Int J Adv Manuf Technol (2011) 52:79–88

DOI 10.1007/s00170-010-2710-7

ORIGINAL ARTICLE

Optimization of drilling parameters on surface roughness

in drilling of AISI 1045 using response surface
methodology and genetic algorithm
Erol Kilickap & Mesut Huseyinoglu & Ahmet Yardimeden

Received: 14 January 2010 / Accepted: 2 May 2010 / Published online: 20 May 2010
# Springer-Verlag London Limited 2010

Abstract Modeling and optimization of cutting parameters 1 Introduction

are one of the most important elements in machining
processes. The present study focused on the influence Drilling is one of the most common and complex
machining parameters on the surface roughness obtained in operations among many kinds of machining methods. It is
drilling of AISI 1045. The matrices of test conditions widely used in a variety of manufacturing industries
consisted of cutting speed, feed rate, and cutting environ- including aerospace and automotive sectors [1]. Surface
ment. A mathematical prediction model of the surface roughness resulting from drilling operations has tradition-
roughness was developed using response surface method- ally received considerable research attention. It has an
ology (RSM). The effects of drilling parameters on the impact on the mechanical properties like fatigue behavior,
surface roughness were evaluated and optimum machining corrosion resistance, creep life, etc. It also affects other
conditions for minimizing the surface roughness were functional attributes of parts like friction, wear, light
determined using RSM and genetic algorithm. As a result, reflection, heat transmission, lubrication, electrical conduc-
the predicted and measured values were quite close, which tivity, etc. [2].
indicates that the developed model can be effectively used Metal removal fluids decrease friction between the
to predict the surface roughness. The given model could be cutting tool and the workpiece material, preventing tool
utilized to select the level of drilling parameters. A wear, and reducing surface roughness [3]. The conventional
noticeable saving in machining time and product cost can cutting fluids utilized in machining are considered a
be obtained by using this model. problem for manufacturer, since these substances can
seriously damage human health and environment. Environ-
Keywords Response surface methodology . Genetic mental concerns have become increasingly important to
algorithm . Box-Behnken design of experiments . Minimum productive processes, allied with their economic and
quantity lubricant . Drilling . Surface roughness technological aspects. Large quantities of emulsion-based
cooling fluids for machining are still widely used in the
metal working industry, generating high consumption and
discard costs and impacting the environment. The increas-
ing need for environmentally friendly production techni-
ques and rapid growth of cutting fluid disposal costs have
justified the demand for an alternative to machining
E. Kilickap (*) : M. Huseyinoglu : A. Yardimeden
processes using fluids. Over the last decade, however, the
Department of Mechanical Engineering, Dicle University,
21280 Diyarbakir, Turkey goal of research in this field has been to restrict as much as
e-mail: ekilickap@dicle.edu.tr possible the use of cooling fluids or lubricants in metal
M. Huseyinoglu production processes. Dry machining and minimum quan-
e-mail: mesuth@dicle.edu.tr tity lubricant (MQL) machining have become the focus of
A. Yardimeden attention of researches and technicians in the field of
e-mail: ayardim@dicle.edu.tr machining as an alternative to traditional fluids [4–9].
80 Int J Adv Manuf Technol (2011) 52:79–88

Table 1 Single point crossover process Table 3 Uniform crossover

process 1 Parent A1 A2 A3 A4
1 Parent A1 A2….An An+1.... Ak 2 Parent B1 B2 B3 B4
2 Parent B1 B2.…Bn Bn +1.....Bk 1 0 0 1
1 Child A1 A2….An Bn +1….Bk 1 Child B1 A2 A3 B4
2 Child B1 B2….Bn An +1….Ak 2 Child A1 B2 B3 A4

Process modeling and optimization are two important 2 Methodology

issues in manufacturing. The manufacturing processes are
characterized by a multiplicity of dynamically interacting 2.1 Response surface methodology
process variables. Surface finish has been an important
factor of machining in predicting performance of any Response surface methodology is a collection of mathe-
machining operation. In order to develop and optimize a matical and statistical techniques, which are useful for the
surface roughness model, it is essential to understand the modeling and analyzing the engineering problems and
current status of work in this area [10]. In order to developing, improving, and optimizing processes. It also
understand the effects of machining parameters in the has important applications in the design, development, and
various machining, a lot of researchers used optimization formulation of new products, as well as in the improvement
techniques [11–13]. Most surface roughness prediction of existing product designs, and it is an effective tool for
models are empirical and are generally based on experi- constructing optimization models [17].
ments in the laboratory. In addition, it is very difficult in RSM consists of the experimental strategy for explor-
practice to keep all factors under control as required to ing the space of the process or input factors, empirical
obtain reproducible results [12]. Generally, these models statistical modeling to develop an appropriate approxi-
have a complex relationship between surface roughness and mating relationship between the yield and the process
operational parameters, work materials, and chip-breaker variables, and optimization methods for finding the levels
types [14]. Optimization of machining parameters not only or values of the process variables that produce desirable
increases the utility for machining economics, but also the values of the response outputs [17]. Response surface
product quality to a great extent [15]. As a result, there have method designs also help in quantifying the relationships
been a great many research developments in modeling between one or more measured responses and the vital
surface roughness and optimisation of controlling parameters input factors.
to obtain a surface finish of desired level since only proper The first step of RSM is to define the limits of the
selection of cutting parameters can produce a better surface experimental domain to be explored. These limits are made
finish. But such studies are far from complete since it is very as wide as possible to obtain a clear response from the
difficult to consider all the parameters that control the surface model [18]. The cutting speed, feed rate, and cutting
roughness for a particular manufacturing process [16]. environment are the drilling variable, selected for our
In this study, an effort has been made to estimate the investigation.
surface roughness using experimental data. The surface In the next step, the planning to accomplish the experi-
roughness model was developed by response surface ments by means of RSM using a Box-Behnken design.
methodology (RSM). It has also been attempted to optimize In many engineering fields, there is a relationship
the surface roughness prediction model using a genetic between an output variable of interest (y) and a set off
algorithmic approach. controllable variables (x1, x2,…xn). The relationship be-

Table 4 Mutation process

Table 2 Double point crossover process
Original child 1 1101111000011110
1 Parent A B C Original child 2 1101100100110110
2 Parent D E F
Child with mutation 1 1100111000011110
1 Child A E C
Child with mutation 2 1101101100110110
2 Child D B F
Int J Adv Manuf Technol (2011) 52:79–88 81

Table 5 Experimental factors

and their levels Symbol Drilling parameter Level 1 (−1) Level 2 (0) Level 3 (1)

A Cutting speed, V (m/min) 5 10 15

B Feed rate, f (mm/rev) 0.1 0.2 0.3
C Cutting environment MQL (1) Comp. air (2) Dry (3)

tween the drilling control parameters and the responses is If there is curvature in the system, then a polynomial of
given as: higher degree must be used, such as the second order
model;
y ¼ f ðx1 ; x2 ; . . . ; xn Þ þ " ð1Þ
X
k X
k k 1 X
X k
where, ε represents the noise or error observed in the y ¼ b0 þ b j xj þ bjj x2j þ bij xi xj þ " ð4Þ
response (y). If we denote the expected response be EðyÞ ¼ j¼1 j¼1 i j
f ðx1 ; x2 ; . . . xn Þ ¼ h and then the surface represented by;
where, i ¼ 1; 2; . . . ; k  1 and j=1,2,…,k also i<j [18].
h ¼ f ð x1 ; x2 ; . . . xn Þ ð2Þ
2.2 Genetic algorithm
is called a response surface. The variable x1, x2,…xn in
Eq. 2 are called natural variables, because they are
GA, which imitates the evolution mechanism of nature, is
expressed in natural units of measurement.
used for finding a particular data in a dataset [19]. GA
In most RSM problems, the form of the relationship
produces ever-improving solutions based on the rule ‘the
between the independent variables and the response is
best one survives’. For this purpose, it uses a fitness
unknown, it is approximated. Thus, the first step in RSM is
function that selects the best and operators like regeneration
to find an appropriate approximation for the true functional
and mutation to produce new solutions. Another feature of
relationship between response and the set of independent
GA is that it involves a group solution. By the way
variables. Usually, a low-order polynomial in some region
optimum solutions among other ones could be picked and
of the independent variables is employed. If the response is
disqualified ones are eliminated.
well modeled by a linear function of the independent
The most important feature that distinguishes GA from
variables, then the approximating function is the first order
other algorithms is selection. Fitness of a solution increases
model;
the chance for it to be selected. However, it does not
y ¼ b 0 þ b 1 x1 þ b 2 x2 þ . . . þ b k xk þ " ð3Þ guarantee that. Formation of first group in selection is

Table 6 Design matrix and

observed values of surface Trial no. Cutting speed Feed rate Cutting environment Surface roughness (Ra)
roughness
1 −1 −1 0 2.33
2 1 −1 0 4.27
3 −1 1 0 4.10
4 1 1 0 6.64
5 −1 0 −1 3.04
6 1 0 −1 6.63
7 −1 0 1 3.48
8 1 0 1 5.37
9 0 −1 −1 2.06
10 0 1 −1 3.37
11 0 −1 1 3.10
12 0 1 1 4.23
13 0 0 0 3.10
14 0 0 0 3.15
15 0 0 0 3.05
16 0 0 0 3.07
17 0 0 0 3.10
82 Int J Adv Manuf Technol (2011) 52:79–88

Table 7 Model summary statis-

tics for surface roughness Source SD R2 Adj. R2 Sum of square Lack of fit

Linear 0.88 0.6400 0.5569 17.96 10.10 Suggested

2FI 0.96 0.6692 0.4707 18.78 9.28
Quadratic 0.45 0.9495 0.8846 26.64 1.41 Suggested

random. However, fitness of solutions determines the regeneration, the chromosomes in coming generations
chance of being selected in that random selection [19–21]. might start to repeat each other after some period of time
GA passes through three phases to form a new generation: and thus, production of different chromosomes might halt
Evaluating the value of fitness for each individual in the old or decrease drastically. For this reason, some of the
generation, selecting individuals based on their value of chromosomes are subject to mutation in order to increase
fitness (using fitness function) and continuing generations the diversity of them. Mutation process is as indicated in
with selected individuals using operators such as crossover Table 4.
and mutation.
The aim of crossover process is to produce child
chromosomes by changing the locations of main chromo- 3 Experimental details
some genes and hence to obtain chromosomes having even
higher value of fitness from the ones with high values of A detailed survey has been carried out to find out how
fitness. There are three types of mostly used crossover for drilling parameters affect surface roughness. The three
binary coding in the literature [20]. Single point, double cutting parameters selected for the experiments are cutting
point, and uniform crossover processes were explained in speed (V), feed rate (f), and cutting environment (dry,
order and following tables are helpful [19–21] (Tables 1 compressed air and MQL). The drilling parameters for the
and 2). experiments are shown in Table 5.
In the uniform crossover, in order to determine displac- Drilling tests were conducted under conditions using a
ing genes, the numbers 0 and 1 are generated randomly. In SX XHMT vertical drilling machine. AISI 1045 was used
this manner, the same numbers of genes are generated. In for the drilling experiments in the study. All the samples are
Table 3 the string ‘1001’ is a randomly generated number the form of 200  40  20 mm3 blocks. The cutting tools
string. For number of ‘1’ genes will be displaced, but for used for experimental study were 8 mm diameter TiN
number of ‘0’ will not be displaced (Table 3). coated HSS drills. Surface roughness (Ra) finish of each
The purpose of this process is to form a new chromo- drilled hole was measured using Taylor-Hobson’s Surtonic
some by changing the place of one or more genes of an 3+ surface roughness tester using a meter cut-off length of
existing chromosome. As a consequence of permanent 0.8 mm. The measurements of surface roughness (Ra) at

Table 8 ANOVA table for re-

sponse surface quadratic model Source SDQ DF MS F value Prop>F

Model 26.64 9 2.96 14.63 0.0009 Significant

A—Cutting speed 12.40 1 12.40 61.27 0.0001
B—Feed rate 5.41 1 5.41 26.74 0.0013
C—Cutting 0.15 1 0.15 0.72 0.4241
environment
AB 0.09 1 0.09 0.44 0.5262
AC 0.72 1 0.72 3.57 0.1008
BC 8.100E-003 1 8.100E-003 0.04 0.8471
A2 7.57 1 7.57 37.38 0.0005
B2 0.042 1 0.042 0.21 0.6637
C2 0.16 1 0.16 0.80 0.4022
Residual 1.42 7 0.20
Total 28.06 16
SD 0.45 C.V. 11.93 R2 0.9495 Pred. R2 0.1951
SDQ sum of squares, Mean 3.77 PRESS 22.59 Adj. R2 0.8846 Adeq. precision
DF degrees of freedom, 12.651
MS mean square
Int J Adv Manuf Technol (2011) 52:79–88 83

Table 9 ANOVA table for

reduced response surface Source SDQ DF MS F value Prop>F
quadratic model
Model 25.63 4 6.41 31.62 <0.0001 Significant
A—Cutting speed 12.40 1 12.40 61.19 <0.0001
B—Feed rate 5.41 1 5.41 26.71 0.0002
C—Cutting environment 0.15 1 0.15 0.72 0.0412
A2 7.67 1 7.67 37.85 <0.0001
Residual 2.43 12 0.20
Lack of Fit 2.43 8 0.30 212.08 <0.0001 Significant
Total 28.06 16
SDQ sum of squares, DF SD 0.45 C.V. 11.94 R2 0.9133 Pred. R2 0.7814
degrees of freedom, MS mean Mean 3.77 PRESS 6.13 Adj. R2 0.8844 Adeq. precision 17.903
square

90o around the drilled hole were taken at four positions 4.2 RSM based surface roughness mathematical model
from three different samples and averaged surface rough-
ness was taken for the analysis. The RSM was performed to predict the surface roughness
in drilling of AISI 1045. Table 7 gives the model summary
statistics for surface roughness.
4 Implementation (methodology) and results Table 7 reveals that quadratic model is the best suggested
model for surface roughness. So, for further analysis this model
4.1 Design of experiments was used. The relative importance of the drilling parameters
with respect to the surface roughness was investigated to
Drilling tests were conducted to assess the influence of determine more accurately the optimum combinations of the
drilling parameters on surface roughness in drilling AISI drilling parameters by using analysis of variance (ANOVA).
1045. The design of experimental was accomplished Table 8 gives the ANOVA results for the response surface
with three input parameters: cutting speed, feed rate, quadratic model for surface roughness in drilling AISI 1045.
and cutting environment. It was performed in Table 6. The model F value of 14.63 in Table 8 implies that the
Table 6 also gives experimental results for surface model is significant for surface roughness. The values of
roughness. "Prob>F" for model is less than 0.1000 which indicates that

Fig. 1 Normal probability plot of residuals for surface roughness Fig. 2 Residuals versus predicted response for surface roughness
84 Int J Adv Manuf Technol (2011) 52:79–88

the model terms are significant. In this case A, B, and A2 that interaction among drilling parameters are in significant
are significant model terms. Other model terms can be said while the independent effects of drilling parameters are also
to be not significant. These insignificant model terms can significant. Therefore, cutting speed (A), feed rate (B), and
be removed and may result in an improved model. By the second-order effect of cutting speed (A2) are the
selecting the backward elimination procedure to automati- significant model terms.
cally reduce the terms that are not significant, the resulting The “Lack of Fit F-value” of 199.84 implies the lack of fit
ANOVA table for the reduced quadratic model for surface is significant. There is only a 0.01% chance that a "Model F-
roughness is shown in Table 9. Value" this large could occur due to noise. The R2 (0.9081)
From Table 9, the model F value of 42.84 indicates that value is high, close to 1, which is desirable. Adequate
the model is still significant. The values of Prob>F less precision measures the S/N ratio. Ratio greater than 4
than 0.1000 show model terms are important. It is noted indicates adequate model discrimination. In this particular

Fig. 3 Variation of surface roughness with cutting speed at different feed rates. a MQL, b compressed air, c dry
Int J Adv Manuf Technol (2011) 52:79–88 85

case, it is 19.611 indicates an adequate signal. Therefore, the rate, and cutting environment) is obtained from coefficients
model can be used to navigate the design space. resulting using the Design Expert software. The following
In analyzing the surface roughness in drilling, statistical equations are the final empirical model in terms of coded
models play an important role. The models are used for factors for surface roughness (Ra);
prediction of results [22, 23]. RSM’s Box-Behnken design
consisting of 17 experiments was calculated for developing
the mathematical model for surface roughness. The math- Ra ¼ 4:115  0:82767A þ 8:225B þ 0:135C
ematical relationship for correlating the surface roughness
þ 0:0538A2 ð5Þ
and the considered drilling parameters (cutting speed, feed

Fig. 4 Variation of surface roughness with cutting environment at different cutting speeds. a Feed rate: 0.1 mm/rev, b feed rate: 0.2 mm/rev, c
feed rate: 0.3 mm/rev
86 Int J Adv Manuf Technol (2011) 52:79–88

The normal probability plots of the residuals and the predict the surface roughness values for any combination of
plots of the residuals versus the predicted response for drilling parameters within the range of experimentation.
surface roughness are shown in Figs. 1 and 2. Figure 1
revealed that the residuals generally fall on a straight line
implying that the errors are normally distributed. Also 5 Optimisation of surface roughness with GA
Fig. 2 shows that the residuals versus predicted responses
for surface roughness data, it is seen that no obvious pattern In this study, optimum drilling parameters for AISI 1045
and unusual structure. This implies that the models material at the lowest possible surface roughness value was
proposed are adequate and there is no reason to suspect calculated using GA. Equation 5 derived by response
any violation of the independence or constant variance surface method was taken as the objective function to be
assumption [24, 25]. minimized for the lowest surface roughness value. The
From the developed RSM-based mathematical model, flowchart of the Basic GA was given in Fig. 6.
the effect of drilling parameter on surface roughness is The variation graph of objective function (5) is shown in
examined. Figures 3a-c and 4a-c show graphs of surface Fig. 7. When Fig. 7 is checked, low values of A, B, and C
roughness as a function of drilling parameters when drillng parameters produced minimum Ra values. This result is
AISI 1045. similar to experimental study results. Moreover, in this
From Figs. 3a-c and 4a-c, it can be noticed that surface study, although it can be seen as simple study, optimum
roughness is affecting by cutting speed ant the best results drilling parameters for minimum surface roughness was
are obtained at lower cutting speeds. The values of hole obtained and Eq. 5 would provide drilling parameter
surface roughnes were much better for the MQL condition condition by using GA for the selected experimental
thn for the compressed air and dry drilling [7]. It appeared material (AISI 1045). This study would provide basic
from these figures that the surface roughness increased information for complex studies considering drilling of
under dry drilling. This case was explained due to more similar materials.
intensive temperature and stress at the hole tips [26]. Taking the minimum and the maximum values of
Additionally, comprasing Figs. 3a-c and 4a-c it can be seen drilling parameters into account, boundary conditions for
tht the surface roughenss deteriorates as feed rates is the objective function (5).
increased. This result was similar with studies cited in Ref
5  A  15 ð7Þ
[7, 27]. From these figures, it can be said that the best
condition used to drill this material is using the cutting
speed of 5 m/min, the feed rate of 0.1 mm/rev and the MQL
0:1  B  0:3 ð8Þ
conditions.

4.3 Validation of experimental results 1C3 ð9Þ

In order to predict and verify the improvement surface Equation 5 was also taken as the fitness function for the
roughness for drilling of AISI 1045 with respect to the optimization of surface roughness value obtained from
chosen initial parameter setting, confirmation tests are used. drilling AISI 1045. The algorithm given in Fig. 6 was run
The predicted optimum value of S/N ratio (ηpred.) is by using single point, double point, uniform crossover, and
calculated from the following expression. different mutation operators. This approach provides

X
k 8
hperd: ¼ hm þ ðhi þ hm Þ ð6Þ 7
Experimental
i¼1 Predicted
Surface roughness, Ra(µm)

6
where, ηm is total mean of S/N ratio, ηi is the mean of S/N 5
ratio at the optimal level, and k is the number of main 4
drilling parameters that significantly affect the performance
3
[23, 28, 29].
Figure 5 shows the validation of experimental results for 2
the surface roughness. As shown in Fig. 5, the difference 1
between the predicted the surface roughness by Eq. 5 and 0
the experimental value is small. Thus, the response 1 4 7 10 13 16 19 22 25
Trial number
equations for the surface roughness evolved through
response surface methodology can be used to successfully Fig. 5 Validation of experimental results for surface roughness
Int J Adv Manuf Technol (2011) 52:79–88 87

Best: 1.8895 Mean: 1.8896

Start 4
Best fitness

Fitness value
3 Mean fitness
Randomly generate first population (P)

Print the End 1

Function tol. <_ 10-3 result 0 5 10 15 20 25 30 35 40 45 50
Generation
Current Best Individual
8

Current best individual

X= 1
i=i+1 6 Y= 7.6239

4
X= 3
Evaluate the fitness of chromosomes in 2 X= 2 Y= 1
Y= 0.1
the generation
0
1 2 3
Number of variables (3)
Select and match qualifiedindividuals
for crossover and mutation (p’) Fig. 8 Variation of fitness function and best individuals

Initial population was randomly produced according to 100

Crossover and mutation matched iteration number 1 elitism and 30 starting number for the
individuals
drilling parameters of upper and lower limits. Function
tolerance was selected 0.001.
Evaluate the fitness of chromosomes Variation of fitness function and best individuals in the
in the population (p’) optimization are given Fig. 8. As seen in the figure, value
of the mean fitness decreases with increasing number of
iteration. Function tolerance was found after 20 iteration
Select parent individuals from (p) and
number. Best individuals taken GA program were given
(p’) populations for the next iteration
Table 10.
Considering the optimum drilling parameters in the GA,
Fig. 6 Flowchart of the basic GA the minimum surface roughness (Ra=1.89 µm) value was
obtained at V=7.62 m/min, f=0.1 mm/rev, and MQL(1).
optimum drilling conditions for corresponding, given Three experiments were performed according to the
maximum and minimum values surface roughness [13]. optimum drilling parameters for testing. As a result of
The best result for minimum surface roughness was these experiments, average surface roughness was mea-
obtained by using double point crossover. Also, various sured 1.97 µm. It was show that the results found by GA
values were examined for mutation and crossover possibil- were in conformity with the experimental and theoretical
ities. The optimum possible result could be produced by ones.
0.80 for crossover and 0.001 mutation possibilities. The
input drilling parameter levels were fed to the GA program.
6 Conclusions

Experiments were conducted on drilling machine using

TiN-coated HSS drills, the data surface roughness was
8
collected under different drilling conditions for various
7
6 f=0.3
combination of cutting speed, feed rate, and cutting
5 f=0.2 environment.
Ra f=0.1
4
3
Table 10 Best individuals for minimum surface roughness
2
1
3 Parameters Variable Value
15
2
10 Cutting speed, V (m/min) A 7.62
Cutting Environment 1 5
Cutting speed Feed rate, f (mm/rev) B 0.1
Cutting environment C MQL (1)
Fig. 7 The variation of objective function
88 Int J Adv Manuf Technol (2011) 52:79–88

Main contribution of the study is to the minimum surface 9. Zeilmann RP, Weingaertner WL (2006) Analysis of temperature
during drilling of Ti6Al4V with minimal quantity of lubricant. J
roughness and to find out optimum drilling condition using
Mater Process Technol 179:124–127
an integration of RSM and GA. RSM and GA approach 10. Suresh KRN, Venkateswara RP (2005) Selection of optimum tool
provide a systematic and effective methodology for the geometry and cutting conditions using a surface roughness
modeling and the optimization. prediction model for end milling. Int J Adv Manuf Technol
26:1202–1210
RSM provides a large amount of information with a
11. Mital A, Mehta M (1988) Surface roughness prediction models for
small amount of experimentation. The RSM based surface fine turning. Int J Produc Res 26:1861–1876
roughness model in terms of cutting speed, feed rate, and 12. Van Luttervelt CA, Childs THC, Jawahir IS, Klocke F, Venuvinod
cutting environment was developed by means of the PK (1998) Present situation and future trends in modelling of
machining operations. Progress Report of the CIRP Working
experimental database as per Box-Behnken design of Group on ‘Modelling of machining operations’. Ann CIRP 47
experiments. The quadratic modes developed using RSM (2):587–626
were reasonable accurate and can be used for prediction 13. Ozel C, Kilickap E (2006) Optimisation of surface roughness with
within the limits of the factors investigated. GA approach in turning 15% SiCp reinforced AlSi7Mg2 MMC
material. Int J Mach Machinability Mater 1(4):476–487
Genetic algorithms have been very useful in optimisation
14. Suresh PVS, Venkateswara RP, Deshmukh SG (2002) A genetic
of the response variable and also in multi-response cases. algorithmic approach for optimization of surface roughness
The RSM based surface roughness model can be optimized prediction model. Int J Mach Tools Manuf 42:675–680
using a genetic algorithm in order to find the optimum 15. Azouzi R, Guillot M (1998) On-line optimization of the turning
using an inverse process neuro controller. J Manuf Sci Eng
values of independent variables.
120:101–107
From the RSM model and the optimization results, the 16. Benardos PG, Vosniakos GC (2003) Predicting surface roughness
predicted and measured values are quite close, which in machining: a review. Int J Mach Tools Manuf 43:833–844
indicates that the developed model can be effectively used 17. Myers RH, Montgomery DC (1995) Response surface methodol-
ogy: process and product optimization using designed experi-
to predict the surface roughness. The given model can be ments. Wiley, New York
utilized to select the level of drilling parameters. Using this 18. Pradhan MK, Biswas CK (2008) Modelling of machining
model, a noticeable saving in time and cost has been parameters for MRR in EDM using response surface methodol-
obtained. ogy. Proceedings of NCMSTA’08 Conference, Hamirpur 535–542
19. David G (1989) Genetic algorithms in search, optimization and
machine learning. Addison, Wesley
20. Mitsuo G, Runwei C (1997) Genetic algorithms and engineering
design. Wiley-Interscience Publication
References 21. Zbigniew M (1996) Genetic algorithms + data structures =
evolution programs. Springer
1. Gaitonde VN, Karnik SR, Siddeswarapa B, Achyutha BT (2008) 22. Palanikumar K (2008) Application of Taguchi and response
Integrating Box-Behnken design with genetic algorithm to surface methodologies for surface roughness in machining glass
determine the optimal parametric combination for minimizing fiber reinforced plastics by PCD tooling. Int J Adv Manuf Technol
burr size in drilling of AISI 316 L stainless steel. Int J Adv Manuf 36:19–27
Technol 37:230–240 23. Palanikumar K, Karthikeyan R (2006) Optimal machining con-
2. Sahoo P, Barman TK, Routara BC (2008) Fractal dimension ditions for turning of particulate metal matrix composites using
modelling of surface profile and optimisation in CNC end milling Taguchi and response surface methodology. Mach Sci Technol
using response surface method. Int J Manuf Res 3:360–377 10:417–433
3. Bhowmick S, Alpas AT (2008) Minimum quantity lubrication 24. Noordin MY, Vankatesh VC, Sharif S, Elting S, Abdullah A
drilling of aluminium-silicon alloys in water using diamond-like (2004) Application of response surface methodology in describing
carbon coated drills. Int J Mach Tools Manuf 48:1429–1443 the performance of coated carbide tools when turning AISI 1045
4. Nandi AK, Davim JP (2009) A study of drilling performance with steel. J Mater Process Technol 145:46–68
minimum quantity of lubricant using fuzzy logic rules. Mecha- 25. Alrabii SA, Zumot LY (2007) Chip thickness and microhardness
tronics 19:218–232 prediction models during turning of medium carbon steel. J App
5. Silva LR, Bianchi EC, Catai RE, Fusse RY, França TV, Aguiar PR Math 2007:1–12
(2005) Study on the behavior of the minimum quantity lubricant- 26. Khan MMA, Dhar NR (2006) Performance evaluation of
MQL technique under different lubricating and cooling conditions minimum quantity lubrication by vegetable oil in terms of cutting
when grinding ABNT 4340 steel. J Braz Soc Mech Sci Eng 27 force, cutting zone temperature, tool wear, job dimension and
(2):192–199 surface finish in turning AISI-1060 steel. J Zhejiang University
6. Diniz AE, Micaroni R (2007) Influence of the direction and flow Science:A 7(11):1790–1799
rate of the cutting fluid on tool life in turning process of AISI 27. Mendes OC, Avila RF, Abrao AM, Reis P, Davim JP (2006) The
1045 steel. Int J Mach Tools Manuf 47:247–254 performance of cutting fluids when machining aluminium alloys.
7. Braga DU, Diniz AE, Miranda GWA, Coppini NL (2002) Using a Ind Lubr Tribol 58(5):260–268
minimum quantity of lubricant (MQL) and a diamond coated tool 28. Tosun N, Cogun C, Tosun G (2004) A study kerf and materials
in the drilling of aluminum–silicon alloys. J Mater Process removal rate in wire electrical discharge machining based on
Technol 122:127–138 Taguchi method. J Mater Process Technol 152:316–322
8. Heinemann R, Hinduja S, Barrow G, Petuelli G (2005) Effect of 29. Davidson MJ, Balasubramanian K, Tagore GRN (2008) Surface
MSS on the tool life of small twist drills in deep-hole drilling. Int roughness prediction of flow-formed AA 6061 alloy by design of
J Mach Tools Manuf 1:1–6 experiments. J Mater Process Technol 202:41–46