Materials and Manufacturing Processes

ISSN: 1042-6914 (Print) 1532-2475 (Online) Journal homepage: http://www.tandfonline.com/loi/lmmp20

A Fuzzy Logic-Based Prediction Model for Kerf

Width in Laser Beam Machining
Anamul Hossain, Altab Hossain, Y. Nukman, M. A. Hassan, M. Z. Harizam, A.
M. Sifullah & P. Parandoush
To cite this article: Anamul Hossain, Altab Hossain, Y. Nukman, M. A. Hassan, M. Z. Harizam,
A. M. Sifullah & P. Parandoush (2016) A Fuzzy Logic-Based Prediction Model for Kerf Width
in Laser Beam Machining, Materials and Manufacturing Processes, 31:5, 679-684, DOI:
10.1080/10426914.2015.1037901
To link to this article: http://dx.doi.org/10.1080/10426914.2015.1037901

Accepted author version posted online: 05

Jun 2015.
Published online: 05 Jun 2015.
Submit your article to this journal

Article views: 170

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=lmmp20
Download by: [King Faisal University]

Date: 16 February 2016, At: 01:00

Materials and Manufacturing Processes, 31: 679684, 2016

Copyright # Taylor & Francis Group, LLC
ISSN: 1042-6914 print=1532-2475 online
DOI: 10.1080/10426914.2015.1037901

A Fuzzy Logic-Based Prediction Model for Kerf Width

in Laser Beam Machining
Anamul Hossain1, Altab Hossain1, Y. Nukman1,2, M. A. Hassan1,2, M. Z. Harizam1,
A. M. Sifullah1, and P. Parandoush1
1

Downloaded by [King Faisal University] at 01:00 16 February 2016

Department of Mechanical Engineering, Faculty of Engineering, University of Malaya, Kuala, Lumpur, Malaysia
2
Department of Mechanical Engineering, Faculty of Engineering, Assiut University, Assiut, Egypt

In laser beam machining, the main concern is the machining quality as kerf width of the end product. It is essential for industrial
applications to cut the workpiece with minimum kerf width. However, it is difcult to develop a precise functional relationship between
input and output variables in laser machining. Therefore, an effort has been conducted to build up an intelligent fuzzy expert system
(FES) model to predict the kerf width in CO2 laser cutting. The employed input parameters were assisting gas pressure, laser power, cutting
speed, and standoff distance. The fuzzy logic was performed on fuzzy toolbox in MATLAB R2009b by employing Mamdani technique. In
total, 81 experiments were carried out and experimental results were used for training and testing of the developed fuzzy model. Relative
error and goodness of t were used to investigate the accuracy of the prediction ability and the values of 3.852% and 0.994, respectively, were
found to be satisfactory. This paper will extend knowledge about the prediction of kerf width by using FES model.
Keywords Fuzzy; Kerf; Laser; Model; PMMA.

frequency [3]. Goeke et al. [4] presented the application of

CO2 laser cutting on carbon ber-reinforced plastics
(CFRPs). They observed that radiation of laser and
absorption into material have an indicative inuence on
heat-affected zone and kerf width. Quintero et al. [5]
applied CO2 laser cutting on phenolic resin boards
(PRBs) and particleboard plates that were covered with
a melamine sheet imitating beechwood. They used a
design of experiment (DOE) and found the output laser
cutting quality as kerf width and surface roughness.
Dubey et al. [6] applied a combined method of Taguchi
Method and Response Surface Method (TMRSM) for
the optimization of silicon alloy steels width of kerf
and material removal rate (MRR). They also developed
a mathematical model of kerf width and MRR, according
to the laser input parameters such as gas pressure, pulse
width, cutting speed, and frequency of the pulse. Pandey
et al. [7] analyzed the kerf quality of Ti alloy in pulsed
laser cutting by an intelligent FES. Sheng et al. [8] proposed an analytical prediction model of kerf width as a
relation to erosion front. Pandey et al. [9] developed
FES model for duralumin sheet to get minimum kerf
width and also deviations on top and bottom sides. Syn
et al. [10] analyzed surface quality of Incoloy alloy by
an intelligent FES model. It has not been found in literature that articial intelligent (AI) such as an FES is
applied on nonmetal like PMMA to predict kerf width.
Therefore, an effort has been carried out to build up a
model of kerf width in CO2 laser cutting by FES. The uses
of PMMA are being increased not only for as a substitute
of glass but also for bone cement, prostheses, intraocular
lens, dentures, biosensors, and biomechanical device [11].
This research will be helpful to nd the optimum cutting

INTRODUCTION
The applications of laser in the industry have been
growing up signicantly because of its outstanding benets, such as machining quality, productivity, saving of
the pre- and post-processing of the materials, and minimum loss of the parent workpiece as well as technical
benets [1, 2]. In laser cutting processes, a kerf width is
formed through the relative movement of the workpiece
and laser beam. This movement allows creating an intricate shape on a at workpiece. Mechanism of laser cutting is one kind of thermal process that depends on
laser power which is divided into conduction, energy of
melting, and vaporization and energy losses to the
environment to balance the input laser energy. The objective of this research to investigate kerf width for different
input parameters such as assisting gas pressure, laser
power, cutting speed, and standoff distance onto 3 mm
thickness of polymethylmethacrylate (PMMA) and an
effort has been carried out to establish a fuzzy expert system (FES) model in order to predict the kerf width.
Laser machining quality is strongly dependent on input
parameters. In laser machining process, laser beam hits
on workpiece which can be restrained by the adjustment
of the assisting gas pressure, laser power, scanning speed,
beam mode (continuous wave (CW) or pulsed), and beam
Received December 15, 2014; Accepted March 8, 2015
Address correspondence to Altab Hossain, Department of
Mechanical Engineering, Faculty of Engineering, University of Malaya,
50603 Kuala Lumpur, Malaysia; E-mail: altab75@um.edu.my,
altab76@gmail.com
Color versions of one or more of the gures in the article can be
found online at www.tandfonline.com/lmmp.

679

680

A. HOSSAIN ET AL.

Downloaded by [King Faisal University] at 01:00 16 February 2016

conditions and take decisions for adjusting the input

parameters of PMMA material in laser machining
industry with minimum loss of the parent material.
MATERIALS AND METHODS
All the experiments have been performed by a Zech
laser machine that can provide up to 500 W CW CO2
laser with the mode structure of TEM01. It consists of
two parts: a laser generation system (ZLX5) and a laser
cutting workstation (ZL 1010). In this experiment, laser
machine was associated with a computer system to control the operation. AutoCAD software was used for
drawing. C-Cut software was applied to synchronize
movements of the laser head with cutting direction.
The material used for this experiment was PMMA with
thickness of 3.0 mm. A digital caliper, resolution of
0.01 mm and range from 0 to 150 mm, was employed
to get the measurement of the lengths H1 and H2 as
shown in Fig. 1. Corners power, nozzle diameter, and
delay time were constant in this experiment. Table 1
shows the list of the input parameters with different
levels. Linear rectangular cuts were made on a PMMA
workpiece. A schematic of the cutting nozzle and optics
is shown in Fig. 2. A converging lens (thickness 6 mm)
with a xed focal length (127 mm) was installed along
the beam path. Standoff distance was varying from 1
to 10 mm to produce laser spot on the workpiece. To
measure the upper kerf width, H1 and H2 side lines were
picked out three times at several locations and were
computed by the following equation:
Kerfwidth

H1  H2
2

The interaction of the laser beam and material is a complex phenomenon. It can be classied as absorption,
reection, conduction, convection, melting, vaporization, radiation, scattering, and transmission [12]. But
laser machining of PMMA is different from other metal
because CO2 laser interaction with PMMA begins
vaporization while it is subjected to melt shearing [13,
14]. Hence, the statement of the laser machining of
PMMA is vague information. The principle of FES
has been vindicated as an efcient tool to manage this

FIGURE 1.Kerf width measurement.

TABLE 1.Input variables and their levels

Input parameters

Level 1

Level 2

Level 3

Units

Assist gas pressure: A

Laser power: B
Cutting speed: C
Standoff distance: D

0.5
100
0.2
1

2.5
300
0.7
5

4.5
500
1.2
10

Bars
Watt
m=min
Mm

kind of information [9, 15]. Generally, researchers use

statistical analysis for evaluation and investigation of
the laser parameters [16, 17] based on experimental result
only. On the other hand, the fuzzy logic model is a
multi-input multi-output modeling technique which combines different heterogeneous data (such as experimental
data, numerical data, mathematical data, recommendation, and suggestions). Also, we have found relative
error in regression analysis for the experimental data
alone was 9.38% (R2 0.907), which is greater than fuzzy
model relative error 3.852% (R2 0.989). This paper concentrates on fuzzy logic because our further implication is
fuzzy controller for autolaser cutting process.
Generally the variables are dened as linguistic variables. Each linguistic variable has linguistic values that
may be expressed as synthetic language, words, and sentences. In this study, 3 linguistic values were used for
input variable (Low, Medium, and High) and 15 linguistic
values were used for the kerf width (kw1, kw2, kw3, kw4,
kw5, kw6, kw7, kw8, kw9, kw10, kw11, kw12, kw13,
kw14, and kw15). PMMAs kerf width prediction of
FES was implemented on MATLAB R2009b software.
For getting more precise results, more membership
functions were used in output. For fuzzication, two
membership functions are common: triangle and
trapezoidal membership function [18]. Triangle membership function was used because it is dened as a simplest
membership function [19]. In this research, IFTHEN
rules have been built up by a set of 64 experimental data,
which is called as training data. Rules are built up in the
MATLAB rule editor and can be accessed in the rule
viewer. In this CO2 laser study fuzzication of the

FIGURE 2.A schematic of the cutting nozzle and optics.

681

A FUZZY LOGIC-BASED PREDICTION MODEL

linguistic parameters were made by the following function:


GPi1

LPi2

CSi3

SDi4

i1 ; 0:5  i1  4:5
0;
otherwise

i2 ; 100  i2  500
0;
otherwise
i3 0:2  i3  1:2
0;
otherwise
i4 ; 1  i4  10
0;
otherwise


Downloaded by [King Faisal University] at 01:00 16 February 2016

KW o1

2

3


4

o1 ; 0  o1  1:9
0;
otherwise

5

6

where i1, i2, i3, and i4 are input variables as assisting gas
pressure, laser power, cutting speed, and standoff
distance, respectively, and o1 is the output variable kerf
width.
By the implication process, fuzzy rules were evaluated.
Mamdani implication was used because there were three
membership functions for each input linguistic variables
which were taken into consideration. The implication
results were analyzed by an aggregation process. In this
process, output results of the rules were aggregated into
a single fuzzy set. The fuzzication process, membership
function, and the linguistic fuzzy sets of assisting gas
pressure (GP), laser power (LP), cutting speed (SD),
standoff distance (SD), and kerf width (KW) interval
can be deduced from Eqs. (2)(6) as follows:

lMedium GP

lMedium LP

lLow CS

8
<
:

lMedium SD

8
<
:
8
<
:

9
100  i2  300 =
300  i2  500 ;
0; i2 > 500

i2 100
300100 ;
500i2
500300 ;

9
=

9
1  i4  5 =
5  i4  10 ;
0; i4 > 10

10

i4 1
51 ;
10i4
105 ;

11

where y is the crisp value of the kerf width and x is the

position of the singleton in the respective universe.
In the stage of defuzzication, truth degree (l) of rules
can be calculated by the aid of minimum for individual
rules and then maximum for between two working rules.
One example is considered, to perceive fuzzication,
input variable GP (i1) 2.5 bars, LP (i2) 300 W, CS
(i3) 0.2 m=min. and SD (i4) 5 mm, then the rule 24
is red. The ring strength a of the rule 24 is obtained
as follows (from Eqs. (7)(10)):
a24 min
flMedium GP; lMedium LP; lLow CS; lMedium SDg
minf1; 1; 1; 1g 1
Thus, the membership function for rule 24 is obtained
as follows:
l24 KW minf1; lkw9 KW g
According to the view of Rajasekaran et al. [20], crisp
decision is more accurate if output is represented as a
single scalar. By using Eq. (11), the crisp output of kerf
width found 1 mm that was close to MATLAB result of
0.995 mm.
The mathematical and statistical methods have been
used to investigate the accuracy of the prediction ability.
Relative error (e) of formation has been established by
using the following formula:

n 
X

yi  yi  100%
e
12
 y  n
i
i1
Another method goodness of t (g) for the prediction
capability of the system is calculated as follows:
v
u
n
P
u
yi  yi 2
u
u
13
g u1  i1
n
P
t
yi  yi 2
i1

1 ; i3  0:2
0:2  i3  0:7
;
0; i3 > 0:7

0:7i3
0:70:2 ;

8
<

9
0:5  i1  2:5 =
2:5  i1  4:5 ;
0; i1 > 4:5

i1 0:5
2:50:5 ;
4:5i1
4:52:5 ;

P 
x lAx
y P
lAx

The output result is obtained as fuzzy variables. By

defuzzication process, the fuzzy output can be converted into a relative value. Centroid method was used
for defuzzication to nd nonfuzzy value y [19] as
expressed in the following equation:

where n is the total experiment number, yi is the experimental value, y^i is the fuzzy logic predicted value, and yi
is the average of the experimental value.
RESULTS AND DISCUSSION
The kerf width prediction of the fuzzy logic model has
been developed in the present study as an effect of
assisted gas pressure, laser power, cutting speed, and
standoff distance. Linguistic variables were fuzzied by
triangle membership function. Centroid method was
used to defuzzication. A total number of 81 experiments were carried out where 64 data were used to train
and 17 data were applied to test the validation of the

Downloaded by [King Faisal University] at 01:00 16 February 2016

682

A. HOSSAIN ET AL.

cutting process to predict kerf width. Table 2 shows

the comparison of prediction error for articial neural
network (ANN), Taguchi ANN, Taguchi ANN with
genetic algorithm, and FES model with the experimental
values [21].
According to the statistical p values presented in
Table 3; its clear that standoff distance, laser power,
and cutting speed are signicant for upper kerf width.
The beam diameter is approximately proportional to
standoff distance. From the surface plot (Fig. 4(a)),
it can be said that the kerf width becomes wider
with the increase of standoff distance. Because with
the increase of standoff distance, the area of the beam
spot also increases, so that it leads to the area of
enhanced thermal energy resulting in wider kerf width.
On the other hand, at higher speeds, kerf width
decreases, which is in agreement with Tamrin et al.
[13] and Cenna et al. [22]. From the surface plot
(Fig. 4(b)), it shows that with increasing of laser power
the kerf width also increases because of this thermal
energy is also agreement with Fu et al. [23].
CONCLUSIONS
In this study, a Mamdani FES model has been proposed to predict kerf width for the inuence of gas pressure, laser power, standoff distance, and cutting speed on
laser beam machining of PMMA sheet. All the experiments were performed by varying the input parameters
with different levels. Experimental results were used
for training and testing of the developed fuzzy model.
The fuzzy logic model was performed on fuzzy toolbox
in MATLAB R2009b. Prediction ability of the model
was satisfactory in terms of relative error and goodness
of t. Hence, the proposed model can be used in PMMA
laser machining simulation.
From the study, the following conclusions can be drawn:

FIGURE 3.Correlation between experimental and fuzzy logic-predicted

kerf width: (a) for training and (b) for testing.

developed intelligent model. Figure 3 shows the accuracy of FES as the goodness of t is 0.994
(R2 0.989), relative error 3.852% and 0.97
(R2 0.966), relative error 5.2% for training and testing,
respectively. Relative errors were found less than 10%
for both training and testing. Nukman et al. [21]
accepted less than 10% prediction error in CO2 laser

1. The prediction capability of the developed model

has been found better than Taguchi ANN model
[21].

TABLE 2.Comparison of prediction error with experimental values [21].

Input Parameters
Predicted
check Experiment

1
2
3
4

Predicted % error

Assist gas pressure

Laser power

Cutting speed

Standoff distance

Normal ANN

Taguchi ANN

Taguchi ANN GA

Fuzzy Model

4.5
0.5
2.5
2.5

100
100
300
500

0.2
0.7
0.7
0.2

1
5
1
5

7.78
4.93
9.21
36.84

8.89
6.17
7.90
21.05

7.78
0
5.26
5.26

3.39
0.18
7.08
0

TABLE 3.Regression analysis

Intercept
Assist gas pressure
Laser power
Cutting speed
Standoff distance

Coefcients

Standard error

t Stat

P-value

Lower 95%

Upper 95%

0.458091017
0.01101022
0.000898931
0.28477947
0.059464585

0.041880916
0.007512731
7.98653E05
0.031977372
0.003558002

10.93794173
1.465541711
11.25559063
8.905655826
16.71291554

1.00827E15
0.148173634
3.2331E16
1.89836E12
7.65538E24

0.374257256
0.026048585
0.000739063
0.348789134
0.05234247

0.541924777
0.004028144
0.001058799
0.220769805
0.066586699

683

Downloaded by [King Faisal University] at 01:00 16 February 2016

A FUZZY LOGIC-BASED PREDICTION MODEL

FIGURE 4.Control surface of the fuzzy inferring system: (a) effects of standoff distance and cutting speed on kerf width (at laser power 300 W and gas
pressure 2.5 bar); (b) effects of laser power and cutting speed on kerf width (at standoff distance 5 mm and gas pressure 2.5 bar); (c) effects of laser power
and gas pressure on kerf width (at cutting speed 0.7 m=min and standoff distance 5 bar).

3. The statistical analysis of the fuzzy result for kerf

width has shown the relative error (3.852%) and
goodness of t 0.994 (R2 0.989) in training,
whereas relative error (5.2%) and goodness of t
0.97 (R2 0.966) were obtained in testing.
4. When cutting a 3.0 mm sheet of PMMA with a CW
CO2 laser, kerf width becomes wider with the
increase of standoff distance and power but at higher
speed kerf width decreases.
5. The minimum kerf width was 0.3167 mm at gas
pressure 0.5 bar, laser power 100 W, standoff distance 1 mm and cutting speed 1.2 m=mm.

ACKNOWLEDGMENTS
The Faculty of Engineering, University of Malaya,
Malaya, is gratefully acknowledged by the authors

FUNDING
Financial and technical support from the University of
Malaya under the Grant RP010A-13AET=RP010B13AET=RP010C-13AET (UMRG) and the faculty of
engineering are gratefully acknowledged by authors.
REFERENCES
1. Rajaram, N.; Sheikh-Ahmad, J.; Cheraghi, S.H. CO2 laser cut
quality of 4130 steel. International Journal of Machine Tools
and Manufacture 2003, 43(4), 351358.
2. Schuocker, D. Laser cutting. Materials and Manufacturing
Processes 1989, 4 (3), 311330.
3. Dubey, A.K.; Yadava, V. Laser beam machiningA review.
International Journal of Machine Tools and Manufacture 2008,
48 (6), 609628.
4. Goeke, A.; Emmelmann, C. Inuence of laser cutting parameters on CFRP part quality. Physics Procedia 2010, 5,
253258.

Downloaded by [King Faisal University] at 01:00 16 February 2016

684
5. Quintero, F.; Riveiro, A.; Lusqui~
nos, F.; Comesa~
na, R.; Pou,
J. CO2 laser cutting of phenolic resin boards. Journal of Materials Processing Technology 2011, 211 (11), 17101718.
6. Kumar Dubey, A.; Yadava, V. Multi-objective optimisation
of laser beam cutting process. Optics & Laser Technology
2008, 40 (3), 562570.
7. Pandey, A.K.; Dubey, A.K. Fuzzy expert system for prediction
of kerf qualities in pulsed laser cutting of titanium alloy sheet.
Machining Science and Technology 2013, 17 (4), 545574.
8. Sheng, P.; Chryssolouris, G. Investigation of acoustic sensing
for laser machining processes Part 2: Laser grooving and
cutting. Journal of Materials Processing Technology 1994, 43
(24), 145163.
9. Pandey, A.K.; Dubey, A.K. Taguchi based fuzzy logic optimization of multiple quality characteristics in laser cutting
of Duralumin sheet. Optics and Lasers in Engineering 2012,
50 (3), 328335.
10. Syn, C.Z.; Mokhtar, M.; Feng, C.J.; Manurung, Y.H.P.
Approach to prediction of laser cutting quality by employing
fuzzy expert system. Expert Systems with Applications 2011,
38 (6), 75587568.
11. Frazer, R.Q.; Byron, R.T.; Osborne, P.B.; West, K.P. PMMA:
An essential material in medicine and dentistry. Journal of
Long-Term Effects of Medical Implants 2005, 15 (6), 629639.
12. Parandoush, P.; Hossain, A. A review of modeling and simulation of laser beam machining. International Journal of
Machine Tools and Manufacture 2014, 85, 135145.
13. Tamrin, K.F.; Nukman, Y.; Sheikh, N.A.; Harizam, M.Z.
Determination of optimum parameters using grey relational
analysis for multi-performance characteristics in CO2 laser
joining of dissimilar materials. Optics and Lasers in
Engineering 2014, 57, 4047.
14. Powell, J. CO2 Laser Cutting; Springer-Verlag: London, UK,
1993.

A. HOSSAIN ET AL.
15. Hassan, M.A.; El-Sharief, M.A.; Aboul-Kasem, A.;
Ramesh, S.; Purbolaksono, J. A fuzzy model for evaluation
and prediction of slurry erosion of 5127 steels. Materials &
Design 2012, 39, 186191.
16. Hernandez-Castaneda, J.; Sezer, H.K.; Li, L. Statistical
analysis of ytterbium-doped bre laser cutting of dry pine
wood. Proceedings of the Institution of Mechanical Engineers,
Part B: Journal of Engineering Manufacture 2009, 223 (7),
775789.
17. Wee, L.; Crouse, P.; Li, L. A statistical analysis of striation
formation during laser cutting of ceramics. The International
Journal of Advanced Manufacturing Technology 2008, 36
(78), 699706.
18. Pedrycz, W. Why triangular membership functions? Fuzzy
Sets and Systems 1994, 64 (1), 2130.
19. George, B.Y.; Klir, J. Fuzzy Sets and Fuzzy Logic: Theory
and Applications; Prentice-Hall: Upper Saddle River, NJ,
1994.
20. Rajasekaran, S.; Pai, G.V. Neural Networks, Fuzzy Logic and
Genetic Algorithm: Synthesis and Applications; Rajkamal
Electric Presss: New Delhi, India, 2003.
21. Nukman, Y.; Hassan, M.A.; Harizam, M.Z. Optimization of
prediction error in CO2 laser cutting process by Taguchi
articial neural network hybrid with genetic algorithm.
Applied Mathematics & Information Sciences 2013, 7 (1),
363370.
22. Cenna, A.A.; Mathew, P. Analysis and prediction of laser
cutting parameters of bre-reinforced plastics (FRP)
composite materials. International Journal of Machine Tools
and Manufacture 2002, 42 (1), 105113.
23. Fu, C.H.; Guo, Y.B.; Sealy, M.P. A predictive model and
validation of laser cutting of nitinol with a novel moving
volumetric pulsed heat ux. Journal of Materials Processing
Technology 2014, 214 (12), 29262934.