Journal of the Institute of Engineering, 2018, 14(1): 170-178

170 TUTA/IOE/PCU © TUTA/IOE/PCU

Printed in Nepal

Optimization of Process Parameters of Abrasive Jet Machining

on Hastelloy through Response Surface Methodology

D. V. Sreekanth1, M. Sreenivasa Rao2

1
Department of Mechanical Engineering, St. Martin's Engineering College, Secunderabad, India
2
Department of Mechanical Engineering, JNT University, Hyderabad, India
Corresponding author: [email protected]

Received: Dec 24, 2017 Revised: Jan 05, 2018 Accepted: Jan 7, 2018

Abstract: The demand for micro products is rapidly increasing in chemical,

marine and aerospace industries. The super alloys which have high strength and
corrosion resistance properties play a major role. Hastelloy which is difficult to
machine by the conventional processes can be machined by using AJM. Hastelloy
C276 sheet of thickness 1mm has been drilled on the AJM test rig using variable
process parameters. In this paper optimization of process parameters of Abrasive
Jet Machining of Hastelloy C276 by RSM methodology is presented. The values
obtained in RSM Analysis were compared with the Analysis of Variance (ANOVA).
Various levels of experiments are conducted using L15 orthogonal array for both
MRR and KERF.

Key words: AJM, Hastelloy C276, MRR, KERF, RSM

1. Introduction
Hastelloy C276 is a Nickel-chromium-molybdenum fashioned combination that is viewed as
the most flexible consumption safe compound accessible. This amalgam is impervious to the
arrangement of grain limit accelerates in the weld heat-influenced zone, in this way making it
reasonable for most concoction procedure applications in an as welded condition. Combination
C-276 likewise has superb imperviousness to setting, stress-erosion making and oxidizing
environments laugh uncontrollably to 1900°F. AJM is a mechanical energy based unconventional
machining process used to remove unwanted material from a given work piece. Material fracture
occurs due to the impact of high velocity air/ gas stream of abrasive particles on the work piece
[6]. Gas used is carbon dioxide or nitrogen or compressed air. The selection of abrasive particles
depends on the hardness and Metal Removal Rate (MRR) of the work piece. Most commonly,
aluminum oxide or silicon carbide particles are used [4]. Abrasive Jet Machining is used for drilling,
deburring, etching, and cleaning of hard and brittle metals, alloys, and non-metallic materials [2].
There are no toxic wastes given off by abrasive water jets, and no oils are necessary in the process
of machining [5]. The Major Process Parameters that affects the MRR & KERF in AJM are Gas
Pressure, Nozzle diameter, Abrasive mass flow rate, Nozzle tip distance [3].
 Sreekanth and Rao 171

Fig.1: Hole generated by Abrasive jet machining process

Fig. 2: WC Nozzles of various sizes

The material removal rate (MRR) can be defined as the volume of material removed divided by the
machining time. Another way to define MRR is to imagine an “instantaneous" material removal
rate as the rate at which the cross-section area of material being removed moves through the work
piece [1]. The formula used to calculate MRR is
ρπd2t
MRR= (1)
4z
The cut or the hole generated in AJM with a width is called as KERF. The top kerf is wider than
the bottom kerf. For maintaining proper KERF the optimal standoff distance is to be maintained.
The formula for kerf is
KERF = Diameter hole at the top-Diameter of the hole at bottom (2)
2. Methodology
Experimental design (DOE) is a useful complement to multivariate data analysis because it
generates “structured" data tables, i.e. data tables that contain an important amount of structured
variation. This underlying Structure will then be used as a basis for multivariate modeling,
which will guarantee stable and robust models. The DOE technique helps to study many factors
simultaneously and most economically.
Response surface methodology (RSM) is a combination and collection of mathematical and
statistical techniques for empirical model building. By careful design of experiments, the objective
is to optimize a response (output variable) which is influenced by several independent variables
(input variables). An experiment is a series of tests, called runs, in which changes are made in
the input variables in order to identify the reasons for changes in the output response. Two basis
methods used in RSM are central composite and Box-Behnken methods. In the present optimization
Box-Behnken design is used, which is a type of response surface design that does not contain an
embedded factorial or fractional factorial design. For finding the response surface regressions and
predictions MINI TAB software is used.
172 Optimization of Process Parameters of Abrasive Jet Machining on Hastelloy through Response Surface Methodology

3. Experimentation Work
The experimentation was carried out in St. Martin's engineering college on the AJM test rig. The
experiment was done keeping constant pressure of 8kg/cm2. This has been done as Hastelloy, which
is very difficult to be machined at lower pressures. The experimental setup consists of the major
components like Air compressor, Air filter, Pressure Regulator and Pressure gauge, Dehumidifier,
Mixing Chamber, Nozzle and Arrangement to hold the work piece etc [9].
The variable levels that are considered based on the parameters are, for first variable Abrasive
flow rate three levels are taken they are 3.5, 4.5, 5.5 and the units are (g/min), for variable standoff
distance three levels are taken (8, 9,10 mm) and for variable Nozzle diameter the three levels taken
are (2, 3,4 mm). Pressure of air, Size and Type of abrasive and are kept constant [8]. Experiments
are carried out based on Response surface methodology is Box–Behnken design. The material used
for experimentation is Hastelloy sheet and the Material Removal Rate (MRR) process parameter
is the measure of performance. The abrasives used are Silicon Carbide of size 40 microns. The
analysis was done using Minitab-17.
Table 1: Machining parameters and their levels

LEVELS AFR (gm/min) SOD (mm) ND (mm)

1 3.5 7 2
2 4.5 8 3
3 5.5 9 4

3.1 Response Surface Regression for Metal Removal Rate

The metal removal rate is calculated based on the standard formulas and obtained the MRR values
for 15 experiments based on random order priority. The run orders and the standard orders are
established based on design of experiments. The Box-Behnken design matrix of three variables
and a response (MRR) is shown in Table 2. The numbers of factors with base runs, total runs with
the center points are mentioned below.
3.1.1 Box-Behnken Design for 15 runs
Factors: 3, Replicates: 1, Base runs: 15, Total runs: 15
Base blocks: 1, Total blocks: 1, Center points: 3
Table 2: The Box-Behnken design matrix of three variables and a response (MRR)

Sl.no ND SOD AFR MRR

1 2 8 3.5 0.019
2 4 7 4.5 0.017
3 3 8 4.5 0.042
4 3 8 4.5 0.042
5 3 9 5.5 0.07
6 2 7 4.5 0.018
7 3 7 3.5 0.052
 Sreekanth and Rao 173

8 4 8 5.5 0.036
9 3 8 4.5 0.042
10 2 8 5.5 0.02
11 3 7 5.5 0.063
12 4 8 3.5 0.036
13 3 9 3.5 0.042
14 2 9 4.5 0.01
15 4 9 4.5 0.029

3.1.2 Response Surface Regression: MRR (gm/s) vs. AFR, SOD, ND

Based on Box-Behnken design the response surface regression has been developed, and the
influence of four variables on response (MRR) is discussed with F-values as shown in Table 3.
Table 3: Analysis of variance table of factors vs. MRR

Source DF Adj SS Adj MS F P Value

Value
Model 9 0.004029 0.00045 10.29 0.01
Linear 3 0.00051 0.00017 3.91 0.088
ND 1 0.000315 0.00032 7.23 0.043
SOD 1 0 0 0 0.985
AFR 1 0.000196 0.0002 4.5 0.087
Square 3 0.003346 0.00112 25.63 0.002
ND*ND 1 0.002554 0.00255 58.68 0.001
SOD*SOD 1 0.000023 2.3E-05 0.52 0.501
AFR*AFR 1 0.000545 0.00055 12.53 0.017
2-Way Interaction 3 0.000173 5.8E-05 1.32 0.365
ND*SOD 1 0.000101 0.0001 2.31 0.189
ND*AFR 1 0 0 0 0.967
SOD*AFR 1 0.000072 7.2E-05 1.66 0.254
Error 5 0.000218 4.4E-05
Lack-of-Fit 3 0.000218 7.3E-05 * *
Pure Error 2 0 0
Total 14 0.004247      

Model Summary
S R-sq R-sq(adj) R-sq(pred)
0.0065968 94.88% 85.65% 18.02%
The regression model of response surface is generated based on linear, square and 2-way interactions.
According to the model summary it is identified that the R-Square value is 94.88 % (any R-sq
174 Optimization of Process Parameters of Abrasive Jet Machining on Hastelloy through Response Surface Methodology

value more the 90% is considered as satisfactory regression model). The Regression equation is
MRR(g/s) = 0.440 +  0.1246 ND -  0.0739 SOD -  0.1380 AFR -  0.02630 ND*ND +  0.00249 SOD*SOD
+  0.01215 AFR*AFR +  0.00501 ND*SOD -  0.00014 ND*AFR +  0.00425 SOD*AFR

Contour Plot of MRR(g/s) vs SOD, ND Contour Plot of MRR(g/s) vs SOD, AFR Contour Plot of MRR(g/s) vs ND, AFR
9.0 9.0 4.0
MRR(g/s) MRR(g/s) MRR(g/s)
< 0.01 < 0.045 < 0.01
0.01 – 0.02 0.045 – 0.050 0.01 – 0.02
0.02 – 0.03 0.050 – 0.055 0.02 – 0.03
0.03 – 0.04 0.055 – 0.060 0.03 – 0.04
8.5 8.5 3.5
> 0.04 0.060 – 0.065 0.04 – 0.05
> 0.065 > 0.05
Hold Values
AFR 4.5 Hold Values Hold Values
ND 3 SOD 8

SOD
SOD

ND
8.0 8.0 3.0

7.5 7.5 2.5

7.0 7.0 2.0

2.0 2.5 3.0 3.5 4.0 3.5 4.0 4.5 5.0 5.5 3.5 4.0 4.5 5.0 5.5
ND AFR AFR

Graph 1: Contour plot of MRR VS SOD, ND &AFR

From Graph 1 contour plot of metal removal rate vs. AFR, SOD, ND are plotted. It is observed
that MRR increases by increase in pressure but the pressure is kept constant for performing the
response regression. The influence of AFR, SOD and ND are plotted ,the hold values of AFR (4.5
g/s) SOD (8mm) and ND (3 mm) are identified and the optimal values of parameters.

Surface Plot of MRR(g/s) vs SOD, ND Surface Plot of MRR(g/s) vs SOD, AFR Surface Plot of MRR(g/s) vs ND, AFR
Hold Values Hold Values Hold Values
AFR 4.5 ND 3 SOD 8

0.06
0.04
0 .0 6

M R R (g/s) 0.03 M R R (g/s)

M R R (g/s) 0.04
0 .0 2 0 .0 5
9 9
4
0.02
0.01
8 SOD 0.04 8 SOD
3 ND
2 4 .0
3 7 4.8 7 4.0
ND 4 AFR 5 .6 4.8 2
AFR 5 .6

Graph 2: Representation of surface plot of MRR VS SOD, ND, AFR

The 3D surface plots shows the combined effect of a set of parameters on the response (MRR).The
hold values of AFR, SOD, ND for the response are mentioned in Graph 2.
3.1.3 Response optimization: MRR (g/s)
Table 4: Response and its levels

Response Goal Lower Target Upper Weight Importance

MRR(g/s) Maximum 0.0096 0.0696 1 1 1

As shown in Table 4, the lower, upper and target values of MRR is selected. Multiple Response
Prediction has been shown.
Table 5: Representation of parameters and their optimal levels

Solution ND SOD AFR Fit Desirability

1 3.21212 9 5.5 0.067 0.955887
 Sreekanth and Rao 175

Graph 3: Optimization plot shows optimized values of parameters

From optimization plot Graph 3 the optimized parametric values and the ranges of optimization
are AFR (5.5 g/min), SOD (9 mm), and ND (4 mm).
3.2 Response surface regression for KERF
The width of cut (KERF) is calculated based on the standard formulas and the difference between
top surface diameter and bottom surface diameter. The obtained KERF values for 15 experiments
are executed based on random order priority. The Box-Behnken design matrix of three variables
and a response (KERF) is shown in Table 4.
3.2.1 Box-Behnken design with 3 factors and 15 runs
Factors: 3, Replicates: 1, Base runs: 15, Total runs: 15
Base blocks: 1, Total blocks: 1, Center points: 3
Table 6: The Box-Behnken design matrix of variables & response (KERF)

SL.NO ND SOD AFR KERF(mm)

1 4 8 3.5 4
2 2 8 3.5 5
3 4 9 4.5 5
4 2 9 4.5 6
5 4 8 5.5 5
6 3 8 4.5 4
7 2 7 4.5 3.8
8 3 9 5.5 6
9 3 8 4.5 3.8
10 3 9 3.5 5
11 2 8 5.5 4
12 3 7 3.5 6
13 3 7 5.5 6
14 3 8 4.5 4
15 4 7 4.5 6
176 Optimization of Process Parameters of Abrasive Jet Machining on Hastelloy through Response Surface Methodology

3.2.2 Response surface regression: KERF (mm) Vs ND, SOD, AFR

Based on Box-Behnken design the response surface regression has been developed, and the
influence of four variables on response (KERF) is discussed with F-values as shown in Table 5.
Table 7: Analysis of variance table of factors vs KERF

Source DF Adj SS Adj MS F-Value P-Value

Model 9 10.81 1.201 6.41 0.027
Linear 3 0.31 0.103 0.55 0.669
ND 1 0.18 0.18 0.96 0.372
SOD 1 0.005 0.005 0.03 0.877
AFR 1 0.125 0.125 0.67 0.451
Square 3 6.693 2.231 11.91 0.01
ND*ND 1 3.00E-04 3.00E-04 0 0.972
SOD*SOD 1 5.846 5.846 31.21 0.003
AFR*AFR 1 1.151 1.151 6.14 0.056
2-Way Interaction 3 3.81 1.27 6.78 0.033
ND*SOD 1 2.56 2.56 13.67 0.014
ND*AFR 1 1 1 5.34 0.069
SOD*AFR 1 0.25 0.25 1.33 0.3
Error 5 0.937 0.187
Lack-of-Fit 3 0.91 0.303 22.75 0.042
Pure Error 2 0.027 0.013
Total 14 11.75      

Model Summary
S R-sq R-sq(adj) R-sq(pred)
0.432820 92.03% 87.68% 0.00%
The regression model of response surface is generated based on linear, square and 2-way
interactions. According to the model summary it is identified that the R-Square value is 92.03
% (any R-sq value more the 90% is considered as satisfactory regression model). The regression
equation is
KERF (mm) = 91.2 + 4.25 ND - 18.83 SOD - 8.40 AFR + 0.008 ND*ND + 1.258 SOD*SOD
+ 0.558 AFR*AFR - 0.800 ND*SOD + 0.500 ND*AFR + 0.250 SOD*AFR
From the contour plot of Graph 4 it is observed that the influence of AFR, SOD and ND on KERF.
The hold values of AFR (4.5 gm/sec), SOD (8 mm) and ND (3 mm) which affects the width of cut
are identified.
 Sreekanth and Rao 177

Contour Plot of KERF(mm) vs SOD, ND Contour Plot of KERF(mm) vs SOD, AFR Contour Plot of KERF(mm) vs ND, AFR
9.0 9.0 4.0
KERF(mm) KERF(mm) KERF(mm)
< 4.0 < 4.0 < 4.0
4.0 – 4.5 4.0 – 4.5 4.0 – 4.4
4.5 – 5.0 4.5 – 5.0 4.4 – 4.8
5.0 – 5.5 5.0 – 5.5 4.8 – 5.2
8.5 8.5 3.5
5.5 – 6.0 5.5 – 6.0 > 5.2
> 6.0 > 6.0
Hold Values
Hold Values Hold Values SOD 8
AFR 4.5 ND 3

SOD
SOD

ND
8.0 8.0 3.0

7.5 7.5 2.5

7.0 7.0 2.0

2.0 2.5 3.0 3.5 4.0 3.5 4.0 4.5 5.0 5.5 3.5 4.0 4.5 5.0 5.5
ND AFR AFR

Graph 4: Contour plot of KERF VS SOD, ND, and AFR

3.3.3 Response optimization: KERF (mm)

Table 8: Response and its levels

Response Goal Lower Target Upper Weight Importance

KERF(mm) Maximum 2.6 3.8 6 1 1

As shown in Table 8, the lower, upper and target values of KERF is selected. The Multiple Response
Prediction has been shown.
Table 9: Representation of parameters and their optimal levels

Solution ND SOD AFR Fit Desirability

1 2 7.62626 4.9141 3.5608 1

The representation of predicted optimal values of parameters, i.e. ND, SOD and AFR with the
response (KERF) is shown in Graph 5. Main effect plot of KERF is indicated in Graph 6. Fig 3
shows the hastelloy plates of different thicknesses machined by AJM process.

Graph 5: Plot representation of predicted optimal values of parameters

Main Effects Plot for KERF(mm)
Fitted Means
ND SOD AFR
5.4

5.2

5.0
Mean of KERF(mm)

4.8

4.6

4.4

4.2

4.0

3.8

2 3 4 7 8 9 4.0 4.8 5.6

Graph 6: Representation of main effect plot of KERF

178 Optimization of Process Parameters of Abrasive Jet Machining on Hastelloy through Response Surface Methodology

Fig. 3: Hastelloy plates of different thicknesses machined by AJM

4. Conclusion
Response surface methods are implemented on machining of glass, ceramics, composites, Hastelloy
by selecting pressure, AFR, SOD,ND as factors and metal removal rate and KERF as objectives.
According to the design of experiments 15 experiments are conducted for glass and Hastelloy and
27 experiments are conducted for ceramics and composites. Plots are developed for showing
the influence of variables on objectives. Regression equations for both MRR & KERF are
developed. The response variable values suitable for machining are also identified. The design
and the experiment have been validated as the R-square value should range between (90-100)
percent.
References
[1] Balasubramaniam R, Krishnan J and Ramakrishnan N (1999), An experimental study on the
abrasive jet deburring of cross-drilled holes, Journal of Materials, Elsevier 91(3) : 178-182.
[2] Bhattacharya A (1973), New Technology, The Institution of Engineers Mechanical Engineering
Division, India.
[3] Finnie I (1960), Erosion of Surface by Solid Particles, Wear 3: 87-103.
[4] Ingulli CN (1967), Abrasive Jet Machining, Tool and Manufacturing 59: 28-33.
[5] Jain VK, Choudhury SK and Ramesh KM (2002), AJM of alumina and glass , Journal of Machine
tool & Manufacture 42 : 1269-1276.
[6] Paul AK and Roy RK (1987), Some studies on Abrasive jet machining, Journal of the Institution
of Engineers, India 68 (2): 27-30.
[7] Sarkar PK and Pandey PC (1976), Some Investigations on the Abrasive Jet Machining, Journal
of the Institution of Engineers, India 56: 284-287.
[8] Srikanth DV, Sreenivasa RM (2014), Abrasive jet machine - Research review, International
Journal of Advanced Engineering Technology, 5(2) : 18-24.
[9] Verma AP and Lal GK (1984), An Experimental Study of Abrasive Jet Machining', International
Journal of Machine Tool Design and Research, 24(1) : 19-29 .