Engineering with Computers (2011) 27:251259

DOI 10.1007/s00366-010-0195-5

ORIGINAL ARTICLE

Genetic Algorithm and Simulated Annealing to estimate optimal

process parameters of the abrasive waterjet machining
Azlan Mohd Zain Habibollah Haron
Safian Sharif

Received: 17 January 2010 / Accepted: 9 July 2010 / Published online: 25 July 2010
Springer-Verlag London Limited 2010

Abstract In this study, two computational approaches,

Genetic Algorithm and Simulated Annealing, are applied to
search for a set of optimal process parameters value that
leads to the minimum value of machining performance.
The objectives of the applied techniques are: (1) to estimate the minimum value of the machining performance
compared to the machining performance value of the
experimental data and regression modeling, (2) to estimate
the optimal process parameters values that has to be within
the range of the minimum and maximum coded values for
process parameters of experimental design that are used for
experimental trial and (3) to evaluate the number of iteration generated by the computational approaches that lead
to the minimum value of machining performance. Set of
the machining process parameters and machining performance considered in this work deal with the real experimental data of the non-conventional machining operation,
abrasive waterjet. The results of this study showed that
both of the computational approaches managed to estimate
the optimal process parameters, leading to the minimum
value of machining performance when compared to the
result of real experimental data.
Keywords Optimization  Optimal process parameters 
Minimum machining performance

A. M. Zain (&)  H. Haron

Faculty of Computer Science and Information System,
Universiti Teknologi Malaysia (UTM),
81310 Skudai, Johor, Malaysia
e-mail: azlanmz@utm.my
S. Sharif
Faculty of Mechanical Engineering,
Universiti Teknologi Malaysia (UTM),
81310 Skudai, Johor, Malaysia

1 Introduction
Recently, some established computational approaches
applied by previous works to suggest the optimal process
parameters in machining problem are such as Genetic
Algorithm (GA), Simulated Annealing (SA), Tabu Search
(TS), Ant Colony Optimization (ACO), and Particle Swarm
Optimization (PSO). Some of the advantages of computational approaches such as GA in optimizing process
parameters for machining problems may include [6, 12, 20,
21]: (1) preferred when near-optimal conditions instead of
the exact optimal solution are cost effective and acceptable
for implementation by the manufacturers; (2) a derivativefree approach for near-optimal point(s) search direction;
(3) able to handle objective functions of any complexity
with both discrete (for example, integer) and continuous
variables; (4) a simple complementation of the model by
new input parameters without modifying the existing
model structure; (5) an automatic search for the non-linear
connection between the inputs and outputs; (6) a fast and
simple optimizing technique.
Application of the GA, literatures show that most of the
works are related to conventional machining operations,
particularly for machining performance of surface roughness (Ra). Focused on estimation of the optimal process
parameters in the conventional machining operations, with
GA, Ra increases with an increase in the depth of cut and
nose radius [16]. High cutting speed during milling process
is preferred for the low Ra value [5]. GA reduces the Ra
value on mild steel from 2.60 to 0.71 lm for cutting speed,
feed rate and depth of cut process parameters [15]. Feed
rate has the greatest influence on Ra in the milling process
when evaluated by the application of the GA optimization
technique [3]. With feed rate, cutting speed, axial depth
of cut, radial depth of cut, machining tolerance process

123

252

parameters, GA reduces the Ra value in the mould cavity

from 0.412 to 0.375 lm, corresponding to about a 10%
improvement [14]. With GA, for the process parameters of
feed rate, cutting speed and axial depth of cut, an Ra value
was obtained that is lower than the values of experimental
results [13]. There is very limited number of studies that
dealt with GA for non-conventional machining optimization, particularly in the abrasive waterjet (AWJ) machining
process. A study dealt with AWJ machining process that
applying GA technique for estimating the optimal solutions
of the process parameters such as waterjet pressure at the
nozzle exit, diameter of abrasive waterjet nozzle, traverse
of the nozzle, mass flow rate of water and mass flow rate of
abrasives was reported by Jain et al. [7].
One of the alternatives in using computational approaches is the application of SA in estimating the optimal
process parameters, particularly for the Ra value in
machining process. Some of the capabilities of SA in
optimizing machining conditions may include [9, 18, 19,
22]: (1) SA algorithm is easy to program and typically
takes only a few hundred lines of the computer code, (2)
implementation of a new problem often only takes very
little modifications to the existing code, (3) SA algorithm
does not need to calculate the gradient descent that is
required for most traditional optimization algorithms,
hence can be applied to various kinds of objectives and
constraint functions, (4) SA algorithm can be used to
determine the global minimum more efficiently instead of
trapping in a local minimum where the objective function
has surrounding barriers, and (5) SA search is independent
of initial conditions. Despite the numerous capabilities of
SA, its application in optimization of process parameters
for various machining performances was given less attention by researchers. SA was used to optimize the cutting
parameters for conventional machining such as multi-pass
milling process [19]. Juan et al. [8] applied SA to the
polynomial network to determine the optimal cutting
parameters for minimum production cost in high-speed
machining (HSM) SKD61 tool steels. SA was also considered in optimization of machining conditions for nonconventional machining such as wire electrical discharge
machining [18].
Reviews so far show that application of SA for optimizing process parameters in AWJ machining was not yet
considered by researchers. In this study, GA and SA
computational approaches are taken up to observe the
possible improvement in the result that may be obtained
particularly for AWJ machining problem. This study
focuses on Regression model to formulate the optimization
fitness function equation. Researches on development of
Ra had been widely reported by previous studies [1, 11].
With reference to previous studies, it was found that

123

Engineering with Computers (2011) 27:251259

optimization of process parameters for Ra in AWJ

machining using the computational approaches, particularly using GA and SA are still lacking. As such optimization of the process parameters, which include traverse
speed, waterjet pressure, standoff distance, abrasive grit
size and abrasive flow rate on the Ra in AWJ process of AA
7075-T6 can be considered as a new contribution to the
machining research. With GA and SA, it is expected that it
can produce more significant result for the Ra value compared to the experimental and Regression modeling results.

2 Methodology
The steps taken to implement the GA and SA in fulfilling
the objectives of this study are:
1.

2.

3.

4.

In the Experimental data module, a set data from

machining process parameters and a set data from
machining performance from a case study [4] were
referred.
In the Regression modeling module, the model was
developed using the machining process parameters
from the case study in the Experimental module. A
multilinear stepwise regression analysis was performed
to predict the surface roughness value, and the secondorder polynomial regression was used to build the
predicted equation for surface roughness [4].
In the GA-based optimization module, the predicted
equation of the Regression model in Regression
modeling module would become the objective function for an optimization solution. The minimum and
maximum coded value for process parameters of the
experimental design would define the boundary for
optimization solution. Based on some of the criteria,
the minimum predicted performance value at the
optimal process parameters was estimated.
In the SA-based optimization module, similar to the
GA, the predicted equation of the Regression model in
Regression modeling module would become the
objective function for an optimization solution and
the minimum and maximum coded value for process
parameters of experimental design would define the
boundary for optimization solution. In the SA, extra
information is required which is the initial point for the
optimization solution. The process parameters value of
the Regression model would define the initial point for
optimization solution. Based on some of the criteria,
the minimum predicted performance value at the
optimal process parameters was estimated.

The strategies used in implementing the SA and GA

techniques are illustrated in Fig. 1.

Engineering with Computers (2011) 27:251259

253
Table 2 Levels of process parameters and coding identification

Experimental data:
- Machining type : Non-traditional
- Machining operation : AWJ
- Workpiece material: Al 7075-T6

Independent variables

Units

Level
1

Regression modeling:
- Multilinear stepwise regression analysis: predict the
machining performance value
- Second-order polynomial regression: Determination of the
machining performance predicted equation to be the GA and
SA fitness function

Traverse speed, V

mm/min

50

100

150

Waterjet pressure, P

MPa

125

175

250

Standoff distance, h

Mm

2.5

Abrasive grit size, d

lm

60

90

120

Abrasive flow rate, m

g/s

0.5

3.5

No. Setting values of experimental process parameters

Ra (lm)

Table 3 Ra values for real machining

GA and SA optimization:
- Formulation of the optimization solution
- Find the combination of the optimal process parameters
- Find the minimum machining performance value

V (m/min)

Evaluation of the GA and SA results:

- The minimum machining performance value of GA and SA
are compared to the result of experimental sample data and
Regression modeling.

Fig. 1 Flow of searching for optimum process parameters

P (MPa)

h (mm)

d (lm)

m (g/s)

50

125

60

0.5

2.124

50

125

60

2.753

50

125

60

3.5

3.352

50

175

2.5

90

0.5

4.311

50

175

2.5

90

4.541

50

175

2.5

90

3.5

5.123

7
8

50
50

250
250

4
4

120
120

0.5
2

6.789
7.524

50

250

120

3.5

9.123

10

100

125

2.5

120

0.5

3.575

11

100

125

2.5

120

4.457

12

100

125

2.5

120

3.5

5.628

13

100

175

60

0.5

7.010

14

100

175

60

7.535

15

100

175

60

3.5

7.893

16

100

250

90

0.5

8.121

17

100

250

90

8.312

18

100

250

90

3.5

9.163

19

150

125

90

0.5

4.328

20

150

125

90

5.120

21

150

125

90

3.5

5.852

According to the design of the experiment for the five

process parameters, the coding level for the AWJ Al 7075T6 are illustrated in Table 2. The measurements were taken
at a distance of 5 mm from the top of the cut surface. A SJ201 portable surface roughness measurement device was
used in order to measure the average surface roughness
value (Ra). A LEO 32 scanning electron microscope (SEM)
was used to investigate the machined surface.

22
23

150
150

175
175

1
1

120
120

0.5
2

6.143
6.721

24

150

175

120

3.5

7.780

25

150

250

2.5

60

0.5

8.890

26

150

250

2.5

60

27

150

250

2.5

60

3.5

Table 1 Chemical composition

of Al 7075 alloy

3.2 Experimental results

3 Materials and experimental setup of the case study

The machining experiment conducted by Caydas and Hascalik [4] to measure the Ra value in the AWJ machining
process was considered in this study. The material of
machined workpiece used in the experiments was Al 7075-T6
wrought alloy (AlZnMgCu1.5). The chemical composition of
the material of machined workpiece is given in Table 1.
3.1 Experimental design for surface roughness
measurements

Al

91.02

Cu

1.65

Mg

2.0

Cr

0.23

Zn

Mn

0.1

Ra (minimum)

9.120
10.035
2.124

Related to the problem investigated, an experiment was conducted that dealt with the Ra measurement. In the experiment,
experimental trials were executed that were based on L27
Taguchis orthogonal array. The Ra values of each experimental trial for process parameters are given in Table 3.

123

254

Engineering with Computers (2011) 27:251259

4 Regression modeling

Table 4 Predicted surface roughness values of Regression model

The measurement of Ra in the problem investigated in

relation to the process parameters is expressed mathematically as in (1):
q r s t

u 0

Ra cV P h d m e

where Ra is the experimental (measured) in lm, V is the

traverse cutting speed in mm/min, P is the waterjet pressure in MPa, h is the standoff distance in mm, d is abrasive
grit size in lm, m is the abrasive flow rate in g/s, e0 is
experimental error, and c, q, r, s, t, and u are the model
parameters to be estimated using the experimental data.
To develop the Regression model for estimating the Ra
value, the mathematical model given in (1) is linearized by
performing a logarithmic transformation as follows:
ln Ra ln c q ln V r ln P s ln h t ln d u ln m
ln e0 :

No. Setting values of experimental process parameters

V (m/min)

P (MPa)

h (mm)

d (lm)

Ra (lm)

m (g/s)

50

125

60

2.62915

50

175

2.5

90

0.5

4.00520

50

175

2.5

3.5

5.42532

50

250

120

7.69815

10

100

125

2.5

120

0.5

3.66819

12

100

125

2.5

120

3.5

5.55233

14
16

100
100

175
250

4
1

60
90

2
0.5

7.36548
7.96455

18

100

250

90

3.5

9.21330

20

150

125

90

4.98615

22

150

175

120

0.5

6.07837

24

150

175

120

3.5

7.79815

26

150

250

2.5

9.23448

Ra (minimum)

90

60

2.62915

Then, (2) can be written as follows:

y b0 x 0 b 1 x 1 b 2 x 2 b 3 x 3 b 4 x 4 b 5 x 5 e

where y is the logarithmic value of the experimental Ra,

x0 = 1 is a dummy variable, x1, x2, x3, x4 and x5 are the
process parameter values (logarithmic transformations) of
V, P, h, d and m, respectively, e is the logarithmic transformation of experimental error e0 and b0, b1, b3, b4 and b5
are the model parameters to be estimated using the experimental data.
Next, (3) can also be written as follows:
y^ y  e b0 x0 b1 x1 b2 x2 b3 x3 b4 x4 b5 x5
4
where y^ is the logarithmic value of the predictive (estimated) Ra.
Equation (4) can be extended to form a second-order
polynomial regression for surface roughness predicted
equation and given as follows:
y^ Ra b0 b1 V b2 P b3 h b4 d b5 m b11 V 2
b22 P2 b33 h2 b44 d 2 b55 m2
b12 VP b13 Vh b14 Vd b15 Vm b23 Ph b24 Pd
b25 Pm b34 hd b35 hm b45 dm:
5
From the results of Caydas and Hascalik [4], the final
regression model for surface roughness obtained is as
follows:
Ra 5:07976 0:08169V 0:07912P  0:34221h
 0:08661d  0:34866m  0:00031V 2  0:00012P2
0:10575h2 0:00041d2 0:07590m2  0:00008Vm
 0:00009Pm 0:03089hm 0:00513dm:
6

123

Subsequently, (6) will be proposed as the objective

function for optimization solution of the SA and GA. The
results of prediction of the Regression model are given in
Table 4 [4].
5 GA optimization solution
The GA is a method for solving both constrained and
unconstrained optimization problems and is based on natural selection, the process that drives the biological evolution. Figure 2 illustrates the flow of how the GA operates
in order to optimize a problem. The solution of an optimization problem with the GA begins with a set of
potential solution that is known as chromosomes. The
entire sets of these chromosomes comprise of populations
which are randomly selected. The chromosomes evolve
during several iterations or generations. New generations
known as offspring are generated by utilizing the crossover
and mutation techniques. Crossover involves the process of
splitting two chromosomes and then combining one-half of
each chromosome with the other pair. Mutation involves
the process of flipping a chromosome. The GA repeatedly
modifies a population of individual solutions. At each step,
the GA selects individuals at random from the current
population to be parents and uses them to produce the
children for the next generation. Over successive generations, the population evolves toward an optimal solution.
The target of the optimization process in this study is to
determine the optimal values of the process parameters that
lead to the minimum value of Ra. To formulate the optimization problem, the Regression model which is proposed

Engineering with Computers (2011) 27:251259

255
Table 5 Combination of GA parameter rates leading to the optimal
solution

Initialize parameters

Mutation

Generate population

Fitness function

Crossover

Selection

No

Parameters

Setting value/function type

Population size

100

Scaling function

Rank

Selection function

Roulette wheel

Crossover function

Heuristic

Crossover rate

0.8

Mutation function

Adaptive feasible

The best
fitness is
obtained?

Optimal solution
Yes

Fig. 2 The flow of GA for optimization

in (6) is taken to be the fitness function of the optimization

solution and is expressed as follows:
Minimize Ra V; P; h; d; m
min5:07976 0:08169V 0:07912P  0:34221h
 0:08661d  0:34866m  0:00031V 2  0:00012P2
0:10575h2 0:00041d2 0:07590m2  0:00008Vm
7

The minimization of the fitness function value of (7) is

subjected to the boundaries (limitations) of the process
parameters. The range of values of experimental process
parameters in Table 2 is selected to present the limitations
of the optimization solution and is given as follows:
50  V  150

8a

125  P  250

8b

1h4

8c

60  d  120

8d

0:5  m  3:5 :

8e

Basically, to obtain the optimal solutions, some criteria

must be considered by the GA algorithm as listed in
Table 5.
By using the fitness function formulated in (7), the
limitations of process parameters formulated in (8a8e)
and the GA parameters given in Table 5, the Matlab
Optimization Toolbox is next applied to find the minimum
values of Ra at the optimal points. The results of the Matlab
Optimization Toolbox are given in Figs. 3 and 4.
From Fig. 3, the set values of optimal process parameters that lead to the minimum Ra value are 50.024 mm/min
for traverse speed, 125.018 MPa for waterjet pressure,

Fig. 3 Optimal solution of GA

Best: 1.5549 Mean: 1.5619

6
5.5

Best fitness
Mean fitness

5
4.5

Fitness value

 0:00009Pm 0:03089hm 0:00513dm:

4
3.5
3
2.5
2
1.5

10

20

30

40

50

60

70

80

90

100

Generation

Fig. 4 Fitness function plot of GA

1.636 mm for standoff distance, 94.973 lm for abrasive

grit size and 0.525 g/s for abrasive flow rate. It is also
indicated that the optimal solution is obtained at the 57th
generation. Figure 4 shows the best fitness value of the GA
is 1.5549 lm with the mean fitness value being 1.5619 lm.

123

256

Engineering with Computers (2011) 27:251259

6 SA optimization solution
SA is a random search technique that is able to escape local
optima using a probability function [10]. Based on the
iterative improvement, the SA algorithm is a heuristic
method with the basic idea of generating random displacement from any feasible solution. This process accepts
not only the generated solutions, which improve the fitness
function but also those which do not improve it with the
probability function; a parameter depending on the fitness
function [17]. SA is a method for solving unconstrained
and bound-constrained optimization problems. It models
the physical process of heating a material and then slowly
lowering the temperature to decrease defects, thus minimizing the system energy. At each iteration of the SA
algorithm, a new point is randomly generated. The distance
of the new point from the current point, or the extent of the
search, is based on a probability distribution with a scale
proportional to the temperature. The algorithm accepts all
new points that lower the objective, but also, with a certain
probability points that raise the objective. By accepting
points that raise the objective, the algorithm avoids being
trapped in a local minimum, and is able to explore globally
for more possible solutions. An annealing schedule is
selected to systematically decrease the temperature as the
algorithm proceeds. As the temperature decreases, the
algorithm reduces the extent of its search to converge to a
minimum. Figure 5 shows the graphical representation in
the SA search of the global minimum [18].
An important part of the SA process is how the inputs
are randomized. The randomization process takes the previous input values and the current temperature as inputs.
The input values are then randomized according to the
temperature. A higher temperature will result in more
randomization; a lower temperature will result in less
randomization. There is no specific method defined by the
SA algorithm for how to randomize the inputs. The exact
nature by which this is done often depends upon the nature
of the problem being solved. Figure 6 illustrates the flow
on how the SA technique operates in order to search the
optimal solution. Figure 7 illustrates the pseudo code of the
SA for searching the optimal solution [2].
The target of the optimization process in this study is to
determine the optimal values of the process parameters that
lead to the minimum value of Ra. Similar to GA, the
Regression model which is proposed in (7) is taken to be the
fitness function of the optimization solution of the SA, and
the range of values given in (8a8e) are selected to present
the limitations of the optimization solution of the SA.
The process parameters that lead to the minimum Ra of
the Regression model as given in Table 4 will be chosen to
be the initial points for the SA solution and are given as
follows:

123

Fig. 5 Graphical illustration of the SA search of the global minimum

[18]

Initial solution

Evaluate the solution

No

Solution
accepted?
Yes
Update the current
solution

Change
tem perature

No

Generate a
new solution

Yes
Decrease

Terminate
the search?

No

Yes
Optimal solution

Fig. 6 The flow of SA for optimization

Initial point of V 50

9a

Initial point of P 125

9b

Initial point of h 1

9c

Initial point of d 60

9d

Initial point of m 2 :

9e

Basically, to obtain the optimal solutions, some criteria

must be considered by the SA algorithm as listed in Table 6.

Engineering with Computers (2011) 27:251259

257

Begin
generate the initial string randomly = q
T=Tmax
Let E(q,T) be the associated energy
while (T Tmin)
for i = i to k
Mutate (flip) a random position in q to yield s
Let E(s,T) be the associated energy
Set q s with probability 1/ ( 1+e-(E(q,T)-E(s,T))/T )
end for
T= rT
end while
Decode the string q to provide the solution of the problem.
End

Fig. 7 Pseudo code of SA [2]

Table 6 Combination of SA parameter rates leading to the optimal

solution

Fig. 8 Optimal solution of SA

Setting value/function type

2.8

Annealing function

Boltzmann annealing

2.6

Reannealing interval

100

Temperature update function

Exponential temperature

Initial temperature

100

Acceptance probability function

Simulated annealing acceptance

Data type

Double

Function value

Parameters

Best Function Value: 1.5355

2.4
2.2
2
1.8
1.6

By using the fitness function formulated in (7), the

limitations of process parameters formulated in (8a8e),
the initial points formulated in (9a9e), and the SA
parameters given in Table 6, the Matlab Optimization
Toolbox is next applied to find the minimum values of Ra
at the optimal points. The results of the Matlab Optimization Toolbox are given in Figs. 8 and 9.
As shown in Figs. 8 and 9, it was observed that the
minimum Ra value is 1.5355 lm. The set values of process
parameters that lead to the minimum Ra value are
50.003 mm/min for traverse speed, 125.029 MPa for
waterjet pressure, 1.486 mm for standoff distance,
107.737 lm for abrasive grit size and 0.5 g/s for abrasive
flow rate. The optimal solution is obtained at the 2,711th
iteration of the SA algorithm.

1.4

500

1000

1500

2000

2500

3000

Iteration

Fig. 9 Fitness function plot of SA

standoff distance, d is an optimal solution of the abrasive

grit size and m is an optimal solution of the abrasive flow
rate of GA and SA, this section discuses the calculation for
validating the result of GA and SA.
Validation of the GA result is given by the following
calculation:
Minimize Ra V; P;h;d;m
5:07976 0:0816950:024 0:07912125:018
 0:342211:636  0:0866194:973  0:348660:525
 0:0003150:0242 0:00012125:0182

7 Validation of the GA and SA results

Theoretically, to validate the result of the proposed
approach in this study, the optimal process parameters
values of the GA and SA will be transferred into the
Regression model equation (7), taken as the objective
function of the optimization solution. With V is an optimal
solution of the traverse speed, P is an optimal solution of
the waterjet pressure, h is an optimal solution of the

0:105751:6362 0:0004194:9732 0:075900:5252

 0:0000850:0240:525  0:00009125:0180:525
0:030891:6360:525 0:0051394:9730:525
1:554742646
1:5547
Validation of the SA result is given by the following
calculation:

123

258

Engineering with Computers (2011) 27:251259

can be concluded that GA and SA have reduced

the Ra value at about 1.0741 (40.86%) and
1.0935 lm (41.59%), respectively.

Minimize Ra V; P;h;d;m
5:07976 0:0816950:003 0:07912125:029
 0:342211:486  0:08661107:737
 0:348660:5  0:0003150:0032 0:00012125:0292

2.

0:105751:4862 0:00041107:7372 0:075900:52

 0:0000850:0030:5  0:00009125:0290:5
0:030891:4860:5 0:00513107:7370:5
1:53548312
1:5355
By transferring the optimal cutting values into the
optimization fitness function equation above, it was
obtained that Ra = 1.5547 lm for GA and Ra = 1.5355 lm
for SA. Next, these values are compared to the minimum
Ra value of the GA and SA. As shown in Figs. 4 and 7, the
minimum Ra value of the GA and SA are 1.5549 and
1.5355 lm, respectively. These are similar to the results of
the transformation processes. This can be taken as the
indicator that the same results (Ra = 1.5549 lm from SA
and Ra = 1.5355 lm from SA) might be obtained when the
set optimal process parameters that were estimated by
using both computational approaches are used in the real
experiment process.

8 Evaluation of the GA and SA results

In this section, discussion is carried out to highlight all the
objectives of the study that stated in Sect. 2. Discussion is
separated into three parts which are evaluation of the
minimum Ra value, evaluation of the optimal process
parameters and evaluation of the number of iteration of the
SA and GA results.
1.

First objective: evaluation of the minimum Ra value

Figures 4 and 7 show the minimum Ra value for both
computational approaches, GA and SA, are 1.5549 and
1.5355 lm, respectively. For fulfilling the first objective of this study, evaluation against the minimum Ra
values of both computational approaches are made by
comparing to the result of experimental data and
Regression model (machining case study by [4].
a.

Experimental data versus GA and SA

As shown in Table 3, the minimum Ra value
among all experimental trials is 2.124 lm. It can
be concluded that GA and SA have reduced the Ra
value at about 0.5691 (26.79%) and 0.5885 lm
(27.71%), respectively.
b. Regression versus GA and SA
As shown in Table 4, the minimum predicted Ra
value of the Regression model is 2.62915 lm. It

123

3.

Second objective: evaluation of the optimal process

parameters
For the second objective of this study, Table 2 shows
that the range values of the coded of experimental
design are 50150 mm/min for traverse speed,
125250 Mpa for waterjet pressure, 14 mm for
standoff distance, 60120 lm for abrasive grit size
and 0.53.5 g/s for abrasive flow rate.
Figure 3 shows that the optimal solution of the GA are
50.024 mm/min for cutting speed, 125.018 Mpa for
waterjet pressure, 1.636 mm for standoff distance,
94.973 lm for abrasive grit size and 0.525 g/s for
abrasive flow rate. From Fig. 8, the optimal solution of
the SA are 50.003 mm/min for cutting speed,
125.029 Mpa for waterjet pressure, 1.486 mm for
standoff distance, 107.737 lm for abrasive grit size
and 0.5 g/s for abrasive flow rate. Since the optimal
values of the GA and SA for each process parameter
are within the range of minimum and maximum value
of experimental design, thus, this study concludes that
the second objective of this study is fulfilled.
Third objective: evaluation of the number of iteration
For the third objective of this study, Fig. 3 shows that
the number of iteration generated by GA that lead to
the minimum value of Ra is 57. From Fig. 8, it was
found that the number of iteration generated by SA
that leads to the minimum value of Ra is 2,711. Since
the number of iteration given by GA is lower than the
number of iteration by SA, thus, this study concludes
that the time of searching the optimal solution can be
made faster by using GA optimization technique.

9 Conclusion
This study proposed two computational approaches, GA
and SA, in order to estimate the optimal solutions of process parameters that lead to the minimum machining performance. The considered machining performance is
surface roughness (Ra). The machined-material was AA
7075 aluminium alloy. Five machining process parameters
for abrasive waterjet operation that considered in this study
are traverse speed, waterjet pressure, standoff distance,
abrasive grit size and abrasive flow rate. Overall, the results
of this study are summarized in Table 7.
From Table 7, for the first issue, it is clear that this study
has found that the GA and SA have been the effective
computational approaches for estimating the minimum
values of Ra compared to the experimental and Regression

Engineering with Computers (2011) 27:251259

259

Table 7 Summary of results of the study

Approach

Issue 1: minimum
Ra (lm)

Issue 2: optimal/non-optimal of V (mm/min),

P (MPa), h (mm), d (lm) and m (g/s)

Issue 3: no.
of iterations

Experimental [4]

2.124

(50, 125, 1, 60, 0.5)

Regression [4]

2.62915

(50, 125, 1, 60, 2)

GA

1.5549

(50.024, 125.018, 1.636, 94.973, 0.525)

57

SA

1.5355

(50.003, 125.029, 1.486, 107.737, 0.5)

2,711

results reported by Caydas and Hascalik [4]. For the second

issue, the optimal value for each of the process parameters
recommended by the GA and SA have satisfied within the
range of minimum and maximum values for process
parameters of experimental design. Focusing on the third
issue, by referring to the number of iteration, it can be
stated that the time of searching the optimal solution can be
made faster by using GA optimization technique. However,
by using SA, the minimum surface roughness value was
much lower than the GA.
Acknowledgments The authors greatly acknowledge Ministry of
Higher Education Malaysia and Universiti Teknologi Malaysia for the
financial support. Special thank you to the reviewer for their valuables
comments and suggestions.

References
1. Arizmendi M, Ferna0 ndez J, de Lopez Lacalle LN, Lamikiz A, Gil
A, Sanchez JA, Campa FJ, Veiga F (2008) Model development for
the prediction of surface topography generated by ball-end mills
taking into account the tool parallel axis offset. Experimental
validation. CIRP Ann Manuf Technol 57(1):101104
2. Bandyopadhyay S, Pal SK, Murthy CA (1998) Simulated
annealing based pattern classification. Inf Sci 109:165184
3. Brezonick M, Kovavic M, Ficko M (2004) Prediction of surface
roughness with genetic programming. J Mater Process Technol
157158:2836
4. Caydas U, Hascalik A (2008) A study on surface roughness in
abrasive waterjet machining process using artificial neural networks and regression analysis method. J Mater Process Technol
202:574582
5. Colak O, Kurbanoglu C, Kayacan MC (2007) Milling surface
roughness prediction using evolutionary programming methods.
J Mater Des 28:657666
6. Cus F, Balic J (2003) Optimization of cutting process by GA
approach. Robot Comput Integr Manuf 19:113121
7. Jain NK, Jain VK, Deb K (2007) Optimization of process
parameters of mechanical type advanced machining processes
using genetic algorithms. Int J Mach Tools Manuf 47(6):900919
8. Juan H, Yu FS, Lee BY (2003) The optimal cutting-parameter
selection of production cost in HSM for SKD61 tool steels. Int J
Mach Tools Manuf 43:679686

9. Khan Z, Prasad LB, Singh T (1997) Machining condition optimization by genetic algorithms and simulated annealing. Comput
Oper Res 24(7):647657
10. Kirkpatrick F, Gelatte CD, Vecchi MP (1983) Optimization by
simulated annealing. Science 220:671780
11. Lopez de Lacalle LN, Lamikiz A, Sanchez JA, Arana JL (2007)
The effect of ball burnishing on heat-treated steel and Inconel 718
milled surfaces. Int J Adv Manuf Technol 32(910):958968
12. Manolas DA, Gialamas TP, Frangopoulos CA, Tsahalis DT
(1996) A genetic algorithm for operation optimization of an
industrial cogeneration system. Comput Chem Eng 20:1107
1112
13. Oktem H, Erzurumlu T, Erzincanli F (2006) Prediction of minimum surface roughness in end milling mold parts using neutral
network and genetic algorithm. J Mater Des 27:735744
14. Oktem H, Erzurumlu T, Kurtaran H (2005) Application of
response surface methodology in the optimization of cutting
conditions for surface roughness. J Mater Process Technol
170:1116
15. Palanisamy P, Rajendaran I, Shanmugasundaram S (2007) Optimization of machining parameters using genetic algorithm and
experimental validation for end-milling operations. Int J Adv
Manuf Technol 32:644655
16. Suresh PVS, Venkateswara P, Deshmukh SG (2002) A genetic
algorithmic approach for optimization of surface roughness prediction model. Int J Mach Tools Manuf 42:675680
17. Swarnkar R, Tiwari MK (2004) Modeling machine loading
problem of FMSs and its solution methodology using a hybrid
tabu search and simulated annealing-based heuristic approach.
Robot Comput Integr Manuf 20:199209
18. Tarng YS, Ma SC, Chung LK (1995) Determination of optimal
cutting parameters in wire electrical discharge machining. Int J
Mach Tools Manuf 35(12):16931701
19. Wang ZG, Rahman M, Wong YS, Sun J (2005) Optimization of
multi-pass milling using parallel genetic algorithm and parallel
genetic simulated annealing. Int J Mach Tools Manuf 45:1726
1734
20. Wong TN, Chan LCF, Lau HCW (2003) Machining process
sequencing with fuzzy expert system and genetic algorithms. Eng
Comput 19:191202
21. Zain AM, Haron H, Sharif S (2010) Application of GA to optimize cutting conditions for minimizing surface roughness in end
milling machining process. Expert Syst Appl 37:46504659
22. Zain AM, Haron H, Sharif S (2010) Simulated annealing to
estimate the optimal cutting conditions for minimizing surface
roughness in end milling Ti-6Al-4V. Mach Sci Technol 14:4362

123