(G) Grain neness number

(GFN)
(B) Binder percentage (wt%
of sand)
(T) Curing time (h)
(CS) Compression strength
(kg/cm2 )
(SS) Shear strength
(kg/cm2 )
(CH) Core hardness
(number)
(SH) Mold shrinkage (m)

40

40

40

2.4

1.6

4
14.81

4
10.11

1
4.04

16.44

14.80

10.30

81.40

59.74

36.81

51.78

24.49

13.07

1.6

its dimensional stability may be critical. In such cases, the responses

can be provided appropriate weights, through their coefcients in
the objective function, as given in Eq. (16).
Maximize f (X) = w1 .CS + w2 .SS + w3 .CH w4 .SH

(16)

here w1 , w2 , w3 and w4 are the weighting factors for compression

strength, shear strength, core hardness and mold shrinkage, respectively. To illustrate further, three cases was designed with different
combination of weights indicated in brackets: case 1 (1,1,1,1), case 2
(1/2,1/2,1/2,1), and case 3 (1/3,1/3,1/3,1). The corresponding objective functions are given in Eqs. (17)(19). The optimal solutions
for the parameters and responses for these cases are reported in
Table 4. It can be seen that in all these cases, the optimal solution for sand grain size takes the lower bound as 40 GFN. In case 2,
where less weightage is given to mechanical properties (compression strength, shear strength, core hardness), the model takes the
lower bound of binder percentages. In case 3, the model takes the
lower bound for all three parameters. The results for the different
cases are reported in Table 4.
f (X1) = CS + SS + CHSH

(17)

f (X2) =

1
1
1
CS + SS + CHSH
2
2
2

(18)

f (X3) =

1
1
1
CS + SS + CH SH
3
3
3

(19)

High quality molds, with the desired values of compression

strength, shear strength, core hardness and shrinkage, are essential
for producing near net shape castings. The experimental investigations enabled studying the effect of molding composition and
curing time on the above properties of chemically bonded sand
molds. The mechanical properties of sand molds or cores were
found to increase with binder content and curing time, but decrease
with sand grain neness. The shrinkage of sand cores increased
with all three parameters (sand grain size, binder percentage, and
curing time). The results were explained in terms of binding and
curing mechanism. The chain reaction of chemicals forms hard
resin bridges, found to be responsible for mechanical properties.
The evaporation of the solvent during curing was found to be
responsible for core shrinkage. Further, the crosslinking mechanism of the core was analysed using SEM analysis. The optimum
mold properties were obtained at 2.4% binder, 40 GFN sand, and
after 4 h of curing time, yielding the optimum compression strength
of 14.80 kg/cm2 , the optimum shear strength of 16.44 kg/cm2 , core
hardness of 81.4 and mold shrinkage of 51 m. The optimization
problem, modelled as multi-objective linear programming problem
enabled obtaining the optimum mold properties for different sets
of requirements. The model was successfully implemented for conducting experiments, showing good correlation between predicted
and observed values (less than 10% error). The model is expected
to be useful for industrial foundries to determine the appropriate
values of molding parameters to achieve the desired values of mold
properties. The work will help to optimize the molding parameters
to obtain high dimensional accurate quality part.
Acknowledgements
This work was carried out in the E-Foundry lab, which was
established with support from the National Knowledge Network
mission of the Government of India, New Delhi. The assistance of
Atul Kumar, Master of Science student in Industrial Engineering and
Operation Research at IIT Bombay is acknowledged in developing
the optimization model. Research associates K.H. Renukananda and
Amit Sata provided useful technical inputs and feedback.
References

The model also enables determining the optimum values of

binder content and curing time for a particular sand grain size. To
validate this, sand of a different grain size (G = 50 GFN) was selected
and the optimum values of binder content and curing time were
obtained from the model, which were 2.4% and 4 h, respectively.
The sand mixture was prepared using the 50 GFN sand and 2.4%
binder, and allowed to cure for 4 h. Samples for all four tests were
prepared and examined. The results derived from the optimization
model and by experiments are reported in Table 5, and the difference was observed to be less than 10%. Thus the usefulness of the
optimization model was conrmed.

Table 5
Comparison between predicted and experimental results.
Parameters/responses

Results using
optimization model

Experimental
results

(G) Grain neness number (GFN)

(B) Binder percentage (wt.% of sand)
(T) Curing time (h)
(CS) Compression strength (kg/cm2 )
(SS) Shear strength (kg/cm2 )
(CH) Core hardness (number)
(SH) Mold shrinkage (m)

50
2.4
4
13.88
15.84
78
59.79

50
2.4
4
14.12
15.28
75
55.6

[1] Ravi B. Metal casting: computer-aided design and analysis. New Delhi: PHI
Learning Pvt. Ltd.; 2005, ISBN 8120327268.
[2] Saikaew C, Wiengwiset S. Optimization of molding sand composition for quality
improvement of iron castings. Appl Clay Sci 2012;6768:2631.
[3] Alonso-Santurde R, Coz A, Viguri JR, Andrs A. Recycling of foundry by-products
in the ceramic industry: green and core sand in clay bricks. Constr Build Mater
2012;27:97106.
[4] Zanetti MC, Fiore S. Foundry processes: the recovery of green moulding sands
for core operations. Resour Conserv Recycl 2003;38:24354.
[5] Campbell J. The concept of net shape for castings. Mater Des 2000;21:37380.
[6] Khandelwal H, Ravi B. 3D printing enabled rapid manufacture of metal parts
at low cost. In: 63rd Indian Foundry Congr. India, Noida: Institute of Indian
Foundrymen; 2015. p. 19.
[7] Surekha B, Rao DH, Rao GKM, Vundavilli PR, Parappagoudar MB. Application of
response surface methodology for modeling the properties of chromite-based
resin bonded sand cores. Int J Mech 2013;7:44358.
[8] Saridikmen H, Kuskonmaz N. Properties of ceramic casting molds produced
with two different binders. Ceram Int 2005;31:8738.
[9] Khandelwal H, Ravi B. Effect of binder composition on the shrinkage of chemically bonded sand cores. Mater Manuf Process 2015;30:146570.
[10] Archibald JJ, Smith RL. Resin binder processes. Casting, ASM Handb
1998;15:21421.
[11] Ayoola WA, Adeosun SO, Sanni OS, Oyetunji A. Effect of casting mould on
mechanical properties of 6063 aluminum alloy. J Eng Sci Technol 2012;7:8996.
[12] Jafari H, Idris MH, Ourdjini A, Karimian M, Payganeh G. Inuence of gating
system, sand grain size, and mould coating on microstructure and mechanical
properties of thin-wall ductile iron. J Iron Steel Res Int 2010;17:3845.
[13] Nwajagu CO, Okafor ICI. A study of the moulding properties of sand bonded by
Ukpor clay. Appl Clay Sci 1989;4:21123.

H. Khandelwal, B. Ravi / Journal of Manufacturing Processes 22 (2016) 127133

[14] Werling JM. Versatile core sand test method. AFS Trans 2004;108:17.
[15] Showman R, Nocera M, Madigan J, Baltz G, Hajduk T, Wohleber J. Re-evaluating
core dimensional changes. Int J Met 2010;4:6374.
[16] Lowe KE, Showman RE. Dimensional changes in chemically bonded molds and
cores. AFS Trans 2011;119:25160.
[17] Kumar S, Satsangi PS, Prajapati DR. Optimization of green sand casting process
parameters of a foundry by using Taguchis method. Int J Adv Manuf Technol
2010;55:2334.
[18] Ganesan T, Vasant P, Elamvazuthi I. Normal-boundary intersection based parametric multi-objective optimization of green sand mould system. J Manuf Syst
2013;32:197205.

133

[19] Kumaravadivel A, Natarajan U. Optimization of sand-casting process variablesa process window approach. Int J Adv Manuf Technol
2012;66:695709.
[20] Niranjan K, Lakshminarayanan PR. Optimization of process parameters for
in situ casting of Al/TiB2 composites through response surface methodology.
Trans Nonferrous Met Soc China 2013;23:126974.
[21] Grabowska B, Sitarz M, Olejnik E, Kaczmarska K. FT-IR and FT-Raman studies of cross-linking processes with Ca2+ ions, glutaraldehyde and microwave
radiation for polymer composition of poly(acrylic acid)/sodium salt of
carboxymethyl starchPart I. Spectrochim Acta A Mol Biomol Spectrosc
2015;135:52935.