Process-induced Residual Stress Analysis

by Resin Transfer Molding

YEONG K. KIM*
Samsung Techwin Co. Ltd
San 14, Nongseo-Ri, Giheung-Eub, Yongin-Si
Kyoungki-Do, Republic of Korea, 449-712

(Received August 11, 2003)

(Accepted January 26, 2004)

ABSTRACT: The objective of this study is to investigate the residual stress

development during the resin transfer molding process. The analysis was based on
the strain measurement tests during the entire cure process of unidirectional carbon
fiber–epoxy matrix composite laminate using aluminum mold. The investigation
efforts were particularly focused on the verification of the strain development
mechanism. Although complex interference between the mold structure and the
composite laminate was observed during the process, it was found that the final
residual strain of the laminate was induced only by thermal mismatch during the
cooldown stage. Based on the observation, the residual stresses were calculated by a
viscoelastic classical laminated plate theory. The viscoelastic modulus in the
calculations was obtained by a series of stress relaxation tests of the material. The
curvatures of [0
/90
] asymmetric laminate were calculated, and the results were
compared with the experimental data. The results showed that the residual stress
developed by the resin transfer molding could be calculated by considering only the
cooldown stage of the process.

KEY WORDS: thermoset composite, resin transfer molding, processing, residual

stress.

INTRODUCTION

ROCESS-INDUCED STRESS OF polymer matrix–fiber-reinforced composite structures

P has always remained an issue of concern. The residual stress directly impacts on the
dimensional stability and the structural performance of the final product. In the analysis
of the residual stress, careful investigation on numerous issues is essential due to
the complex mechanism involved during the process. When a tool is employed in the
manufacturing, the situation becomes more complicated. The tool may be a caul plate for
the autoclave process, a mandrel for filament winding, or a mold for the resin transfer
molding process. The interference between the composite structure and the tool could have
considerable influence on the stress development during the manufacturing process.
*E-mail: [email protected]

Journal of COMPOSITE MATERIALS, Vol. 38, No. 11/2004 959

0021-9983/04/11 0959–14 $10.00/0 DOI: 10.1177/0021998304043746
ß 2004 Sage Publications

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

960 Y. K. KIM

Tremendous efforts were made to examine the residual stress development of polymer
matrix fiber-reinforced composite structure [1–15]. A typical approach was to calculate
the residual stress analytically, and compare the results with the experimental data to
verify the analysis. As the role of the chemical shrinkage strain during the process was
presumed to be important, the modeling has been expanded to the modulus change of the
polymer matrix to cover the entire cure process analysis [5–8]. A few experimental studies
showed an additional issue of the interference effect between the composite structure and
the tool structure such as the boundary condition and the tool surface on the final residual
stress, which exposed the importance of the environmental factors in the stress
development [9–15].
In spite of the contributions of the previous studies, the mechanism of the stress
development seems to be not well defined. The main reason exists in the fact that it is very
difficult to simulate the real situation encountered in the process, especially when
the complex boundary conditions are involved. Usually, the verification of the analysis
is achieved by simply comparing the calculation results with the experimental data
(curvatures of the asymmetric laminated samples, for example). Occasionally, the
analytical results do not reflect the lack of the proper quantification of the detailed
process condition. Therefore, good agreements with the data do not necessarily guarantee
the rightness of the analysis. Probably, one of the best approaches is the in-situ strain
measurement during the entire cure process, which may provide the detailed information
of actual happening during the cure process. This is especially important for the case of the
resin transfer molding process, where the effect of the environmental factor is more
evidential due to the nature of the process using the mold surrounding the composite
structure.
A comprehensive study on the strain development of the resin transfer molding process
was reported during the entire cure process [16]. In this study, the strains were measured
using a conventional strain gage and fiber optic strain gage, and significant strain
developments were measured under the influence of the mold structure. The experimental
data showed that the strains which occurred during the heat-up stage were fixed, and
remained permanent. Consequently, although the data showed a strong interference of
the mold structure during the cure process, the final residual strain in the composite
structure was developed only during the cooldown stage. The purpose of this paper is to
verify this observation, and to define the mechanism of the residual stress development by
the RTM process. The investigation is very important, because if the final residual strain
is developed only during the cooldown stage, it is not necessary to consider the entire cure
process in the RTM process. The investigation is especially focused on the interference
induced by the mold structure to interpret the strain data. Based on the analysis results,
the curvatures of [0
/90
] asymmetric laminates were numerically calculated. A classical
laminated plate theory was employed for viscoelastic formulation. Tests for glass transi-
tion temperature and stress relaxation modulus were performed to obtain the material
properties. The calculation results were compared with the experimental data to verify the
analysis.

EXPERIMENTS

As was mentioned previously, the analysis was based on the experimental data
already obtained and published [16].

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

Process-induced Residual Stress Analysis by RTM 961

In the previous work [16], three different cure cycles were designed for the RTM
process. The resin material consisted of Alardite GY 6010 epoxy, HY 917 hardener, and
DY 070 accelerator (Ciba Specialty Chemical Corporation). For the reinforcing material,
AS4 unidirectional carbon fiber was selected, and the mold material was aluminum.
The peak cure temperatures were 160
C (Cure Cycle 1), 135
C (Cure Cycle 2), and 115
C
(Cure Cycle 3) respectively, and corresponding cure times were determined by the cure
kinetic model. The laminate was unidirectional, and the strains were measured to 0o- and
90o-directions with respect to the fiber. The results are repeated in Figures 1–3. Although
fiber optic gages were used at the same time in the measurements [16], the gages could not
measure the strains until the resin was gelled. Therefore, the data from the conventional
strain gages are shown in the figures.

180 3000

160 2000

140
1000
Temperature ( C)
o

120

Strain (µε)
0
100
Temp −1000
80
o
0 Mold −2000
60 o
90
40 −3000

20 −4000
0 20 40 60 80 100
Time (min)

Figure 1. Strain developments during resin transfer molding process (Cure Cycle 1) [2].

140 3000

120 2000

100 1000
Temperature ( C)

Strain (µε)
o

80 Temp 0

o Mold
0
60 o
−1000
90

40 −2000

20 −3000
0 20 40 60 80 100 120 140
Time (min)

Figure 2. Strain developments during resin transfer molding process (Cure Cycle 2) [2].

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

962 Y. K. KIM
120 2000

100 1000

Temperature ( C)
o

Strain (µε)
80 0

Temp.
60 o −1000
0 Mold

o
90
40 −2000

20 −3000
0 50 100 150 200
Time (min)

Figure 3. Strain developments during resin transfer molding process (Cure Cycle 3) [2].

During the heat-up, the strain gage readings were almost identical to the mold strain.
This implies that the composite structure, especially preform, was fully constrained in both
directions, although the resin was presumably liquid before gellation. In Figure 1, the
constraint was released at about 140
C. Evidentially, the release occurred earlier in the 0
-
direction than in the 90
-direction. At the peak cure temperature, the strains remained
either constant (Figures 2 and 3), or slightly increased (Figure 1) due to the weak
constraint of the mold. When the laminate was removed from the mold after cooldown,
sudden changes in the strains were observed due to the release from the constraint of
the mold. Here, most of the strain to 0
-direction returned to the value at the peak cure
temperature. The data indicate the permanent stretch of the composite laminate in this
direction. The permanent strain (or constraint-induced strain) was verified by the meas-
urement of the maximum strain in the previous work [16]. Presumably, this constraint-
induced strain also occurred in the 90
-direction, but the final residual strain was negative
due to the high coefficient of thermal expansion (CTE) value in this direction. Similar
results can be found in Figures 2 and 3.

DATA ANALYSIS

In addition to the observation of the constraint-induced strain, the data provide

valuable information about the strain development during the cure process. As all the
strains developed during the heat-up and the peak cure temperature dwell remained as
(fixed and irreversible) constrained-induced strain, we can consider that the laminate was
effectively in a strain-free state at the peak cure temperature. Therefore, the strain
developments to the 0
- and 90
-directions during cooldown represent the total residual
strain in the composite structure. In Figure 1, the strain development during cooldown is
5700
" to the 90
-direction, and about 100
" to the 0
-direction, respectively. The CTEs
to the 90
- and the 0
-directions were already found as 43
"/
C and 0.5–1
"/
C in the
previous study, respectively [16]. As the temperature change was about 135
C from the

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

Process-induced Residual Stress Analysis by RTM 963

180 1

160
0.8
140

Normalized Storage
Modulus E' (MPa)
Temperature ( C)
o

0.6
120
Cure cycle 1

100 Cure cycle 2

0.4

80
0.2
60

40 0
0 20 40 60 80 100 120
Time (min)

Figure 4. Storage modulus development during Cure Cycles 1 and 2.

peak cure temperature to room temperature, we could conclude that the strains were
induced only by the temperature drop during the cooldown. The results do not show any
other type of strain development such as chemical shrinkage strain. Same results were
found in Figures 2 and 3. As stated earlier, the strain induced by the chemical shrinkage
has been presumably significant, and regarded as the main contribution to the residual
stress development before cooldown. The data represent that, although there was chemical
shrinkage strain during the cure process, the strain vanished in the constraint-induced
strain, and did not show any contribution to the effective strain development.
For further investigation, the storage modulus development during the cure progress
was examined. The modulus change was measured by a dynamic mechanical analyzer
(DMA 2980, TA Instruments), and the results are shown in Figure 4 for 160
C peak
temperature and 135
C peak temperature cure cycles. The onset of gellation was found at
125
C. The data indicate that the stress development before 125
C was not significant due
to low stiffness. After the gel point, the resin cured rapidly and achieved full cure by
reaching the plateau at the end of the peak cure temperature. We can easily guess that the
chemical shrinkage strain should be detected by the strain gages if the strain development
was strong enough to overcome the constraint effect of the mold. Since the mold structure
strongly constrained the composite laminate during this period, the chemical shrinkage
could not establish effective strain development.
The strains induced by the interference with the mold were significant from the
beginning of the cure cycle to the end of the peak temperature dwell. Again, the strains
were fixed and irreversible, they did not make any contribution to the residual stress
development.

NUMERICAL MODELING

To verify the results discussed in the previous section, a viscoelastic formulation was
developed based on the classical laminated plate theory [17]. Since the strains developed

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

964 Y. K. KIM

before the cooldown are effectively zero, the formulation can be simplified to consider
only the cooldown stage, when the composite laminate is fully cured. The time–
temperature dependent forces N and bending moments M per unit length can be
calculated by the convolution integral as
Zt
X n @½"kx,y 0
½NðtÞx,y ¼ hk ½Qð  0 Þkx,y dt
k¼1 0 @t0
Zt ð1Þ
X n
0 k
@½"kx,y 0
½MðtÞx,y ¼ hk z k ½Qð   Þx,y dt
k¼1 0 @t0

where ½Qð  0 Þkx,y is the time-dependent stiffness matrix of kth lamina, hk is the thickness
of the kth lamina, zk is the distance of the kth lamina from the midplane, ½"kx,y is the strain
matrix of the kth lamina, t is the present time, and t0 is the past time.
The reduced parameters  and 0 are defined as
Zt
1
¼ dt
0 aðTÞ
Z t0 ð2Þ
0 1 0
 ¼ dt
0 aðTÞ

Here, a is shift function of the master curve.

If the midplane strain "0 and the curvature  exist, the corresponding forces and
moments are obtained as
Zt

@ "0 x,y ðt0 Þ
0
½NðtÞx,y ¼ ½Að   Þx,y dt0
0 @t0
Zt
@½x,y ðt0 Þ 0
þ ½Bð  0 Þx,y dt ð3Þ
0 @t0

Z

t @ "0 x,y ðt0 Þ
½MðtÞx,y ¼ ½Bð  0 Þx,y dt0
0 @t0
Zt
@½x,y ðt0 Þ
þ ½Dð  0 Þx,y dt
0 @t0
where

X
n
½Að  0 Þx,y ¼ hk ½Qð  0 Þkx,y
k¼1

Xn
1 2  k
½Bð  0 Þx,y ¼ hk  h2k1 ½Qð  0 Þx,y
k¼1
2

X
n
1  k
½Dð  0 Þx,y ¼ h3k  h3k1 ½Qð  0 Þx,y
k¼1
3

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

Process-induced Residual Stress Analysis by RTM 965

As was discussed in the previous section, the entire strain developments before cooldown
were represented by the constraint-induced strains. Therefore, the constraint-induced
strains were used as initial strains, and the strain developments during cooldown were
considered to calculate the residual stress. Then Equation (1) can be rewritten as

Z
X
n t
k @½"kx,y
½NðtÞx,y ¼ hk ½Qð  0 Þx,y dt0
k¼1 0þ @t0
Xn
þ hk ½Qð  0Þkx,y ½"jt ¼ 0 kx,y ð4Þ
k¼1

Z
X
n t
k @½"kx,y
½MðtÞx,y ¼ hk z k ½Qð  0 Þx,y dt0
k¼1 0þ @t0
Xn
þ hk zk ½Qð  0Þkx,y ½"jt¼0 kx,y
k¼1

Here, ½"jt ¼ 0 kx,y are the initial strain matrices at t ¼ 0. Similarly, Equation (3) can be
written as

Z
 Z
t @ "0 ðt0 Þ x,y t @½ðt0 Þx,y
0 0
½NðtÞx,y ¼ ½Að   Þx,y dt þ ½Bð  0 Þx,y
dt0
0þ @t0 0þ @t0
Xn
  Xn
1 2 
þ hk ½Qð  0Þkx,y "0 t ¼ 0 x,y þ hk  h2k1 ½Qð  0Þkx,y ½jt ¼ 0 x,y ð5Þ
k¼1 k¼1
2

Z
 Z
t @ "0 ðt0 Þ x,y t @½ðt0 Þx,y
0
½MðtÞx,y ¼ ½Bð   Þx,y dt0 þ ½Dð  0 Þx,y dt0
0þ @t0 0þ @t0
X n
1 2 
  Xn
1 3 
þ hk  h2k1 ½Qð  0Þkx,y "0 t ¼ 0 x,y þ hk  h3k1 ½Qð  0Þkx,y ½jt ¼ 0 x,y
k¼1
2 k¼1
3


Here, "0 t ¼ 0 and jt ¼ 0 are the initial midplane strains and the initial curvatures,
respectively.

MATERIAL PROPERTIES

To calculate the residual stress, the stress relaxation master curve for the fully cured
composite material was obtained. A series of stress relaxation tests for the unidirectional
laminate were performed by the DMA to the transverse direction to the fiber. The starting
temperature was 34
C, and from 45
C, the temperature was increased by 15
C. The test
was performed for 25 min at each temperature. Figure 5 represents the result of the stress
relaxation of transverse modulus E2 versus time at each temperature. The material showed
strong viscoelastic behavior at the different temperatures. And the modulus was rapidly
relaxed when the temperature was 135oC. To obtain the master curve the modulus curves

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

966 Y. K. KIM
7
34°C
6 45°C
60°C
75°C
5 90°C
105°C
120°C
4
E (GPa)

135°C
150°C
2

0
0.01 0.1 1 10 100
Time (min)
Figure 5. Stress relaxation data at each temperature.

Figure 6. Master curve and its modeling.

were shifted along the time frame. The reference temperature was chosen to be 34oC. The
master curve and the corresponding shift curve are shown in Figures 6 and 7, respectively.
To implement the modulus in the calculation, the shift function was modeled by a
polynomial function, and the master curve was modeled by a series of exponential
functions.

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

Process-induced Residual Stress Analysis by RTM 967

Figure 7. Shift function and its modeling.

Table 1. Stiffness of unidirectional composite laminate.

Q11 (GPa) Q12 (GPa) ! Q!22 (MPa) ! (min)
1 202.3 0.23
2 249.5 3.27
3 339.7 24.9
4 446.9 254.1
102.2 4.1 5 319.3 15.6E2
6 823.7 91.5E2
7 2022.6 10.2E4
8 2490.4 8.5E5
9 1372.9 7.2E6
1 984.8

For the stiffness, it was assumed that the material properties in the fiber direction were
essentially time-independent. Therefore, Q11 and Q12 were constants. The transverse
stiffness was modeled by the exponential function series based on the stress relaxation
curve such as
X n  
! 
Q22 ¼ Q1
22 þ Q 22 exp ð6Þ
!¼1
!

!
where Q1 22 is the fully relaxed stiffness, Q22 are weight parameters for discrete exponential
function, and  is the discrete relaxation time.
Nine terms were chosen for the discrete terms and the parameters for the stiffness are
listed in Table 1. Since the analysis was for the [0
/90
] asymmetric laminate, shear
stiffness Q33 is not shown.

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

968 Y. K. KIM

DETERMINATION OF LINEAR VISCOELASTIC REGION

Generally, the linear viscoelastic region is determined by obtaining an isochronal from a

series of creep (or relaxation) tests under different magnitudes of stress (or strain) [18]. In
this research, a dynamic mechanical analysis was conducted. In this approach, cyclic levels
of stress and strain are increased at a constant frequency. The point where the complex
modulus is deviated by more than 10% from a plateau value is regarded as a departure
from the linear viscoelastic region [19]. Unidirectional laminate was fabricated and cured
by the RTM process. After the cure, the sample was cut to the size of 1.6  12  38 mm3.
Since the material behavior to the fiber direction is elastic, only the storage modulus E20
to the direction transverse to the fiber direction was measured by the dynamic mechanical
analyzer under I Hz oscillation. The temperature for the measurements was chosen to be
30, 60, 90, 120 and 150
C, respectively. Figures 8 and 9 illustrate the storage modulus

5
Storage Modulus (GPa)

3 30°C
60°C
90°C
2 120°C
150°C

0
0.01 0.1 1 10 100
Stress (MPa)

Figure 8. Storage modulus (E02 ) variations by strain changes at different temperatures.

5
Stotage Modulus (GPa)

4
30°C
60°C
3 90°C
120°C
150°C
2

0
1 10 100 1000 10
Strain (µε)

Figure 9. Storage modulus (E02 ) variations by strain changes at different temperatures.

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

Process-induced Residual Stress Analysis by RTM 969

changes as the oscillatory stress and strain levels change. It was found that the material
behavior was within the linear viscoelastic range up to 5000
" of the strain and 35 MPa of
the stress. The rapid drop of the modulus at 150
C was attributed to the test temperature
higher than Tg. The linear range was found to be identical regardless of the test
temperatures below Tg.

RESULTS

To solve the curvatures k in Equation (5), an incremental numerical procedure proposed

by Wang et al. [20] was employed. Zero curvatures were considered during the calculation
for the period of cooldown when the laminate was in the mold. When the laminate was
removed from the mold, zero curvatures did not hold any more. The situation was
simulated by replacing the Young’s modulus of the aluminum mold by a small value
(1.0E-7, in this study).
We found two different Tg’s of the material depending on the measurement method.
In the stress relaxation test in the previous section, 135
C was selected because of the
strong relaxation behavior at the temperature. Another Tg was found by a dynamic
mechanical analysis. Figure 10 shows the result of the DMA results, and 145
C was
chosen at the peak of the loss modulus. For the case of Cure Cycle 1, both 145
C and
135
C were chosen as the initial temperature for the calculation. This was based on the
assumption that the stress development is elastic in the range above Tg due to the full
relaxation of the modulus. Therefore, any stress induced by increasing temperature during
cure process above Tg is compensated by decreasing the temperature in this range. For the
case of Cure Cycle 2, 135
C was the initial temperature according to the cure temperature
profile. The stress calculation for Cure Cycle 3 was not considered, because the material
could not be fully cured [16].

Figure 10. Storage and loss modulus at 1 Hz vs. temperature for fully cured AS4/CYBA unidirectional
composite.

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

970 Y. K. KIM

4.5

Curvature (m )
−1

3.5

3
o
: Viscoelastic calculations (135 C Tg)
o
: Viscoelastic calculations (145 C Tg)
2.5
: Elastic calculations

2
Cure cycle 1 Cure cycle 2
Cure Cycle Type

Figure 11. Results of the calculations and the data.

In the previous report [16], asymmetric cross-ply laminate specimens [04/904] of

dimensions 5  25.4  0.1 cm3 were manufactured by the RTM process according to the
Cure Cycles 1–3. After the cures were completed, the samples were taken out from the
mold, and the curvatures were measured. Figure 11 shows the comparisons of the sample
data and the calculation results. Again, Cure Cycle 3 was not shown here. When the
initial temperature was 145
C for the case of Cure Cycle 1, the final curvature was
3.35 m1. For the case of Cure Cycle 2, the result was 3.26 m1. The calculation results
are also shown in Figure 11 including the elastic solutions. The elastic solutions were based
on the elastic modulus measured separately in the previous report [16]. The initial
temperatures of the elastic calculations for Cure Cycle 1 and Cure Cycle 2 were 160
C and
135
C, respectively. For the viscoelastic calculations, excellent agreement was found for
the Cure Cycle 2, and the results were fair for the Cure Cycle 1.
As seen in Figure 11, the curvatures of Cure Cycle 1 were slightly lower than those of
Cure Cycle 2. Although we assumed the strain-free state at the peak cure temperature in
this analysis, there might be a slight recovery of the tensile strain, especially when the cure
temperature is high (Cure Cycle 1, in this study). In the light of the nature of the
unidirectional laminate, the slight recovery of the fiber to the 0
-direction in Cure Cycle 1
could lead to decreasing the curvature. Nevertheless, the data is not enough to draw a
conclusive statement about this observation.
The maximum stress and strain occurred to the transverse direction to the fiber during
the calculations were 2170
" and 20 MPa when the maximum temperature was 145
C for
Cure Cycle 1, which ensured that the material behaved in the linear viscoelastic manner in
these analyses.

CONCLUSIONS

In this paper, the residual stress analysis of composite laminates manufactured by the
resin transfer molding technique was presented. The data obtained in the previous work

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

Process-induced Residual Stress Analysis by RTM 971

were analyzed to investigate the effect of the constraint-induced strain on the residual
stress development during the cure process. It was found that, due to the strong constraint
effect of the mold structure, the thermal expansion and chemical shrinkage could not build
effective strains during the cure process, and eventually vanished as the constraint-induced
strains. Since the constraint-induced strains were permanently fixed (irreversible), the
laminate was assumed to be effectively strain-free at the peak cure temperature. Therefore,
the only strains developed during the entire process were at the cooldown stage. Based on
the analysis, the viscoelastic numerical formulation was set-up, and the residual stress was
calculated for the [0
/90
] asymmetric laminate. Excellent agreements were found between
the calculation results and the experimental data, especially when the peak cure tempera-
ture was 135
C. The analyses showed that the residual stress could be calculated by
considering only cooldown stage with viscoelastic material properties. As the constraint
effect is strongly dependent on the various parameters such as the mold surface, fiber
volume ratio, fabric structure of preform, and the structural geometry, further inves-
tigations are necessary to define the detail of the mechanism.

REFERENCES

1. Kim, Y.K. and White, S.R. (1998). Cure-Dependent Viscoelastic Residual Stress Analysis of
Filament Wound Composite Cylinder, Mechanics of Composite Materials and Structures, 5(4):
327–354.
2. Bogetti, T.A. and Gillespie Jr, J.W. (1992). Process-induced Stresses and Deformation in Thick-
Section Thermoset Composite Laminates, Journal of Composite Materials, 26(5): 626–660.
3. Kim, Y.K. and White, S.R. (1996). Stress Relaxation Behavior of 3501-6 Epoxy Resin During
Cure, Polymer Engineering and Science, 36(23): 2852–2862.
4. White, S.R. and Kim, Y.K. (1998). Process-Induced Residual Stress Analysis of AS4/3501-6
Composite Material, Mechanics of Composite Materials and Structures, 5(2): 153–186.
5. Adolf, D. and Martin, J.E. (1996). Calculation of Stresses in Crosslinking Polymer, Journal of
Composite Materials, 30(1): 13–34.
6. Eom, Y., Boogh, L., Michaud, V., Sunderland, P. and Manson, J. (2000). Time-Cure-
Temperature Superposition for the Prediction of Instantaneous Viscoelastic Properties During
Cure, Polymer Engineering and Science, 40(6): 1281–1292.
7. Adolf, D., Martin, J.E., Chambers, R.S., Burchett, S.N. and Guess, T.R. (1998). Stresses during
thermoset cure, Journal of Materials Research, 13(3): 530–550.
8. Monk, D.J., He, Y. and Soane, D.S. (1991). Stresses in Polyimide Films During Cure and
Thermal Cycling, In: Manufacturing Processes and Materials Challenges in Microelectronic
Packaging, AMD-Vol. 131/EEP-Vol. 1, pp. 87–93, ASME.
9. Cho, M., Kim, M.H., Choi, H.S., Chung, C.H. Ahn, K.J. and Eom, Y.S. (1998). A Study on the
Room Temperature Curvature Shapes of Unsymmetric Laminates Including Slippage Effects,
Journal of Composite Materials, 32: 460–482.
10. Fernlund, G., Rahman, N., Courdji, R., Bresslauer, M., Poursartip, A., Willden, K. and
Nelson, K. (2002). Experimental and Numerical Study of the Effect of Cure Cycle, Tool Surface,
Geometry, and lay-up on the Dimensional Fidelity of Autoclave-processed Composite Parts,
Composites, Part A – Applied Science and Manufacturing, 33(3): 341–351.
11. Wiersma, H.W., Peeters, J.B. and Akkerman, R. (1998). Prediction of Spring Forward
in Continuous-fiber/polymer L-shaped Parts, Composites, Part A – Applied Science and
Manufacturing, 29A: 1333–1342.
12. Zhu, Q., Geubelle, P.H., Li, M. and Tucker, C.L. (2001). Dimensional Accuracy of Thermoset
Composites: Simulation of Process-induced Residual Stresses, Journal of Composite Materials,
35(24): 2171–2205.

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015

972 Y. K. KIM

13. Tarsha-Kurdi, K.E. and Oliver, P. (2002). Thermoviscoelastic Analysis of Residual Stresses and
the Influence of Autoclave Pressure on these Stresses in Carbon/epoxy Laminates, Composites
Science and Technology, 62(4): 559–565.
14. Kim, Y.K. and White, S.R. (1998). Cure-Dependent Viscoelastic Residual Stress Analysis of
Filament Wound Composite Cylinder, Mechanics of Composite Materials and Structures, 5(4):
327–354.
15. White, S.R. and Zhang, Z. (1993). The Effect of Mandrel Material on the Processing-Induced
Residual Stresses in Thick Filament Wound Composite Cylinder, J. of Reinforced Plastics and
Composites, 12: 698–711.
16. Kim, Y.K. and Daniel, I.M. (2002). Cure Cycle Effects on Composite Structures by Resin
Transfer Molding, Journal of Composite Materials, 36(14): 1725–1743.
17. Daniel, I.M. and Ishai, O. (1994). Engineering Mechanics of Composite Materials, Oxford
University Press, New York.
18. McCrum, N.G., Buckley, C.P. and Bucknell, C.B. (1988). Principles of Polymer Engineering,
Oxford University Press, New York.
19. TA Instruments, Thermal Analysis & Rheology Application Library RS-23, (1999). TA
Instruments, Inc.
20. Wang, T.M., Daniel, I.M. and Gotro, J.T. (1992). Thermoviscoelastic Analysis of Residual
Stresses and Warpage in Composite Laminates, Journal of Composite Materials, 26(6): 883–899.

Downloaded from jcm.sagepub.com at HOWARD UNIV UNDERGRAD LIBRARY on March 4, 2015