G Model

JMP-373; No. of Pages 9 ARTICLE IN PRESS

JournalofManufacturingProcessesxxx(2017)xxx–xxx

Contents lists available at ScienceDirect

Journal of Manufacturing Processes

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

Technical Paper

Investigations on dimensional accuracy of the components prepared

by hybrid investment casting
Parlad Kumar a , Rupinder Singh b,∗ , I.P.S. Ahuja a
a
Department of Mechanical Engineering, Punjabi University, Patiala, Punjab, India
b
Department of Production Engineering, Guru Nanak Dev Engineering College, Ludhiana, Punjab, India

a r t i c l e i n f o a b s t r a c t

Article history: The rapid prototyping technologies (RP) are very useful when a limited number of pieces are promptly
Received12February2017 required like: in making/testing prototypes, design iterations and design optimisations. The fused depo-
Receivedinrevisedform15July2017
sition modelling (FDM) is one of the RP technologies that can use thermoplastic for making sacrificial
Accepted15July2017
patterns used in investment casting (IC) applications. Investment casting, often known as precision
Available online xxx
casting process, is used to produce parts with controlled dimensional tolerances. In the present study,
dimensional accuracy of hip joints prepared with FDM assisted IC process has been optimised using a
Keywords:
design of experiments technique. The effect of process parameters on FDM assisted IC process has been
Fused deposition modelling
Investment casting
evaluated and studied using ANOVA. The results of the present study indicated that hip joint castings are
Additive manufacturing acceptable as per the international tolerance grades.
Rapid prototyping ©2017TheSocietyofManufacturingEngineers.PublishedbyElsevierLtd.Allrightsreserved.
Dimensional accuracy
Tolerance grades

1. Introduction Dimensional accuracy of a component represents the degree of

In the recent past, rapid prototyping (RP) technologies have
gained importance due to their various advantages [1,2]. Fused
Deposition Modelling (FDM) is one of the major RP techniques.
In this method, a three dimensional CAD model is prepared and
given as an input to the FDM machine [3]. The FDM machine pro-
cesses this file and makes the solid Acrylonitrile-butadiene-styrene
(ABS) plastic model. The properties of this plastic model can be
controlled by controlling FDM machine parameters [4]. This plastic
model can be used to replace wax patterns in traditional invest-
ment casting (IC) process without any major modification in the
process [5,6]. The plastic based sacrificial patterns benefit the IC
process by eliminating the expensive tooling required for making
wax patterns. Further, it enables the engineers to make design iter-
ations, readily [7,8]. Therefore, this method is very time and cost
effective when only limited number of pieces of a particular size
are required, such as: in design iterations, missile parts, space craft
parts, patient specific implants etc. The application of FDM based
pattern in IC process to produce metal cast parts can be termed as
rapid investment casting [9]. The FDM assisted IC process can be
used to make functionally graded materials [10].
∗ Corresponding author. Tel.: +91 9872257575; fax: +91 1612490339.
E-mail addresses: pkgarg4050@gmail.com (P. Kumar),
rupindersingh78@yahoo.com (R. Singh), ahujaips@yahoo.co.in (I.P.S. Ahuja).
agreement between the manufactured dimension and its designed
specification [11]. Dimensional accuracy is the most critical fac-
tor in IC and it is directly affected by the accuracy of the pattern.
If the pattern is inaccurate, there are little chances of producing
an accurate casting. But, the dimensional accuracy achieved with
the FDM process is less as compared to conventional wax patterns
[12]. The less accuracy of FDM is due to many factors such as: pro-
cess parameters of the machine [13]; shrinkage of model during the
transition from a semi-liquid state to solid and cooling at solid state
[13,14]; and stair case effect on curved and taper surfaces [15]. The
temperature and build speed during the build process also affects
the accuracy of the parts [16]. Another main factor for dimensional
inaccuracy is layer thickness [17]. In addition to this, there are some
other factors which influence the dimensional accuracy of FDM
independently or in interaction with others [18].
The review of literature reveals that a lot of work has been done
by various researcher to improve the dimensional accuracy of FDM
parts. Masood et al. [19] described a methodology for computing
the volumetric error for any orientation. A mathematical technique
was developed to determine the optimum part orientation of the
parts built by the FDM. The technique was verified by comparing
the results with the experimental measurements of the volumetric
errors of the parts built at different orientations. Gregorian et al.
[16] found an optimal shrinkage compensation factor for improv-
ing the accuracy of FDM models prepared by using FDM-1650
machine. A shrinkage compensation factor of 0.7% was suggested.

http://dx.doi.org/10.1016/j.jmapro.2017.07.008
1526-6125/©2017TheSocietyofManufacturingEngineers.PublishedbyElsevierLtd.Allrightsreserved.

Please cite this article in press as: Kumar P, et al. Investigations on dimensional accuracy of the components prepared by hybrid
investment casting. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.07.008
G Model
JMP-373; No. of Pages 9 ARTICLE IN PRESS
2 P.Kumaretal./JournalofManufacturingProcessesxxx(2017)xxx–xxx

Tong et al. [20] suggested an approach of software error compen-

Identification of Problem
sation to improve the accuracy of a FDM machine by considering
correction of both STL and SSL file formats. The STL file compensa-
tion method was applied to the FDM-3000 machine and it reduced Selection of Benchmarks
the average volumetric error by around 30%. Sood et al. [21] con-
ducted a study for improving dimensional accuracy of FDM built
parts. It is found that the shrinkage of the FDM parts adversely Identification of the process parameters
affect the dimensions. Optimum parameters setting was found out
using Taguchi’s parameter design in order to minimise percentage
Trial runs to establish the range of parameters
change in length, width and thickness of standard test specimen.
Bakar et al. [22] showed that FDM is less accurate when making cir-
cular shape. The researchers recommended that for making small Design of Experiments by using Taguchi L18
parts with a good accuracy and appearance, appropriate values of
contour width and internal raster should be applied, besides apply-
Making Patterns on FDM machine
ing thin layer. Percoco et al. [23] found that shrinkage compensation
factor gives correct results with solid parts but poor results for
hollow and sparse parts. Galantucci et al. [13] verified the adequate- Producing investment castings
ness of the shrinkage compensation factor for shafts and holes. It
was found that shrinkage compensation factor gives better results
on shafts as compared to holes. Noriega et al. [24] proposed a Measurements of dimensions of the castings
method for increasing accuracy of the distance between parallel
faces on FDM manufactured prismatic parts by replacing the theo- Calculation of tolerances grades
retical values of CAD model dimensions by new optimised values.
A reduction in manufacturing error up to 50% was reported. Guo
and Zhou [25] discussed about possible errors like machine error, Analysis of results
STL model error and shrinkage error during FDM pattern making
and silicone mould casting. The authors identified various process
Conclusions
parameters influencing these errors and suggested that CAD model
compensation and process improvement should be used to elim-
Fig. 1. Methodology of the study.
inate the errors. Gurrala and Regalla [26] conducted experiments
for multi-objective optimisation of FDM machine parameters for
optimum strength and volumetric shrinkage. It was found that the the components produced by using hybrid investment casting pro-
strength of the part has to be compromised in order to obtain better cess are consistent with the permissible range of tolerance grades
dimensional stability. Boschetto and Bottini [27] tried to predict the as per ISO standards.
obtainable accuracy of the FDM parts and reported that the dimen-
sional deviations of FDM parts strongly depend upon the deposition 2. Methodology
angle. Volpato et al. [28] found that the dimensional accuracy in the
Z direction of an FDM part can be increased by improving the sur- The various steps of the study are represented in Fig. 1 in the
face quality of the support base. It was reported that reductions in form of a flow chart.
Z-axis error of up to 50% can be achieved. However, the manufac-
turing time and volume of support material required are adversely 2.1. Selection of benchmarks
affected with the proposed configurations. Singh et al. [29] tried to
improve the dimensional accuracy of the FDM assisted castings by Three different parts of practical importance and having differ-
changing the shell moulding process. It was proposed to add some ent volume to surface area ratios have been selected as shown in
fibrous material in the slurry material during shell making in the IC Fig. 1. These parts are the different types of prosthetic biomedical
process. implants that are used in total hip replacement joint of human body.
The different methods suggested by various researchers for Three different grades of stainless steel have been used to make
improving the dimensional accuracy are mainly focused on error these parts by hybrid investment casting process. The drawings of
correction by software and parametric optimisation of FDM these parts have been shown in Fig. 2(a–c).
machine. These methods have been proved to be successful for
improving the dimensional accuracy of FDM models. But for IC 2.2. Identification of process parameters
applications the dimensional accuracy can be further increased by
the hybridisation of conventional wax pattern based investment The various input factors that may influence the dimensional
casting with the modern plastic pattern based investment casting. accuracy have been identified. These factors are: type of pattern,
In this method, instead of using only wax or plastic pattern, a wax volume to area ratio of the casting, orientation of pattern in FDM
coated plastic pattern can be used in investment casting process machine, density of the pattern, mould thickness and grade of
and this process can be termed as hybrid investment casting. the material. For experimentation, two or three levels have been
In this paper, it is proposed that a thin layer of wax on plastic selected for each input factor. The factors and their levels have been
pattern will reduce inaccuracy due to staircase effect and shrinkage. decided on the basis of preliminary study and pilot experiments.
The effect of some other casting process parameters like: volume The list of various input factors along with their levels is given in
to area ratio of the casting; orientation of the part during pattern Table 1.
making; density of the pattern; mould wall thickness; and grade A brief description of these input factors has been given below:
of material has also been analysed on the dimensional accuracy
of the castings. In order to get the best dimensional accuracy, the (i) Type of pattern: Two types of patterns have been selected. The
optimisation of process parameters has also been done and it has first pattern is made of ABS plastic and coated with a thin layer
been checked that whether international tolerance (IT) grades of of wax whereas the second is uncoated ABS plastic pattern.

Please cite this article in press as: Kumar P, et al. Investigations on dimensional accuracy of the components prepared by hybrid
investment casting. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.07.008
G Model
JMP-373; No. of Pages 9 ARTICLE IN PRESS
P.Kumaretl/JonfMcigsx(2017)–
Fig. 2. Drawings of three benchmarks.
(ii) Volume to Area (V/A) ratio: This ratio is obtained by dividing
the volume of the casting by the surface area of the casting. The
volume to area ratio is a factor which affects the cooling time of
the casting. Three different parts as shown in Fig. 2 have been
chosen with three different V/A ratios. The calculated values
of three V/A ratios are 2.75, 3.34 and 4.25 for the drawing one,
two and three, respectively.
(iii) Orientation of the pattern: Orientation of making the pattern
in FDM machine also affects the dimensional accuracy of the
pattern. Three different orientations 0◦ , 45◦ and 90◦ have been
selected.
(iv) Density of the pattern: The FDM machine can create solid or
hollow parts according to the requirement. Three different pat-
tern interiors i.e. low density, high density and solid have been
selected on the basis of availability in FDM machine.
(v) Mould thickness: Mould thickness is prepared by coating a
number of slurry layers on the casting tree. The increase in
number of slurry layers would increase the thickness of the
mould. The mould thickness may affect the cooling rate of
the casting. There is no direct way to control the thickness
of the mould except controlling the number of slurry layers.
Seven, 8 and 9 layers have been selected as three levels for this
factor.
(vi) Grade of material: In order to study any possible effect of
different grades of material on dimensional accuracy three dif-
ferent grades of metal have been selected. In the present study,
SS304L, SS316 and SS304 have been selected as three grades of
stainless steel.
2.3. Design of experiments
In order to reduce the number of experiments, Taguchi’s ortho-
gonal array has been used for the design of experiments. In this
study, the sum of number of degrees of freedom for the selected six
parameters is 11 and there is one factor with two levels and five fac-
tors with three levels. Therefore, the appropriate orthogonal array
for this case is L18 (21 × 37 ). This array consists of 8 columns for
assigning factors or interactions and 18 rows for designating the
Table 1
Different input factors and their levels.
S. no. Input factor Designation Levels
Level-1
1 Type of pattern A Wax coated ABS
2 Volume to area ratio B 2.75
3 Orientation of the pattern C 0◦
4 Density of the pattern D Low
5 Mould thickness (no. of layers) E 7
6 Grade of material F SS304L
Please cite this article in press as: Kumar P, et al. Investigations on dimensional accuracy of the components prepared by hybrid
investment casting. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.07.008
3
trial or experiment conditions. The assignment of factors is made
based on linear graph as shown in Fig. 3.
Each dot in the linear graph represents the factor column num-
ber and line joining two dots represents the interaction between the
factors assigned to these columns. Numbers on line shows the col-
umn numbers to which the interaction can be assigned. In this study
factor A has two levels, therefore it has been assigned to column
number 1. The other factors have been assigned to next columns as
indicated by linear graph in Fig. 3. The interaction between factor
A and B has been studied. The final L18 orthogonal array with fac-
tors and their selected levels used for the experiments is shown in
Table 2.
2.4. Praparation of FDM patterns
Three dimensional solid models of the parts (Fig. 2) were
prepared using Solidworks Office Premium 2010 software and
exported as STL file to FDM machine software. The FDM machine
used for this purpose was Stratasys uPrint SE 3D printer. The mate-
rial used for making patterns was ABS plastic supplied by Stratasys.
The layer thickness in FDM machine was kept at 0.254 mm for all
the pieces. The 18 patterns were prepared according to different
trial runs on an FDM machine. Out of the eighteen patterns, nine
were preheated with hot air then coated with thin layer of wax by
just dipping and removing from a pool of molten wax maintained
at 100 ◦ C. Nine sprues were also prepared by using FDM machine.
A liquid adhesive was used to attach two patterns on each sprue as
shown in Fig. 4.
2.5. Investment casting
All the 18 patterns were coated with different number of slurry
layers according to the trial runs of L18 array. The patterns were
burnt out in a furnace at 1080 ◦ C to get moulds for pouring the
molten metal. Different grades of metal were poured in the moulds
to get 18 pieces of castings as shown in Fig. 5. Three repetitions of
experiments were conducted in order to check the repeatability of
the experiments and to minimise the error.
Degree of freedom
Level-2 Level-3
Uncoated ABS –
3.34 4.25 2
45◦ 90◦ 2
High Solid 2
8 9 2
SS316 SS304 2
G Model
JMP-373; No. of Pages 9 ARTICLE IN PRESS
4 P.Kumaretal./JournalofManufacturingProcessesxxx(2017)xxx–xxx

1 2 3 4 5 6 7 8

(A) (B) (C) (D) (E) (F)

Fig. 3. Linear graph for L18 .

Table 2
L18 experimental design for selected input parameters and their levels.

Exp. no. A B C D E F
Type of pattern V/A ratio Orientation (in degree) Density of pattern Mould thickness Grade of material

1 Wax Coated 2.75 0 Low 7 SS 304 L

2 Wax Coated 2.75 45 High 8 SS 316
3 Wax Coated 2.75 90 Solid 9 SS 304
4 Wax Coated 3.34 0 Low 8 SS 316
5 Wax Coated 3.34 45 High 9 SS 304
6 Wax Coated 3.34 90 Solid 7 SS 304 L
7 Wax Coated 4.25 0 High 7 SS 304
8 Wax Coated 4.25 45 Solid 8 SS 304 L
9 Wax Coated 4.25 90 Low 9 SS 316
10 Uncoated 2.75 0 Solid 9 SS 316
11 Uncoated 2.75 45 Low 7 SS 304
12 Uncoated 2.75 90 High 8 SS 304 L
13 Uncoated 3.34 0 High 9 SS 304 L
14 Uncoated 3.34 45 Solid 7 SS 316
15 Uncoated 3.34 90 Low 8 SS 304
16 Uncoated 4.25 0 Solid 8 SS 304
17 Uncoated 4.25 45 Low 9 SS 304 L
18 Uncoated 4.25 90 High 7 SS 316

3. Results and discussion
The important dimensions of all the fifty four castings were mea-
sured by using a CMM. One common dimension (thickness) in all
the specimens has been identified and compared with the dimen-
sion on drawing in Fig. 2. In this way dimensional deviation for
each casting was calculated. Dimensional deviation has been taken
to represent the dimensional accuracy. Therefore, less dimensional
deviation means more dimensional accuracy.
Signal to noise (S/N) ratio is used to determine the influence and
variation caused by each factor and interaction relative to the total
variation observed in the result. The advantage of using S/N ratio is
that it uses a single measure, mean square deviation (MSD), which
incorporates the effect of changes in mean as well as the varia-
tion (standard deviation) with equal priority. Moreover, the results
behave linearly when expressed in terms of S/N ratios. The linear
behaviour of the results is an assumption necessary to express per-
formance in optimal conditions [21]. The objective of the study is
to get the minimum dimensional deviation and to get the maxi-
mum dimensional accuracy, therefore, S/N ratios for dimensional
deviations were calculated on the basis of ‘lower is better’ approach
which is logarithmic function based on MSD, given by the following
equation:
 
1 2
r
S
= −10 log (MSD) = −10 log yi (1)
NLB r
i=0

where MSD is the mean square deviation, yi is the observed value

of the response characteristic, r is the number of repetitions.
The experimental analysis is made by using MINITAB 16 sta-
tistical software package [31]. The dimensional deviation and
calculated signal to noise ratio for dimensional deviation has been
shown in Table 3. The value of S/N ratio is more for less dimensional
deviation values.

3.1. Analysis of S/N ratio

Fig. 4. Patterns attached with sprue.
Mean signal-to-noise ratios for all the factors and all the lev-
els have been calculated and shown in Table 4. The parameters
are ranked according to their effect on dimensional accuracy with
higher ranks given to the parameters having higher delta value.
Higher is the rank, higher is the significance. The delta value is
calculated by finding the difference between the maximum and

Please cite this article in press as: Kumar P, et al. Investigations on dimensional accuracy of the components prepared by hybrid
investment casting. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.07.008
G Model
JMP-373; No. of Pages 9 ARTICLE IN PRESS
P.Kumaretl/JonfMcigsx(2017)– 5

Fig. 5. Specimens after casting.

Table 3
S/N ratio for dimensional deviation.

Exp. no. Dimensional deviation (mm) S/N ratio (dB)

R1 R2 R3 Average

1 0.129 0.131 0.130 0.130 17.721

2 0.110 0.111 0.109 0.110 19.172
3 0.048 0.050 0.052 0.050 26.016
4 0.111 0.110 0.110 0.110 19.146
5 0.047 0.052 0.051 0.050 26.013
6 0.039 0.041 0.041 0.040 27.884
7 0.089 0.090 0.091 0.090 20.915
8 0.020 0.020 0.025 0.022 33.233
9 0.011 0.010 0.010 0.010 39.706
10 0.320 0.340 0.330 0.330 9.627
11 0.230 0.245 0.245 0.240 12.392
12 0.070 0.080 0.070 0.073 22.676
13 0.330 0.340 0.330 0.333 9.542
14 0.163 0.170 0.160 0.164 15.683
15 0.020 0.020 0.030 0.023 32.467
16 0.105 0.103 0.100 0.103 19.770
17 0.051 0.052 0.050 0.051 25.847
18 0.025 0.026 0.025 0.025 31.925

Average 22.763

Table 4
Response table for mean S/N ratios of dimensional accuracy.

Level A B C D E F
Type of pattern V/A ratio Orientation Density of pattern Mould thickness Grade of material

1. 25.53* 17.93 16.12 24.55* 21.09 22.82

2. 19.99 21.79 22.06 21.71 24.41* 22.54
3. – 28.57* 30.11* 22.04 22.79 22.93*
Delta 5.54 10.63 13.99 2.84 3.32 0.39
Rank 3 2 1 5 4 6
*
Represents higher S/N ratio.

Please cite this article in press as: Kumar P, et al. Investigations on dimensional accuracy of the components prepared by hybrid
investment casting. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.07.008
G Model
JMP-373; No. of Pages 9 ARTICLE IN PRESS
6 P.Kumaretal./JournalofManufacturingProcessesxxx(2017)xxx–xxx
Fig. 6. Main effect plots for mean S/N ratios.
minimum values of mean S/N ratios for each factor. Mean is calcu-
lated by using the following equation:
X1 + X2 + X3 + · · · + Xi · · · + Xn
A= (2)
n
where DN is the S/N ratio of ith experimental run, DM is the mean
of S/N ratios, D is the number of experiments.
Perusal of Table 4 represents that orientation of the pattern in
FDM machine is having the highest rank 1 and is the most signifi-
cant factor followed by the V/A ratio with rank 2 and type of pattern
with rank 3. The mould thickness, density of the pattern and grade
of material have been ranked at 4, 5 and 6, respectively. The effect
of different input factors on mean S/N ratios is shown in Fig. 6. The
higher value of S/N ratio represents less dimensional deviation or
more dimensional accuracy.
Fig. 6 shows that the dimensional accuracy increases when wax
coated plastic pattern is used instead of non-coated plastic pattern.
This is due to the fact that the coating of wax increases the size of the
pattern slightly and this increase in dimension balances the shrink-
age loss at the stage of pattern making and solidification. Therefore,
wax coated plastic patterns are better than uncoated plastic pat-
terns in terms of dimensional accuracy. It is also found that the
dimensional accuracy increases with increase in V/A ratio. It may
be due to the fact that as the V/A ratio increases the solidification
time of the casting also increases [32]. Generally, directional solidi-
fication can be achieved with proper solidification time and this will
enable the feeding channels of riser during the last few seconds of
casting solidification to back support the shrinkages produced. The
orientation of pattern during FDM process also plays very impor-
tant role. The maximum dimensional accuracy was observed at 90◦
orientation while the same was minimum at 0◦ orientation. It is the
limitation of the FDM machine that it cannot produce same dimen-
sional accuracy for all orientations. The effect of density of pattern
is less. However, the low density pattern gives better dimensional
accuracy as compared to high density or solid patterns. This may be
due to the fact that with high density patterns more ash is formed
inside the cavity and some remaining ash can decrease the dimen-
sional accuracy. The effect of mould thickness is also less, however
the little effect may be due to the change on cooling rate with
change in shell thickness. The effect of type of grade material is
Please cite this article in press as: Kumar P, et al. Investigations on dimensional accuracy of the components prepared by hybrid
investment casting. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.07.008
negligible because there is only small difference in physical prop-
erties of all the three grades and probably this has not affected the
dimensional accuracy. The interaction plot for factor A and B has
been shown in Fig. 7. It represents parallel lines which represents
no interaction between factor A and B.
3.2. Analysis of variance
Results obtained from the experiments were analysed using
ANOVA which helps in predicting the significance of input factors
taken for any desired response function. It indicates the most influ-
encing factors. The response function in this work is dimensional
accuracy. The knowledge of the contribution of individual factor
taken is very important for the control of the final response. For
this, the sum of squares, degree of freedom, variance, F-Ratio and
percentage contribution of each factor are calculated. The sum of
squares is a measure of the deviations of the experimental data
from the mean value of data. The percentage contribution of a factor
defines its significance rate on the dimensional deviation. Normally,
the higher the percentage contribution the more will be the effect
of the input parameter on the output response. By controlling the
factors with high contribution, the total variation can be reduced
leading to improvement of process performance.
ANOVA was carried out for obtained S/N ratios and the results
are shown in Table 5. The percentage contribution of various input
parameters is also calculated and shown.
Table 5 shows the calculated F-values for all the factors. Larger
F-value indicates that the variation of the process parameter makes
a big change on the performance characteristics. It shows that the
factor A, B and C are significant at 5% (P < 0.05) level of significance.
It can be observed from table that the percentage contribution of
factor A is 11.51% whereas the percentage contributions of factor B
and C are approximately 29% and 49%, respectively. Percentage con-
tribution represents that the influence of orientation of the pattern
in FDM is significantly higher than the other parameters followed
by V/A ratio and type of pattern. The error variation due to uncon-
trollable parameters is approximately 5%. The factors D, E and F are
not statistically significant. There effect of interaction of factor A
and B is also not significant.
G Model
JMP-373; No. of Pages 9 ARTICLE IN PRESS
P.Kumaretl/JonfMcigsx(2017)– 7

Interaction Plot for SN ratios

Data Means

1 2 3
A
30
1
2
25
A
20

15
B
30
1
2
25 3
B
20

15
1 2

Signal-to-noise: Smaller is better

Fig. 7. Interaction plot for factor A and B.

Table 5
Results of ANOVA for dimensional accuracy.

Source DOF Sum of squares Variance F P Percentage contribution

A 1 138.21 138.210 9.29 0.038 11.51

B 2 347.66 173.828 11.68 0.021 28.96
C 2 591.85 295.924 19.89 0.008 49.30
D 2 28.95 14.477 0.97 0.453 2.41
E 2 33.15 16.577 1.11 0.412 2.76
F 2 0.47 0.236 0.02 0.984 0.04
A×B 2 0.71 0.354 0.02 0.977 0.06
Residual error 4 59.51 14.877 4.96

Total 17 1200.51 100

3.3. Optimum set of parameters
The higher the signal to noise ratio, the more favourable is the
effect of input variable on the output. From Table 4, the optimum
levels (factor levels with highest S/N ratio) have been identified for
the significant factors in order to achieve maximum dimensional
accuracy. These optimum levels and their corresponding values are
shown in Table 6.
Perusal of Table 6 shows that the best dimensional accuracy can
be achieved if the FDM pattern is wax coated; volume to area ratio of
the casting is high and orientation of pattern in FDM machine is 90◦ .
The effect of other factors like density of pattern, mould thickness
and grade of material is statistically insignificant and their levels
can be selected on the basis of economic criteria.
confirmatory experiment is highly recommended to verify the
experimental results. The purpose of the confirmation experiment
is to validate the conclusions drawn during the analysis phase. The
confirmation experiment is performed by conducting a test with
specific combination of the factors and levels previously evaluated.
In this study, two confirmatory experiments were conducted by
using the optimal levels of the optimal process parameters.
In this study there are six input factors and only three are sig-
nificant. From Table 5, it can be seen that factors C, B and A are
most significant. The predicted optimum response in terms of S/N
ratio (pre ) is given by using Eq. (3) [21,33]. In this equation ¯ is
the average of S/N ratio for all the observations (Table 3), whereas
Ā1 , B̄3 , and C̄3 are average response (S/N ratio) for factor A, B and C
at respective level of 1, 3, 3, respectively (Table 6).

    
3.4. Confirmatory experiments pre = ¯ + C̄3 − ¯ + B̄3 − ¯ + Ā1 − ¯
      
pre = C̄3 + B̄3 + Ā1 − 2¯
The experimental confirmatory test is the final step in verify-
ing the results drawn based on Taguchi’s design approach. The (3)
= (30.11) + (28.57) + (25.53) − 2 (22.76)
= 85.65 − 46.18
Table 6 = 38.69 dB
Optimum set of parameters for maximum dimensional accuracy.

Input factor Highest mean Optimum Optimum value By using Eq. (1), predicted dimensional deviation = 0.0116 mm.
S/N ratio level Confirmatory experiments were performed by using the
Type of pattern 25.53 A1 Wax coated obtained set of optimum factor values. The confirmation test values
V/A ratio of the casting 28.57 B3 4.25 of the dimensional deviation are 0.010 and 0.012 mm, close to the
Orientation of pattern 30.11 C3 90◦ predicted value.

Please cite this article in press as: Kumar P, et al. Investigations on dimensional accuracy of the components prepared by hybrid
investment casting. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.07.008
G Model
JMP-373; No. of Pages 9 ARTICLE IN PRESS
8 P.Kumaretal./JournalofManufacturingProcessesxxx(2017)xxx–xxx

Table 7
Calculations of IT grades according to ISO standards.

Exp no. Drawing no. Nominal Dia (Dn ) in mm Measured Dia (Dm ) in mm Tolerance unit (i) Tolerance value (n) in ␮m IT grade

1 1 12.000 11.667 1.0827 307.872 IT14

2 1 12.000 11.627 1.0827 344.817 IT14
3 1 12.000 11.593 1.0827 375.604 IT14
4 2 19.500 19.063 1.3074 333.996 IT14
5 2 19.500 18.933 1.3074 433.430 IT14
6 2 19.500 18.887 1.3074 469.124 IT14
7 3 35.250 34.287 1.5612 617.047 IT14
8 3 35.250 34.337 1.5612 585.020 IT14
9 3 35.250 34.310 1.5612 602.101 IT14
10 1 12.000 11.640 1.0827 332.502 IT14
11 1 12.000 11.550 1.0827 415.628 IT14
12 1 12.000 11.550 1.0827 415.628 IT14
13 2 19.500 18.937 1.3074 430.881 IT14
14 2 19.500 18.827 1.3074 515.017 IT14
15 2 19.500 18.990 1.3074 390.087 IT14
16 3 35.250 34.573 1.5612 433.427 IT14
17 3 35.250 34.287 1.5612 617.047 IT14
18 3 35.250 34.287 1.5612 617.047 IT14

4. Calculations of tolerance grades
The results of dimensional measurements have been used to
evaluate the tolerance unit n that derives starting from the standard
tolerance factor i, defined in ISO standard UNI EN 20286-I (1995).
In fact, the standard tolerance are not evaluated separately for each
nominal size, but for a range of nominal size, for the generic nomi-
nal dimension (DN ), the number of tolerance unit n is evaluated as
follows [34–36].
n = 1000 (DN − DM ) /i
i = 0.45 (D) 1/3 + 0.001 D
where DN is the is a nominal dimension, DM is the is a measured
dimension, i is the is a tolerance factor, D is the is the geometric
mean of the limiting values of basic step.
In present case the diameter of round portion on the drawings
(Fig. 2) has been taken as nominal dimension. For experiment run
1, the standard basic step is 10–18 mm. Therefore,
D = (10 × 18) 1/2 = 13.42 mm
i = 0.45 (13.42) 1/3 + 0.001 (13.42)
= 1.0827
n = 1000 (12 − 11.667) /1.0827
= 307.872
Similarly, the values of n for all the experimental runs were
calculated and the corresponding IT grades have been obtained as
shown in Table 7.
It has been observed in Table 7 that the measured diameter is
less than the nominal diameter for all the specimens. This is due
to the fact that in hybrid IC process there is two stage shrinkage.
First, some shrinkage occurs in the FDM model during the transition
from a semi-liquid state to solid state and cooling at solid state
[16,23]. Second, during solidification of the casting some shrinkage
of metal occurs [13]. However, in order to get the desired dimension
and to improve the process the shrinkage allowance can be added
in the dimension of the pattern. The IT grades of the components
produced by using hybrid investment casting process are consistent
with the permissible range of tolerance grades as per ISO standard
UNI EN 20286-I (1995).
5. Conclusions
In this study, the relationship between dimensional accuracy
and various process parameters of hybrid investment casting has
been investigated. The FDM process has been used to prepare the
investment casting patterns by using ABS plastic. These patterns
were coated with a layer of wax and used for investment casting to
make a biomedical implant. It has been found that:
(1) A thin coating on plastic patterns can improve the dimensional
accuracy of the investment castings made by using FDM based
patterns.
(2) The geometries with more volume to area ratios are more
accurate as compared to less volume to area ratios. For good
dimensional accuracy the orientation of the pattern in FDM
machine should be 90◦ .
(3) The IT grades of the components produced by using hybrid
investment casting process are consistent with the permissible
range of tolerance grades as per ISO standard UNI EN 20286-I
(1995).
Acknowledgements
The authors are thankful to UGC (Minor Research Project), New
Delhi for providing research grant under Minor Research Project
scheme. The authors are also thankful to Manufacturing Research
Lab (GNDEC, Ludhiana) and Department of Mechanical Engineer-
ing, Punjabi University, Patiala for providing research facilities. The
authors are also thankful to Mr. Puran Singh (lab technician) and
Er. Sunpreet Singh (SRF) for providing their technical support.
References
[1] Williams CB, Mistree F, Rosen DW. A functional classification framework for
the conceptual design of additive manufacturing technologies. J Mech Des
2011;113:1–11.
[2] Wei C, Dong J. Hybrid hierarchical fabrication of three-dimensional scaffolds. J
Manuf Process 2014;16:257–63.
[3] Greul M, Pintat T, Greulich M. Rapid prototyping of functional metallic parts.
Comput Ind 1995;28:23–8.
[4] Li L, Sun Q, Bellehumeur C, Gu P. Composite modeling and analysis for fabri-
cation of FDM prototypes with locally controlled properties. J Manuf Process
2002;4:129–41.
[5] Upcraft S, Fletcher R. The rapid prototyping technologies. Assembly Autom
2003;23:318–30.
[6] Kumar P, Ahuja IPS, Singh R. Application of fusion deposition modelling for
rapid investment casting—a review. Int J Mater Eng Innov 2012;3:204–27.
[7] Chua CK, Chew TH, Eu KH. Integrating rapid prototyping and tooling with vac-
uum casting for connectors. Int J Adv Manuf Technol 1998;14:617–23.
[8] Chua CK, Feng C, Lee CW, Ang GQ. Rapid investment casting: direct and indirect
approaches via model maker II. Int J Adv Manuf Technol 2005;25:26–32.
[9] Cheah CM, Chua CK, Lee CW, Feng C, Totong K. Rapid prototyping and tooling
techniques: a review of applications for rapid investment casting. Int J Adv
Manuf Technol 2005;25:308–20.
[10] Singh S, Singh R. Wear modelling of Al–Al2 O3 functionally graded material
prepared by FDM assisted investment castings using dimensionless analysis. J
Manuf Process 2015, http://dx.doi.org/10.1016/j.jmapro.2015.01.007.

Please cite this article in press as: Kumar P, et al. Investigations on dimensional accuracy of the components prepared by hybrid
investment casting. J Manuf Process (2017), http://dx.doi.org/10.1016/j.jmapro.2017.07.008
G Model
JMP-373; No. of Pages 9 ARTICLE IN PRESS
P. Kumar et al. / Journal of Manufacturing Processes xxx (2015) xxx–xxx
[11] Farzadi A, Waran V, Solati-Hashjin M, Rahman ZAA, Asadi M, Osman NAA.
Effect of layer printing delay on mechanical properties and dimensional accu-
racy of 3D printed porous prototypes in bone tissue engineering. Ceram Int
2015;41:8320–30.
[12] Rahmati S, Akbari J, Barati E. Dimensional accuracy analysis of wax patterns cre-
ated by RTV silicon rubber molding using Taguchi approach. Rapid Prototyping
J 2007;13:115–22.
[13] Galantucci LM, Lavecchia F. Direct digital manufacturing of ABS parts: an
experimental study on effectiveness of proprietary software for shrinkage com-
pensation. Int J Digital Content Technol Appl 2012;6:546–55.
[14] Macků M, Horáček M. Applying RP-FDM technology to produce prototype
castings using the investment casting method. Arch Foundry Eng 2012;12:
75–82.
[15] Idris MH, Sharif S, Harun WSW. Evaluation of ABS patterns produced from
FDM for investment casting process. In: Proceedings of the Ninth Asia
Pacific Industrial Engineering & Management Systems Conference. 2008.
p. 1299–304.
[16] Gregorian A, Elliott B, Navarro R, Ochoa F, Singh H, Monge E, et al. Accuracy
Improvement in Rapid Prototyping Machine (FDM-1650). In: Solid Freeform
Fabrication Proceedings, 12th, Solid Freeform Fabrication Symposium. 2001. p.
77–84.
[17] Ahn D, Kweon JH, Kwon S, Song J, Lee S. Representation of surface roughness
in fused deposition modeling. J Mater Process Technol 2009;209:5593–600.
[18] Garg A. A Fuzzy Taguchi Approach for Improving Dimensional Accuracy of
Fused Deposition Modeling (FDM) Built Parts. Rourkela, India: Department of
Mechanical Engineering, NIT; 2005 (B.Tech. Thesis).
[19] Masood SH, Rattanawong W, Iovenitti P. Part build orientations based on
volumetric error in fused deposition modeling. Int J Adv Manuf Technol
2000;16:162–8.
[20] Tong K, Joshi S, Lehtihet EA. Error compensation for fused deposition modeling
(FDM) machine by correcting slice files. Rapid Prototyping J 2008;14:4–14.
[21] Sood AK, Ohdar RK, Mahapatra SS. Improving dimensional accuracy of fused
deposition modelling processed part using grey Taguchi method. Mater Des
2009;30:4243–52.
Please cite this article in press as: Kumar P, et al. Investigations on dimensional accuracy of the components prepared by hybrid
investment casting. J Manuf Process (2015), http://dx.doi.org/10.1016/j.jmapro.2015.07.008
9
[22] Bakar NSA, Alkahari MR, Boejang H. Analysis on fused deposition modelling
performance. J Zhejiang Univ Sci, A 2010;11:972–7.
[23] Percoco G, Galantucci LM, Lavecchia F. Validation study of an analytical model
of FDM accuracy. In: Katalinic B, editor. DAAAM International Scientific Book.
Vienna, Austria: Published by DAAAM International; 2011. p. 585–92 (Chapter
48).
[24] Noriega A, Blanco D, Alvarez BJ, Garcia A. Dimensional accuracy improvement
of FDM square cross-section parts using artificial neural networks and an opti-
mization algorithm. Int J Adv Manuf Technol 2013;69:2301–13.
[25] Guo H, Zhou Y. Research on the impacting factors of dimensional accu-
racy with silicone mold casting products based on FDM. Adv Mater Res
2014;912–914:309–13.
[26] Gurrala PK, Regalla SP. Multi-objective optimisation of strength and volumetric
shrinkage of FDM parts. Virtual Phys Prototyping 2014;9:127–38.
[27] Boschetto A, Bottini L. Accuracy prediction in fused deposition modeling. Int J
Adv Manuf Technol 2014;73:913–28.
[28] Volpato N, Foggiatto JA, Schwarz DC. The influence of support base on FDM
accuracy in Z. Rapid Prototyping J 2014;20:182–91.
[29] Singh R, Singh S, Mahajan V. Investigations for dimensional accuracy of invest-
ment casting process after cycle time reduction by advancements in shell
moulding. Proc Mater Sci 2014;6:859–65.
[31] MINITAB 16 Statistical software package, User Manual. Copyright © 2015
Minitab Inc.
[32] Flemings MC. Solidification process. Metallur Trans 1974;5:2134.
[33] Krishnaiah K, Shahabudeen P. Applied Design of Experiments and Taguchi
methods. 4th ed. New Delhi: PHI Learning Pvt. Ltd; 2012. p. 248.
[34] Gill SS, Kaplas M. Efficacy of powder-based three-dimensional printing (3DP)
technologies for rapid casting of light alloys. Int J Adv Manuf Technol
2011;52:53–64.
[35] Singh R. Process capability study of polyjet printing for plastic components. J
Mech Sci Technol 2011;25:1011–5.
[36] Gill PS. A Text Book of Machine Drawing. 18th ed. New Delhi: S.K. Kataria and
Sons; 2014. p. 288.