Professional Documents
Culture Documents
www.elsevier.com/locate/tws
a,*
Department of Civil and Environmental Engineering, The University of Western Ontario, London,
Ontario, Canada N6A 5B9
b
Structural Engineer, Mccavour Eng. Ltd, Mississauga, Ontario, Canada
Received 16 June 1999
Abstract
Due to their high corrosion and chemical resistance, fiber reinforced plastic (FRP) materials
are increasingly being used in the construction of industrial chimneys. The design of a chimney
is governed by wind loads as well as thermal loads resulting from the differences among the
ambient, the operating and the curing temperatures. This study involves an investigation for
the thermal stresses induced in angle-ply laminated FRP chimneys, using an in-house
developed laminated shell element model. The finite element model is verified by performing
thermal analysis of a number of plate and shell problems and comparing the results to those
available in the literature. An extensive parametric study is then conducted using the shell
element model to identify the parameters which significantly affect thermal stresses induced
in FRP chimneys.
The study indicates that the thermal stresses are only affected by the inclination of the
lamina plies, the percentage of fibers content and the through thickness temperature distribution. Analyses also show that localized cracks in the direction perpendicular to the fibers
are expected to occur due to the thermal loads. Finally, thermal stress values that can be used
in the design of FRP chimneys, when cracking is considered, are presented as function of the
through thickness temperature distributions. 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Laminated; Composite; Shell; Finite element; Thermal; FRP chimneys; Design
58
1. Introduction
Due to their high corrosive and chemical resistances, fiber reinforced plastic
materials (FRP) are becoming widely used in the construction of industrial chimneys.
In 1984, Plecnik et al. [1] reported a 52 m height free-standing fiber glass stack
which was manufactured for a sugar processing plant in Moses Lake, Washington,
USA. During the past decade, a large number of FRP chimneys have been built in
North America and in various places around the world. FRP chimneys are commonly
applied in the pulp and paper and the chemical industries where highly corrosive
gases are produced. FRP flue gases have been also built in power generation stations.
FRP stacks are usually constructed from a large number of angle-ply layers. The
constituents of the layers are typically Vinyl Ester resin and E-glass fibers. A 70%
fiber content (based on weight) is typically used in the construction of FRP stacks
(as well as pipes).
No national code currently exists for the design of FRP chimneys. As stated by
Pritchard [2], an attempt is currently undertaken by the International Committee on
Industrial Chimneys, CICIND, to develop such a code. This study is a part of
research program to study the structural performance of FRP chimneys. This investigation focuses on the effect of thermal loads.
This study starts by briefly describing a laminated shell element model which is
employed in performing thermal stress analysis of FRP chimneys. The finite element
model is extended to include thermal loads and is verified using results of thermal
stress analysis for a number of plate and shell problems that are available in the
literature. The finite element model is then used to perform an extensive parametric
study in order to identify the main parameters affecting thermal stresses induced in
FRP chimneys. Practical considerations which should be taken into account in the
thermal stress analysis of FRP chimneys, are discussed. Finally, charts predicting
thermal stresses induced in FRP chimneys as functions of the parameters defining
the through thickness temperature distribution are presented.
59
Fig. 1.
60
sxx
syy
(1)
txz
tyz
where o and so are the initial local strain and stress vectors, respectively. The
matrix [D] is related to the constitutive matrix for orthotropic material [D] using
the following transformation:
[D][Te]T[D][Te]
(2)
The constitutive matrix [D], given by Jones [9], is defined in the material axes
system 123. The matrix [T] represents the transformation matrix relating the local
axes system (x, y and z) to the material axes system (1, 2 and 3). An expression
for [T] is given by Cook et al. [10].
A temperature change T induces initial thermal strains {e12
oL} (in the material
axes 12) which are given by:
e1
a1LT
e2
a2LT
{e } g12 0
12
oL
g13
g23
(3)
where a1L and a2L are the thermal expansion coefficients of the Lth layer in the
direction of material axes 1 and 2, respectively. The transformation matrix [Te] is
applied to {e12
oL} to obtain the local initial strains {eoL} expressed relative to the
local axes x, y and z i.e.
{eoL}[Te]1{e12
oL}
61
(4)
p(1/2) {s}T{e}dv
(5)
and substituting Eq. (1) into Eq. (5) (putting {soL}={0}), the load vector {f} due
to temperature change is obtained using the principle of minimum potential energy
and is then incorporated into the shell element model.
3. Verification of the model
In order to verify the accuracy of the developed finite element model, thermal
analysis of a number of plate and shell problems is performed using the consistent
laminated shell element.
3.1. Isotropic plate subjected to linearly varying temperature change
A simply supported isotropic plate is analyzed under a linearly varying through
thickness temperature distribution. The temperature variation at any point within the
plate is expressed as: T(x,y,z)=TL z/H, where TL is the value of the temperature at
the top and bottom fibers of the plate; z is the coordinate normal to the plate and
measured from the mid-surface; and H is the thickness of the plate. The plate is
simply supported at its four edges.
Results of the analysis are presented using the dimensionless parameter wL, which
is defined as:
wLHw/a1TLA2
where: wL is the central deflection of the plate, a1 is the coefficient of thermal expansion, and A is the length of the plate along the x-axis.
The analyses are conducted for different A/B and H/A ratios; where B is the length
of the plate along the y-axis. Values of the dimensionless parameter wL resulting
from these analyses together with those predicted by Timoshenko et al. [11] are
presented in Table 1 showing an excellent agreement. It should be noted that the
displacements resulting from the thermal analysis of isotropic plate are independent
of the modulus of elasticity of the plate. In the analyses, the Poissons ratio of the
plate is assumed to be 0.3.
3.2. Anti-symmetric angle ply laminate plate subjected to linearly varying
temperature change
An angle ply () square plate is considered for thermal stress analysis using the
consistent shell element. The plate has the same boundary conditions and is subjected
62
Table 1
Results of the analysis of an isotropic plate subjected to linearly varying temperature
wL
H/A
A/B=1.0
0.100
0.075
0.050
0.010
[11]
0.09578
0.09578
0.09578
0.09578
A/B=1.5
CLT
0.09598
0.09601
0.09623
0.09783
[11]
0.05824
0.05824
0.05824
0.05824
A/B=2.0
CLT
0.05832
0.05835
0.05844
0.05928
[11]
0.03702
0.03702
0.03702
0.03702
CLT
0.03706
0.03708
0.03712
0.03749
2 layers
q
0
15
30
45
[12]
0.16711
0.16792
0.16221
0.16071
4 layers
CLS
0.17141
0.16659
0.15525
0.14944
[12]
0.16711
0.15921
0.14419
0.13817
CLS
0.17141
0.16213
0.14573
0.13857
sxxsyy
Ea(T1T2)
2(1n)
63
(6)
T1 and T2 are the temperatures at the inside and outside faces of the shell, respectively. In the above equation, the stresses at the outer face are tensile if T1T2. An
isotropic free standing cylinder having a modulus of elasticity E=36.85 GPa, a Poissons ratio n=0.3, a coefficient of thermal expansion a=7.7106 m/m/C and a
diameter D=3.0 m has been modeled using the consistent laminated shell element.
The cylinder is subjected to the through thickness temperature distribution shown in
Fig. 2. The above parameters are substituted into Eq. (6) to obtain the stresses at a
cross section away from the boundaries. According to Eq. (6), such a section is
subjected to pure circumferential and longitudinal bending stresses (i.e. stresses at
the mid-surface equal zero) which are equal to 2.027 MPa and 2.027 MPa at the
inner and outer faces, respectively. Results of the finite element analysis together
with those predicted by Timoshenko et. al. [11] are presented in Fig. 2 by plotting
the stresses sbx, sby and sax along the length of the cylinder (in this figure y=0
corresponds to the free end), where: sbx is the outer circumferential bending stress,
sby is the outer longitudinal bending stress and sax is the mid-plane circumferential
stress. Fig. 2 shows that the longitudinal bending stress sby vanishes at the free end
and that both sbx, sby approach the exact value (2.027 MPa) away from the boundaries. It is also clear from the figure that a full agreement between the finite element
and the closed form solution is achieved.
Fig. 2. Variation of outer and mid-surface longitudinal and circumferential stresses at free end of cylinder
due to linearly varying temperature change.
64
Source
WL
BC1
[15]
CLS
[15]
CLS
1.4106
1.4470
1.4326
1.4883
BC2
65
the consistent laminated shell element provides a very good agreement with the
analyses conducted by Chandrashekhara et al. [13].
66
Fig. 4. Circumferential and longitudinal stresses of five layers angle-ply (55) FRP chimney versus
the laminate thickness at a section away from the boundaries of the chimney.
67
Fig. 5. Circumferential and longitudinal stresses of five layers angle-ply (55) FRP chimney versus
the laminate thickness at the base of the chimney.
analysis in order to asses the effect of the diameter of the chimney. The temperature
variation follows the linear distribution previously described when studying the effect
of the thickness. The parametric study is performed by varying the diameter of the
chimney in the range between 1.5 m and 6 m. The variation of the thermal stresses
induced at the base of the chimney versus the diameter is presented in Fig. 6. It
could be concluded from the figure that the change in the diameter has no significant
effect on the thermal stresses. In Fig. 7 both the hoop thermal stresses (sx) and the
axial (meridional) thermal stresses (sy) are plotted along the height of one of the
analyzed chimneys. As might be expected, both the hoop and the axial thermal
stresses have rapid fluctuations near the boundaries (for both the fixed and the free
boundaries). In general, the thermal stress distributions show high stress values which
occur very close to the boundaries and are localized in a narrow region.
Fig. 6.
The effect of the diameter on the thermal stresses induced at the base of a FRP chimney.
68
Fig. 7. The hoop and the axial stresses at the inside face along the height of a FRP chimney subjected
to linearly varying temperature.
69
() and the 5 layer laminate is a symmetric angle-ply laminate. For each laminate
configuration, the angle of orientation has been varied between 0 and 90. In Fig.
8, the variations of the maximum along fibers stresses s1 and transverse fibers
stresses s2 (occurring near the base) for the outside face of the chimney are plotted
versus the angle of orientation for different laminate configurations. Fig. 9 shows
similar graphs plotted for the inside face. Both figures indicate that the number of
layers has no significant effect on both the along fibers and the transverse fibers
stresses. It was found that for a certain thickness, the increase in the number of
layers would lead to a decrease in the transverse (interlaminar) shear stresses. As
such, a conservative approach can be adopted by using 10 layers (usually the number
of layers are larger than that) to develop charts for thermal stresses induced in FRP
chimneys. At the inside face of the shell, the increase of the fiber orientation leads
to an increase in the along fibers stresses reaching maximum values at =90 and
also leads to a decrease of the across fibers stresses which reach minimum values
at =90. For the outside face, the increase of the angle ply leads to a slight
decrease in the stresses which is then followed by a significant increase of the stresses
with the angle (at =37.5 for the case of s1).
4.5. Summary of the results of the parametric study
From the above conducted parametric study, it can be concluded that the height,
the diameter, the thickness and the number of layers used to achieve the thickness
have almost no effect on the maximum thermal stresses induced in FRP chimneys.
Fig. 8. The maximum longitudinal and transverse stresses at the outside face of the laminate vs the
angle of orientation.
70
Fig. 9. The maximum longitudinal and transverse stresses at the outside face of the laminate vs the
angle of orientation.
Such stresses are usually very localized in a narrow region near the base of the
chimney. The main parameters affecting the values of the stresses are the temperature
profile, the angle of the orientation of the fibers, the coefficient of thermal expansion
and the modulus of elasticity along the fibers direction. For practical FRP chimneys
consisting of glass fibers and vinyl ester resin, the last two parameters depend mainly
on the percentage of the fibers content.
The practical range for the angle of inclination is between 35 and 55. Examining the stress values shown in Figs. 5 and 6 (these figures represent results for chimney having =55), it can be concluded that the maximum value for the stresses s1
(along the fibers direction) and s2 (perpendicular to the fibers direction) are approximately 100 MPa and 80 MPa, respectively. Typical ultimate bending strengths
(which are double the ultimate axial strengths) along the fibers s1u and perpendicular
to the fibers s2u have approximately the following values s1u=1100 MPa and
s2u=33.5 MPa (for 70% E-glass content based on weight). Comparison between the
induced stresses and the ultimate bending strength indicates that although large factor
of safety is achieved along the fibers direction, the cross fibers direction is unsafe.
As such, one would expect that cracks localized at the bottom part of the chimneys,
parallel to the fibers direction, would occur (independent of the value of the
thickness) due to thermal stresses.
4.6. Practical considerations for attempting design procedure of FRP chimneys
From the above discussions, it is clear that the temperature distribution assumed
in the analysis results in across fibers stresses which are approximately 2.5 times
71
the ultimate stresses in that direction. As such, it is very hard to avoid cracking in
the across fibers direction unless insulation layer has been provided to reduce the
temperature gradient. Moreover, if the design is governed by preventing such cracks,
the fiber reinforcement would be redundant. Knowing that cracks will occur, it has
been decided to analyze the FRP chimneys under thermal loads by assuming that
the stiffness in the direction perpendicular to the fibers almost vanishes (i.e. E2 is
very small). This assumption is made for all layers along the height of the chimney.
The authors believes that this assumption is conservative because, in practice, cracks
will not occur in all layers and not necessary along the whole height of the chimney.
The safety of an FRP chimney analyzed under such an assumption can be checked
by assuring that the stresses along the fibers do not exceed the ultimate strength
divided by a suitable factor of safety and also that the interlaminar shear stresses
are also well below the ultimate shear strength. By assuring that the interlaminar
shear stresses are safe and using an angle-ply configuration, it is expected that the
cracks in one layer will be very much controlled by the stiffness of the two adjacent
layers along the fibers direction. Fig. 10 shows the variation of the along fibers
stresses s1 with the angle ply for a typical FRP chimney using the temperature
distribution described above (after degrading the across fibers stiffness). It should
be noted that the analysis has been performed for a practical range of varying
between 35 and 60. It should be mentioned that the circumferential forces (close
to the boundaries) have changed from tensile to compressive forces after degrading
the lateral modulus of the plies. Consequently, the compressive stresses in the inside
face of the shell have significantly increased compared to the increase of the tensile
stresses at the outside face of the shell. It can be concluded from Fig. 10 that the
Fig. 10. The maximum along fibers stresses at the inner and outer face of the laminate vs the angle of
orientation after degrading the across fibers stiffness of the layers.
72
maximum stresses s1 do not exceed a value of 180 MPa. This value leads to approximately a factor of safety of six when compared to the ultimate strength. In order to
check safety against shear failure, the in-plane shear stress t12 as well as the transverse shear stresses t13 and t23 resulted from the same analyses are plotted in Fig.
11 versus the angle of orientation . The typical values for the ultimate shear strength
in-plane and transverse are given by t12=70.6 MPa, t13=70.6 MPa and t23=18.85
MPa. Comparison between the induced shear stresses and the ultimate ones reveals
that factor of safety of approximately 3.5, 15 and 5.6 are achieved for the in-plane
and the transverse shear stresses, respectively.
4.7. Thermal stress values to be used in practical design of FRP chimneys
As mentioned before, the thermal stresses induced from temperature variation in
FRP chimney depend on the following factors:
1.
2.
3.
4.
Fibers content
Angle of inclination of the fibers
Temperature profile
Type of fibers and resin
Restraining the design to FRP chimneys constructed from vinyl ester resin reinforced
by 70% (based on weight) E-glass fibers, for a certain angle of inclination of the
Fig. 11. The in-plane shear stress t12, transverse shear stress t13, t23, of 10 layer laminate at the bottom
of the chimney after degrading the across fibers stiffness of the layers.
73
fibers, the thermal stresses depend only on the temperature profile. This profile is
governed by two parameters which are:
1. The variation of mid-surface temperature with respect to the curing temperature
Tm.
2. The difference between the temperature at the inside and the outside faces
(T); T=TinsidToutsid.
Using the approach described in the previous section, analyses have been conducted
to determine the maximum stresses s1 as function of Tm and T for two angle
configurations, =35 and 55, respectively. Figs. 12 and 13 show the variation
of the maximum along fibers stresses s1 versus the temperature variation T for
different values of Tm and for =35 and 55, respectively. These graphs can be
used to estimate the stresses induced in a FRP chimney, having the above-described
properties under various temperature variations. Comparison between the two graphs
indicates that in general higher thermal stresses are introduced when the fibers
become more vertically inclined (i.e. =55 leads to higher thermal stresses than
Fig. 12. The along fibers thermal stress of 35 angle-ply FRP chimney for different temperature fields
(degraded across fibers modulus E2=E1/1000).
74
Fig. 13. The along fibers thermal stress of 55 angle-ply FRP chimney for different temperature fields
(degraded across fibers modulus E2=E1/1000).
=35). The shear stresses associated with various temperature profiles are shown
in Tables 4 and 5 for =35 and 55, respectively. It should be noted that the
shear stresses vary linearly with the parameter Tm and independently of T. The
designer of FRP chimney has to assure that a sufficient factor of safety is achieved
against shear failure.
5. Conclusions
In this study, the formulation of the consistent laminated shell element is extended
to include thermal stress analysis. A number of plate and shell structures are modeled
for thermal stress analysis and the results are compared with those available in the
literature. In all examples, the element gives adequate predictions for thermal
stresses. The developed finite element formulation is then used to study the effect
of various parameters which might influence the thermal stresses induced in angleply laminated fiber reinforced plastic chimneys. Results of the parametric study indi-
75
Table 4
The in-plane and transverse shear stresses associated with the along fibers stresses in Fig. 12 for an angleply laminate 35
Middle surface temperature
Tm (C)
t12 (MPa)
t13 (MPa)
t23 (MPa)
0
20
40
60
80
100
0.00
5.22
10.44
15.66
21.88
26.10
0.00
0.14
0.27
0.41
0.54
0.68
0.00
0.25
0.50
0.75
1.00
1.25
Table 5
The in-plane and transverse shear stresses associated with the along fibers stresses in Fig. 13 for an angleply laminate 55
Middle surface temperature
Tm (C)
t12 (MPa)
t13 (MPa)
t23 (MPa)
0
20
40
60
80
100
0.00
5.25
10.5
15.75
21.00
26.25
0.00
0.85
1.70
2.55
3.40
4.25
0.00
0.77
1.52
2.28
3.04
3.80
cate that the thickness, the diameter, the height and the number of laminae have no
significant effect on the induced thermal stresses. Analyses indicate that the thermal
stresses depend mainly on the through thickness temperature distribution (relative to
the curing temperature), the angle of orientation of the fibers, the coefficient of thermal expansion and the modulus of elasticity along the fibers direction. The last two
parameters depend on the fiber content in the matrix. The thermal stress analysis of
typical FRP chimneys shows high stress concentration near the boundaries with inplane across fibers stresses exceeding the typical ultimate strength in this direction.
As such, cracks are expected to occur in FRP chimneys as a result of through thickness temperature variations. However, it is believed that these cracks will be controlled if the interlaminar shear stresses are less than the ultimate shear strength
divided by an appropriate factor of safety.
The analysis then proceeds by assuming a negligible value for the modulus of
elasticity in the direction perpendicular to the fibers. Results of this last set of analysis
indicate that for the practical range of the early mentioned influential parameters,
the along fiber direction stresses as well as the shear stresses of cracked chimneys
are within acceptable values. Finally charts predicting the along fiber thermal stresses
76
induced in typical cracked FRP chimneys (but limited to 70% fiber content and
angles of inclination = 35 and 55) as a function of the through thickness temperature distribution are presented. These stress values can be considered when the
design of a FRP chimney is attempted.
References
[1] Plecnik JM, Whitman WE, Baker TE, Pham MP. Design concepts for the tallest free-standing fiberglass stack. Polymer Composites 1984;5(3):1869.
[2] Pritchard BN. Industrial chimneys: a review of the current state of art. Proc Instn Civ Engrs Structs
Bldgs 1996;116:6981.
[3] Padovan J. Thermoelasticity of cylindrical anisotropic generally laminated cylinders. J Appl Mech
1976;43:12430.
[4] Fettahlioglu OA, Wang PC. Asymptotic solutions for thermal stress and deformation in orthotropic
nonhomogeneous shells of revolution. J Thermal Stresses 1988;11:30524.
[5] Lin TD, Boyd DE. Thermal stresses in multilayer anisotropic shells. J Engng Mech Div, Proc ASCE
1971;97:82945.
[6] Ahmed BMI, Zienkiewicz OC. Analysis of thick and thin shell structures by curved finite elements.
Int J Num Methods Engng 1970;2:41951.
[7] Koziey BL, Mirza FA. Consistent thick shell element. Comp Struct 1997;65(4):53149.
[8] Koziey BL. Formulation and applications of consistent shell and beam elements. PhD thesis, McMaster University, Hamilton, Canada, 1993.
[9] Jones RM. Mechanics of composite materials. New York (NY): McGraw Hill Book Company, 1975.
[10] Cook RD, Malkus DS, Plesha ME. Concepts and applications of finite element analysis. New York
(NY): John Wiley and Sons Inc, 1989.
[11] Timoshenko S, Woinowsky-Krieger W. Theory of plates and shells. New York: McGraw-Hill, 1959.
[12] Wu H, Tauchert TR. Thermoelastic analysis of laminated plates. 2: anti symmetric cross-ply and
angle-ply laminates. J Therm Stresses 1980;3:36578.
[13] Chandrashekhara K, Bhimaraddi A. Thermal stress analysis of laminated doubly curved shells using
a shear flexible finite element. Comput Struct 1993;52(5):102330.
[14] Mallick PK. Composites engineering handbook. New York: Macel Dekker, Inc, 1997.
[15] Thangaratnam RK, Palaninathan and Ramachandran J. Thermal stress analysis of laminated plates
and shells. Comput. Struct 1988;30(6):14031411.