N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
813–822
OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF
GALVANIZED STEEL USING ARTIFICIAL NEURAL NETWORK
COUPLED WITH GENETIC ALGORITHM
OPTIMALNA DOL@INA PODPORE ([ABLONE, VODILA) PRI
TERMI^NEM VRTANJU GALVANIZIRANEGA JEKLA Z UPORABO
UMETNE NEVRONSKE MRE@E IN GENETSKEGA ALGORITMA
Navasingh Rajesh Jesudoss Hynes1, Ramar Kumar1, Jebaraj Angela Jennifa Sujana2
1Mepco Schlenk Engineering College, Department of Mechanical Engineering, Sivakasi, Vindhunagan, 626005 Tamil Nadu, India
2Mepco Schlenk Engineering College, Department of Information Technology, Sivakasi, Vindhunagan, 626005 Tamil Nadu, India
findhynes@yahoo.co.in, mepcokumar@gmail.com, ang_jenefa@mepcoeng.ac.in
Prejem rokopisa – received: 2016-09-27; sprejem za objavo – accepted for publication: 2017-01-24
doi:10.17222/mit.2016.290
Thermal drilling is a novel sheet-metal-hole-making technique that utilizes the heat produced at the interface of the rotating
conical tool and workpiece in order to soften the workpiece and pierce a hole into it. In this work, experiments with thermal
drilling of galvanized steel were conducted based on the Taguchi L27 orthogonal array. Significant process parameters such as
rotational speed, tool angle and workpiece thickness were varied during the experimentation. In thermal drilling, the
thermal-drill tool pushes aside a large amount of workpiece material to form a sleeve, which is often referred to as the bushing
length. A predictive model for the bushing length was developed using a feed-forward artificial neural network based on experi-
mental data. As the bushing length is closely associated with the tapping process, the influences of the input process parameters
play a vital role in fastening galvanized steel with threaded fasteners in diverse engineering applications. The optimization
problem was solved by implementing a genetic algorithm under constraint limits to maximize the bushing length. Further, a
confirmation test was conducted with the intention to compare the optimum value and its corresponding bushing length
predicted by the genetic algorithm. Good agreement was observed between the predicted and the experimental values.
Keywords: thermal drilling, artificial neural network, genetic algorithm, galvanized steel, bushing length
Termi~no vrtanje je nova, za vrtanje lukenj v plo~evino, uporabljena tehnika, ki izkori{~a toploto, proizvedeno na povr{ini
vrte~e se konice orodja na obdelovancu z namenom, da ga zmeh~a in vanj naredi luknjo. V delu so bili izvedeni preizkusi na
osnovi metode Taguchi L27 z ortogonalno matriko s termi~nim vrtanjem galvaniziranega jekla. Pomembni parametri postopka,
kot so: hitrost vrtenja, kot orodja in debelina obdelovanca, so se med eksperimentiranjem spreminjali. Pri toplotnem vrtanju,
vrtanje orodja potisne stran ve~ materiala obdelovanca tako, da se tvori navarek (rokav) okoli luknje, ki se pogosto omenja kot
dol`ina {ablone. Napovedni model za dol`ino {ablone, je razvit z uporabo umetne nevronske mre`e, ki temelji na znanstvenih
podatkih. Ker je dol`ina {ablone precej povezana s procesom izdelave navoja, vplivi teh vhodnih procesnih parametrov igrajo
klju~no vlogo pri pritrditvi galvaniziranega jekla z navojem pritrdilnih elementov v razli~nih in`enirskih aplikacijah. Problem
optimizacije je bil re{en z implementacijo genetskega algoritma na podlagi omejitev za pove~anje dol`ine {ablone. Ugotovljeno
je bilo dobro ujemanje med napovedano in eksperimentalno vrednostjo.
Klju~ne besede: termi~no vrtanje, umetna nevronska mre`a, genetski algoritem, galvanizirano jeklo, dol`ina {ablone
1 INTRODUCTION
Thermal drilling is an emerging hole-making process
with significant breakthroughs in many drilling situ-
ations in both automobile and aerospace applications.
This process uses the heat energy produced at the inter-
face in order to soften and then pierce the workpiece.1
Moreover, this heat energy enhances the flow ability of
the workpiece material, which is extruded onto both the
front and back sides of a drilled hole. Finally, the
extruded or deformed material forms a bushing shape,
which surrounds the drilled hole.2 The length of the
bushing formed on the workpiece after a thermal-drilling
operation is called the bushing length. Achieving a
sufficient bushing length is very important since the
bushing length can increase the threading depth and the
clamp-load-bearing capability in various engineering
applications.
Contrary to the conventional drilling process, in
thermal drilling, there is no chip and wastage of the ma-
terial, since all the deformed material contributes to
developing a bushing. As it abolishes chip generation, it
can be seen as a clean, eco-friendly and chipless hole-
making technique. When joining metal sheets or thin-
walled structures by drilling holes using the conven-
tional-drilling process, just one or two threads can be
made in them; however, reliability is a matter of great
concern in such a situation. Alternatively, weld nuts and
threaded rivet nuts are used, but due to thermal
distortion, they get jammed and twisted during the
fastening of the structures.3 On the other hand, during the
thermal-drilling process, the bushing formed is about 3
to 4 times thicker than the workpiece. The bushing thus
allows a greater contact area and high structural rigidity
of fasteners in joining situations.4
Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822 813
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
UDK 519.6:620.1:67.017 ISSN 1580-2949
Original scientific article/Izvirni znanstveni ~lanek MTAEC9, 51(5)813(2017)
A. H. Streppel and H. J. J Kals5 suggested that the
thermal-drilling process could be applied to different
materials like carbon steels, stainless steel, copper, brass
and aluminium. They carried out experimentation on the
thermal drilling of aluminium. Due to strong pre-strain
hardening, the process resulted in bad quality of the
bushing. M. Kerkhofs and M. Van Stappen6 compared
the performance of (Ti, Al) N coated tungsten carbide
thermal drills with uncoated drills. They reported that the
coated thermal drills allowed a longer tool life than the
uncoated flow drills. S. F. Miller et al.7 conducted
experiments with the thermal-drilling process on steel,
aluminum and titanium. They studied the properties such
as the hardness and microstructure of drilled holes. They
concluded that in thermal drilling large deformation and
high frictional heat are generated, resulting in the
changes in the material properties and microstructure of
a workpiece. They reported that the thermal conductivity
of the material drilled affects the quality and integrity of
the hole.
S. F. Miller et al.8 conducted an experimental and
numerical analysis of a thermal-drilling process on the
AISI 1020 cold-rolled carbon steel under a constant feed
rate. Moreover, they developed two classic models for
the thermal-drilling process. They predicted the tempe-
rature distribution using a finite-element-based thermal
model and in another attempt, they predicted the thrust
force and torque using the basic principles of physics. S.
F. Miller et al.9 thermally drilled into aluminium and
magnesium alloys using tungsten carbide and titanium
carbide in a cobalt matrix under different rotational
speeds and feed rates. They analyzed the thrust force,
torque, energy, power and peak power required for
drilling. They also evaluated the formation of the
bushing shape of cast metals. The bushings produced
during the thermal drilling of brittle cast metals demon-
strated a radial fracture. In their conclusions, they
suggested to pre-heat a cast metal workpiece and create a
high rotational speed for obtaining a cylindrically shaped
bushing without a significant radial fracture.
S. F. Miller et al.10 developed a three-dimensional
finite-element model for the thermal-drilling process in
order to evaluate the plastic strain and deformation of a
workpiece. S. F. Miller et al.11 investigated thermal-drill
tool-wear characteristics during the thermal drilling of an
AISI 1015 carbon steel workpiece. Furthermore, the
thrust force and torque were analyzed. They concluded
that the carbide tool is durable and it demonstrates the
least tool wear after drilling 11.000 holes. However,
progressively severe abrasive grooving on the tool tip
was observed. S. M. Lee et al.12 carried out thermal
drilling into the IN-713LC super alloy under different
rotation speeds and feed rates. They investigated the
material properties such as hardness, roundness, and
surface roughness of the holes drilled into the IN-713LC
super alloy. They reported that the hardness is higher
near the wall of a drilled hole and it decreases with the
increasing distance from the edge of the hole. In addition
to that, higher rotational speeds and feed rates demon-
strate better roundness and lower surface roughness.
H. M. Chow et al.13 optimized the process parameters
of a thermally drilled austenite stainless-steel workpiece
using the Taguchi method.14 They considered the input
parameters such as friction angle, friction/contact-area
ratio, feed rate and drilling speed, and studied their in-
fluence on the response parameters like surface rough-
ness. Moreover, the hardness and microstructural aspects
of drilled holes were studied. It was observed that the
surrounding area of a drilled hole acquired a fine grain
size and a compact structure with a higher microhardness
than that of the area away from the drilled region. S. M.
Lee et al.15 employed thermal drilling for the AISI 304
stainless steel using tungsten carbide drills with and
without coating. Their results illustrated that at the same
rotational speed and for the same number of holes
drilled, the coated drills experienced less tool wear than
the uncoated drills. Furthermore, they investigated the
changes in the relationship between the drilled-surface
temperature, tool wear and axial thrust force.
W. L. Ku et al.16 optimized the parameters of thermal
drilling into austenite stainless steel using the Taguchi
method. They studied the effects of the friction angle,
friction/contact-area ratio, feed rate and rotational speed
on the response-quality characteristics such as the
surface roughness and bushing length. They revealed that
at optimized drilling conditions, the bushing length of a
drilled hole was nearly three times longer than the plate
thickness, and a mirror-like quality wall surface of the
drilled hole could be obtained. M. Folea et al.17 investi-
gated the thermal drilling of a maraging-steel workpiece.
The authors reported that the temperature was the most
important factor in the thermal-drilling process. They
also revealed that a greater quality of the holes drilled
into maraging steel could be achieved with higher rota-
tional speeds. T. K. Mehmet et al.18 studied the effects of
the thermal-drilling parameters such as friction angle,
friction/contact-area ratio, feed rate and rotational speed
on workpiece temperature, thrust force and torque in
thermal drilling of the ST12 material. They revealed that
the thrust force and torque reduce with the increasing
rotational speed and increase with the friction angle, feed
rate and friction/contact-area ratio.
D. Biermann and Y. Liu19 demonstrated the feasibi-
lity of the thermal drilling of a magnesium wrought alloy
and analyzed the thrust forces and torque. They
measured the process temperature online and examined
the strength of the joint through tapping and thread
forming. B. B. Mehmet20 experimentally and numerically
investigated the thermal drilling of the AISI 1020 steel.
In their study, an analytical model for the torque, axial
power and heat-transfer coefficient was developed. Good
agreement was observed between experimental and
numerical values.
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
814 Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Similar work was performed by P. Krasauskas et al.21
on the thermal drilling into the AISI 304 steel. P. D.
Pantawane and B. B. Ahuja22 carried out regression
modelling of the thermal drilling of the AISI 1015 steel
using the Taguchi method. They studied the effects of the
tool-diameter ratio and rotational speed on the material
thickness and the effect of the feed on the response
parameters including the thrust force, torque and surface
roughness of drilled holes. G. Somasundaram et al.23
carried thermal drilling on the Al-SiC composite
material and studied the roundness errors on drilled
holes. They considered the input process parameters
such as the composition of workpiece, workpiece thick-
ness, rotational speed and feed rate. They developed an
empirical relation between the process parameters and
the roundness error using the response surface methodo-
logy.
Artificial neural network (ANN) is a biologically in-
spired computer program designed to simulate the way,
in which the human brain processes information.24 ANN
is widely applied in modeling many machining ope-
rations like turning, milling and drilling. S. R. Karnik
and V. N. Gaitone25 used a multilayer feed-forward ANN
to predict the influence of the process parameters such as
the cutting speed, feed, drill diameter, point angle and
lip-clearance angle on the burr height and burr thickness
in drilling the AISI 316L stainless steel. R. S. Mamilla et
al.26 applied a multilayer ANN for modeling an abrasive
flow-finishing process on the AISI 1040 and AISI 4340
steel. S. Assarzadeh and M. Ghoreishi27 successfully
used a multilayer ANN for modeling the metal-removal
rate and surface roughness in an electrical discharge
machining process involving the BD3 steel. O. Babur et
al.28 developed an ANN model according to experimental
measurement data for the end milling of Inconel 718.
They coupled a genetic algorithm (GA) with an ANN to
forecast the surface roughness. S. Sarkar et al.29 pro-
duced a multilayer feed-forward-ANN model to predict
the process parameters of the machining of  titanium
aluminide with a wire-electrical-discharge machine. A.
K. Singh et al.30 used a multilayer feed-forward ANN to
predict the flank wear of high-speed steel drill bits for
drilling holes into a copper workpiece.
The literature review on thermal drilling reveals that
experimental investigations and numerical simulations of
the effects of mechanical and physical properties on the
thermal-drilling process was mainly carried out on
copper, brass, aluminium, magnesium and stainless-steel
alloys, maraging steel, ST12 steel and IN-713LC super
alloy. Galvanized steel is being widely used for car body
structures and it is a good candidate for the implemen-
tation of the thermal-drilling process. The application of
galvanized steel is widely used in the areas of roofing
material, doors, ship’s ducts and panels, electrical-
appliance automobile parts, etc.31
In the present work, thermal drilling was carried out
on galvanized steel. Based on experimental measurement
data, an ANN model was developed to predict the for-
mation of the bushing length in thermal drilling of
galvanized steel. This model considers rotational speed,
tool angle and workpiece thickness as significant ther-
mal-drilling process parameters. Then the optimum
values of these drilling parameters were computed, with
the aim of achieving the maximum bushing length, by
solving the optimization problem implementing a GA.
Thus, the present work allows us to achieve the maxi-
mum bushing length, which is highly desirable for the
subsequent tapping and joining in automotive applica-
tions. A schematic diagram of the combined ANN-GA
optimization24 is shown in Figure 1. It indicates the
stages of the GA process and its relation with the ANN
process.
2 THERMAL DRILLING
2.1 Physical description of the process
Figure 2 shows a schematic representation of the
thermal-drilling process. Based on the thermal-drill
geometry, there are four steps involved in thermal drill-
ing. Initially, the center point zone of a drill approaches
and pierces the workpiece. Secondly, the produced heat
softens the workpiece as a result of the friction between
the thermal drill and the workpiece. Thirdly, the softened
material is pushed downward and the drill moves for-
ward to form the bushing using the cylindrical zone of
the tool. Fourthly, the extruded burr on the workpiece
surface is pressed to form a boss with the shoulder zone
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822 815
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 1: Flow chart of the integrated ANN-GA optimization process
A. H. Streppel and H. J. J Kals5 suggested that the
thermal-drilling process could be applied to different
materials like carbon steels, stainless steel, copper, brass
and aluminium. They carried out experimentation on the
thermal drilling of aluminium. Due to strong pre-strain
hardening, the process resulted in bad quality of the
bushing. M. Kerkhofs and M. Van Stappen6 compared
the performance of (Ti, Al) N coated tungsten carbide
thermal drills with uncoated drills. They reported that the
coated thermal drills allowed a longer tool life than the
uncoated flow drills. S. F. Miller et al.7 conducted
experiments with the thermal-drilling process on steel,
aluminum and titanium. They studied the properties such
as the hardness and microstructure of drilled holes. They
concluded that in thermal drilling large deformation and
high frictional heat are generated, resulting in the
changes in the material properties and microstructure of
a workpiece. They reported that the thermal conductivity
of the material drilled affects the quality and integrity of
the hole.
S. F. Miller et al.8 conducted an experimental and
numerical analysis of a thermal-drilling process on the
AISI 1020 cold-rolled carbon steel under a constant feed
rate. Moreover, they developed two classic models for
the thermal-drilling process. They predicted the tempe-
rature distribution using a finite-element-based thermal
model and in another attempt, they predicted the thrust
force and torque using the basic principles of physics. S.
F. Miller et al.9 thermally drilled into aluminium and
magnesium alloys using tungsten carbide and titanium
carbide in a cobalt matrix under different rotational
speeds and feed rates. They analyzed the thrust force,
torque, energy, power and peak power required for
drilling. They also evaluated the formation of the
bushing shape of cast metals. The bushings produced
during the thermal drilling of brittle cast metals demon-
strated a radial fracture. In their conclusions, they
suggested to pre-heat a cast metal workpiece and create a
high rotational speed for obtaining a cylindrically shaped
bushing without a significant radial fracture.
S. F. Miller et al.10 developed a three-dimensional
finite-element model for the thermal-drilling process in
order to evaluate the plastic strain and deformation of a
workpiece. S. F. Miller et al.11 investigated thermal-drill
tool-wear characteristics during the thermal drilling of an
AISI 1015 carbon steel workpiece. Furthermore, the
thrust force and torque were analyzed. They concluded
that the carbide tool is durable and it demonstrates the
least tool wear after drilling 11.000 holes. However,
progressively severe abrasive grooving on the tool tip
was observed. S. M. Lee et al.12 carried out thermal
drilling into the IN-713LC super alloy under different
rotation speeds and feed rates. They investigated the
material properties such as hardness, roundness, and
surface roughness of the holes drilled into the IN-713LC
super alloy. They reported that the hardness is higher
near the wall of a drilled hole and it decreases with the
increasing distance from the edge of the hole. In addition
to that, higher rotational speeds and feed rates demon-
strate better roundness and lower surface roughness.
H. M. Chow et al.13 optimized the process parameters
of a thermally drilled austenite stainless-steel workpiece
using the Taguchi method.14 They considered the input
parameters such as friction angle, friction/contact-area
ratio, feed rate and drilling speed, and studied their in-
fluence on the response parameters like surface rough-
ness. Moreover, the hardness and microstructural aspects
of drilled holes were studied. It was observed that the
surrounding area of a drilled hole acquired a fine grain
size and a compact structure with a higher microhardness
than that of the area away from the drilled region. S. M.
Lee et al.15 employed thermal drilling for the AISI 304
stainless steel using tungsten carbide drills with and
without coating. Their results illustrated that at the same
rotational speed and for the same number of holes
drilled, the coated drills experienced less tool wear than
the uncoated drills. Furthermore, they investigated the
changes in the relationship between the drilled-surface
temperature, tool wear and axial thrust force.
W. L. Ku et al.16 optimized the parameters of thermal
drilling into austenite stainless steel using the Taguchi
method. They studied the effects of the friction angle,
friction/contact-area ratio, feed rate and rotational speed
on the response-quality characteristics such as the
surface roughness and bushing length. They revealed that
at optimized drilling conditions, the bushing length of a
drilled hole was nearly three times longer than the plate
thickness, and a mirror-like quality wall surface of the
drilled hole could be obtained. M. Folea et al.17 investi-
gated the thermal drilling of a maraging-steel workpiece.
The authors reported that the temperature was the most
important factor in the thermal-drilling process. They
also revealed that a greater quality of the holes drilled
into maraging steel could be achieved with higher rota-
tional speeds. T. K. Mehmet et al.18 studied the effects of
the thermal-drilling parameters such as friction angle,
friction/contact-area ratio, feed rate and rotational speed
on workpiece temperature, thrust force and torque in
thermal drilling of the ST12 material. They revealed that
the thrust force and torque reduce with the increasing
rotational speed and increase with the friction angle, feed
rate and friction/contact-area ratio.
D. Biermann and Y. Liu19 demonstrated the feasibi-
lity of the thermal drilling of a magnesium wrought alloy
and analyzed the thrust forces and torque. They
measured the process temperature online and examined
the strength of the joint through tapping and thread
forming. B. B. Mehmet20 experimentally and numerically
investigated the thermal drilling of the AISI 1020 steel.
In their study, an analytical model for the torque, axial
power and heat-transfer coefficient was developed. Good
agreement was observed between experimental and
numerical values.
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
814 Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Similar work was performed by P. Krasauskas et al.21
on the thermal drilling into the AISI 304 steel. P. D.
Pantawane and B. B. Ahuja22 carried out regression
modelling of the thermal drilling of the AISI 1015 steel
using the Taguchi method. They studied the effects of the
tool-diameter ratio and rotational speed on the material
thickness and the effect of the feed on the response
parameters including the thrust force, torque and surface
roughness of drilled holes. G. Somasundaram et al.23
carried thermal drilling on the Al-SiC composite
material and studied the roundness errors on drilled
holes. They considered the input process parameters
such as the composition of workpiece, workpiece thick-
ness, rotational speed and feed rate. They developed an
empirical relation between the process parameters and
the roundness error using the response surface methodo-
logy.
Artificial neural network (ANN) is a biologically in-
spired computer program designed to simulate the way,
in which the human brain processes information.24 ANN
is widely applied in modeling many machining ope-
rations like turning, milling and drilling. S. R. Karnik
and V. N. Gaitone25 used a multilayer feed-forward ANN
to predict the influence of the process parameters such as
the cutting speed, feed, drill diameter, point angle and
lip-clearance angle on the burr height and burr thickness
in drilling the AISI 316L stainless steel. R. S. Mamilla et
al.26 applied a multilayer ANN for modeling an abrasive
flow-finishing process on the AISI 1040 and AISI 4340
steel. S. Assarzadeh and M. Ghoreishi27 successfully
used a multilayer ANN for modeling the metal-removal
rate and surface roughness in an electrical discharge
machining process involving the BD3 steel. O. Babur et
al.28 developed an ANN model according to experimental
measurement data for the end milling of Inconel 718.
They coupled a genetic algorithm (GA) with an ANN to
forecast the surface roughness. S. Sarkar et al.29 pro-
duced a multilayer feed-forward-ANN model to predict
the process parameters of the machining of  titanium
aluminide with a wire-electrical-discharge machine. A.
K. Singh et al.30 used a multilayer feed-forward ANN to
predict the flank wear of high-speed steel drill bits for
drilling holes into a copper workpiece.
The literature review on thermal drilling reveals that
experimental investigations and numerical simulations of
the effects of mechanical and physical properties on the
thermal-drilling process was mainly carried out on
copper, brass, aluminium, magnesium and stainless-steel
alloys, maraging steel, ST12 steel and IN-713LC super
alloy. Galvanized steel is being widely used for car body
structures and it is a good candidate for the implemen-
tation of the thermal-drilling process. The application of
galvanized steel is widely used in the areas of roofing
material, doors, ship’s ducts and panels, electrical-
appliance automobile parts, etc.31
In the present work, thermal drilling was carried out
on galvanized steel. Based on experimental measurement
data, an ANN model was developed to predict the for-
mation of the bushing length in thermal drilling of
galvanized steel. This model considers rotational speed,
tool angle and workpiece thickness as significant ther-
mal-drilling process parameters. Then the optimum
values of these drilling parameters were computed, with
the aim of achieving the maximum bushing length, by
solving the optimization problem implementing a GA.
Thus, the present work allows us to achieve the maxi-
mum bushing length, which is highly desirable for the
subsequent tapping and joining in automotive applica-
tions. A schematic diagram of the combined ANN-GA
optimization24 is shown in Figure 1. It indicates the
stages of the GA process and its relation with the ANN
process.
2 THERMAL DRILLING
2.1 Physical description of the process
Figure 2 shows a schematic representation of the
thermal-drilling process. Based on the thermal-drill
geometry, there are four steps involved in thermal drill-
ing. Initially, the center point zone of a drill approaches
and pierces the workpiece. Secondly, the produced heat
softens the workpiece as a result of the friction between
the thermal drill and the workpiece. Thirdly, the softened
material is pushed downward and the drill moves for-
ward to form the bushing using the cylindrical zone of
the tool. Fourthly, the extruded burr on the workpiece
surface is pressed to form a boss with the shoulder zone
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822 815
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 1: Flow chart of the integrated ANN-GA optimization process
of the thermal drill. Finally, the thermal drill retracts
leaving a hole with a bushing length.37
2.2 Formation of the bushing length
The initial volume of the material (Vi) available to
produce a hole by thermal drilling is given in Equation
(1).33 As shown in Figure 2, D1 and Pt represent the
inner-hole diameter in mm and the thickness of the
workpiece in mm, respectively.
V D Pi =
π
4 1
2
t (1)
During thermal drilling, the volume of the material to
be displaced (Vf) in order to produce a bushing is given
in Equation (2).33 As shown in Figure 2, D2 and L
represent the outer-bushing diameter in mm and the
bushing length in mm, respectively.
[ ]V D D Lf = −
π
4 2
2
1
2 (2)
It is known that the initial approximation of the
volume of the material for producing a hole is the same
as the volume of the material to be displaced during the
formation of the bushing. Therefore, Equations (1) and
(2) are assumed to be equal. Equating Equations (1) and
(2), Equation (3) is obtained:
[ ]π π
4 41
2
2
2
1
2D P D D Lt = − (3)
While rearranging the terms in Equation (3), the
bushing length (L) can be determined as shown in
Equation (4):
[ ]L
D
D D
P=
−
1
2
2
2
1
2 t
(4)
3 EXPERIMENTAL PART
In the present work, the experiments were carried out
at in a vertical drilling machine. A hot-dip-galvanized
steel workpiece with thickness values of 1, 1.5, 2 mm
and with dimensions of 50 mm × 150 mm was used.
6-mm-diameter thermal drills with different tool angles
were used in the experimentation. They were machined
out of high-speed steel rods, having the required
dimensions as shown in Figure 3. The feed rate of the
thermal-drilling process is kept constant at 240 mm/min.
The chemical composition of the galvanized steel is
given in Table 1. Axial forces were measured using a
digital drilling-tool dynamometer. Temperature measure-
ments were done on-line using a non-contact-type
infra-red thermometer and the temperature profile of the
process was obtained using the DATA TEMP MX
software with a time increment of 0.016 s. The micro-
structure of the holes drilled into galvanized steel was
measured using a scanning electron microscope (Carl
Zeiss EVO18, Germany). Three different levels of the
input process parameters were used, as shown in Table
2.
Based on the proper selection of the input process
parameters, the desirable output parameters can be
obtained during the experimentation. The selected input
process parameters such as the rotational speed, tool
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
816 Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 3: Obtained thermal drills with different tool angles: a) 35 de-
gree tool angle b) 37.5 degree tool angle, c) 45 degree tool angle
Table 1: Chemical composition of galvanized steel (w/%)
C Si S P Mn Ni Cr
0.003 0.006 0.005 0.018 0.173 0.011 0.031
Mo V Cu Al Nb Zn Ti Fe
0.001 0.002 0.017 0.035 0.001 0.004 0.05 Balance
Table 2: Input process parameters and their levels
Factors Levels
1 2 3
Rotational speed (min–1) 1944 2772 3600
Tool angle (°) 30 37.5 45
Workpiece thickness (mm) 1 1.5 2Figure 2: Schematic representation of the thermal-drilling process
angle and workpiece thickness, and the output parameter,
the bushing length of the thermally drilled galvanized
steel sheet, were analyzed in this work. Based on the
Taguchi’s L27 orthogonal array, experiments were con-
ducted at three levels of the drilling-process parameters
and the corresponding bushing length was measured
using a digital vernier caliper (Model No CD-12C, Mitu-
toya Corporation, Japan). All the experimental values of
the bushing length are displayed in Table 3. Samples of
the thermally drilled holes are shown in Figure 4.
During the thermal drilling, the burr is extruded onto the
top surface, and it is leveled when it comes in contact
with the shoulder of the thermal-drill tool as shown in
Figure 5a. The formation of the bushing at the rear side
of the galvanized steel plate is shown in Figure 5b. It
can be used to act as a cylindrical sleeve bearing and it
can be taped to fabricate an internal screw. Additionally,
it widely reduces the complexity of joining components
in the aerospace and automotive industries.
Table 3: Experimental results of the thermal-drilling process
Expt No.
Rotational
speed
(min–1)
Tool angle
(°)
Workpiece
thickness
(mm)
Bushing
length
(mm)
1 1944 30 1 2.54
2 1944 30 1.5 3.42
3 1944 30 2 4.38
4 1944 37.5 1 3.24
5 1944 37.5 1.5 3.21
6 1944 37.5 2 4.41
7 1944 45 1 2.97
8 1944 45 1.5 4.87
9 1944 45 2 5.26
10 2772 30 1 2.67
11 2772 30 1.5 3.84
12 2772 30 2 4.02
13 2772 37.5 1 2.91
14 2772 37.5 1.5 3.83
15 2772 37.5 2 4.42
16 2772 45 1 2.94
17 2772 45 1.5 4.31
18 2772 45 2 5.57
19 3600 30 1 3.55
20 3600 30 1.5 4.64
21 3600 30 2 5.84
22 3600 37.5 1 2.78
23 3600 37.5 1.5 3.85
24 3600 37.5 2 5.92
25 3600 45 1 2.42
26 3600 45 1.5 4.42
27 3600 45 2 5.61
4 ARTIFICIAL-NEURAL-NETWORK
MODELING
Recently, the use of the artificial intelligence perfor-
mance has been remarkable in the entire engineering
field.24 To understand and manage any industrial course
of action, modeling and optimization of the process are
essential. An accurate control is a requirement necessary
to accomplish better quality and productivity. ANN plays
a significant role in forecasting the solutions for non-
linear problems in all engineering fields. Using statistical
techniques, numerous researchers endeavored to develop
a model based on experimental data. In general, a neural
network represents a network of many processors ope-
rating in parallel. Each processor contains a small size of
the local memory. Then, these units are coupled through
communication channels, which typically carry numeric
data. An excellent instance of a biological neural net-
work is the brain of a human. It comprises the most
demanding and controlling arrangement, in which
learning along with training controls the behavior of a
human to take action to solve any problem met in
day-by-day life.
ANN can be successfully employed to predict the
output parameters for any given input parameters, based
on the training set for a given complex problem.28 In the
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822 817
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 5: Appearance of a hole made by thermal drilling: a) top view
and b) rear view
Figure 4: Holes made by thermal drilling in galvanized steel plates
of the thermal drill. Finally, the thermal drill retracts
leaving a hole with a bushing length.37
2.2 Formation of the bushing length
The initial volume of the material (Vi) available to
produce a hole by thermal drilling is given in Equation
(1).33 As shown in Figure 2, D1 and Pt represent the
inner-hole diameter in mm and the thickness of the
workpiece in mm, respectively.
V D Pi =
π
4 1
2
t (1)
During thermal drilling, the volume of the material to
be displaced (Vf) in order to produce a bushing is given
in Equation (2).33 As shown in Figure 2, D2 and L
represent the outer-bushing diameter in mm and the
bushing length in mm, respectively.
[ ]V D D Lf = −
π
4 2
2
1
2 (2)
It is known that the initial approximation of the
volume of the material for producing a hole is the same
as the volume of the material to be displaced during the
formation of the bushing. Therefore, Equations (1) and
(2) are assumed to be equal. Equating Equations (1) and
(2), Equation (3) is obtained:
[ ]π π
4 41
2
2
2
1
2D P D D Lt = − (3)
While rearranging the terms in Equation (3), the
bushing length (L) can be determined as shown in
Equation (4):
[ ]L
D
D D
P=
−
1
2
2
2
1
2 t
(4)
3 EXPERIMENTAL PART
In the present work, the experiments were carried out
at in a vertical drilling machine. A hot-dip-galvanized
steel workpiece with thickness values of 1, 1.5, 2 mm
and with dimensions of 50 mm × 150 mm was used.
6-mm-diameter thermal drills with different tool angles
were used in the experimentation. They were machined
out of high-speed steel rods, having the required
dimensions as shown in Figure 3. The feed rate of the
thermal-drilling process is kept constant at 240 mm/min.
The chemical composition of the galvanized steel is
given in Table 1. Axial forces were measured using a
digital drilling-tool dynamometer. Temperature measure-
ments were done on-line using a non-contact-type
infra-red thermometer and the temperature profile of the
process was obtained using the DATA TEMP MX
software with a time increment of 0.016 s. The micro-
structure of the holes drilled into galvanized steel was
measured using a scanning electron microscope (Carl
Zeiss EVO18, Germany). Three different levels of the
input process parameters were used, as shown in Table
2.
Based on the proper selection of the input process
parameters, the desirable output parameters can be
obtained during the experimentation. The selected input
process parameters such as the rotational speed, tool
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
816 Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 3: Obtained thermal drills with different tool angles: a) 35 de-
gree tool angle b) 37.5 degree tool angle, c) 45 degree tool angle
Table 1: Chemical composition of galvanized steel (w/%)
C Si S P Mn Ni Cr
0.003 0.006 0.005 0.018 0.173 0.011 0.031
Mo V Cu Al Nb Zn Ti Fe
0.001 0.002 0.017 0.035 0.001 0.004 0.05 Balance
Table 2: Input process parameters and their levels
Factors Levels
1 2 3
Rotational speed (min–1) 1944 2772 3600
Tool angle (°) 30 37.5 45
Workpiece thickness (mm) 1 1.5 2Figure 2: Schematic representation of the thermal-drilling process
angle and workpiece thickness, and the output parameter,
the bushing length of the thermally drilled galvanized
steel sheet, were analyzed in this work. Based on the
Taguchi’s L27 orthogonal array, experiments were con-
ducted at three levels of the drilling-process parameters
and the corresponding bushing length was measured
using a digital vernier caliper (Model No CD-12C, Mitu-
toya Corporation, Japan). All the experimental values of
the bushing length are displayed in Table 3. Samples of
the thermally drilled holes are shown in Figure 4.
During the thermal drilling, the burr is extruded onto the
top surface, and it is leveled when it comes in contact
with the shoulder of the thermal-drill tool as shown in
Figure 5a. The formation of the bushing at the rear side
of the galvanized steel plate is shown in Figure 5b. It
can be used to act as a cylindrical sleeve bearing and it
can be taped to fabricate an internal screw. Additionally,
it widely reduces the complexity of joining components
in the aerospace and automotive industries.
Table 3: Experimental results of the thermal-drilling process
Expt No.
Rotational
speed
(min–1)
Tool angle
(°)
Workpiece
thickness
(mm)
Bushing
length
(mm)
1 1944 30 1 2.54
2 1944 30 1.5 3.42
3 1944 30 2 4.38
4 1944 37.5 1 3.24
5 1944 37.5 1.5 3.21
6 1944 37.5 2 4.41
7 1944 45 1 2.97
8 1944 45 1.5 4.87
9 1944 45 2 5.26
10 2772 30 1 2.67
11 2772 30 1.5 3.84
12 2772 30 2 4.02
13 2772 37.5 1 2.91
14 2772 37.5 1.5 3.83
15 2772 37.5 2 4.42
16 2772 45 1 2.94
17 2772 45 1.5 4.31
18 2772 45 2 5.57
19 3600 30 1 3.55
20 3600 30 1.5 4.64
21 3600 30 2 5.84
22 3600 37.5 1 2.78
23 3600 37.5 1.5 3.85
24 3600 37.5 2 5.92
25 3600 45 1 2.42
26 3600 45 1.5 4.42
27 3600 45 2 5.61
4 ARTIFICIAL-NEURAL-NETWORK
MODELING
Recently, the use of the artificial intelligence perfor-
mance has been remarkable in the entire engineering
field.24 To understand and manage any industrial course
of action, modeling and optimization of the process are
essential. An accurate control is a requirement necessary
to accomplish better quality and productivity. ANN plays
a significant role in forecasting the solutions for non-
linear problems in all engineering fields. Using statistical
techniques, numerous researchers endeavored to develop
a model based on experimental data. In general, a neural
network represents a network of many processors ope-
rating in parallel. Each processor contains a small size of
the local memory. Then, these units are coupled through
communication channels, which typically carry numeric
data. An excellent instance of a biological neural net-
work is the brain of a human. It comprises the most
demanding and controlling arrangement, in which
learning along with training controls the behavior of a
human to take action to solve any problem met in
day-by-day life.
ANN can be successfully employed to predict the
output parameters for any given input parameters, based
on the training set for a given complex problem.28 In the
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822 817
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 5: Appearance of a hole made by thermal drilling: a) top view
and b) rear view
Figure 4: Holes made by thermal drilling in galvanized steel plates
present work, an ANN is developed using MATLAB
version 7.10, neural network toolbox, with the intention
of predicting the bushing length as a function of three
input process parameters including the rotational speed,
tool angle and workpiece thickness. Figure 6 shows the
neural-network architecture of the bushing length ob-
tained with thermal drilling. The input data is provided
through input layer neurons, and it is fed forward
through the hidden layer. Then the response is obtained
at the output layer neurons. The neurons are linked by
weights that take part in the learning process. The
weights help in attaining the optimum solution and
escape from the local minima. To establish the optimum
structural design, we apply the trial-and-error method
with the appropriate activation function and the most
excellent training algorithm. A number of models were
created and tested. A hyperbolic tangent sigmoid transfer
function is considered as the hidden-layer activation
function, and it is given in Equation (5):
f x x
e x
( ) ( )= =
−
−−tansig
2
1
12 (5)
The most important selection criteria used for the
suitable model are the mean squared error and the
regression-coefficient value. The mean sum of the
squared error (MSE) is the average squared difference
between the predicted value and the actual experimental
value as shown in Equation (6). A lower value of the
MSE indicates a better model. The regression coefficient
(R) was measured by correlating the predicted and the
actual values as given in Equation (7). An R value of 1
indicates a close relationship between the predicted and
actual experimental value, i.e., a very small error.
MSE
n
A Pi i
i
= −∑1 2 (6)
R
A P
P
i i
i
n
i
i
n= −
−
⎛
⎝
⎜
⎜
⎜
⎞
⎠
⎟
⎟
⎟
=
=
∑
∑
1
2
1
2
1
(7)
where Ai is the actual experimental value, Pi is the
predicted value and n is the number of patterns. With
the intention of measuring the accuracy of the prediction
model, the percentage of error is calculated with
Equation (8):
z
E P
E
=
−
×100 (8)
where z = % of prediction error, E = experimental value,
P = predicted value
5 RESULTS AND DISCUSSION
The results of the ANN coupled with the GA34 are
utilized to predict and optimize the bushing length based
on the input process parameters such as the rotational
speed, tool angle and workpiece thickness in the ther-
mal-drilling process, and they are discussed below.
5.1 Prediction of the bushing length by the ANN
Table 4 displays the parameters used with this
modeling technique. The developed ANN model of the
bushing length was trained by means of the selected
parameters. During the training process, a 19 set input
for training trials were presented each time and the
resultant output was acquired. The number of the
hidden-layer neurons varied from 1 to 20. It was deter-
mined based on trial and error, through step-by-step
increasing the number of neurons and examining their
result against the forecast value. The final neuron
number of the hidden layer obtained was 10. The ulti-
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
818 Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 7: Correlation between the predicted values of the ANN model
and the experimental data for the prediction of the bushing length
using: a) training data, b) validation data, c) test data and d) entire data
Figure 6: ANN modeling of the bushing length in thermal drilling
mate weights between the input layer and hidden layer
(w1, w2, w3), the bias (b) and weights between the
hidden layer and the output layer (w4) are shown in
Table 5. Finally, the lowest MSE value achieved was
0.00415 when the configuration of the ANN was 3-10-1.
There are three input neurons, each indicating one input
variable form the input layer, 10 neurons in the hidden
layer and the output layer with one neuron correspond to
the bushing length.
Table 4: Parameters used in the ANN modeling
Input variables Rotational speed, tool angle,workpiece thickness
Output variable Bushing length
Network type Feed-forward backpropagation
Algorithm
Levenberg-Marquardt
back-propagation algorithm
(trainlm)
Activation function Hyperbolic tangent sigmoidtransfer function
Data division 70 % (training); 15 %(validation); 15 % (testing)
Number of neurons in the
hidden layer 10
Table 5: Final weights in between the layers
No. of
neurons
Weights
Input to hidden layer Hidden tooutput layer
w1 w2 w3 b w4
1 -0.682 -1.572 1.929 3.529 0.680
2 -0.865 1.620 -2.115 2.483 -0.576
3 3.045 0.934 -0.191 -1.179 0.111
4 1.722 -2.097 1.179 -1.184 -0.253
5 -1.826 -1.470 1.863 0.273 -0.183
6 -0.027 -0.448 3.025 -0.131 1.113
7 -0.619 2.955 0.332 -0.694 0.268
8 0.855 -1.318 -2.373 1.895 0.036
9 1.645 0.306 -2.767 2.418 0.939
10 1.822 -1.597 -1.613 3.188 -0.638
The output performance of the established ANN is
examined on the basis of the regression-correlation
coefficient (R value) between the predicted values and
the experimental values. The predicted responses of the
ANN model are in excellent conformity with the expe-
rimental values, i.e., the correlation regression coeffi-
cients of 100 % for the training set, 98.41 % for the
validation set and 94.80 % for the testing set are
achieved as shown in Figure 7. It was concluded that the
feed-forward back-propagation algorithm of the configu-
ration of 3-10-1 gives the most excellent results for the
prediction of the bushing length as shown in Table 6.
The graphical representation of the experimental and
predicted values of the bushing length by the ANN is
shown in Figure 8. It shows that the ANN prediction
values are very close to the experimental values of the
bushing length. The average prediction error between the
experimental and predicted values of the bushing length
is 1.843 %. Therefore, it is demonstrated that the
developed ANN model is suitable for forecasting the
bushing length of drilled holes in thermal drilling of
galvanized steel, having the highest accuracy of
98.157 %.
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822 819
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 8: Correlation between experimental and predicted values
Table 6: Comparison of experimental and predicted values of the
bushing length
Expt No. Bushing length (mm)
Prediction
error (%)Experimental
value
Predicted ANN
value
1 2.54 2.485 2.165
2 3.42 3.427 0.204
3 4.38 4.252 2.922
4 3.24 2.618 19.197
5 3.21 3.336 3.925
6 4.41 4.425 0.340
7 2.97 2.843 4.276
8 4.87 4.677 3.963
9 5.26 5.517 4.886
10 2.67 2.385 10.674
11 3.84 3.714 3.281
12 4.02 4.275 6.343
13 2.91 2.905 0.172
14 3.83 3.814 0.418
15 4.42 4.615 4.412
16 2.94 2.964 0.816
17 4.31 4.284 0.603
18 5.57 5.599 0.521
19 3.55 3.738 5.296
20 4.64 4.585 1.185
21 5.84 5.893 0.907
22 2.78 2.760 0.719
23 3.85 3.875 0.649
24 5.92 5.836 1.419
25 2.42 2.755 13.843
26 4.42 4.450 0.679
27 5.61 6.007 7.077
present work, an ANN is developed using MATLAB
version 7.10, neural network toolbox, with the intention
of predicting the bushing length as a function of three
input process parameters including the rotational speed,
tool angle and workpiece thickness. Figure 6 shows the
neural-network architecture of the bushing length ob-
tained with thermal drilling. The input data is provided
through input layer neurons, and it is fed forward
through the hidden layer. Then the response is obtained
at the output layer neurons. The neurons are linked by
weights that take part in the learning process. The
weights help in attaining the optimum solution and
escape from the local minima. To establish the optimum
structural design, we apply the trial-and-error method
with the appropriate activation function and the most
excellent training algorithm. A number of models were
created and tested. A hyperbolic tangent sigmoid transfer
function is considered as the hidden-layer activation
function, and it is given in Equation (5):
f x x
e x
( ) ( )= =
−
−−tansig
2
1
12 (5)
The most important selection criteria used for the
suitable model are the mean squared error and the
regression-coefficient value. The mean sum of the
squared error (MSE) is the average squared difference
between the predicted value and the actual experimental
value as shown in Equation (6). A lower value of the
MSE indicates a better model. The regression coefficient
(R) was measured by correlating the predicted and the
actual values as given in Equation (7). An R value of 1
indicates a close relationship between the predicted and
actual experimental value, i.e., a very small error.
MSE
n
A Pi i
i
= −∑1 2 (6)
R
A P
P
i i
i
n
i
i
n= −
−
⎛
⎝
⎜
⎜
⎜
⎞
⎠
⎟
⎟
⎟
=
=
∑
∑
1
2
1
2
1
(7)
where Ai is the actual experimental value, Pi is the
predicted value and n is the number of patterns. With
the intention of measuring the accuracy of the prediction
model, the percentage of error is calculated with
Equation (8):
z
E P
E
=
−
×100 (8)
where z = % of prediction error, E = experimental value,
P = predicted value
5 RESULTS AND DISCUSSION
The results of the ANN coupled with the GA34 are
utilized to predict and optimize the bushing length based
on the input process parameters such as the rotational
speed, tool angle and workpiece thickness in the ther-
mal-drilling process, and they are discussed below.
5.1 Prediction of the bushing length by the ANN
Table 4 displays the parameters used with this
modeling technique. The developed ANN model of the
bushing length was trained by means of the selected
parameters. During the training process, a 19 set input
for training trials were presented each time and the
resultant output was acquired. The number of the
hidden-layer neurons varied from 1 to 20. It was deter-
mined based on trial and error, through step-by-step
increasing the number of neurons and examining their
result against the forecast value. The final neuron
number of the hidden layer obtained was 10. The ulti-
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
818 Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 7: Correlation between the predicted values of the ANN model
and the experimental data for the prediction of the bushing length
using: a) training data, b) validation data, c) test data and d) entire data
Figure 6: ANN modeling of the bushing length in thermal drilling
mate weights between the input layer and hidden layer
(w1, w2, w3), the bias (b) and weights between the
hidden layer and the output layer (w4) are shown in
Table 5. Finally, the lowest MSE value achieved was
0.00415 when the configuration of the ANN was 3-10-1.
There are three input neurons, each indicating one input
variable form the input layer, 10 neurons in the hidden
layer and the output layer with one neuron correspond to
the bushing length.
Table 4: Parameters used in the ANN modeling
Input variables Rotational speed, tool angle,workpiece thickness
Output variable Bushing length
Network type Feed-forward backpropagation
Algorithm
Levenberg-Marquardt
back-propagation algorithm
(trainlm)
Activation function Hyperbolic tangent sigmoidtransfer function
Data division 70 % (training); 15 %(validation); 15 % (testing)
Number of neurons in the
hidden layer 10
Table 5: Final weights in between the layers
No. of
neurons
Weights
Input to hidden layer Hidden tooutput layer
w1 w2 w3 b w4
1 -0.682 -1.572 1.929 3.529 0.680
2 -0.865 1.620 -2.115 2.483 -0.576
3 3.045 0.934 -0.191 -1.179 0.111
4 1.722 -2.097 1.179 -1.184 -0.253
5 -1.826 -1.470 1.863 0.273 -0.183
6 -0.027 -0.448 3.025 -0.131 1.113
7 -0.619 2.955 0.332 -0.694 0.268
8 0.855 -1.318 -2.373 1.895 0.036
9 1.645 0.306 -2.767 2.418 0.939
10 1.822 -1.597 -1.613 3.188 -0.638
The output performance of the established ANN is
examined on the basis of the regression-correlation
coefficient (R value) between the predicted values and
the experimental values. The predicted responses of the
ANN model are in excellent conformity with the expe-
rimental values, i.e., the correlation regression coeffi-
cients of 100 % for the training set, 98.41 % for the
validation set and 94.80 % for the testing set are
achieved as shown in Figure 7. It was concluded that the
feed-forward back-propagation algorithm of the configu-
ration of 3-10-1 gives the most excellent results for the
prediction of the bushing length as shown in Table 6.
The graphical representation of the experimental and
predicted values of the bushing length by the ANN is
shown in Figure 8. It shows that the ANN prediction
values are very close to the experimental values of the
bushing length. The average prediction error between the
experimental and predicted values of the bushing length
is 1.843 %. Therefore, it is demonstrated that the
developed ANN model is suitable for forecasting the
bushing length of drilled holes in thermal drilling of
galvanized steel, having the highest accuracy of
98.157 %.
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822 819
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 8: Correlation between experimental and predicted values
Table 6: Comparison of experimental and predicted values of the
bushing length
Expt No. Bushing length (mm)
Prediction
error (%)Experimental
value
Predicted ANN
value
1 2.54 2.485 2.165
2 3.42 3.427 0.204
3 4.38 4.252 2.922
4 3.24 2.618 19.197
5 3.21 3.336 3.925
6 4.41 4.425 0.340
7 2.97 2.843 4.276
8 4.87 4.677 3.963
9 5.26 5.517 4.886
10 2.67 2.385 10.674
11 3.84 3.714 3.281
12 4.02 4.275 6.343
13 2.91 2.905 0.172
14 3.83 3.814 0.418
15 4.42 4.615 4.412
16 2.94 2.964 0.816
17 4.31 4.284 0.603
18 5.57 5.599 0.521
19 3.55 3.738 5.296
20 4.64 4.585 1.185
21 5.84 5.893 0.907
22 2.78 2.760 0.719
23 3.85 3.875 0.649
24 5.92 5.836 1.419
25 2.42 2.755 13.843
26 4.42 4.450 0.679
27 5.61 6.007 7.077
5.2 Optimization of the bushing length using a GA
The optimization of the thermal drilling process is
performed using the GA in MATLAB toolbox, with the
intention of enhancing the effectiveness of the drilling
process in order to achieve high structural rigidity for
fasteners in joining situations.35,36 This algorithm makes
a binary coding system to characterize the variables such
as rotational speed (RS), tool angle (TA) and workpiece
thickness (WT). All of the process variables are symbo-
lized by a ten-bit binary equivalent. In the chromosome,
the process variables are represented as a substring. The
GA employs different types of crossover and mutation
operators to predict the maximum values of the bushing
length. At this time, the second-order mathematical
model is considered as an objective function with the
aim of maximizing the output bushing length (BL). In
this model, the rotational speed (RS), tool angle (TA)
and workpiece thickness (WT) are considered as the
input parameters. The aim of the optimization is to
maximize the bushing length; however, usually, a GA is
used to achieve the minimum function value for a mini-
mization problem. Hence, in the present situation, the
objective function was converted into a minimization
problem. A unity negative factor is multiplied to the
objective function in the larger-the-best-type bushing
length (L) of the responses characteristic of thermal
drilling to make them minimize the type objective. The
parameters used in the GA technique are shown in Table
7. Equation (9) is a developed regression model of the
bushing length and it is used as the fitness function or
objective function for this problem. It is developed with
the knowledge gained form ANN values.37,38
Table 7: GA parameters
Population type Double vector
Population size 100
Scaling function Rank
Selection function Roulette wheel
Elite count 2
Crossover fraction 0.8
Mutation function Adaptive feasible
Crossover function Heuristic
Stall generations 50
Maximum bushing length (L) = 6.56203 – 4.64305 ×
10–4 × RS – 0.210894 × TA – 1.04080 × WT +
3.54118 × 10–7 × RS2 – 0.00413827 – TA2 – 0.282222
× WT2 – 5.82394 × 10–5 × RS × TA + 0.000668277 ×
RS × WT + 0.0584444 × TA × WT (9)
is subjected to constrained variables such as:
1944 min–1 = RS = 3600 min–1 (RS – rotational speed)
30° = TA = 45° (TA – tool angle)
1 mm = WT = 2 mm (WT – workpiece thickness).
Figure 9 shows the bushing-length fitness-function
plot obtained with the genetic algorithm. The negative
sign from the final result can be eliminated to get the
maximum value of the bushing length. The optimum
result is achieved at the 52nd iteration of the genetic
algorithm. It is found that the mean fitness value is
5.7472 mm, whereas the best fitness value is 5.7584 mm.
Table 8 presents the optimized and experimental value
of the bushing length. Very close agreement between the
optimized and experimental values of the bushing length
is obtained. It confirms the potential applicability of
these GA techniques for the industry-related problems.
Figure 10 shows the cross-section of the bushing length
of a hole drilled in galvanized steel under the condition
of the optimal process parameter. The bushing length can
offer a longer contact-surface area, which can uphold a
shaft securely, and the drilled-hole inner surface has a
thermally affected zone.
Table 8: Comparison of predicted and experimental values of the
bushing length
Optimum drilling input
process parameters
Output
para-
meters
Rota-
tional
speed
(min–1)
Tool
angle
(°)
Work-
piece
thickness
(mm)
Bushing
length
(mm)
Parameters optimized
with genetic algorithm 3552.461 45 2 5.758
Experimentally used
parameters 3600 45 2 5.6
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
820 Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 9: Fitness-function plot of the bushing length from the genetic
algorithm
Figure 10: Cross-section of a thermally drilled hole
5.3 Microstructural investigation
Figure 11 shows a macrograph obtained with
scanning electron microscopy at a magnification of 17×.
The scanning electron microscope was utilized to
observe and investigate the inside surface of a hole
drilled in galvanized steel. During the thermal-drilling
operation, the rubbing action between the interface of the
thermal drill and the galvanized-steel workpiece led to a
rise in the temperature of up to 798 °C as recorded
on-line during the experimentation. Due to such fric-
tional heating and application of the axial tensile force,
Luders bands were formed to relieve the high amount of
the internal stresses experienced in the galvanized
steel.39,40 Luders bands started to form at the tail end of
the bushing at a distance of 914 μm, after the piercing of
a hole with the center zone of the thermal drill. Until a
distance of 1725 μm, stretcher-strain markings such as
Luders bands are observed in the micrograph shown in
Figure 11. The width of each band thus formed is
different because of non-uniform yielding of the galva-
nized steel during the contact with the cylindrical zone of
the thermal drill.
6 CONCLUSIONS
The present work demonstrates the possibility of
thermal drilling of galvanized steel that has tremendous
applications in the domain of automobile and aerospace
engineering. Owing to its importance, the mechanism
and formation of the bushing length is studied.
Due to frictional heating and applied axial force,
internal stresses tend to increase in that region and
subsequently lead to the formation of Luders bands at
the tail end of the bushing. The formation of these
stretcher-strain marks is due to discontinuous non-uni-
form yielding of the galvanized steel.
The relationship between the input parameters such
as the rotational speed, tool angle and workpiece thick-
ness, and the output parameters like the bushing length is
modeled through an ANN technique. The developed
ANN model is appropriately incorporated with the GA to
optimize the thermal-drilling process parameters.
A good correlation was observed between the
experimental measurements and the predicted optimum
values. This shows that the ANN model combined with
the GA can be successfully applied to find the optimum
conditions for achieving the maximum bushing length in
the thermal drilling of galvanized steel.
The modeling and optimization are valid for one
material and coating only. For different materials, the
building of a new ANN model is required, and the gene-
tic optimization is to be performed again.
Acknowledgements
The authors are grateful for the financial grant by the
Mepco Schlenk Engineering College (Autonomous),
Sivakasi, under the Students Project Scheme (Letter No.
OF/EDC/2840/2015-2016 dated 24.10.2015) for establi-
shing an experimental set-up. The authors acknowledge
the support and encouragement by Dr. S. Arivazhagan,
Principal and Dr. P. Nagaraj, Head of Mechanical
Engineering, towards this work.
7 REFERENCES
1 N. R. J. Hynes, M. V. Maheshwaran, Numerical analysis on thermal
drilling of aluminum metal matrix composite, AIP. Conf. Proc., 1728
(2016), 1–5, doi:10.1063/1.4946597
2 O. Cebeli, D. Zulkuf, Investigate the friction drilling of aluminium
alloys according to the thermal conductivity, Tem. J., 2 (2013) 1,
93–101
3 S. F. Miller, Experimental analysis and numerical modeling of the
friction drilling process, Thesis, University of Michigan, 2006
4 N. R. J. Hynes, M. Muthukumaran, N. Rakesh, C. K. Gurubaran,
Numerical analysis in friction drilling of AISI1020 steel and AA
6061 T6 alloy, Recent Advances in Environmental and Earth
Sciences and Economics, 39 (2015), 145–149
5 A. H. Streppel, H. J. J Kals, Flow drilling: a preliminary analysis of a
new bush-making operation, CIRP Ann. Manuf. Techn., 32 (1983) 1,
167–171, doi:10.1016/S0007-8506(07)63383-6
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822 821
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 11: Microstructure of a thermally drilled hole
5.2 Optimization of the bushing length using a GA
The optimization of the thermal drilling process is
performed using the GA in MATLAB toolbox, with the
intention of enhancing the effectiveness of the drilling
process in order to achieve high structural rigidity for
fasteners in joining situations.35,36 This algorithm makes
a binary coding system to characterize the variables such
as rotational speed (RS), tool angle (TA) and workpiece
thickness (WT). All of the process variables are symbo-
lized by a ten-bit binary equivalent. In the chromosome,
the process variables are represented as a substring. The
GA employs different types of crossover and mutation
operators to predict the maximum values of the bushing
length. At this time, the second-order mathematical
model is considered as an objective function with the
aim of maximizing the output bushing length (BL). In
this model, the rotational speed (RS), tool angle (TA)
and workpiece thickness (WT) are considered as the
input parameters. The aim of the optimization is to
maximize the bushing length; however, usually, a GA is
used to achieve the minimum function value for a mini-
mization problem. Hence, in the present situation, the
objective function was converted into a minimization
problem. A unity negative factor is multiplied to the
objective function in the larger-the-best-type bushing
length (L) of the responses characteristic of thermal
drilling to make them minimize the type objective. The
parameters used in the GA technique are shown in Table
7. Equation (9) is a developed regression model of the
bushing length and it is used as the fitness function or
objective function for this problem. It is developed with
the knowledge gained form ANN values.37,38
Table 7: GA parameters
Population type Double vector
Population size 100
Scaling function Rank
Selection function Roulette wheel
Elite count 2
Crossover fraction 0.8
Mutation function Adaptive feasible
Crossover function Heuristic
Stall generations 50
Maximum bushing length (L) = 6.56203 – 4.64305 ×
10–4 × RS – 0.210894 × TA – 1.04080 × WT +
3.54118 × 10–7 × RS2 – 0.00413827 – TA2 – 0.282222
× WT2 – 5.82394 × 10–5 × RS × TA + 0.000668277 ×
RS × WT + 0.0584444 × TA × WT (9)
is subjected to constrained variables such as:
1944 min–1 = RS = 3600 min–1 (RS – rotational speed)
30° = TA = 45° (TA – tool angle)
1 mm = WT = 2 mm (WT – workpiece thickness).
Figure 9 shows the bushing-length fitness-function
plot obtained with the genetic algorithm. The negative
sign from the final result can be eliminated to get the
maximum value of the bushing length. The optimum
result is achieved at the 52nd iteration of the genetic
algorithm. It is found that the mean fitness value is
5.7472 mm, whereas the best fitness value is 5.7584 mm.
Table 8 presents the optimized and experimental value
of the bushing length. Very close agreement between the
optimized and experimental values of the bushing length
is obtained. It confirms the potential applicability of
these GA techniques for the industry-related problems.
Figure 10 shows the cross-section of the bushing length
of a hole drilled in galvanized steel under the condition
of the optimal process parameter. The bushing length can
offer a longer contact-surface area, which can uphold a
shaft securely, and the drilled-hole inner surface has a
thermally affected zone.
Table 8: Comparison of predicted and experimental values of the
bushing length
Optimum drilling input
process parameters
Output
para-
meters
Rota-
tional
speed
(min–1)
Tool
angle
(°)
Work-
piece
thickness
(mm)
Bushing
length
(mm)
Parameters optimized
with genetic algorithm 3552.461 45 2 5.758
Experimentally used
parameters 3600 45 2 5.6
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
820 Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 9: Fitness-function plot of the bushing length from the genetic
algorithm
Figure 10: Cross-section of a thermally drilled hole
5.3 Microstructural investigation
Figure 11 shows a macrograph obtained with
scanning electron microscopy at a magnification of 17×.
The scanning electron microscope was utilized to
observe and investigate the inside surface of a hole
drilled in galvanized steel. During the thermal-drilling
operation, the rubbing action between the interface of the
thermal drill and the galvanized-steel workpiece led to a
rise in the temperature of up to 798 °C as recorded
on-line during the experimentation. Due to such fric-
tional heating and application of the axial tensile force,
Luders bands were formed to relieve the high amount of
the internal stresses experienced in the galvanized
steel.39,40 Luders bands started to form at the tail end of
the bushing at a distance of 914 μm, after the piercing of
a hole with the center zone of the thermal drill. Until a
distance of 1725 μm, stretcher-strain markings such as
Luders bands are observed in the micrograph shown in
Figure 11. The width of each band thus formed is
different because of non-uniform yielding of the galva-
nized steel during the contact with the cylindrical zone of
the thermal drill.
6 CONCLUSIONS
The present work demonstrates the possibility of
thermal drilling of galvanized steel that has tremendous
applications in the domain of automobile and aerospace
engineering. Owing to its importance, the mechanism
and formation of the bushing length is studied.
Due to frictional heating and applied axial force,
internal stresses tend to increase in that region and
subsequently lead to the formation of Luders bands at
the tail end of the bushing. The formation of these
stretcher-strain marks is due to discontinuous non-uni-
form yielding of the galvanized steel.
The relationship between the input parameters such
as the rotational speed, tool angle and workpiece thick-
ness, and the output parameters like the bushing length is
modeled through an ANN technique. The developed
ANN model is appropriately incorporated with the GA to
optimize the thermal-drilling process parameters.
A good correlation was observed between the
experimental measurements and the predicted optimum
values. This shows that the ANN model combined with
the GA can be successfully applied to find the optimum
conditions for achieving the maximum bushing length in
the thermal drilling of galvanized steel.
The modeling and optimization are valid for one
material and coating only. For different materials, the
building of a new ANN model is required, and the gene-
tic optimization is to be performed again.
Acknowledgements
The authors are grateful for the financial grant by the
Mepco Schlenk Engineering College (Autonomous),
Sivakasi, under the Students Project Scheme (Letter No.
OF/EDC/2840/2015-2016 dated 24.10.2015) for establi-
shing an experimental set-up. The authors acknowledge
the support and encouragement by Dr. S. Arivazhagan,
Principal and Dr. P. Nagaraj, Head of Mechanical
Engineering, towards this work.
7 REFERENCES
1 N. R. J. Hynes, M. V. Maheshwaran, Numerical analysis on thermal
drilling of aluminum metal matrix composite, AIP. Conf. Proc., 1728
(2016), 1–5, doi:10.1063/1.4946597
2 O. Cebeli, D. Zulkuf, Investigate the friction drilling of aluminium
alloys according to the thermal conductivity, Tem. J., 2 (2013) 1,
93–101
3 S. F. Miller, Experimental analysis and numerical modeling of the
friction drilling process, Thesis, University of Michigan, 2006
4 N. R. J. Hynes, M. Muthukumaran, N. Rakesh, C. K. Gurubaran,
Numerical analysis in friction drilling of AISI1020 steel and AA
6061 T6 alloy, Recent Advances in Environmental and Earth
Sciences and Economics, 39 (2015), 145–149
5 A. H. Streppel, H. J. J Kals, Flow drilling: a preliminary analysis of a
new bush-making operation, CIRP Ann. Manuf. Techn., 32 (1983) 1,
167–171, doi:10.1016/S0007-8506(07)63383-6
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822 821
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
Figure 11: Microstructure of a thermally drilled hole
6 M. Kerkhofs, M. Van Stappen, The performance of (Ti, Al) N-coated
flow drills, Surf. Coat. Technol., 68 (1994) 69, 741–746
7 S. F. Miller, P. J. Blau, A. J. Shih, Microstructural alterations
associated with friction drilling of steel, aluminum, and titanium, J.
Mater. Eng. Perform., 14 (2005), 647–653, doi:10.1361/
105994905X64558
8 S. F. Miller, R. Li, H. Wang, A. J. Shih, Experimental and numerical
analysis of the friction drilling process, J. Manuf. Sci. Eng., 128
(2006), 802–810, doi:10.1115/1.2193554
9 S. F. Miller, J. Tao, A. J. Shih, Friction drilling of cast metals, Int. J.
Mach. Tools Manuf., 46 (2006), 1526–1535, doi:10.1016/
j.ijmachtools.2005.09.003
10 S. F. Miller, A. J. Shih, Thermo-mechanical finite element modeling
of the friction drilling process, J. Manuf. Sci. Eng., 129 (2007),
531–538
11 S. F. Miller, P. J. Blau, A. J. Shih, Tool wear in friction drilling, Int.
J. Mach. Tools Manuf., 47 (2007), 1636–1645, doi:10.1016/
j.ijmachtools.2006.10.009
12 S. M. Lee, H. M. Chow, B. H. Yan, Friction drilling of IN-713LC
cast superalloy, Mater. Manuf. Proces., 22 (2007), 893–897,
doi:10.1080/10426910701451697
13 H. M. Chow, S. M. Lee, L. D. Yang, Machining characteristic study
of friction drilling on AISI 304 stainless steel, J. Mater. Process.
Technol., 207 (2008), 180–186, doi:10.1016/j.jmatprotec.2007.
12.064
14 B. Mehmet, C. Ibrahim, G. Mustafa, O. Feridun, Application of the
taguchi method to optimize the cutting conditions in hard turning of
a ring bore, Mater. Tech., 49 (2015) 5, 765–772, doi:10.17222/mit.
2014.246
15 S. M. Lee, H. M. Chow, F. Y. Huang, B. H. Yan, Friction drilling of
austenitic stainless steel by uncoated and PVD AlCrN- and
TiAlN-coated tungsten carbide tools, Int. J. Mach. Tools Manuf., 49
(2009), 81–88,doi:10.1016/j.ijmachtools.2008.07.012
16 W. L. Ku, C. L. Hung, S. M. Lee, H. M. Chow, Optimization in ther-
mal friction drilling for SUS 304 stainless steel, Int. J. Adv. Manuf.
Technol., 53 (2011), 935–944, doi:10.1007/s00170-010-2899-5
17 M. Folea, D. Schlegel, E. Gete, C. Langlade, A. Roman, Preliminary
tests on flow drilling of maraging steels, Acad. J. Manuf. Eng. 10
(2012) 4, 42–47
18 T. K. Mehmet, A. Alaattin, B. Bertan, K. A. Hamza, An experimental
study on friction drilling of ST12 steel, T. Can. Soc. Mech. Eng., 38
(2014), 319–329
19 D. Biermann, Y. Liu, Innovative flow drilling on magnesium wrought
alloy AZ31, Proc. CIRP., 18 (2014), 209–214, doi:10.1016/j.procir.
2014.06.133
20 B. B. Mehmet, G. Kadir, G. Arif, Three-dimensional finite element
model of friction drilling process in hot forming processes, J. Proc.
Mech. Eng., (2015), 1–7, doi:10.1177/0954408915614300
21 P. Krasauskas, S. Kilikevi~ius, R. ^esnavi~ius, D. Pa~enga, Expe-
rimental analysis and numerical simulation of the stainless AISI 304
steel friction drilling process, Mecha., 20 (2014) 6, 590–595,
doi:10.5755/j01.mech.20.6.8664
22 P. D. Pantawane, B. B. Ahuja, Parametric analysis and modelling of
friction drilling process on AISI 1015, Int. J. Mecha. Manuf. Sys., 7
(2014) 1, 60–79, doi:10.1504/IJMMS.2014.062771
23 G. Somasundaram, B. S. Rajendra, K. Palanikumar, Modeling and
analysis of roundness error in friction drilling of aluminum silicon
carbide metal matrix composite, J. Compos. Mater., 46 (2011) 2,
169–181, doi:10.1177/0021998311410493
24 S. Changyu, W. Lixia, C. Wei, W. Jinxing, Optimization for injection
molding process conditions of the refrigeratory top cover using com-
bination method of artificial neural network and genetic algorithms,
Poly. Plast. Tech. Eng., 46 (2007), 105–112, doi:10.1080/
03602550601152853
25 S. R. Karnik, V. N. Gaitone, Development of artificial neural network
models to study the effect of process parameters on burr size in
drilling, Int. J. Adv. Manuf. Technol., 39 (2008), 439–453,
doi:10.1007/s00170-007-1231-5
26 R. S. Mamilla, S. Mondal, J. Ramkumar, V. K. Jain, Experimental
investigations and modeling of drill bit-guided abrasive flow
finishing (DBG-AFF) process, Int. J. Adv. Manuf. Technol., 42
(2009), 678–688, doi:10.1007/s00170-008-1642-y
27 S. Assarzadeh, M. Ghoreishi, Neural-network-based modeling and
optimization of the electro-discharge machining process, Int. J. Adv.
Manuf. Technol., 39 (2008), 488–500, doi:10.1007/s00170-007-
1235-1
28 O. Babur, O. Hasan, K. Hasan, Optimum surface roughness in end
milling Inconel 718 by coupling neural network model and genetic
algorithm, Int. J. Adv. Manuf. Technol., 27 (2005), 234–241,
doi:10.1007/s00170-004-2175-7
29 S. Sarkar, S. Mitra, B. Bhattacharyya, Parametric optimisation of
wire electrical discharge machining of ã titanium aluminide alloy
through an artificial neural network model, Int. J. Adv. Manuf.
Technol., 27 (2006), 501–508, doi:10.1007/s00170-004-2203-7
30 A. K. Singh, S. S. Panda, S. K. Pal, D. Chakraborty, Predicting drill
wear using an artificial neural network, Int. J. Adv. Manuf. Technol.,
28 (2006), 456–462, doi:10.1007/s00170-004-2376-0
31 S. M. Hamidinejad, F. Kolahan, A. H. Kokabi, The modeling and
process analysis of resistance spot welding on galvanized steel sheets
used in car body manufacturing, Mater. Des., 34 (2012), 759–767,
doi:10.1016/j.matdes.2011.06.064
32 Y. Lieh-Dai, K. Wei-Liang, C. Han-Ming, W. Der-An, L. Yan-
Cherng, Mar-M247, Haynes-230 & Inconel-718 study of machining
characteristics for Ni-based super alloys on friction drilling, Adv.
Mat. Res., 459 (2012), 632–637, doi:10.4028/www.scientific.net/
AMR.459.632
33 J. B. Peter, C. J. Brian, Q. Jun, Feasibility of thermally drilling
automotive alloy sheet, castings, and hydro formed shapes, (2007),
http://web.ornl.gov/info/reports/2007/3445605662084.pdf, 08/2016
34 P. Sathiya, K. Panneerselvam, M. Y. Abdul Jaleel, Optimization of
laser welding process parameters for super austenitic stainless steel
using artificial neural networks and genetic algorithm, Mater. Des.,
36 (2012), 490–498, doi:10.1016/j.matdes.2011.11.028
35 K. Girish, S. S. Kuldip, Predictive modelling and optimization of
machining parameters to minimize surface roughness using artificial
neural network coupled with genetic algorithm, Proc. CIRP., 31
(2015), 453–458, doi:10.1016/j.procir.2015.03.043
36 S. S. Kuldip, S. Sachin, K. Girish, Optimization of machining
parameters to minimize surface roughness using integrated ANN-GA
approach, Proc. CIRP., 29 (2015), 305–310, doi:10.1016/j.procir.
2015.02.002
37 D. S. Nagesh, G. L. Datta, Genetic algorithm for optimization of
welding variables for height to width ratio and application of ANN
for prediction of bead geometry for TIG welding process, App. Soft.
Comp., 10 (2010), 897–907, doi:10.1016/j.asoc.2009.10.007
38 M. Z. Azlan, H. Habibollah, S. Safian, Genetic algorithm and
simulated annealing to estimate optimal process parameters of the
abrasive water jet machining, Eng. Comp., 27 (2011), 251–259,
doi:10.1007/s00366-010-0195-5
39 V. S. Ananthan, E. O. Hall, Macroscopic aspects of Luders band
deformation in mild steel, Actametall. Mater., 39 (1991) 12,
3153–3160, doi:10.1016/0956-7151(91)90049-7
40 F. H. Julian, K. Stelios, On the effect of Luders bands on the bending
of steel tubes. Part II: Analysis, Int. J. Solids. Struct., 48 (2011),
3285–3298, doi:10.1016/j.ijsolstr.2011.07.012
N. R. J. HYNES et al.: OPTIMUM BUSHING LENGTH IN THERMAL DRILLING OF GALVANIZED STEEL ...
822 Materiali in tehnologije / Materials and technology 51 (2017) 5, 813–822
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
J. P. BANTANG, D. CAMACHO: GELLING POLYSACCHARIDE AS THE ELECTROLYTE MATRIX ...
823–829
GELLING POLYSACCHARIDE AS THE ELECTROLYTE MATRIX
IN A DYE-SENSITIZED SOLAR CELL
@ELIRNI POLISAHARID KOT ELEKTROLITNA OSNOVA V
SOLARNIH CELICAH, OB^UTLJIVIH NA BARVILA
Jose Paolo Bantang1, Drexel Camacho1,2
1De La Salle University, Chemistry Department, 2401 Taft Avenue, 1004 Manila, Philippines
2De La Salle University, Organic Materials and Interfaces Unit, CENSER, 2401 Taft Avenue, 1004 Manila, Philippines
drexel.camacho@dlsu.edu.ph
Prejem rokopisa – received: 2016-09-30; sprejem za objavo – accepted for publication: 2017-03-16
doi:10.17222/mit.2016.294
Hydrophilic polysaccharide, -carrageenan, was utilized as the polymer matrix in gel-electrolyte systems for dye-sensitized
solar-cell (DSSC) applications. The influence of the solvent system was investigated to optimize the solubility of -carrageenan
and tetrabutylammonium-iodide (TBAI)/I2 electrolytes by minimizing the water content because of its unfavorable effect on
DSSCs. We report herein that two solvent systems, a water/acetonitrile mixed solvent and DMSO, were found to effectively
dissolve the components. The composite natures of the -carrageenan-electrolyte systems in these solvents were confirmed with
an FTIR analysis. The presence of -carrageenan did not impede the electrochemical properties of the electrolytes, as confirmed
with cyclic voltammetry, electrochemical impedance spectroscopy and linear sweep voltammetry. The incorporation of the gel
electrolytes in DSSCs showed that the DMSO system exhibited better solar-cell efficiency compared to the mixed-solvent
system.
Keywords: dye-sensitized solar cell, -carrageenan, gel electrolyte, electrochemical impedance spectroscopy, ionic conductivity
Hidrofilni polisaharid, -karagen, je bil uporabljen kot polimerna osnova v `elirno elektrolitskih sistemih za uporabo v barvno
ob~utljivih son~nih celicah (angl. DSSC). Da bi izbolj{ali topnost -karagena in elektrolitov tetrabutilamonijevega iodida
(TBAI)/I2 z zmanj{anjem vsebnosti vode, zaradi njenega ne`elenega u~inka na DSSC, je bila preiskovana ob~utljivost sistema
topil. ^lanek poro~a, da sta bila najdena dva nova sistema topil, me{anica topila voda/acetonitril in DMSO, ki u~inkovito
raztopita komponente. Lastnosti oz. obna{anje kompozitov elektrolitskega sistema -karagen v teh raztopinah, so bile potrjene z
FTIR-analizo. Prisotnost -karagenskega elektrolitnega sistem ni predstavljala ovire za imepdan~no spektroskopijo in lienarno
"sweep" voltametrijo. Vklju~itev gel-elektrolitov v DSSC je pokazala, da DMSO-sistem ka`e bolj{o solarno celi~no
u~inkovitost kot sistem me{anih topil.
Klju~ne besede: barvno ob~utljive son~ne celice, rastlinska `elatina -karagen, gel-elektrolit, elektrokemi~na impedan~na
spekroskopija, ionska prevodnost
1 INTRODUCTION
Gel-polymer electrolytes are promising materials for
a potential incorporation in electrochemical devices1
such as dye-sensitized solar cells (DSC). This is because
their mechanical behavior is that of solids, yet their inter-
nal structure is flexible and their conductivity behavior
resembles that of a liquid state allowing a good elec-
trode/electrolyte contact.2,3 Moreover, its ease of fabrica-
tion allows more tunability in designing systems for
specific applications.
Natural polysaccharide is a potential matrix for
polymer electrolytes owing to its hydrophilic and
gel-forming capacity that traps the solvent together with
the redox couple inside the polymer matrix. The high
water-retention property of polysaccharide-based poly-
mer-electrolyte systems was shown to promote good
ionic conductivities and thermal stability.4,5 Agarose,6,7
cellulose and its derivatives8,9 and -carrageenan10 are the
gel-forming natural polysaccharides that have been used
as polymer-electrolyte (PE) systems for dye-sensi-
tized-solar-cell (DSSC) applications. To maximize the
water-retention property of these polysaccharide-based
gel systems, an aqueous medium is necessary. However,
we can see that water disturbs the interfacial attachment
of dyes to TiO2. Thus, it is highly desirable to develop a
polysaccharide-based electrolyte system in a smaller
amount of water or in a non-aqueous medium.
-Carrageenan, which is composed mainly of
disaccharide units of -D-galactopyranose with either an
-D-galactopyranose or 3,6-anhydrogalactose is a pro-
mising polysaccharide matrix for electrolytes due to its
hydrophilic, linear and sulfated properties. The electro-
static interactions of ions with hydroxyl groups and
sulfate groups in the main chain are essential to the
conduction mechanism of a polysaccharide electrolyte
system.11,12 The promising features of -carrageenan as a
polymer-electrolyte system can be potentially applied to
DSSCs. Solid-state DSSCs using carrageenan-gel elec-
trolyte systems were fabricated by soaking the aqueous
polymer gel with the redox electrolytes11 and forming
thin polymer membranes13 leading to decent efficiencies.
To date, studies on the preparation and characterization
of carrageenan-electrolyte systems have been limited
Materiali in tehnologije / Materials and technology 51 (2017) 5, 823–829 823
MATERIALI IN TEHNOLOGIJE/MATERIALS AND TECHNOLOGY (1967–2017) – 50 LET/50 YEARS
UDK 67.017:621.3.035.22:544.623 ISSN 1580-2949
Original scientific article/Izvirni znanstveni ~lanek MTAEC9, 51(5)823(2017)