applied

sciences
Article
Integrated Optimum Layout of Conformal Cooling
Channels and Optimal Injection Molding Process
Parameters for Optical Lenses
Chen-Yuan Chung
Department of Mechanical Engineering, National Central University, Taoyuan City 32001, Taiwan;
cxc474@case.edu or cychung@ncu.edu.tw; Tel.: +886-34267331; Fax: +886-34254501




Received: 20 August 2019; Accepted: 12 October 2019; Published: 15 October 2019


Abstract: Plastic lenses are light and can be mass-produced. Large-diameter aspheric plastic lenses
play a substantial role in the optical industry. Injection molding is a popular technology for plastic
optical manufacturing because it can achieve a high production rate. Highly efficient cooling channels
are required for obtaining a uniform temperature distribution in mold cavities. With the recent
advent of laser additive manufacturing, highly efficient three-dimensional spiral channels can be
realized for conformal cooling technique. However, the design of conformal cooling channels is very
complex and requires optimization analyses. In this study, finite element analysis is combined with
a gradient-based algorithm and robust genetic algorithm to determine the optimum layout of cooling
channels. According to the simulation results, the use of conformal cooling channels can reduce the
surface temperature difference of the melt, ejection time, and warpage. Moreover, the optimal process
parameters (such as melt temperature, mold temperature, filling time, and packing time) obtained
from the design of experiments improved the fringe pattern and eliminated the local variation of
birefringence. Thus, this study indicates how the optical properties of plastic lenses can be improved.
The major contribution of present proposed methods can be applied to a mold core containing the
conformal cooling channels by metal additive manufacturing.

Keywords: gradient-based algorithm; robust genetic algorithm; warpage; design of experiments;

fringe pattern; birefringence

1. Introduction
The demand for plastic optical lenses has been increasing in the industry. The precision
requirements for high-tech optical products have become stringent, which has led to the growth of
global markets for high-precision optical articles [1,2]. Although the optical properties of glass, such as
the refractive index and dispersion, are quite stable, the glass fabrication process is very complicated
and difficult. Moreover, plastic products are lightweight, colorable, robust, and low cost. They can be
manufactured using a one-step process regardless of their geometric complexity. Thus, plastic products
have become crucial in contemporary industrial development.
Despite the aforementioned advantages, plastic optical lenses may encounter volumetric shrinkage,
which leads to the formation of thermally induced residual stress during the cooling process of injection
molding. This residual stress slightly results in local variations in the birefringence, which affects the
image quality [3]. Achieving a uniform temperature distribution for removing the residual stress is
difficult in conventional cooling channels. Uneven shrinkage occurs if conventional cooling channels
are used during the cooling process [4]. For designing conformal cooling channels, the geometric
shape of conventional cooling channels can be appropriately adjusted through injection molding
simulation. This helps to improve the defects caused by conventional cooling channels. The use of

Appl. Sci. 2019, 9, 4341; doi:10.3390/app9204341 www.mdpi.com/journal/applsci

Appl. Sci. 2019, 9, 4341 2 of 17

conformal cooling channels facilitates an even distribution of the surface temperature of the mold
cavities, thereby reducing the thermally induced residual stress formed during the cooling process [5],
effectively shortening the cooling time, and improving the cooling efficiency [6].
Researchers have devoted considerable attention to hybrid manufacturing processes,
which combine metallic powder-based laser additive processes and subtractive machining processes.
These hybrid manufacturing processes are considered the most promising technology for fabricating
conformal cooling channels [7]. With metal additive manufacturing technology, complex conformal
cooling channels can be fabricated to closely fit the shape of the mold cavity and core.
Thus, uniform cooling can be achieved for products even in narrow regions or areas that may
easily accumulate heat. Consequently, the quality of the products improves, and the cycle time
decreases. This technology can be applied to products that vary in thickness. It allows the products to
achieve uniform heat dissipation with a high cooling efficiency [8].
In recent years, optimization algorithms have been used to design cooling channels for plastic
injection molds. Qiao [9] combined the advantages of the Davidon–Fletcher–Powell (DFP) method and
the simulated annealing (SA) algorithm to optimize a cooling system layout. First, the DFP method
was used to find the local optimum layout of the cooling channel. Then, the SA algorithm was adopted
to determine the global optimum layout of the cooling channel, which allowed the surfaces of the
mold cavities to possess a uniform temperature distribution. Park and Dang [10] used design of
experiments (DOE) and response surface methodology for designing an array of baffles in cooling
channels. They established a mathematical model for obtaining the optimal configuration of cooling
channels with an array of baffles. This technique can effectively improve the heat removal performance
and is applicable to large-sized molds and molds with complex cavity shapes. Dang and Park [11]
also proposed an optimization method for the design of U-shaped milled groove cooling channels.
This method aimed to achieve temperature uniformity for the mold and utilized computer-aided
engineering (CAE) for design modification. The mold of a car fender was selected to analyze its
cooling channel design and verify the theoretical calculation. The quality levels of the products
were compared before and after the optimization. After the optimization of the cooling channels,
the warpage decreased, and the temperature uniformity increased.
Optimization strategies involving the use of various algorithms for designing the shape and layout
of cooling channels have been frequently discussed in the literature [12–14]. Moreover, many studies
have used DOE to set optimal process parameters for injection molding [15,16]. However, these two
issues have rarely been integrated in the literature. This study combined finite element
analysis with optimization algorithms to analyze the temperature field during the cooling stage.
Subsequently, the entire injection molding process was conducted using DOE to obtain the best process
parameters. The aim of this study was to uniformly cool melt within a cavity. An optimization was
conducted to design conformal cooling channels, which alleviated the thermally induced residual
stress formed during the manufacturing of optical lenses and solved the uneven shrinkage problem.
Such optimization can also effectively shorten the cooling time and enhance the cooling efficiency
during the manufacturing process, which can improve the image quality of plastic optical lenses.

2. Methods

2.1. Materials
The lens employed in this study was a plastic optical lens commonly used in projectors.
This projector lens was a large-diameter aspheric lens [17] with a diameter of 46 mm (Figure 1).
The lens was composed of cyclo-olefin polymer (COP; Zeonex 480R) from the Zeon Corporation (Tokyo,
Japan). This material had low water absorption, high optical transmittance, and low birefringence [18].
 Appl. Sci. 2019, 9, x FOR PEER REVIEW 3 of 18
Appl.
Appl. Sci.
Sci.2019,
2019, 9,
9, x4341
FOR PEER REVIEW 3 3of
of 18
17

91
91
92 Figure 1.
Figure Specifications of
1. Specifications of the
the large-diameter
large-diameter aspheric
aspheric plastic
plastic lens
lens provided
provided by
by Glory
Glory Science
Science
92
93 Company
Figure 1. Limited.
Specifications
Company Limited. of the large-diameter aspheric plastic lens provided by Glory Science
93 Company Limited.
2.2. Design of the Runner and Gate System
94 2.2. Design of the runner and gate system
94 2.2. Design
Figureof2the runnerthe
displays andshape
gate system
and dimensions of the cold runner and gate system. The volume of
95 Figure 2 displays the shape
the runner and gate system was 8.984 and cmdimensions
3 . A uniformof the cold
melt flowrunner
front and
was gate system.
injected Themold
into the volume of
cavity
95
96 the Figure and
runner 2 displays
gate the shape
system was and
8.984dimensions
cm of the cold
3. A uniform melt runner
flow and gate
front was system.
injected The
intovolume
the moldof
by using the wide cross-sectional inlet of the3 fan gate. This enabled a molded product with large width
96
97 the runner
cavity and gate
by using system
the The
wide was 8.984 cm
cross-sectional . A of
uniform melt flow
Thisfront was injected into the mold
to be filled quickly. warpage and sizeinlet
stabilitytheoffan
widegate.
molded enabled
productsa are molded
majorproduct
concerns with
[4].
97
98 cavity
large by
widthusing
to the
be wide
filled cross-sectional
quickly. The inlet
warpage of
and the fan
size gate.
stability This
of enabled
wide a
molded molded
productsproduct
are with
major
Although the fan gate necessitated manual trimming, it reduced the formation of flow-induced residual
98
99 large width
concerns toAlthough
[4].the be filled the
quickly. Thenecessitated
fanpassed
gate warpage and size stability of wide molded theproducts areofmajor
stress when melt polymer through the manual
gate intotrimming, it reduced
the mold cavity. Therefore,formation
the fan gateflow-
was
99
100 concerns
induced [4]. Although
residual stressthe fan
when gate
the necessitated
melt polymer manual
passed trimming,
through it reduced
the gate the
intoformation
the mold of flow-
cavity.
adopted in this study. Residual stress may lead to poor optical properties, such as uneven distribution
100
101 induced
Therefore, residual
thepatternstresswas
fan gate when the melt
adopted in this polymer
study. passed stress
Residual throughmaythe gate
lead into optical
to poor the mold cavity.
properties,
of the fringe and local variation in the birefringence.
101
102 Therefore, the fan gate was adopted in this study. Residual stress may lead
such as uneven distribution of the fringe pattern and local variation in the birefringence. to poor optical properties,
102 such as uneven distribution of the fringe pattern and local variation in the birefringence.

103
103
104 Figure
Figure 2. Shape and
2. Shape and dimensions
dimensions of
of the
the runner–gate
runner–gate system.
system
104 Figure 2. Shape and dimensions of the runner–gate system
2.3. Mold Design for Conventional and Planar Conformal Cooling Channels
105 2.3. Mold design for conventional and planar conformal cooling channels
105 2.3. Mold
The design for conventional
core and cavity plates and planar
used in conformal
this studycooling
were channels
made of NAK80, which is prehardened
106 The
steel.The
Thecore and cavity
dimensions plates
ofplates used
the mold in
werethis study
150 were
mmwere made
(L) ×made
150 mm of NAK80,
(W) × 205which
mmis is(H).
prehardened steel.
A single-cavity
106
107 The core
dimensionsandofcavity
the mold used150
were in this
mm study
(L) × 150 mm (W) of×NAK80,
205 mm which
(H). A prehardened
single-cavity steel.
mold
107 molddimensions
The containingofboththe conventional
mold were 150cooling
mm (L) channels and(W)
× 150planar
mm planar
× 205conformal
mm cooling
(H).channels channelsmold
A single-cavity was
108 containing
adopted both conventional
(Figure 3) because thecooling
lens was channels
large. and
The total conformal
number of cooling
elements in the cavitywas
wasadopted
222,720.
108
109 containing
(Figure both conventional
3) because the lens cooling
was large. channels
The total and planarofconformal
number elements cooling
in the channels
cavity was was adopted
222,720. The
109 The conventional
(Figure 3) because cooling
the lenschannel
was diameter
large. The was number
total suggested of in the design
elements in [4] to
the be 10was
cavity mm. The layout
222,720. The
110 conventional cooling channel diameter was suggested in the design [4] to be 10 mm. The layout of
110 conventional cooling channel diameter was suggested in the design [4] to be 10 mm. The layout of
 Appl. Sci. 2019, 9, 4341 4 of 17
Appl. Sci. 2019, 9, x FOR PEER REVIEW 4 of 18
Appl. Sci. 2019, 9, x FOR PEER REVIEW 4 of 18
111 of the
the planar
planar conformal
conformal cooling
cooling channeladhered
channel adheredtotothe
thechannel
channeldesign
designguidelines
guidelines proposed
proposed in in the
111
112 the planar conformal
literature cooling
[19] for laser channel
additive adhered toofthe
manufacturing channel
metals. design
These rulesguidelines
These rules werewere used proposed
to in
determine thethe
112
113 literature [19] for laser additive manufacturing of metals. These rules were used
appropriate distance between channels or between the channel and mold cavity surface according toto determine the
113
114 appropriate
the coolingdistance
channelbetween
diameterchannels
(Figure or
4). between thethese
Following
Following channel
these and
rules
rules was
was mold cavity
way tosurface
the way
the ensure according
that the moldto
114
115 the cooling
cavity andchannel
core havediameter
enough(Figure 4). Following
mechanical strength.these rules was
To alleviate
To alleviate thethe way to ensure
thermally
thermally induced
induced that the mold
warpage
warpage [20],
[20],
115
116 cavity and core have
parameters of
the parameters oftheenough
theplanar mechanical
planarconformal strength.
conformalcooling To alleviate
coolingchannel
channelwere
werethe
setsetthermally
asas induced
b =b4 =mm,
follows:
follows: warpage
4 mm, [20],
= 8 mm,
a = 8amm, and
116
117 the =parameters
cand = 8 mm,
8cmm, ofasthe planar
displayed
as displayed conformal
in Figure
in Figure 5. 5.cooling channel were set as follows: b = 4 mm, a = 8 mm, and
117 c = 8 mm, as displayed in Figure 5.

118
118
119 Figure
Figure 3.3. Internal
Internalarrangement
arrangementofof thethe mold
mold used
used in this
in this study.
study. DeepDeep
blue, blue,
brightbright blue, and
blue, green, green, and
orange
119
120 Figure
orange3. components
Internal arrangement
represent of the
the mold usedcooling
conventional in this channels,
study. Deep blue,
planar bright blue,
conformal green,
cooling and
channels,
components represent the conventional cooling channels, planar conformal cooling channels, lens,
120
121 orange
lens, components
and represent
runner–gate, the conventional cooling channels, planar conformal cooling channels,
respectively.
and runner–gate, respectively.
121 lens, and runner–gate, respectively.

122
122
Figure 4. Design rules for the three-dimensional conformal cooling channel [19].
123 Figure 4. Design rules for the three-dimensional conformal cooling channel [19]
123 Figure 4. Design rules for the three-dimensional conformal cooling channel [19]
 Appl. Sci. 2019, 9, x FOR PEER REVIEW 5 of 18
Appl. Sci. 2019, 9, 4341 5 of 17
Appl. Sci. 2019, 9, x FOR PEER REVIEW 5 of 18

124
124
125 Figure 5. Schematic of the planar conformal cooling channel
Figure 5. Schematic of the planar conformal cooling channel.
125 Figure 5. Schematic of the planar conformal cooling channel
126 2.4. Simulation of the filling and packing stages
2.4. Simulation of the Filling and Packing Stages
126
127 2.4. Simulation
In this study,of thethefilling
CAE andmodepacking stages
of Moldex3D was used to simulate injection molding because no
In this study, the CAE mode of Moldex3D was used to simulate injection molding because no
128
127 empirical
In this data
study, were CAE available for the machine settings. to The maximum injection pressure and
empirical data werethe availablemode for theofmachine
Moldex3D was
settings. used
The maximum simulate injection
injection molding
pressure andbecause
maximum no
129
128 maximum
empirical packing
data were pressure
available were forset as
the 250 MPa.
machine The initial
settings. melt
The temperature
maximum was◦
injection set as 270 ℃,and
pressure and
packing pressure were set as 250 MPa. The initial melt temperature was set as 270 C, and the initial
130
129 the initial mold temperature was setset as
as250 ℃ according
100MPa. to melt
the process parameters [21,22] from
℃, andthe
maximum
mold packing
temperature pressure
was 100 ◦ C
set aswere according Theprocess
to the initial temperature
parameters [21,22]wasfromset as 270
the Moldex3D
131
130 Moldex3D
the initial mold (CoreTech
temperature System was Corporation, Taiwan)
100 ℃ according
set as databank. databank.
to the The
process filling time
parameters was
[21,22] set as 0.79
(CoreTech System Corporation, Taiwan) The filling time was set as 0.79 s, and the from the
packing
132
131 seconds, and
Moldex3D the packing
(CoreTech time was
System set as 7 seconds.
Corporation, Taiwan) The filling flow
databank. The rate was time
filling set in was
six steps
set (Figure
as 0.79
time was set as 7 s. The filling flow rate was set in six steps (Figure 6a). The first step involved
133
132 6a). The and
seconds, first step involvedtimethe wasprocess 7 of filling Thethe filling
runner. Therate subsequent five steps(Figure
varied
the process ofthe packing
filling the runner. set as
The seconds.
subsequent five steps flow
varied was settointhe
according six variations
steps in
134
133 according
6a). The to
first the
step variations
involved in the
the cross-sectional
process of fillingareathe of the
runner.plastic
The lens. The
subsequent packing
five pressure
steps was
varied
the cross-sectional area of the plastic lens. The packing pressure was set in three steps (Figure 6b).
135
134 set in three
according steps
topressure (Figure 6b).
the variations in the Thecross-sectional
packing pressure area of the
the plastic
first step was setpacking
as 85%pressure
of the filling
The packing of the first step was set as 85% of of
the filling lens.
pressure The
at the end of filling. Inwas the
136
135 pressure
set in three at steps
the end of filling.
(Figure 6b). The In the following
packing two steps,
pressure of to the
firstpacking
therelease step was pressure
set as 85%was of decreased to
the filling
following two steps, the packing pressure was decreased stress.
137
136 release stress.
pressure at the end of filling. In the following two steps, the packing pressure was decreased to
137 release stress.

138
138 Figure 6. (a) Multistep setting of the filling flow rate profile and (b) multistep setting of the packing
139 Figure 6.profile.
pressure (a) Multistep setting of the filling flow rate profile and (b) multistep setting of the packing
140
139 pressure profile.
Figure 6. (a) Multistep setting of the filling flow rate profile and (b) multistep setting of the packing
2.5. Geometric Optimization of the Conformal Cooling Channels
140 pressure profile.
141 2.5. Geometric optimization
The average of the conformal
mold temperature cooling
was 89.5 channels
◦ C and the average melt temperature was 215.6 ◦ C at the
141
142 completion
2.5. Geometric of optimization
The average the mold
Moldex3D simulation
of the
temperature was of
conformal the℃filling
cooling
89.5 channels
and and
the packing
average stages.
melt These temperatures
temperature was 215.6 ℃ were at the
143
142
input into
completion COMSOL
The average of themoldMultiphysics
Moldex3D
temperature
(Version
simulation
was 89.5
5.2,
of theCOMSOL
filling
℃ and theand
Inc., Burlington,
packing
average meltstages.
MA, USA) and served
These temperatures
temperature
as
was 215.6 ℃ atwere
the
the
144
143
initial
input conditions
into COMSOL
completion
of the
of the Moldex3D
cooling stage.
Multiphysics simulation
According
(Versionof the
to the default
5.2,filling
COMSOL and Inc., settings
packing Burlington,
stages.
in Moldex3D,
MA, USA)
These
the
andsuggested
temperatures servedwereas
◦ C, and the suggested cooling time was 18.6 s. However, the cooling
145
144
channel
the initial
input
temperature
conditions
into COMSOL
was 100
of the cooling
Multiphysics stage.5.2,
(Version According
COMSOLto theBurlington,
Inc., default settings
MA, USA) in Moldex3D,
and served the as
146
145
time in the COMSOL
suggested channel software waswas
temperature set as10020 s,℃,andand neither
the the runner cooling
suggested
the initial conditions of the cooling stage. According to the default settings in Moldex3D,
nor the moldtime base
was was
18.6 included
seconds.
the
147
146
in the simulation.
However,
suggested the
channel
The simulation
cooling time in thewas
temperature
only
COMSOLfocused on the heat
100 ℃,software
and thewas
transfer
set as 20between
suggested seconds,time
cooling
the cooling
and neither
was
channel
18.6the
and
runner
seconds.
148
147
lens. The
nor the mold
However,
cooling
the base
channel
coolingwastime was
included assumed
in thein
to have
the simulation.
COMSOL software
a turbulent
The was
flow.
simulation
set as 20 only
By using
seconds,focusedthe
and on
Reynolds
the heat
neither
number
the transfer
runner
149
148
formula,
between the
the volumetric
cooling
nor the mold base was
flow rate
channel
included and of
in
theThe
lens.
the
conventional
cooling
simulation.
cooling
channel
The waschannel
simulation assumed
only
withtoahave
focused
diameter
a
on
of 10 mm
turbulent
the heat flow.was
transferBy
150 derived
using the 28.81 cm /sec
as Reynolds 3
number and that of thethe
formula, planar conformal
volumetric flow cooling
rate of channel
the with a diameter
conventional cooling ofchannel
4 mm
149 between the cooling channel 3 /sec.and lens. The cooling channel was assumed to have a turbulent flow. By
151
150
was
withderived
using athe as 11.53
diameter
Reynolds of 10 cmmm
number was Moreover,
derived
formula,
COMSOL
28.81 cmwas
theasvolumetric /sec
flow
adopted
and of
rate
to couple
thatthe
the non-isothermal
ofconventional
the planar conformal pipe
cooling
cooling channel
152
151
flow
with
interface
channel with with
a diameter
the heat transfer
a diameter
of 10 mmofwas 4 mm in solids
was
derived derivedinterface
as 28.81ascm 11.53[23]cm for /sec.
/sec and
simulating
thatMoreover, the temperature
of the planar COMSOL
conformal
distribution
was adopted
cooling
153
152 to couple the non-isothermal pipe flow interface with the
channel with a diameter of 4 mm was derived as 11.53 cm /sec. Moreover, COMSOL was adopted heat transfer in solids interface [23] for
153 to couple the non-isothermal pipe flow interface with the heat transfer in solids interface [23] for
 Appl. Sci. 2019, 9, 4341 6 of 17
Appl. Sci. 2019, 9, x FOR PEER REVIEW 6 of 18

154 simulating
of thethe
the melt in temperature
mold cavitydistribution of the melt
during the cooling in the
process. Themold cavity during
solidification of thethe coolingmelt
polymer process.
flow
155 The the
near solidification
cold cavityof wallthewas
polymer melt flow
not considered. near the cold
Furthermore, cavity wall inwas
two approaches not considered.
SmartDO (FEA-Opt
156 Furthermore,
Technology Inc.,two approaches
Miaoli in SmartDO
County; Taiwan), (FEA-Opt
namely Technologyalgorithm
the gradient-based Inc., Miaoli
(GBA)County; Taiwan),
[24] and robust
157 namely algorithm
genetic the gradient-based algorithm
(RGA) [25,26], (GBA) [24]integrated
were separately and robustwithgenetic algorithm
COMSOL (RGA) [25,26],
to optimally designwere
the
158 separately
layout integrated
of the conformalwith COMSOL
cooling to The
channels. optimally design cooling
conventional the layout of the
channels conformal
were cooling
not involved in
159 channels. The conventional cooling channels were not involved in optimization. The
optimization. The integration framework is displayed in Figure 7. The script loop is available in the integration
160 framework
online is displayed
supplementary in Figure
data. 7. The
The pros andscript loop
cons of theistwo
available in the online
optimization supplementary
algorithms were compareddata.
161 The pros and cons of the two optimization algorithms
according to the temperature distributions on the lens surfaces.were compared according to the temperature
162 distributions on the lens surfaces.

163
164 Figure Flow
7. Flow
Figure 7. chart
chart of finite
of finite element
element analysis
analysis integrated
integrated with thewith the optimization
optimization algorithms,
algorithms, gradient-
165 gradient-based algorithm (GBA) and robust genetic algorithm (RGA), for the design
based algorithm (GBA) and robust genetic algorithm (RGA), for the design of conformal of conformal
cooling
166 cooling
channelschannels.

167 The
The value
value ofof the
the distance
distance between
between the the conformal
conformal cooling
cooling channel
channel and
and mold
mold cavity
cavity surface
surface must
must
168 satisfy the rules suggested in Figure 4 so that the design variables (DVs) have
satisfy the rules suggested in Figure 4 so that the design variables (DVs) have a reasonable range, as a reasonable range,
169 as displayed
displayed in in Figure
Figure 8. Moreover,
8. Moreover, at at
thethe suggested
suggested cooling
cooling time
time of of
18.618.6 s, COMSOL
seconds, COMSOL indicated that
indicated
170 the maximal and minimal surface temperatures of the mold cavity for
that the maximal and minimal surface temperatures of the mold cavity for the planar conformalthe planar conformal cooling
0
were Tsur,max 0
channels 0T
171 cooling channels wereand Tsur and
sur,min
, max
, respectively.
0
Tsur , min
After each iteration
, respectively. After eachof the optimization,
iteration the maximal
of the optimization,
and minimal surface temperatures of the mold cavity for the modified conformal cooling channels
172 the maximal and minimal surface temperatures of the mold cavity for the modified conformal cooling
were Tsur,max and Tsur,min , respectively. To ensure that the temperature of mold cavity surface was
173 channels
evenly were Tsur(i.e.,
distributed andvalue
, max the Tsur , min
of T, respectively. To ensure that the temperature of mold cavity
sur,max − Tsur,min was small), the objective function was set as
174 following
surface wasEquation (1):
evenly distributed (i.e., the value of Tsur , max − Tsur , min was small), the objective function
175 was set as following equation (1): [1 − (Tsur,min /Tsur,max )] × 10, 000 (1)

The scaling factor of 10,000 in Equation (1) increases the accuracy of the calculation. To ensure that the
[1 − (Tsur ,min / Tsur ,max )] × 10, 000 (1)
overall temperature of the mold cavity surface after optimization was lower than that of the initial
state, the following constraints were set:
176 The scaling factor of 10,000 in equation (1) increases the accuracy of the calculation. To ensure that
177 the overall temperature of the mold cavity surface − Tafter
0 optimization was lower than that of the initial
Tsur,max sur,max < 0 (2)
178 state, the following constraints were set:
0
Tsur,min − Tsur,min
0 <0 (3)
Tsur ,max − Tsur ,max < 0 (2)

179
 Appl. Sci. 2019, 9, x FOR PEER REVIEW 7 of 18
Appl. Sci. 2019, 9, 4341 7 of 17

0
Tsur,min − Tsur,min < 0 (3)
After setting the DVs, objective function, and constraints, the planar conformal cooling channels
180 After
couldsetting the DVs,
be designed andobjective
modified function, and constraints,
as three-dimensional the planar
channels conformal
by using cooling algorithm.
the optimization channels
181 could be designed and modified as three-dimensional channels by using the optimization algorithm.
This modification improved the cooling efficiency and allowed the temperature of the melt in the mold
182 This modification
cavity to be evenlyimproved the cooling efficiency and allowed the temperature of the melt in the
distributed.
183 mold cavity to be evenly distributed.

184
185 Lowerand
Figure8.8.Lower
Figure andupper
upperbounds
boundsononthe
thedesign
designvariables
variables(DVs)
(DVs)according
accordingtotothe
thedistances
distancessuggested
suggested
in Figure 4 between the cooling channel and mold cavity surface, where the initial value represents the
186 in Figure 4 between the cooling channel and mold cavity surface, where the initial value represents
geometric layout of the planar conformal cooling channels.
187 the geometric layout of the planar conformal cooling channels
2.6. Molding Process Optimization for the Optimized Conformal Cooling Channels
188 2.6. Molding process optimization for the optimized conformal cooling channels
After obtaining the optimized DVs for the conformal cooling channels by using the GBA and
189 After obtaining
RGA algorithms, thethe optimized
geometric DVs for
layouts the conformal
of these two typescooling
of cooling channels
channelsby using the GBA and
were converted into
190 RGA
solidalgorithms,
mesh models theby geometric
using the layouts of these
Rhinoceros 3Dtwo types of
software and cooling
were thenchannels were converted
imported into Moldex3D. into
191 solid
The mesh
processmodels by using
conditions forthethe Rhinoceros
filling and 3Dpacking
softwarestages
and werewerethentheimported
same as intothoseMoldex3D.
mentioned Thein
192 process
Sectionconditions for the time,
2.4. The cooling fillingejection
and packing stages were
temperature, andthe same as those
mold-open time mentioned
were set asin Section
18.6 s, 1392.4.
◦ C,
193 The
and 5 s, respectively. After the entire injection molding process had been simulated, the three types 5of
cooling time, ejection temperature, and mold-open time were set as 18.6 seconds, 139 ℃, and
194 seconds,
conformal respectively. After the
cooling channels (theentire injection
planar, GBA-molding process had been
and RGA-optimized simulated,
channels) were the three types
compared with
195 ofregards
conformal cooling channels (the planar,
to warpage and temperature distribution. GBA- and RGA-optimized channels) were compared with
196 regards to warpage
Finally, and temperature
the geometric layout withdistribution.
the highest cooling efficiency was selected for DOE to determine
197 Finally, the geometric layout
the optimal combination of process parameters with the highest cooling
and key efficiency
parameters was selected
impacting for DOE to
the manufacturing
198 determine the optimal
process. Studies combination
[21,22] have indicated that of process
lenses made parameters
of Zeonexand 480Rkey parameters
material have high impacting
flow-inducedthe
199 manufacturing process. Studies [21,22] have indicated that lenses made
birefringence and low thermally induced birefringence. This study focused on the shear stress related of Zeonex 480R material have
200 high flow-induced
to flow birefringence
and the flow-induced and low
residual thermally
stress. Residualinduced
stressbirefringence.
leads to defective This optical
study focused
properties on the
[16].
201 shear stress shear
Therefore, relatedstresses
to flowatand thethe endflow-induced
of the fillingresidual stress.stresses
stage, shear Residualatstress
the endleadsof to
thedefective
packing
202 optical
stage, properties
and the total [16].warpage
Therefore, were shear
set stresses at the end
as the quality of the
factors, andfilling
eachstage,
qualityshear stresses at the
characteristic was
203 end of the packing stage, and the total warpage were set as the
based on the smaller-the-better approach. The traditional trial and error method for predictingquality factors, and each quality
204 characteristic
and controlling wasinjection
based onmolding
the smaller-the-better
conditions is approach.
inefficient The andtraditional
costly because trial and error
of the method
complexity
205 for predicting and controlling injection molding conditions is inefficient
of interactions between multiple manufacturing process parameters. Therefore, the DOE module and costly because of the
206 complexity
provided by of interactions
Moldex3D isbetween more suitablemultiplethan manufacturing
the trial and process parameters.
error method Therefore, the
for evaluating theDOE
ideal
207 module provided by Moldex3D is more suitable than the trial and error
molding conditions [15]. Moreover, the melt temperature, mold temperature, filling time, and packing method for evaluating the
208 ideal molding conditions [15]. Moreover, the melt temperature, mold temperature,
time were selected as the control factors in the DOE. It was assumed that each control factor contained filling time, and
209 packing time
five levels ofwere selected
variation (Tableas the
1). A control factors
Taguchi’s in the DOE.
orthogonal arrayIt was
L25 (5assumed
4 ) [27] wasthatadopted
each control
for the factor
DOE.
4
210 contained
Subsequently,five levels of variation
statistical analysis(Table 1). A to
was used Taguchi’s
determine orthogonal
the optimum L25 (5 ) [27]
arraycombination ofwas
levelsadopted
for the
control factors.
 Appl. Sci. 2019, 9, x FOR PEER REVIEW 8 of 18

211 for the DOE. Subsequently, statistical analysis was used to determine the optimum combination of
212 levels
Appl. for the
Sci. 2019, control
factors.
9, 4341 8 of 17

213 Table 1. Four control factors and five levels used for the design of experiments (DOE) in Moldex3D
Table 1. Four control factors and five levels used for the design of experiments (DOE) in Moldex3D.
Control factors Level 1 Level 2 Level 3 Level 4 Level 5
Control factors Level 1 Level 2 Level 3 Level 4 Level 5
A Melt temperature (℃) 240 255 270 285 300
A Melt temperature (◦ C) 240 255 270 285 300
B Mold
B temperature (℃)
Mold temperature (◦ C) 80.0080.00 89.75
89.75 99.50 109.25 109.25
99.50 119.00 119.00
C CFilling time Filling
(sec) time (sec) 0.69 0.69 0.79
0.79 0.890.89 0.99 0.99 1.09 1.09
D Packing time (sec) 6 7 8 9 10
D Packing time (sec) 6 7 8 9 10

214 3.3.Results
Resultsand
andDiscussion
discussion
215 InInthis
thisresearch,
research,COMSOL
COMSOLsoftware
softwaresimulated
simulatedthe thecooling
coolingprocess
processofofinjection
injectionmolding.
molding.
216 Moreover, SmartDO optimization software was used to design
Moreover, SmartDO optimization software was used to design the conformal cooling channels the conformal coolingfor
217 channels for improving the temperature distribution of the melt and
improving the temperature distribution of the melt and the cooling efficiency. Furthermore, the cooling efficiency.
218 Furthermore,
Moldex3D Moldex3D
software was software
usedwasto used to simulate
simulate a complete
a complete injection
injection moldingcycle
molding cycle at
at the
thesame
same
219 manufacturing
manufacturing conditions. The temperature difference between the inlet and outlet, averagesurface
conditions. The temperature difference between the inlet and outlet, average surface
220 temperature
temperature ofof
thethe
lens, and
lens, warpage
and warpage deformation
deformation of the lenslens
of the were investigated
were investigatedfor the
for three types
the three of
types
221 conformal cooling channels (the planar, GBA- and RGA-optimized channels). Finally,
of conformal cooling channels (the planar, GBA- and RGA-optimized channels). Finally, the optimum the optimum
222 layout
layoutwith
withthe
thehighest
highestcooling
coolingefficiency
efficiencywas
wasselected
selectedtotooptimize
optimizethe themolding
moldingconditions
conditionsbybyusing using
223 a aDOE
DOEmodule.
module.ThisThisoptimization
optimizationimproved
improvedthe thebirefringence
birefringence and
andthe fringe
the fringepattern
pattern ofof
the
thelens.
lens.
3.1. Optimum Layout of Conformal Cooling Channels for the Cooling Stage
224 3.1. Optimum layout of conformal cooling channels for the cooling stage
According to the Moldex3D simulation, the average mold temperature was 89.5 ◦ C and the average
225 According to the Moldex3D simulation, the average mold temperature was 89.5 ℃ and the
melt temperature was 215.6 ◦ C at the end of the packing stage. These temperatures were input into
226 average melt temperature was 215.6 ℃ at the end of the packing stage. These temperatures were input
COMSOL and served as the initial conditions of the cooling stage. By simulating the cooling process
227 into COMSOL and served as the initial conditions of the cooling stage. By simulating the cooling
for the planar conformal cooling channels, the variations in the maximum and minimum surface
228 process for the planar conformal cooling channels, the variations in the maximum and minimum
temperature of the melt in the mold cavity were determined (Figure 9a). According to the cooling time
229 surface temperature of the melt in the mold cavity were determined (Figure 9a). According to the
suggested by Moldex3D, the lens was ejected at 18.6 s. Figure 9b displays the surface temperature
230 cooling time suggested by Moldex3D, the lens was ejected at 18.6 seconds. Figure 9b displays the
distribution for the lens, with a maximum temperature of 359.713 K, a minimum temperature of
231 surface temperature distribution for the lens, with a maximum temperature of 359.713 K, a minimum
358.601 K, and a temperature difference of 1.112 K.
232 temperature of 358.601 K, and a temperature difference of 1.112 K.

233
Figure 9. (a) Surface temperature of the melt during the cooling stage and (b) surface temperature of
234 Figure 9. (a) Surface temperature of the melt during the cooling stage and (b) surface temperature of
the melt at the default ejection time of 18.6 s when using the planar conformal cooling channels.
235 the melt at the default ejection time of 18.6 seconds when using the planar conformal cooling channels.
The GBA and RGA algorithms of the SmartDO software were employed for evenly reducing
236 the surface
The GBA and RGAofalgorithms
temperature the melt inofthe
themold
SmartDO software
cavity. were employed
The objective forthe
function of evenly
two reducing
algorithmsthe
237 converged
surface temperature of the melt in the mold cavity. The objective function of the two algorithms
to a minimum value in the optimization procedure (Figure 10). The objective function value
of the GBA was 19.78, whereas that of the RGA was 16.54. The optimized design parameters obtained
using these two algorithms are displayed in Figure 11. By using these design parameters (Table 2),
249 However, according to the cooling time suggested by Moldex3D, the lens was ejected at 18.6 seconds.
250 The temperature distribution for the surface of the lens at 18.6 seconds is displayed in Figure 12b,
251 with a maximum temperature of 358.906 K, a minimum temperature of 358.173 K, and a temperature
252 difference of 0.733 K. Moreover, the ejection time for the RGA-optimized conformal cooling channels
253 was Sci.
Appl. 14.55 seconds
2019, 9, 4341 (Figure 13a). The temperature distribution on the lens surface at 18.6 seconds
9 offor
17
254 the RGA-optimized channels is displayed in Figure 13b, with a maximum temperature of 357.861 K,
255 a minimum temperature of 357.255 K, and a temperature difference of 0.606 K. Thus, after the planar
two conformal cooling channels with different geometric layouts could be created and imported
256 conformal cooling channels were modified, the GBA- and RGA-optimized three-dimensional
into COMSOL to simulate the cooling process. The cooling efficiency and improvement in the lens
257 conformal cooling channels enhanced the cooling efficiency and improved the temperature
temperature distribution were compared between the GBA- and RGA-optimized channels.
258 distribution of the melt in the mold cavity.

259
Appl. Sci. 2019, 9, x FOR PEER REVIEW 10 of 18
Figure 10. History of the objective function evolved from: (a) the GBA and (b) RGA during optimization.
260 Figure 10. History of the objective function evolved from: (a) the GBA and (b) RGA during
261 optimization

262
Figure 11. Optimized DVs obtained from: (a) the GBA and (b) RGA.
263 Figure 11. Optimized DVs obtained from: (a) the GBA and (b) RGA
Table 2. The optimized DVs in Figure 11 were obtained from the GBA and RGA.

264 Table 2. The Parameters

Optimized optimized DVs
DV1in Figure
DV2 11DV3
were obtained
DV4 from DV6
DV5 the GBA and RGA.
DV7 DV8
Results of GBA 16.59 15.15 16.89 16.98 −7.98 −8.00 −8.00 −6.51
Optimized parameters DV1
Results of RGA DV2
13.51 14.02DV314.65 DV4 −1.99DV5
17.16 DV6 −2.47 DV7
−8.00 −6.31 DV8
Results of GBA 16.59 15.15 16.89 16.98 -7.98 -8.00 -8.00 -6.51
Results of RGA 13.51 14.02 14.65 17.16 -1.99 -8.00 -6.31 -2.47
The conformal cooling channels designed using the GBA were used to simulate the cooling process.
265 The results indicated the variations in the maximum and minimum surface temperature of the melt
in the mold cavity, as displayed in Figure 12a. For the planar conformal cooling channel, the default
maximum surface temperature for the ejected lens was set as 359.713 K. For a temperature of 359.713 K,
the ejection time for the GBA-optimized conformal cooling channels was 16.9 s. However, according to
the cooling time suggested by Moldex3D, the lens was ejected at 18.6 s. The temperature distribution
for the surface of the lens at 18.6 s is displayed in Figure 12b, with a maximum temperature of 358.906 K,
 Appl. Sci. 2019, 9, 4341 10 of 17
262
263 a minimum temperature of 358.173
Figure K, andDVs
11. Optimized a temperature difference
obtained from: of 0.733
(a) the GBA K.RGA
and (b) Moreover, the ejection
time for the RGA-optimized conformal cooling channels was 14.55 s (Figure 13a). The temperature
264 distribution on Table
the lens surface
2. The at 18.6
optimized DVss for the RGA-optimized
in Figure 11 were obtainedchannels
from the is
GBAdisplayed in Figure 13b,
and RGA.
with a maximum temperature of 357.861 K, a minimum temperature of 357.255 K, and a temperature
Optimized
difference parameters
of 0.606 DV1the planar
K. Thus, after DV2 conformal
DV3 cooling
DV4channelsDV5were modified,
DV6 DV7
the DV8
GBA- and
Results of GBA
RGA-optimized 16.59 conformal
three-dimensional 15.15 cooling
16.89 channels
16.98 enhanced
-7.98 the-8.00 -8.00
cooling efficiency -6.51
and
Results of RGA 13.51 14.02 14.65 17.16
improved the temperature distribution of the melt in the mold cavity. -1.99 -8.00 -6.31 -2.47
265

266
267 Figure
Figure 12.12.
(a)(a) Surface
Surface temperature
temperature of of
thethe melt
melt during
during thethe cooling
cooling stage
stage andand (b) surface
(b) surface temperature
temperature of
268 ofmelt
Appl.the
Sci. the at
melt
2019, 9,the at the
default
x FOR default
PEERejection
REVIEW ejection timesof
time of 18.6 18.6using
when seconds when using thecooling
the GBA-optimized GBA-optimized
channels. cooling
11 of 18
269 channels

270
271 Figure 13.(a)(a)Surface
Figure13. temperature
Surface temperature of of
thethe
melt during
melt thethe
during cooling stage
cooling andand
stage (b) surface temperature
(b) surface of
temperature
272 the
of melt at theat
the melt default ejectionejection
the default time of time
18.6 sof
when
18.6using the RGA-optimized
seconds when using the cooling channels. cooling
RGA-optimized
273 channels.
3.2. Comparison of the Conformal Cooling Channels Designed by Using Different Algorithms for the Entire
Injection Molding Process
274 3.2. Comparison of the conformal cooling channels designed by using different algorithms for the entire
275 The conformal
injection cooling channels designed using the GBA and RGA optimization algorithms
molding process
were imported into Rhinoceros 3D software, which was used to construct the solid mesh models.
276 The conformal cooling channels designed using the GBA and RGA optimization algorithms
The models were then used to execute the entire injection molding simulation with the Moldex3D
277 were imported into Rhinoceros 3D software, which was used to construct the solid mesh models. The
software. The computation time was approximately 45 min when using two 2.4-GHz Intel Xeon
278 models were then used to execute the entire injection molding simulation with the Moldex3D
E5-2620 CPUs with 64 GB of RAM. The results indicated that the optimized conformal cooling channels
279 software. The computation time was approximately 45 minutes when using two 2.4-GHz Intel Xeon
280 E5-2620 CPUs with 64 GB of RAM. The results indicated that the optimized conformal cooling
281 channels exhibited a lower average surface temperature for the melt compared with that exhibited
282 by the planar conformal cooling channels (Figure 14). At the end of cooling, the surface temperatures
283 simulated using Moldex3D were consistent with those simulated using COMSOL (Figures 9b, 12b,
284 and 13b). If the temperature on the lens surface could be evenly distributed, the thermal-
 Appl. Sci. 2019, 9, 4341 11 of 17

exhibited a lower average surface temperature for the melt compared with that exhibited by the planar
conformal cooling channels (Figure 14). At the end of cooling, the surface temperatures simulated using
Moldex3D were consistent with those simulated using COMSOL (Figure 9b, Figure 12b, and Figure 13b).
If the temperature on the lens surface could be evenly distributed, the thermal-displacement-induced
warpage could be improved. Therefore, to evaluate the thermally induced warpage, 18 measured nodes
were selected on the lens surface. The locations and numbering of these measured nodes are displayed
in Figure 15a. Measured nodes 1–16 were located on the top and bottom edges of the lens and used to
calculate the total thermal warpage. The results were used to evaluate the variation in the circularity
(roundness) due to the warpage. The average thermal warpage values for the planar conformal
cooling channels, GBA-optimized conformal cooling channels, and RGA-optimized conformal cooling
channels were 98.296, 98.214, and 98.121 µm, respectively (Figure 15b,c). Thus, the optimized conformal
cooling channels exhibited a lower thermal warpage than the planar conformal cooling channels.
Moreover, measured nodes 17 and 18 were located at the centers of the top and bottom surfaces of the
lens, respectively. These two measured nodes were used to observe the relative thermal warpage in the
vertical direction
Appl. Sci. 2019, (z-direction).
9, x FOR PEER REVIEW The vertical thermal warpage values for the GBA- and RGA-optimized
12 of 18
conformal cooling channels were 21.617 and 21.565 µm, respectively. Both these values were lower
306 16). the
than Thevertical
GBA- thermal
and RGA-optimized conformal
warpage of the planar cooling
conformal channels
cooling improved
channels the temperature
(21.638 µm).
307 uniformity of the mold.

308
309 14. Average surface
Figure 14. surface temperature
temperature of the
the melt
melt in
in cavity
cavity during
during the
the complete
complete injection
injection molding
molding
310 cycle with different
different conformal
conformal cooling
cooling channels.
channels

311
312 Figure 15. (a) Reference numbers and locations of the 18 measured nodes on the lens; (b) total thermal
313 warpage values measured using the upper nodes (1–8); and (c) total thermal warpage values
314 measured using the bottom nodes (9–16).
308
309 Figure 14. Average surface temperature of the melt in cavity during the complete injection molding
Appl. Sci. 2019, 9, 4341 12 of 17
310 cycle with different conformal cooling channels

308
309 Figure 14. Average surface temperature of the melt in cavity during the complete injection molding
310 cycle with different conformal cooling channels

311
Figure 15. (a) Reference numbers and locations of the 18 measured nodes on the lens; (b) total thermal
312 Figure 15.values
warpage (a) Reference
measurednumbers and
using the locations
upper nodesof(1–8);
the 18and
measured
(c) total nodes onwarpage
thermal the lens;values
(b) total thermal
measured
313 warpage
using values nodes
the bottom measured using the upper nodes (1–8); and (c) total thermal warpage values
(9–16).
314 measured using the bottom nodes (9–16).
When designing cooling channels, the rules for their geometric layout should be considered and
the temperature difference between the inlet and outlet should be minimized to prevent the warpage
of the lens due to uneven temperature distribution in the mold. Generally, if the product requires high
311 accuracy, the temperature difference between the inlet and outlet should be lower than 2.5 ◦ C [28].
The simulation results of the GBA- and RGA-optimized cooling channels indicated that the temperature
312 difference
Figurebetween the inlet
15. (a) Reference and outlet
numbers of the optimized
and locations cooling channels
of the 18 measured was
nodes on the lower
lens; thanthermal
(b) total that of the
313 planarwarpage values
conformal measured
cooling using
channels the upper
regardless of nodes (1–8);
the upper andand (c) total
bottom thermal(Figure
channels warpage16).values
The GBA-
314 measured using the
and RGA-optimized bottom nodes
conformal (9–16).
cooling channels improved the temperature uniformity of the mold.

315

315
Figure 16. Temperature difference between the inlet and outlet for: (a) the upper cooling channel and
(b) bottom cooling channel.

3.3. Optimal Process Parameters for Injection Molding

After the geometric optimization, the cooling channel with the highest cooling efficiency
for the mold and the most even temperature distribution for the lens surface was adopted.
Thus, the RGA-optimized cooling channel layout was selected for the DOE to determine the most
suitable molding conditions. The optimal process parameters were selected on the basis of the
signal/noise (S/N) ratio response. The optical properties of the lenses were compared before and after
performing the DOE. Figure 17 illustrates the calculation results of the S/N ratio response from the DOE
module of Moldex3D. The total S/N ratio response was obtained by adding the S/N ratio responses
from three quality factors under four control factors and five levels (Figure 17d). A high S/N ratio
Appl. Sci. 2019, 9, 4341 13 of 17

indicates low noise (external influence). According to the Taguchi method, the level with the maximum
S/N ratio was selected as the optimal condition [16,29]. Therefore, the optimal combination of control
factors, reported in Table 1, and their corresponding levels was A4B2C4D1. The optimal control factor
level settings were A4 (melt temperature: 285 ◦ C), B2 (mold temperature: 89.75 ◦ C), C4 (filling time:
0.99 s), and D1 (packing time: 6 s). This combination of process parameters (A4B2C4D1) was used
in the injection molding simulation. The optics module of Moldex3D was adopted to predict the
birefringence and fringe pattern of the lens. After the DOE analysis, the birefringence of the lens
considerably decreased (Figure 18a), which indicated that the residual stress within the material also
diminished [16]. Moreover, after the processing optimization, the fringe pattern on the lens resembled
concentric circles (Figure 18b). The distributions of the fringe pattern on the optimized molded lens
were rarefied near the gate, which implied that the optical properties of the lens had been improved.
Appl. Sci. 2019, 9, x FOR PEER REVIEW 14 of 18

338
339 Figure 17. S/N ratios for: (a) shear stress at the end of the filling stage; (b) shear stress at the end of
Figure 17. S/N ratios for: (a) shear stress at the end of the filling stage; (b) shear stress at the end of the
340 the packing stage; (c) total warpage; and (d) sum of responses from (a) to (c)
packing stage; (c) total warpage; and (d) sum of responses from (a) to (c).
 Appl. Sci. 2019, 9, 4341 14 of 17
Appl. Sci. 2019, 9, x FOR PEER REVIEW 15 of 18

341
Figure 18. Comparison of the (a) birefringence properties and (b) fringe patterns before and after the
342 Figure 18. Comparison of the (a) birefringence properties and (b) fringe patterns before and after the
optimization of process parameters.
343 optimization of process parameters.
4. Conclusions
344 4. Conclusion
In this study, the GBA and RGA optimization algorithms were used to design the geometric
345 In this
layout study, thecooling
of conformal GBA and RGA optimization
channels. algorithms cooling
The planar conformal were used to design
channels had the geometric
a lens surface
346 layout of conformal cooling channels. The planar conformal cooling channels had a
temperature difference of 1.112 K with an ejection time of 18.6 s. The GBA-optimized conformal cooling lens surface
347 temperature
channels reduceddifference of 1.112
the ejection K to
time with
16.9an ejection
s and time the
improved of 18.6 seconds.
cooling ThebyGBA-optimized
efficiency approximately
348 conformal
9.14%. Thecooling channels conformal
RGA-optimized reduced the ejection
cooling time to
channels had16.9 secondstime
an ejection andofimproved theimproved
14.55 s and cooling
349 efficiency by approximately 9.14%. The RGA-optimized conformal cooling channels
the cooling efficiency by approximately 21.77%. The GBA-optimized conformal cooling channels had an ejection
350 time of 14.55
exhibited seconds
a decrease in and improved the
the temperature coolingonefficiency
difference by approximately
the lens surface from 1.112 to21.77%.
0.733 K, The
whichGBA-
was
351 optimized conformal
an improvement cooling Furthermore,
of 34.08%. channels exhibited a decrease in the
the RGA-optimized temperature
conformal difference
cooling channelonexhibited
the lens
352 surface from
a decrease 1.112
in the to 0.733 difference
temperature K, whichonwaslensan improvement
surface from 1.112 of 34.08%.
to 0.606 Furthermore,
K, which the RGA-
was an improvement
353 optimized
of 45.5%. conformal cooling channel exhibited a decrease in the temperature difference on lens
354 surface from 1.112
During to 0.606
the entire K, which
injection was an
molding improvement
cycle, the planarof 45.5%. cooling channels exhibited the
conformal
355 During
highest the entire
average surfaceinjection molding
temperature cycle,
of the melt,thefollowed
planar conformal cooling channels
by the GBA-optimized exhibited
conformal the
cooling
356 highest
channels.average
The surface temperature
RGA-optimized of the melt,
conformal followed
cooling by the
channels GBA-optimized
exhibited the lowestconformal
average cooling
surface
357 channels. The RGA-optimized conformal cooling channels exhibited the lowest average surface
358 temperature of the melt. Furthermore, the distribution of the thermal-displacement-induced warpage
Appl. Sci. 2019, 9, 4341 15 of 17

temperature of the melt. Furthermore, the distribution of the thermal-displacement-induced warpage

of the lens could be used as an index to evaluate whether the cooling channel system was suitably
designed. The GBA- and RGA-optimized conformal cooling channels exhibited improvements of
0.18% and 0.52%, respectively, in average thermal warpage of measured nodes as compared with the
planar conformal cooling channels. However, thermally induced warpage still could not be compared
with experimental results because of insufficiently available experimental data.
In summary, the RGA-optimized conformal cooling channels had the highest cooling efficiency
for the mold and a superior temperature distribution for the melt in the mold cavity, which reduced
thermally induced warpage. Therefore, using RGA-optimized conformal cooling channels shortened
the development time and considerably improved the quality of the plastic lens. DOE was used to
evaluate the effect of manufacturing process parameters on the optical properties. The optimal process
parameters reduced the birefringence and improved the shape of the fringe pattern.
Recently, three-dimensional printing and additive manufacturing have attracted tremendous
attention worldwide. Among these technologies, the additive manufacturing of metal powder has
quickly developed into a feasible technique for fabricating metal parts. Additive manufacturing
typically uses a high-energy electron beam or laser beam to sinter or melt the metal powder to form the
solid parts [30,31]. This method can be used to fabricate mold cores for plastic product manufacturing
and construct three-dimensional conformal cooling channels [32,33]. The methods presented in this
study can serve as useful references to obtain optimized cooling channel layouts and injection molding
conditions for the initial stages of mold development. Mold tooling with conformal cooling channels
can be used to improve the cooling efficiency and temperature distribution of the melt, which solves
warpage issues. Optimal injection molding conditions can improve product quality. By controlling key
molding conditions, the product yield rate can be enhanced, and the product cost can be decreased.

Supplementary Materials: The following are available online at http://www.mdpi.com/2076-3417/9/20/4341/s1.

The integration of the COMSOL finite element model and SmartDO genetic optimization algorithm is available in
the online version.
Author Contributions: Conceptualization, C.-Y.C.; methodology, C.-Y.C.; formal analysis, C.-Y.C.; investigation,
C.-Y.C.; data curation, C.-Y.C.; writing—original draft preparation, C.-Y.C.; writing—review and editing, C.-Y.C.;
supervision, C.-Y.C.; project administration, C.-Y.C.; funding acquisition, C.-Y.C.
Funding: This research was funded by the Ministry of Science and Technology in Taiwan under the grant number
MOST 104-2218-E-008-006.
Acknowledgments: The author would like to thank his student, Po-Yang Chang, was of great help in generating
some of the pictures shown in this paper. Special thanks for manager of EOS Singapore, Stephanie Cheong,
providing an approval of reproducing Figure 4 with permission of the copyright owner. Also, thanks for Shih-Ying
Ke from Glory Science Company providing the AutoCAD file of Figure 1.
Conflicts of Interest: The finite element software COMSOL was run under the license number 5084742. The author
has no conflicts of interest related to this manuscript, commercial or otherwise.

References
1. Garrette, B.; Karnani, A. Challenges in marketing socially useful goods to the poor. Calif. Manag. Rev.
2010, 52, 29–47. [CrossRef]
2. Kohara, T. Development of new cyclic olefin polymers for optical uses. Macromol. Symp. 1996, 101, 571–579.
[CrossRef]
3. Lee, Y.B.; Kwon, T.H. Modeling and numerical of residual stresses and birefringence in injection molded
center-gated disks. J. Mater. Process. Technol. 2001, 111, 214–218. [CrossRef]
4. AC Technology Inc. C-MOLD Design Guide: A Resource for Plastics Engineers; Advanced CAE Technology:
Ithaca, NY, USA, 1995.
5. Sachs, E.; Wylonis, E.; Allen, S.; Cima, M.; Guo, H. Production of injection molding tooling with conformal
cooling channels using the three dimensional printing process. Polym. Eng. Sci. 2000, 40, 1232–1247.
[CrossRef]
Appl. Sci. 2019, 9, 4341 16 of 17

6. Dalgarno, K.W.; Stewart, T.D. Manufacture of production injection mould tooling incorporating conformal
cooling channels via indirect selective laser sintering. Proc. Inst. Mech. Eng. Part B 2001, 215, 1323–1332.
[CrossRef]
7. Yan, C.; Hsu, A. Introduction of Composite Technology, Combining Machining with Selective Laser Melting
for Metal Powder Forming. In Molding Innovation Newsletter Feb. 2012; CoreTech System Co., Ltd.: Chupei City,
Taiwan, 2012; pp. 5–10.
8. Hsu, F.H.; Wang, K.; Huang, C.T.; Chang, R.Y. Investigation on conformal cooling system design in injection
molding. Adv. Prod. Eng. Manag. 2013, 8, 107–115. [CrossRef]
9. Qiao, H. A systematic computer-aided approach to cooling system optimal design in plastic injection molding.
Int. J. Mech. Sci. 2006, 48, 430–439. [CrossRef]
10. Park, H.-S.; Dang, X.-P. Optimization of conformal cooling channels with array of baffles for plastic injection
mold. Int. J. Precis. Eng. Manuf. 2010, 11, 879–890. [CrossRef]
11. Dang, X.-P.; Park, H.-S. Design of U-shape milled groove conformal cooling channels for plastic injection
mold. Int. J. Precis. Eng. Manuf. 2011, 12, 73–84. [CrossRef]
12. Jahan, S.A.; Wu, T.; Zhang, Y.; Zhang, J.; Tovar, A.; El-Mounayri, H. Thermo-mechanical design optimization
of conformal cooling channels using design of experiments approach. Procedia Manuf. 2017, 10, 898–911.
[CrossRef]
13. Li, Z.; Wang, X.; Gu, J.; Ruan, S.; Shen, C.; Lyu, Y.; Zhao, Y. Topology optimization for the design of conformal
cooling system in thin-wall injection molding based on BEM. Int. J. Adv. Manuf. Technol. 2018, 94, 1041–1059.
[CrossRef]
14. Jahan, S.A.; El-Mounayri, H. A thermomechanical analysis of conformal cooling channels in 3D printed
plastic injection molds. Appl. Sci. 2018, 8, 2567. [CrossRef]
15. Chen, C.-C.A.; Vu, L.T.; Qiu, Y.-T. Study on injection molding of shell mold for aspheric contact lens
fabrication. Procedia Eng. 2017, 184, 344–349. [CrossRef]
16. Lin, C.-M.; Hsieh, H.-K. Processing optimization of Fresnel lenses manufacturing in the injection molding
considering birefringence effect. Microsyst. Technol. 2017, 23, 5689–5695. [CrossRef]
17. Shieh, J.-Y.; Wang, L.K.; Ke, S.-Y. A feasible injection molding technique for the manufacturing of large
diameter aspheric plastic lenses. Opt. Rev. 2010, 17, 399–403. [CrossRef]
18. Chen, C.-C.A.; Tang, J.-C.; Teng, L.-M. Effects of mold design of aspheric projector lens for head up display. In
Proceedings of the Polymer Optics Design, Fabrication, and Materials, San Diego, CA, USA, 12 August 2010;
Volume 7788, p. 778806.
19. Mayer, S. Optimised Mould Temperature Control Procedure Using DMLS; EOS Whitepaper 2005; EOS GmbH
Ltd.: Krailling, Germany, 2005; pp. 1–10.
20. G-Plast Pvt. Ltd. Synergy of True & Full 3D Simulation and Conformal Cooling. In Molding Innovation
Newsletter Feb. 2012; CoreTech System Co., Ltd.: Chupei City, Taiwan, 2012; pp. 18–20.
21. Wang, P.-J.; Lai, H.-E. Study of residual birefringence in injection molded lenses. In Annual Technical Conference
(ANTEC); SPE- the Society of Plastics Engineers: Cincinnati, OH, USA, 2007; pp. 2494–2498.
22. Lai, H.-E.; Wang, P.-J. Study of process parameters on optical qualities for injection-molded plastic lenses.
Appl. Opt. 2008, 47, 2017–2027. [CrossRef]
23. COMSOL Inc. Application ID: 12371, Cooling of an Injection Mold. Available online: http://www.comsol.
com/model/12371 (accessed on 1 July 2016).
24. Chung, C.-Y.; Mansour, J.M. Determination of poroelastic properties of cartilage using constrained
optimization coupled with finite element analysis. J. Mech. Behav. Biomed. Mater. 2015, 42, 10–18.
[CrossRef]
25. Chen, Y.-C.; Huang, B.-K.; You, Z.-T.; Chan, C.-Y.; Huang, T.-M. Optimization of lightweight structure and
supporting bipod flexure for a space mirror. Appl. Opt. 2016, 55, 10382–10391. [CrossRef]
26. Chen, S.-Y. An approach for impact structure optimization using the robust genetic algorithm. Finite Elem.
Anal. Des. 2001, 37, 431–446. [CrossRef]
27. Lin, C.-M.; Wang, C.-K. Processing optimization of optical lens in the injection molding. Adv. Mater. Res.
2013, 813, 161–164. [CrossRef]
28. Himasekhar, K.; Lottey, J.; Wang, K.K. CAE of mold cooling in injection molding using a three-dimensional
numerical simulation. J. Eng. Ind. 1992, 114, 213–221. [CrossRef]
Appl. Sci. 2019, 9, 4341 17 of 17

29. Chen, D.-C.; Huang, C.-K. Study of injection molding warpage using analytic hierarchy process and Taguchi
method. Adv. Technol. Innov. 2016, 1, 46–49.
30. Mazur, M.; Brincat, P.; Leary, M.; Brandt, M. Numerical and experimental evaluation of a conformally cooled H13
steel injection mould manufactured with selective laser melting. Int. J. Adv. Manuf. Technol. 2017, 93, 881–900.
[CrossRef]
31. Abbès, B.; Abbès, F.; Abdessalam, H.; Upganlawar, A. Finite element cooling simulations of conformal
cooling hybrid injection molding tools manufactured by selective laser melting. Int. J. Adv. Manuf. Technol.
2019, 103, 2515–2522. [CrossRef]
32. Kitayama, S.; Miyakawa, H.; Takano, M.; Aiba, S. Multi-objective optimization of injection molding process
parameters for short cycle time and warpage reduction using conformal cooling channel. Int. J. Adv.
Manuf. Technol. 2017, 88, 1735–1744. [CrossRef]
33. Kuo, C.-C.; Jiang, Z.-F. Numerical and experimental investigations of a conformally cooled maraging steel
injection molding tool fabricated by direct metal printing. Int. J. Adv. Manuf. Technol. 2019. [CrossRef]

© 2019 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).