Applied Soft Computing 13 (2013) 14331439

Contents lists available at SciVerse ScienceDirect

Applied Soft Computing

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

Hybrid Taguchi-differential evolution algorithm for optimization of multi-pass

turning operations
Ali R. Yildiz
Bursa Technical University, Department of Mechanical Engineering, Bursa, Turkey

a r t i c l e

i n f o

Article history:
Received 18 December 2011
Received in revised form 14 January 2012
Accepted 22 January 2012
Available online 31 January 2012
Keywords:
Hybrid optimization
Differential evolution algorithm
Taguchis method
Manufacturing
Turning

a b s t r a c t
Hybridizing of the optimization algorithms provides a scope to improve the searching abilities of the
resulting method. The purpose of this paper is to develop a novel hybrid optimization algorithm entitled
hybrid robust differential evolution (HRDE) by adding positive properties of the Taguchis method to the
differential evolution algorithm for minimizing the production cost associated with multi-pass turning
problems. The proposed optimization approach is applied to two case studies for multi-pass turning operations to illustrate the effectiveness and robustness of the proposed algorithm in machining operations.
The results reveal that the proposed hybrid algorithm is more effective than particle swarm optimization
algorithm, immune algorithm, hybrid harmony search algorithm, hybrid genetic algorithm, scatter search
algorithm, genetic algorithm and integration of simulated annealing and Hooke-Jeevespatter search.
2012 Elsevier B.V. All rights reserved.

1. Introduction
The machining processes are commonly used by manufacturing industries in order to produce high quality and very complex
products in a short time. These machining processes include large
number of input parameters which may affect the cost and quality of the products. Selection of optimum machining parameters
in such machining processes is very important to satisfy all the
conicting objectives of the process.
The present paper focuses on the well-known multi-pass turning optimization problem which consists of selecting the optimal
cutting parameters, i.e. cutting speeds, feed rates, depths of cut and
number of passes, for minimizing a production cost-based objective function. There are two options to choose the optimal cutting
parameters for a given economic objective. The rst one is concerned with the need of a machine expert that manually selects
the machining parameters on the basis of its own experience and by
means of a proper machining handbook. That way generates many
uncertainties and drawbacks in terms of efciency of solutions and
time/cost requirements. As an alternative to the above mentioned
approach, many research efforts have been made to state a comprehensive mathematical model of a turning process that, in practice,
entails a set of cutting constraints to be handled [114].
Machining optimization problems become tricky whenever a
given objective function must be optimized with respect to a

E-mail address: [email protected]

1568-4946/$ see front matter 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.asoc.2012.01.012

large number of constraints. In the rst study on the machining

economics problems, a theoretical analysis of the optimization
of the machining process is presented by Gilbert [2]. The earlier
approaches tackled the single-pass turning problems [3,4], though
in machining practice it is quite rare to meet cases where the total
depth of cut should be removed just by a single pass. Despite a
higher problem complexity, nowadays, multi-pass operations are
preferred because of the better machining economies [5]. In the
some of the research works, dynamic programming [57], geometric programming [9,10], linear programming [11], sequential
unconstrained minimization technique [12] and linear programming and branch-and-bound [13] were considered for multi-pass
operations. These optimization techniques are either stuck at local
optimum or take a long time to converge to a reasonable result [14].
The convergence speed of evolutionary optimization techniques to the optimal results is better than that of conventional
approaches. Therefore, evolutionary algorithms such as genetic
algorithm (GA), differential evolution algorithm, ant colony
algorithm, immune algorithm and particle swarm optimization
algorithm have been used in many applications instead of conventional techniques [8,1526].
Chen and Tsai [8] used a simulated annealing powered with a
Hooke-Jeeves pattern search (SA-PS) algorithm for optimization of
machine parameters in multi-pass turning.
Yildiz and Saitou [25] developed a novel approach for multicomponent topology optimization of continuum structures using
a multi-objective genetic algorithm to obtain Pareto optimal
solutions that exhibits trade-offs among stiffness, weight, manufacturability and assemble ability.

1434

C0
CI
CM
CR
CT
dr , ds

A.R. Yildiz / Applied Soft Computing 13 (2013) 14331439

constant pertaining to tool-life equation

machine idle cost ($/piece)
cutting cost by actual time in cut ($/piece)
tool replacement cost ($/piece)
tool cost ($/piece)
depths of cut for each pass of rough and nish
machining (mm)
drL , drU lower and upper bounds of depth of rough cut (mm)
dsL , dsU lower and upper bounds of depth of nish cut (mm)
dt
total depth of metal to be removed (mm)
D
diameter of work piece (mm)
fr , fs
feeds in rough and nish machining (mm/rev)
frL , frU
lower and upper bounds of feed in rough machining
(mm/rev)
fsL , fsU
lower and upper bounds of feed in nish machining
(mm/rev)
cutting forces during rough and nish machining
Fr , Fs
(kg f)
maximum allowable cutting force (kg f)
FU
h1 , h2
constants pertaining to tool travel and
approach/depart time (min)
k1 , k2 , k3 constants for roughing and nishing parameter
relations
coefcient pertaining to specic tool-work piece
kf
combination
ko
direct labor cost overhead ($/min)
coefcient pertaining to equation of chip-tool interkq
face temperature
kt
cutting edge cost ($/edge)
L
length of work piece (mm)
number of rough passes
n
p, q, r
constants pertaining to the tool-life equation
cutting power during roughing and nishing (kW)
Pr , Ps
PU
maximum allowable cutting power (kW)
temperatures during roughing and nishing ( C)
Qr , Qs
QU
maximum allowable temperature ( C)
maximum allowable surface roughness (mm)
Ra
Rn
nose radius of cutting tool (mm)
Sc
limit of stable cutting region
t
tool life (min)
constant term of machine idling time (min)
tc
te
tool exchange time (min)
tp
tool life (min) considering roughing and nishing
tr , ts
tool lives (min) for roughing and nishing
variable term of machine idling time (min)
tv
TI
machine idling time (min)
TL , TU
lower and upper bounds of tool life
TM
cutting time by actual machining (min)
TMr , TMs cutting time by actual machining for roughing and
nishing (min)
TR
tool replacement time (min)
unit production cost except material cost ($/piece)
UC
Vr , Vs
cutting speeds in rough and nish machining
(m/min)
VrL , VrU lower and upper bounds of cutting speed in rough
machining (m/min)
VsL , VsU lower and upper bounds of cutting speed in nish
machining (m/min)
vector of machining parameters
X
,,
constants pertaining to expression of chip-tool
interface temperature
power efciency


,
,

constants pertaining to expression of stable cutting

region
constants of cutting force equation

Onwubolu and Kumalo [27] used the GA to optimize multi-pass

turning problem based on the mathematical model of Chen and
Tsai [8]. Vijayakumar et al. [28] focused on the same problem using
ant colony optimization technique. Chen and Chen [29] argued that
Onwubolu and Kumalo [27] did not consider the equality constraint
based on total depth of cut. Similarly, Vijayakumar et al. [28] did
not consider the equality constraint on total depth of cut. Since
Onwubolu and Kumalo [27] and Vijayakumar et al. [28] did not
consider the equality constraint based on total depth of cut, their
results are not compared with those of Chen and Tsai [8]. A scatter
search algorithm for optimization of machining economics models
in multi-pass turning operations is used by Chen [30].
Evolutionary optimization algorithms are efcient at exploring
entire search space; however, they are relatively poor at nding
the precise optimum solution and may have premature convergence towards a local minimum. To compensate deciencies of
the individual algorithms and obtain a more robust optimization
technique, it is common to hybridize different optimization techniques. The hybrid approaches have attracted most of the research
attention in the literature [3135]. Since hybrid optimization algorithms integrate constructive properties of different methods, they
are very effective to nd the optimal results. In some of the research
works on optimization of machine parameters in the literature,
hybrid optimization algorithms were developed and used efciently.
The articial immune algorithm and hill climbing local search
algorithm was hybridized to solve manufacturing and design optimization problems by Yildiz [33].
To improve the efciency of the GA, in our previous work [35],
we hybridized the GA with Taguchis method to optimize machining parameters for multi-pass turning operations. The results of the
proposed robust approach (HRGA) were better than those of scatter
search, the GA and the simulated annealing powered with a HookeJeeves Pattern Search (SA-PS) algorithm for turning operations.
Wang et al. [36] proposed a new hybrid approach based on the
GA and the SA for obtaining optimal machine parameters in milling
operations. In their approach, the GA and the SA were hybridized.
The results of the proposed hybrid approach were better than
genetic algorithms and geometric programming.
Although some improvements regarding optimization of cutting
parameters in multi-pass turning operations have been achieved,
due to the complexity of machine parameters with conicting
objective and constraints, optimization of the machining economic
problems still presents a matter of investigation. Therefore, in
recent years, there has been a growing interest in the new hybrid
optimization approaches to optimize machining economic problems.
The aim of this research is to further develop and strengthen
the differential evolution algorithm (DE) and to optimize cutting
parameters for multi-pass turning operation. In this research, a new
hybrid optimization approach based on differential evolution algorithm and Taguchis method is developed. The proposed hybrid
approach is applied to the two case studies to optimize cutting
parameters in multi-pass turning operations.
2. Optimization model of multi-pass turning operation
The goal of the multi-pass turning operations is to minimize
unit production cost (CU ). The unit production cost is the sum of

A.R. Yildiz / Applied Soft Computing 13 (2013) 14331439

the cutting cost (CM ), machine idle cost (CI ), tool replacement cost
(CR ) and tool cost (CT ), respectively. The developed hybrid optimization approach is used to optimize multi-pass turning operation
for the determination of cutting parameters considering minimum
production cost under a set of machining constraints which are presented and adopted in the references of Shin and Joo [5], Chen and
Tsai [8], and Chen [30].
The cost function
CU = CM + CI + CR + CT
CU = k0

DL
1000Vr fr

(1)

d d 
t
s
dr

+ k0 tc + (h1 L + h2 )
te
+ k0
Tp
+


kt

Tp

d d
t
s



+1

dr

dr

d d 
t
s
dr

DL
1000Vs fs

DL
1000Vs fs

fs2
SRU
8R

Surface nish constraint

(19)

2.1.3. Parameter relations

Vs k3 Vr

(20)

fr k4 fs

(21)

dr k5 d5

(22)

dr =

(dt ds )
n

(23)

In addition to these constraints, the total depth of cut is another

important constraint for the case study. The total depth of cut (dt )
is the sum of the depth of nish cut (ds ) and the total depth of
rough cut (ndr ). The optimization algorithm does not determine the
optimal depth of roughing since it can be given by the mathematical
manipulation as expressed in Eq. (24). Therefore, one can eliminate
the equality constraint (Eq. (23)) and the decision variable (dr ) in
the optimization procedure [8].

1000Vs fs

d d 
t
s

DL
1000Vr fr

DL
1000Vr fr


DL

1435

(2)

2.1. Parameter bounds and cutting condition constraints

In multi-pass turning operations, the unit production cost (CU )
is imposed by different constraints which are (i) parameter bounds
cover depth of cut, cutting speed and feed; (ii) tool-life constraint;
(iii) cutting force constraint; (iv) power constraint; (v) stable
cutting region constraint; (vi) chip-tool interface temperature constraint; (vii) surface nish constraint (only for nish machining);
and (viii) parameter relations. These constraints are as follow [5]:

ds = dt ndr

(24)

Therefore, the equality constraint and the decision variable (dr )

and (n) in the optimization procedure can be eliminated. The ve
machining parameters (Vr , fr , ds , Vs , fs ) are determined for turning
model optimization. Further details about the turning mathematical model and data with respect to machining can be obtained
from Shin and Joo [5] and Chen and Tsai [8], Chen [30] and Yildiz
[31,34,35].
3. The proposed hybrid approach
This section describes the proposed hybrid optimization technique. First, a brief overview of the Taguchis method and
differential evolution algorithm is provided. At the end of the section the hybrid strategy of the proposed HRDE is presented.

2.1.1. Rough machining

Depth of cut drL dr drU

(3)

Feed frL fr frU

(4)

Cutting speed VrL Vr VrU

(5)

3.1. Taguchis method

(6)

The Taguchis method is a universal approach, which is widely

used in robust design [37]. There are three stages to achieve
Taguchis objective: (1) concept design, (2) robust parameter
design, and (3) tolerance design. The robust parameter design is
used to determine the levels of factors and to minimize the sensitivity of noise. That is, a parameter setting should be determined
with the intention that the product response has minimum variation while its mean is close to the desired target. Taguchis method
is based on statistical and sensitivity analysis for determining the
optimal setting of parameters to achieve robust performance. The
responses at each setting of parameters were treated as a measure
that would be indicative of not only the mean of some quality characteristic, but also the variance of that characteristic. The mean and
the variance would be combined into a single performance measure
known as the signal-to-noise (S/N) ratio. Taguchi classies robust
parameter design problems into different categories depending on
the goal of the problem and for each category as follows:
Smaller the better. For these kind of problems, the target value
of y, that is, quality variable, is zero. In this situation, S/N ratio is
dened as follows:

Tool-life constraint

TL tr TU

Cutting force constraint


k1 fr dr

FU

(7)

Power constraint

k1 fr dr Vr
PU
6120

Stable cutting region constraint

(8)
Vr fr dr SC

(9)

Chip-tool interface temperature constraint

Qr = kq VR fr dr QU

(10)

2.1.2. Finish machining

Depth of cut dsL ds dsU

(11)

Feed fsL fs fsU

(12)

Cutting speed VsL Vs VsU

(13)

Tool-life constraint

TL ts TU

Cutting force constraint

Power constraint


k1 fs ds


k1 fS ds VS

6120

(14)
FU

S/N Ratio = 10 log

PU

Stable cutting region constraint

(15)

yi2

VS fs ds SC

(17)

(25)

Larger the better. In this situation, the target value of y, that is,
quality variable, is innite and S/N ratio is dened as follows:

S/N Ratio = 10 log

(18)

(16)

Chip-tool interface temperature constraint

QS = k2 Vs fs ds QU



1/yi2
n

(26)

1436

A.R. Yildiz / Applied Soft Computing 13 (2013) 14331439

Nominal the best. For these kind of problems, the certain target
value is given for y value. In this situation S/N ratio is dened as
follows:
S/N Ratio = 10 log



yi2

s2

(27)

Taguchis method uses an orthogonal array and analysis of mean

to analyze the effects of parameters based on statistical analysis of
experiments. To compare performances of parameters, the statistical test known as the analysis of variance (ANOVA) is used. Further
details and technical merits about robust parameter design can be
found in the [37,38].
3.2. Differential evolution algorithm
Differential evolution algorithm (DE) was rst proposed by Price
and Storn [39]. DE has received an important attention concerning
its potential as a new optimization technique for complex nonlinear problems and has been successfully used in various areas
[4050]. DE uses three important operations: mutation, crossover
and selection, and it utilize the three operators to evolve from randomly generated initial population to nal individual solution [40].
Mutation and crossover are used to generate the trial vectors, and
selection is then used to determine whether the new generated
vectors can survive the next generation. The procedure of the DE
works as follows:
Step 1: Initial algorithm parameters
They are: scale factor F, crossover rate CR, the population size M
and the maximum number of iterations K.
Step 2: Randomly generate M candidate solutions
The initial candidate solutions are generated from a uniform
distribution in the ranges [xjL , xjU ] (j = 1, 2, . . . , N), where, N is the
number of variables.
Step 3: Mutation
The mutation operator can not only increase the diversity
of solution vectors, but also enhance the exploration capability
of solution space for the DE algorithm. For each parent, xik (i =

1, 2, . . . , M) of generation k(k = 1, 2, . . . , K), a trial vector, ik+1 is

created by mutating a target vector. According to the mutation
operator, the trial vector is then calculated using the following
equation:
ik+1 = xik + F(xik xik )
s

(28)

Step 4: Crossover
DE follows a discrete recombination approach where elements
from the parent vector xik , are combined with elements from the

trial vector ik , to produce the offspring uki .

uk+1
=
i,j

k+1
i,j
,

if r and < Cr or j = r1D ;

k
xi,j,

otherwise

(29)

where, r1D is a random integer in [1, D]. CR represents crossover

rate, and CR [0, 1].
Step 5: Selection
The generated offspring uk+1
replaces the parent xik , only if the
i
tness of the offspring is better than that of the parent.

Step 6: Check the stopping criterion.

If the stopping criterion (maximum number of iterations K)
is satised, computation is terminated. Otherwise, Steps 35 are
repeated.
It is decided to use a common set of parameter values, specically F = 0.8, CR = 0.95 for the case study.
3.3. The proposed optimization approach
In this paper, a new hybrid optimization approach (HRDE) is presented to select the optimal machining parameters for multi-pass
turning operations. The proposed approach hybridizes differential evolution algorithm and Taguchi method. It has an important
advantage to consider hybridizing DE with other techniques to
develop a new approach that improves the performance of DE to
solve optimization problems.
A larger population makes the algorithm more likely to locate
a good masking string, but also increases the time taken by the
algorithm. Therefore, there is a need to dene the efcient range of
population size to achieve better optimal solutions in shorter times.
In this research, this shortcoming is eliminated by introducing
Taguchis robust parameter design through the initial population
generation for DE.
The Taguchi method chooses the most suitable combination
of the levels of factors by using S/N table and orthogonal arrays
against the factors that form the variation and are uncontrollable
in product and process [37]. Hence, it tries to reduce the variation in product and process as much as possible. Taguchis robust
parameter design uses statistical performance measure which is
known as S/N ratio that takes both medium and variation into
consideration. Therefore, the current approach uses the issues of
robustness to emphasize the statistical and sensitivity analysis of
robust parameter design to achieve an efcient exploration using
a small population by avoiding the use of a large search space
for the evolution process. The current proposed hybrid approach
involves two stages of optimization: (a) renement of search space
of solutions using Taguchis method and (b) DE search process using
rened population size. The pseudo code of the proposed hybrid
algorithm is given as follows.
Program: HRDE
Begin
Taguchis method
Set the input parameters
Select suitable orthogonal array
Select suitable S/N ratio type
While (not termination condition for experiments) do
Compute S/N ratios
Compute objective function values
Conduct matrix of experiments
End
While (not termination condition for parameters) do
Compute contributions
Generate analysis of variance (ANOVA table)
End
Dene intervals for parameters
Differential evolution algorithm
Begin
Generate randomly initial population of solutions
Repeat
Mutation
Crossover
Until a termination is satised,
End
End.

A.R. Yildiz / Applied Soft Computing 13 (2013) 14331439

4. Example of computational machining optimization

As explained in the above sections, the metal cutting operation has a complex nature, the objectives are usually in conict
with each other, and they have uncontrollable variations in their
design parameters with complex nature. For example, increasing
production rate may increase the production cost by increasing
the rate of tool wear. As stated in Section 2, the complex nature
and high non-linearity of machining optimization problems may
present some shortcomings for optimization approaches. There is
a crucial need to overcome the limitations owing to the traditional
optimization methods and also further to improve the strength of
recent approaches to achieve better results for the machining problems in industry. In the present research, the search space is rened
based on the effect of the various design variables on objective functions. An aim was to reach optimum solutions by using Taguchis
approach coupled with DE.
4.1. Metal cutting example

Table 1
Data for the example of multi pass turning.
L = 300 mm
VrL =50 m/min
drU = 3.0 mm
VsL = 50 m/min
dsU = 3.0 mm
kt =2.5 $/edge
tc = 0.75 min/piece
q = 1.75
TU = 45 min
 = 0.75
FU = 200 kg f
 = 1
 = 0.4
QU = 1000 C

Table 2
Experimental results and S/N ratio.
Ex. No

X1

X2

X3

X4

X5

S/N

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

50
50
50
50
200
200
200
200
350
350
350
350
500
500
500
500

0.1
0.27
0.53
0.9
0.1
0.27
0.53
0.9
0.1
0.27
0.53
0.9
0.1
0.27
0.53
0.9

50
200
350
500
200
50
350
500
350
500
50
200
500
350
200
50

0.1
0.27
0.53
0.9
0.53
0.9
0.1
0.27
0.9
0.53
0.27
0.1
0.27
0.1
0.9
0.27

1
1.66
2.32
3
3
2.32
1.66
1
1.66
1
3
2.32
2.32
3
1
1.66

10.3
3.1
1.9
1.4
3.6
1.8
3.1
9.5
17.9
26.7
2.7
2.2
33.4
3.8
1.8
2.6

20.2
9.7
5.7
3.5
11.1
5.3
10.1
19.5
25.1
28.5
8.8
7.2
30.4
11.8
5.1
8.5

Table 3
Results of the analysis of variance for objective (case 1 with dt = 6 mm).

In this section HRDE is conducted to compute the optimum machining parameters for the metal cutting model,
which is described in Section 2. The metal cutting model
is a non-linear problem subject to constraints with multiple
machining variables. First, Taguchis method is employed to
nd rened population space of design parameters to achieve
better initialize DE search. The machining variables (factors)
x1 (Vr ), x2 (fr ), x3 (Vs ), x4 (fs ) and x5 (ds ) are selected as feed, cutting
speed and depth of cut in rough and nish turning. The suitable
orthogonal array with regard to ve machining variables at four levels each is chosen as L16 matrix. Matrix experiments are designed
using L16 orthogonal array and S/N ratios are conducted. There
are 16 combinations of the design variables in which each row
corresponds to one trial. Machining data for the rst example of
multi-pass turning are shown in Table 1.
The equations for calculating S/N ratios for quality characteristics are logarithmic functions based on the mean square quality
characteristics. For this problem, S/N ratios for objective function
of rst example are computed using smaller-the-better (Eq. (25))
as given in Table 2 since the objective is the minimization of cost.
The relative effect of the different factors can be obtained by
the decomposition of variance, which is called an analysis of variance (ANOVA). The purpose of ANOVA is to investigate the design
parameters that affect signicantly the quality characteristic. It is
designed using S/N ratios as shown in Table 3 for (case 1 with
dt = 6 mm). The intervals of the design parameters are found regarding the effects of factors on the objective.

D = 50 mm
VrU = 500 m/min
frL = 0.1 mm/rev
VsU = 500 m/min
fsL = 0.1 mm/rev
ko = 0.5 $/min
h2 = 0.3
p=5
C0 = 6 1011
kf = 108
 = 0.85
=2
kq = 132
= 0.105
Ra = 10
k1 = 1.0
k2 = 2.5
k3 = 1.0

1437

dt = 6.0 mm
frU = 0.9 mm/rev
drL = 1.0 mm
fsU = 0.9 mm/rev
dsL = 1.0 mm
h1 = 7 104
te = 1.5 min/edge
r = 0.75
TL = 25 min
 = 0.95
PU = 5 kW
Sc = 140
= 0.2
Rn = 1.2 mm

x1
x2
x3
x4
x5

Level 1

Level 2

Level 3

Level 4

SS

% Cont.

9.8
21.7
10.7
12.3
18.3

11.5
13.8
8.3
15.4
13.3

17.4
7.4
13.1
15.1
12.1

14
9.7
20.5
9.7
8.8

127
472.4
258.6
71.7
132.8

11.95
44.45
24.33
6.74
12.44

It can be seen that x2 is the most effective variable for affecting

the metal cutting optimization performance of objective function
with 44.45 per cent contributions as shown in Table 3. Therefore,
the level 3 and level 4 are selected for the search space limits for
x2 . It is dened as 0.53 < x2 < 0.9. The same procedure is applied for
x3 and x5 . The most affecting variables are x2 , x3 , x5 , and x1 , with
contributions 44.45 percent, 24.33 percent, 12.44 percent, and
11.95 percent. The less affecting variable is x4 with contribution
6.74 per cent. The search space limits for x4 are taken as level 1
(lower bound) and level 4 (upper bound) since its contributions is
not high for affecting the metal cutting optimization performance.
Therefore, the intervals of design variables for case 1 are found
as 50 < x1 < 200, 0.53 < x2 < 0.9, 50 < x3 < 200, 0.1 < x4 < 0.9, and
2.32 < x5 < 3. The computed levels are used to generate the initial
population in DE. The analysis of variance is applied for case 2 with
dt = 8 mm and results are given in Table 4.
It can be seen that the most effective variables and their levels
are the same as in the case 1. Therefore, the search space limits of the parameters are 50 < x1 < 200, 0.53 < x2 < 0.9, 50 < x3 < 200,
0.1 < x4 < 0.9, and 2.32 < x5 <3. The initial population of DE is randomly generated for solutions within the range of 50 < x1 < 200,
0.53 < x2 < 0.9, 50 < x3 < 200, 0.1 < x4 <0.9, and 2.32 <x5 <3. Then differential evolution is processed to nd the best result for objective
using the rened search space of solutions.
The efciency of each approach can be measured by comparing
function evaluation number, which is equal to the population size
multiplied by the generation number as each solution is evaluated
once in every generation. The main aim of the evolutionary

Table 4
Results of the analysis of variance for objective (case 2 with dt = 8 mm).

x1
x2
x3
x4
x5

Level 1

Level 2

Level 3

Level 4

SS

% Cont.

13.6
25.7
12.8
14.4
21.40

14.3
17
11.4
18
16.1

19.9
9.8
16.1
19.1
15.2

16.3
11.7
24
13
11.5

94.5
599.1
294.5
58
139.8

7.9
50.5
24.8
4.8
11.7

1438

A.R. Yildiz / Applied Soft Computing 13 (2013) 14331439

Table 5
Comparison of the best computed optimum results for turning problem (HRDE: hybrid robust differential evolution algorithm; PSO: particle swarm optimization algorithm,
HRGA: hybrid robust genetic algorithm, SS: scatter search; SA/PS: simulated annealing and Hooke-Jeeves pattern search; FEGA: oat encoding genetic algorithm; HTHSA:
hybrid harmony search algorithm; articial immune algorithm).

Case 1: cost ($) (dt = 6 mm)

Case 2: cost ($) (dt = 8 mm)

HRDE

AIA

PSO [31]

HRGA [35]

SS [30]

FEGA [29]

HTHSA [23]

SA/PS [8]

2.0461
2.4791

2.12
2.51

2.047
2.479

2.048
2.486

2.066
2.541

2.298
2.817

2.047

2.279
2.741

optimization techniques is to nd the optimum results with

minimum function evaluation number.
The parameters used by the current HRDE for the search process
in this problem are the following:
(a) Population size: 25;
(b) Maximum number of iteration: 200;
(c) Number of objective function evaluations: 5000.
From the comparison of best results given in Table 5, it is seen
that the minimization of the unit production cost in multi-pass
turning operation is achieved by proposed hybrid approach. The
comparison of the results obtained by the proposed approach,
against other techniques, is given in Table 5.
It can be seen that better results for the best computed solutions
are achieved for the turning optimization problem compared to PSO
(particle swarm optimization algorithm), HRGA (hybrid genetic
algorithm), SS (scatter search), FEGA (oat encoding GA) and SA/PS
(Simulated Annealing and Hooke-Jeeves pattern search) as shown
in Table 5. PSO, HRGA and FEGA required 10,000, 27,000, 40,000,
60,000 function evaluations to nd the best solutions, respectively.
The oat encoding GA (FEGA) of Chen and Chen [29] required
60,000 function evaluations to nd the best solutions 2.2988 and
2.8170 for case 1 and case 2, respectively. The use of the proposed
hybrid approach (HRDE) improves the convergence rate by computing the best values 2.0461 and 2.4791 for case 1 and case 2,
respectively, and maintaining the less function evaluations 5000.
5. Conclusions
In this paper, the machining economies problem concerning the
multi-pass turning has been addressed by a hybrid differential evolution optimization technique. The specic issue of this research
is to introduce a new approach to solve machining optimization
problems. In the current paper, a new improved two-stage DE is
presented for the optimization of machining parameters considering minimum production cost under a set of machining constraints
in turning operation. The results of the turning optimization problem indicate the strong inuence of using a rened population of
solutions over a DE without rened search space. It is seen that
present approach performed quite well on the optimization of
machining parameters of turning operation problem nding better
solutions compared to other approaches. From the above computational results and discussions, it is demonstrated that HRDE can
yield effective heuristic solutions from the machining economies
viewpoint. The signicant outcomes of this research highlight that
HRDE can be taken into account as a useful and powerful technique
for optimizing machining problems.
References
[1] A. Costa, G. Celano, S. Fichera, Optimization of multi-pass turning economies
through a hybrid particle swarm optimization technique, International Journal
of Advanced Manufacturing Technology 53 (2011) 421433.
[2] W.W. Gilbert, Economics of Machining Theory and Practice, American Society
Metals, Cleveland, OH, 1950.
[3] D.S. Ermer, Optimization of the constrained machining economics problem
by geometric programming, Journal of Engineering for Industry 93 (1971)
10671072.

[4] G. Boothroyd, P. Rusek, Maximum rate of prot criteria in machining, Journal

of Engineering for Industry 98 (1976) 217220.
[5] Y.C Shin, Y.S. Joo, Optimization of machining conditions with practical constraints, International Journal of Production Research 30 (1992) 29072919.
[6] K. Iwata, Y. Murotsu, T. Iwatsubo, F. Oba, Optimization of cutting conditions
for multi-pass operations considering probabilistic nature in machining conditions, Journal of Engineering for Industry 99 (1977) 211217.
[7] B.K. Lambert, A. Walvekar, Optimization of multi pass machining operations,
International Journal of Production Research 16 (1978) 259265.
[8] M.C. Chen, D.M. Tsai, A simulated annealing approach for optimization of multipass turning operations, International Journal of Production Research 34 (1996)
28032825.
[9] D.S. Ermer, S. Kromodihardo, Optimization of multi pass turning with constraints, Journal of Engineering for Industry 103 (1981) 462468.
[10] B. Gopalkrishan, F.A. Khayyal, Machining parameter selection for turning with
constraints: an analytical approach based on geometric programming, International Journal of Production Research 29 (1991) 18971908.
[11] R. Gupta, J.L. Batra, G.K. Lal, Determination of optimal subdivision of depth of
cut in multi-pass turning with constraints, International Journal of Production
Research 33 (1995) 25552565.
[12] S.K. Hati, S.S. Rao, Determination of optimum machining conditions deterministic probabilistic approaches, Journal of Engineering for Industry 98 (1976)
354359.
[13] F.P. Tan, R.C. Creese, A generalized multi-pass machining model for machining
parameter selection in turning, International Journal of Production Research
33 (1995) 14671487.
[14] Z.G. Wang, M. Rahman, Y.S. Wong, J. Sun, Optimization of multi-pass milling
using parallel genetic algorithm parallel genetic simulated annealing, International Journal of Machine Tools & Manufacture 45 (15) (2005) 17261734.
[15] H.J. Holland, Adaptation in Natural Articial Systems, An Introductory Analysis
with Application to Biology, Control Articial Intelligence, The University of
Michigan Press, Ann Arbor, USA, 1975.
[16] H.M. Azmathulla, A.A. Ghani, Genetic programming for longitudinal dispersion
coefcients in streams, Water Resources Management 25 (2011) 15371544.
[17] A.R. Yildiz, A new design optimization framework based on immune algorithm
and taguchi method, Computers in Industry 60 (8) (2009) 613620.
[18] A.R. Yildiz, N. Ozturk, N. Kaya, F. Ozturk, Hybrid multi-objective shape design
optimization using Taguchis method and genetic algorithm, Structural and
Multidisciplinary Optimization 34 (4) (2007) 277365.
[19] K.Y. Chan, K.W. Chan, G.T.Y. Pong, M.E. Aydin, T.C. Fogarty, S.H. Ling, A statisticsbased genetic algorithm for quality improvements of power supplies, European
Journal of Industrial Engineering 3 (4) (2009) 468492.
[20] Modeling a liquid epoxy molding process using a particle swarm optimization
based fuzzy regression, IEEE Transactions on Industrial Informatics 7 (1) (2011)
148158.
[21] T. Kunakote, S. Bureerat, Multi-objective topology optimization using evolutionary algorithms, Engineerng Optimization 43 (5) (2011) 541557.
[22] K.Y. Chan, C.K. Kwong, T.S. Dillon, Y.C. Tsim, Reducing over tting in manufacturing process modeling using a backward elimination based genetic
programming, Applied Soft Computing 11 (2) (2011) 16481656.
[23] A.R. Yildiz, Hybrid Taguchi-harmony search algorithm for solving engineering
optimization problems, International Journal of Industrial Engineering Theory,
Applications and Practice 15 (3) (2008) 286293.
[24] K.Y. Chan, C.K. Kwong, H. Jiang, M.E. Aydin, T.C. Fogarty, A new orthogonal array
based crossover, with analysis of gene interactions, for evolutionary algorithms
and its application to car door design, Expert Systems with Applications 37 (5)
(2010) 38533862.
[25] A.R. Yildiz, K. Saitou, Topology synthesis of multi-component structural assemblies in continuum domains, Journal of Mechanical Design 133 (1) (2011)
01100811009.
[26] K.Y. Chan, T.S. Dillon, C.K. Kwong, Handling Uncertainties in Modelling Manufacturing Processes with Hybrid Swarm Intelligence, International Journal of
Production Research, 2011.
[27] G.C. Onwubolu, T.T. Kumalo, Optimization of multi-pass turning operation
with genetic algorithm, International Journal of Production Research 39 (2001)
37273745.
[28] K. Vijayakumar, G. Prabhaharan, P. Asokan, R. Saravanan, Optimization of
multi-pass turning operation using ant colony system, International Journal
of Machine Tools & Manufacture 43 (2003) 16331639.
[29] M.C. Chen, K.Y. Chen, Optimization of multipass turning operations with genetic
algorithms: a note, International Journal of Production Research 41 (2003)
33853388.
[30] M.C. Chen, Optimizing machining economics models of turning operations
using the scatter search approach, International Journal of Production Research
42 (2004) 26112625.

A.R. Yildiz / Applied Soft Computing 13 (2013) 14331439

[31] A.R. Yildiz, A novel particle swarm optimization approach for product design
and manufacturing, International Journal of Advanced Manufacturing Technology 40 (5-6) (2009) 617628.
[32] A.R. Yildiz, K. Solanki, Multi-objective optimization of vehicle crashworthiness
using a new particle swarm based approach, International Journal of Advanced
Manufacturing Technology, doi:10.1007/s00170-011-3496-y, in press.
[33] A.R. Yildiz, A novel hybrid immune algorithm for global optimization in design
and manufacturing, Robotics and Computer-Integrated Manufacturing 25 (2)
(2009) 261270.
[34] A.R. Yildiz, Hybrid immune-simulated annealing algorithm for optimal design
and manufacturing, International Journal of Materials and Product Technology
34 (3) (2009) 217226.
[35] A.R. Yildiz, F. Ozturk, Hybrid enhanced genetic algorithm to select optimal
machining parameters in turning operation, Proceedings of the Institution of
Mechanical Engineers, Part B, Journal of Engineering Manufacture 220 (12)
(2006) 20412053.
[36] Z.G. Wang, Y.S. Wong, M. Rahman, Optimisation of multi-pass milling
using genetic algorithm genetic simulated annealing, International Journal of
Advanced Manufacturing Technology 24 (9-10) (2004) 727732.
[37] S.M. Phadke, Quality Engineering Using Robust Design, Prentice Hall, New York,
1989.
[38] G.G. Taguchi, S. Chowdhury, S. Taguchi, Robust Engineering, McGraw-Hill,
2000.
[39] R. Storn, K. Price, Differential evolution - a simple and efcient heuristic for
global optimization over continuous spaces, Journal of Global Optimization 11
(4) (1997) 341359.
[40] D. Zoua, H. Liu, L. Gaob, S. Li, A novel modied differential evolution algorithm for constrained optimization problems, Computers and Mathematics
with Applications 61 (2011) 16081623.

1439

[41] A.G. Krishna, Optimization of surface grinding operations using a differential

evolution approach, Journal of Materials Processing Technology 183 (2007)
202209.
[42] L. Wang, Y. Xu, L. Li, Parameter identication of chaotic systems by hybrid
NelderMead simplex search and differential Evolution Algorithm, Expert Systems with Applications 38 (4) (2011) 32383245.
[43] S. Kitayamaa, M. Arakaw, K. Yamazaki, Differential evolution as the global optimization technique and its application to structural optimization, Applied Soft
Computing 11 (2011) 37923803.
[44] A.R. Yildiz, Hybrid Taguchi-differential evolution algorithm for optimization of multi-pass turning operations, Applied Soft Computing (2012),
doi:10.1016/j.asoc.2011.12.016.
[45] P.V. Hull, M.L. Tinker, G. Dozier, Evolutionary optimization of a geometrically rened truss, Structural and Multidisciplinary Optimization 31 (2006)
311319.
[46] E.K. Silva, H.J.C. Barbosa, A.C.C. Lemonge, An adaptive constraint handling technique for differential evolution with dynamic use of variants in engineering
optimization, Optimization and Engineering 12 (2011) 3154.
[47] T.W. Liao, Two hybrid differential evolution algorithms for engineering design
optimization, Applied Soft Computing 10 (2010) 11881199.
[48] K. Jarmai, J.A. Snyman, J. Farkas, G. Gondos, Optimal design of a welded I-section
frame using four conceptually different optimization algorithms, Structural and
Multidisciplinary Optimization 25 (2003) 5461.
[49] V.V. Truong, Minimum weight design for toroidal pressure vessels using
differential evolution and particle swarm optimization, Structural and Multidisciplinary Optimization 42 (2010) 351369.
[50] N.K. Madavan, Turbomachinery airfoil design optimization using differential
evolution, in: 2nd International Conference on Computational Fluid Dynamics,
Sydney, Australia, July 1519, 2002.