See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318648779

Hygro-thermo-mechanical analysis of fiber-steered composite conical panels

Article  in  Composite Structures · July 2017

DOI: 10.1016/j.compstruct.2017.07.055

CITATIONS READS
12 709

3 authors, including:

Hamidreza Yazdani Sarvestani

National Research Council Canada
64 PUBLICATIONS   452 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Composite helicopter landing gears View project

Nonlocal continuum theories View project

All content following this page was uploaded by Hamidreza Yazdani Sarvestani on 13 November 2018.

The user has requested enhancement of the downloaded file.

 Hygro-thermo-mechanical analysis of fiber-steered composite conical panels

H. Yazdani Sarvestani1, A.H. Akbarzadeh1*, M. Hojjati2

1
AM3L Laboratory, Department of Bioresource Engineering, McGill University, Island of
Montreal, QC, Canada H9X 3V9
2
Department of Mechanical and Industrial Engineering, Concordia University, Montreal, QC,
Canada H3G 1M8

Abstract
Automated fiber placement (AFP) is one of the recently developed robot-assisted composite
manufacturing techniques. Steering fibers along optimal curvilinear paths enables tailoring the
multifunctional properties of laminated composites to fully exploit all potentials of composite
materials. This type of advanced composites, called variable stiffness laminates, can optimize the
structural performance of composites. In this paper, we aim at developing a novel semi-
analytical methodology to conduct hygro-thermo-mechanical analysis on thin to relatively-thick
fiber-steered composite panels. The principle of minimum total potential energy, and Layer-wise
method are employed to explore the effects of fiber steering on the hygro-thermal and
mechanical buckling loads, natural frequency, bending deformation, and stress distributions of
conical and cylindrical panels as well as circular plates made by AFP technology without
considering manufacturing defects. The numerical results show that the buckling loads and
fundamental frequencies of composite conical panels could be improved up to 57% and 44%,
respectively, by using a constant curvature fiber-steered lay-up instead of a corresponding quasi-
isotropic composite.

Keywords: Fiber-steered laminated composite; Conical panel; Layer-wise method; Hygro-

thermo-mechanical analysis; Curvilinear fiber path.

*
Address correspondence to: hamid.akbarzadeh@mcgill.ca, Tel: +1 (514) 398-7680, and Fax: +1 (514) 398-7990.
Email: hamidreza.yazdani@mcgill.ca (H. Yazdani Sarvestani), hamid.akbarzadeh@mcgill.ca (A.H. Akbarzadeh),
mehdi.hojjati@concordia.ca (M. Hojjati)

1
1. Introduction
Over the past few decades, laminated composite structures have been extensively used in a
wide range of applications from aerospace and automotive to naval, as well as in construction
because of their high specific stiffness and strength properties and capability. Composite
laminates are traditionally manufactured by stacking layers of dissimilar fiber orientations, or
material constituents [1]. This conventional method of composite manufacturing is recently
being replaced by advanced manufacturing technologies, e.g. Automated Fiber Placement (AFP)
and Additive Manufacturing (3D printing). The AFP technology has been used for the
manufacturing of aircraft components such as the F/A-18E/F horizontal stabilizer skins [2], the
Bell/Boeing V-22 Osprey aft fuselage [2], the V-22 grip [3], the fuselage sections of the Boeing
787 Dreamliner, and Airbus A380 and A350 [4]. The advantages of AFP technology include, but
are not limited to, improved quality, reduced waste material, reduced labour costs, increased
accuracy and repeatability of the production process, and reduced manufacturing time [5].
Another benefit of AFP technology is its ability to steer fibres on the plane of a ply to fabricate
programmable curvilinear paths, a process that was initially developed in [6]. This capability
enables designers to tailor the structural behavior of composite parts not only through the
thickness to change their stiffness and strength properties (straight fiber composites design) but
also by spatially varying the point wise fiber orientations through the active steering of a band of
tows called course. The resulting fiber-steered composites, so-called variable stiffness (VS)
composites, have been reported to have potentially higher structural performance without weight
penalties compared to their constant stiffness (CS) counterparts in buckling loads [7-9], natural
frequencies [10-11], and flexural stiffness [12]. One early application of VS composites
considered by researchers was to improve stress concentrations surrounding holes by an
optimized steering of fibres around the geometrical discontinuity [13]. Recently, these advanced
composites have been demonstrated to have benefits over the CS composites in aircraft
structures, e.g. as fuselage-type [14] and wing-type structures [15].
In the literature, most of researchers have mainly used two main approaches, finite element
modeling using commercial software and equivalent single-layer (ESL) theories using numerical
or theoretical algorithm, to optimize the structural and thermal performance of fiber-steered
composite structures subjected to multiple load scenarios [16]. The structures studied and
designed to this end include different shapes, such as flat, cylindrical and conical panels. The

2
first approach, finite element methods using commercial software, has been employed in several
research studies for design and optimization of fiber-steered flat plates for the dynamic problem
[17, 18], improvement of buckling loads [19, 20], analysis of progressive damage and failure
[21], and thermal stress analysis [22, 23]. Cylindrical and conical panels are also widely used in
aerospace applications such as in the centre and aft fuselages. In addition, use of VS cylindrical
and conical panels can result in more efficient and thus lighter structures. A few examples of the
numerical and theoretical studies on VS cylindrical and conical composite structures can be
mentioned as: bending-induced buckling analysis [24-27], axial buckling analysis [28-31],
vibration response [32-34], thermal behavior [35, 36], postbuckling simulation [37, 38], and
multi-objective design [39, 40]. Among challenges of using the commercial FE analysis of VS
composites, we can refer to the computational cost and meh sensitivity of a detailed three-
dimensional modelling. The two-dimensional modelling through commercial FE software is also
limited to conventional plate and shell theories with controversial plane stress assumption. In the
second approach, ESL theories such as classical laminated (CLT), first-order shear deformation
(FSDT) and third-order shear deformation (TSDT) theories have also been employed for the
structural analysis of fiber-steered composite flat panels [41, 42], cylindrical panels [43], and
conical panels [44]. Nevertheless, CLT, FSDT and TSDT models can fail to predict accurately
the response of considerably thick composite laminates, requiring more refined theories with low
computational cost to reliably predict the cross-sectional warping and the three-dimensional (3D)
stress distributions through the thickness. In this research, Layer-wise theory (LWT) is of
interest. In LWT, a finite number of independent numerical layers [45, 46] simulates each
physical layer. Layer-wise theory, compared to ESL, can lead to accurate stress field in thick
composite without a need for solving complicated three-dimensional elasticity equations. In
addition, research studies have confirmed the efficiency of LWT, in terms of computational cost,
compared to the finite element simulations [47-50]. Although a few investigations have used
LWT for deflection responses [51] and free vibration analyses [52] of fiber-steered plates and
static analysis of doubly-curved fiber-steered panels [53], less attention has been paid to fully
implementing LWT for the other shapes or structural analyses of fiber-steered composites under
thermal or hygroscopic loads.
In this study, we develop an advanced mathematical framework based on LWT to analyze the
structural performance of fiber-steered composite conical panels subjected to a combined hygro-

3
thermo-mechanical load. This semi-analytical framework is able to accurately predict hygro-
thermo-mechanical structural responses, including the thermo-mechanical buckling loads, natural
frequency, thermo-mechanical deformation, and three-dimensional thermo-mechanical stress
distributions of thin to thick fiber-steered laminated composite conical panels. The solution is
obtained for fiber-steered conical panels with arbitrary boundary conditions where this solution
can be deduced for cylindrical shells and circular plates if the appropriate parameters are selected
for the conical panel geometry. It should be noted that manufacturing defects are not considered
in this study. The major contributions of this study can be summarized as:
(1) Formulating governing equations for hygro-thermo-mechanical analysis of thick variable
stiffness conical panels with arbitrary boundary conditions, hybrid of simply-supported,
clamped, and free boundary conditions.
(2) Obtaining design charts for variable stiffness conical panels illustrating hygro-thermal and
mechanical buckling loads, natural frequency, hygro-thermo-mechanical bending
deformations, and hygro-thermo-mechanical stress distributions.
(3) Extending the derived solution for the structural analysis of fiber-steered conical panels to
cylindrical shells and circular plates.
(4) Introducing a robust model to capture the free-edge effects for thick variable stiffness
conical panels.
The paper is organized as follows: Section 2 introduces geometric parameters used in the
equations describing fiber paths in the laminated composite conical panels. Sections 3, 4, and 5
employ LWT and hybrid Fourier-Galerkin methods to develop governing equations and obtain
solutions for fiber-steered conical panels. Section 6.1 presents the validation of the introduced
method. Sections 6.1 and 6.2 describe the effects of length-to-thickness ratio, and fiber paths on
hygro-thermo-mechanical responses with particular focus on the search for finding optimal fiber
paths. Finally, Section 6.3 presents effects of geometrical parameters on hygro-thermo-
mechanical responses of VS conical panels.

2. Fiber-steered Laminated Composite Conical Panels

A fiber-steered composite conical panel, shown in Fig. 1, is considered in the cylindrical
coordinate system (x, θ, z ) located on the mid-surface and mid-span of the VS laminate where
x, θ, and z are the axial, circumferential, and radial coordinates, respectively. The conical panel

4
 is subjected to an axial load F0, torque T0, bending moment M0, transverse load q, uniform
temperature change ΔT, and uniform moisture concentration change ΔM as shown in Fig. 1. The
dimensions of the composite conical panel are L (length), h (thickness), Rb (minor reference
radius), α (semi-vertex angle), and 2θ0 (span or revolving angle). In addition, and

. It is worth mentioning that the conical panel can be reduced to a cylindrical

shell for α = 0 or to a circular plate for α = π/2.

α z
θ

Rb x (N+1)th surface
N

k+1
k
k-1
L
Ro zk+2 2 kth surface
zk+1 h 1 h
θ0 zk k
θ0
1st surface
(a)
M0

F0
T0 q

ΔT, ΔM M0
F0
T0
(b)
Figure 1: Geometry of an N-layered fiber-steered composite conical panel: (a) coordinate system, and ply sequence,
(b) different loading types applied to the conical panel.

2.1. Fiber-steering Paths for VS Conical Panels

To model a fiber-steered composite, we need to define a reference path along which the AFP
machine places [3] the prepreg composites. The direction of subsequent fiber paths are
determined by shifting the reference path along a given direction, e.g. x- or θ-directions. The
reference fiber path can be defined by a continuous mathematical function. A complex

5
mathematical function can offer high flexibility for steering fiber path; however, it increases the
number of design variables and comes with additional computational costs. Moreover, highly
curved paths modelled with a complex function might not be necessarily manufacturable with
AFP, as a result of the possibility of violating the manufacturing limitations such as lay-up
speed, process temperature, and material thickness. Two types of manufacturing limitations are
shown in Fig. 2: (1) Sharp change in the angle (Δφ) between neighboring segments (Figs. 2 and
3) which leads to fail in satisfying G1 continuity condition between fiber paths in neighboring
two segments and (2) Minimum fiber steering radius of AFP (rmin) as shown in Fig. 2. These
limitations need to be considered to avoid local buckling of manufactured composites and the
fiber wrinkling defects. According to earlier studies [24, 30, and 43], several methods have been
used for varying the fiber path in a curvilinear pattern changing the reference path orientation
along the x- or θ-directions. For instance, for VS cylinders, the reference path along the x-
direction [43] and along the θ-direction [24, 30] has been considered, while the reference path
along the x-direction [28, 33, and 44] has been considered for fiber-steered conical panels. In this
study, we consider two different paths; the first is a fiber path with a linear angle variation and
the second one is a fiber path with a constant curvature [54].

Manufacturable

T0

Δϕ
Non-manufacturable
T1 (Sharp point)

T2
rmin
Non-manufacturable
(Minimum radius)
Figure 2: Manufacturable and non-manufacturable fiber paths on a flattened VS conical panel. AFP head moves in a
continuous curve path with a minimum fiber steering radius (rmin).

2.1.1. Linearly varying fiber path

6
 The fiber angle on a conical panel is defined to vary linearly along the axial direction from Tn-
1 at the small radius to Tn at the large radius, where n is segment number (see Fig. 3) [54]:
x
  x   Tn1  Tn  Tn1  1
L
where φ(x) is the fiber orientation along the fiber path as shown in Fig. 3, in which constant
angle refers to the constant stiffness composites.

2θ0
Constant angle
Constant curvature
Tn-1
Linear variation

Segment n

Tn
Figure 3: Fiber-steering paths on a flattened segment of a conical panel.

2.1.2. Constant curvature fiber path

A path with a constant curvature has also been presented in Fig. 3. The fiber path begins from
an angle T0 at Rb and ends at an angle Tn at Rn. As results, the fiber angle can be defined as [54]:
 R sin T   R0  x   Rb  
2 2

  x   arcsin  b 0
    2a 
 R0  x  sin   2 R0  x   
  
where the curvature of fiber path is:

1  2 Rn 2 Rb 
  sin Tn  sin T0   2b 
L  Rn  Rb Rn  Rb 
It is worth mentioning that the standard notation to define a particular VS laminate with the
aforementioned fiber path definitions is <T0|Tn> for a single layer, where T0 = Tn represents a

7
straight fiber or constant stiffness lamina. The shifting direction of the reference fiber path,
affecting both fiber angle and defect distribution, can play a major role on the optimum structural
performance achievable by fiber-steering [41]. An extension to the fiber-steering path definition
is the variation of angle in multiple stages. This means that the conical panel is divided into
multiple segments. To ensure the manufacturability of fiber-steered panels with different
segments, it is necessary to satisfy the continuity condition at the boundaries of neighbouring
segments; G0 condition refers to the continuity of values of angle fibers in neighbouring
segments while G1 condition refers to the continuity of the slope of angle fibers.
The role of manufacturing defects, which are typically gaps and overlaps, on the structural
analysis of fiber-steered laminated plates has also been investigated in a few papers [42, 55].
Gaps are resin-rich areas with lower stiffness, while overlaps can be modelled as thickness build-
ups. It has been shown that the panel made with a complete-gap strategy has a higher amplitude
of dynamic deflection and a lower response frequency than that of a defect-free panel. A
complete overlap strategy, on the other hand, leads to a lower amplitude of deflection and a
higher frequency [41, 42]. However, to be able to focus on the effects of fiber steering on the
hygro-thermo-mechanical responses, we have ignored the possible manufacturing defects in the
formulation.

3. Displacement Field
Several displacement-based methods have been proposed to study the behavior of laminated
composite structures. In general, these methods may be divided into two groups: (1) equivalent
single-layer (ESL) theories and (2) Layer-wise theories (LWTs) which are based on the variation
of the displacement field through the thickness. For some applications, ESL theories provide a
sufficiently accurate behavior; however, ESL theories are mostly insufficient to determine 3D
stress and strain fields at the ply level. This weakness is perceptible in moderately thick to thick
laminated composite structures. LWT allows each layer of the laminate to act as a real 3D layer
able to present accurate results, especially for local quantities such as interlaminar stresses. The
displacement components of a generic point in the laminate, based on LWT, are expressed as [1]:
u  x,  , z , t   U k  x ,  , t   k  z   3a 
v  x, , z, t   Vk  x,  , t   k  z   3b 
w  x, , z, t   Wk  x, , t   k  z   k  1, 2,..., N LWT
 1  3c 

8
where k is a dummy index. In Eqs. (3), u (x, θ, z, t), v (x, θ, z, t), and w (x, θ, z, t) represent the
displacement components in x, θ, and z directions, respectively. Furthermore, Uk (x, θ, t), Vk (x, θ,
t), and Wk (x, θ, t) represent the displacements of the points initially located on the kth
mathematical surface of VS composite conical panels in the x, θ, and z directions, respectively,
as seen in Fig. 1. Moreover, Φk (z) is the global Lagrangian interpolation function used for the
discretization of the displacement through the thickness, which can be chosen to be linear,
quadratic, or higher-order polynomial in terms of the thickness coordinate z. It is worth
emphasizing that the accuracy of LWT is enhanced by subdividing each lamina into a finite
number of numerical layers (i.e., p). The higher the number of subdivision through the thickness,
the higher accuracy of the structural predictions will be. Therefore, NLWT and k in Eqs. (3) are
referring to the virtual layers of the composite for Layer-wise method where NLWT = p × N. LWT
is general in that the number of subdivisions through the thickness and can be equal to, or greater
than the number of physical layers through the thickness. In addition, the number of subdivisions
can be different from layer to layer. The reduction in the number of layers makes the calculation
in thick laminates more efficient. Therefore, the convergence study should be performed to
determine an optimum value for p in order to obtain accurate results can be obtain with the
lowest computational cost. The linear global interpolation function is defined as [1, 49]:
 0 z  zk 1
 2 z
   zk 1  z  zk
 k ( z )   k11  k  1, 2,..., N LWT
 1 4
 k  z  zk  z  zk 1

 0 z  zk 1

where represents the local Lagrangian linear interpolation functions within the
kth mathematical layer defined as [49]:
1 1
1k  z    zk 1  z  and  k2  z    z  zk   5
hk hk
where hk is the thickness of the kth mathematical layer. We should note that by using linear global
interpolation functions, the continuity of displacement components through the thickness of the
laminate is identically satisfied. On the other hand, the transverse strain components remain
discontinuous at the interfaces, which eventually reinforce the feasibility of having continuous
interlaminar normal and shear stresses at the interfaces of neighboring layers by increasing the
number of numerical layers (p) through the laminate. However, increasing the number of

9
numerical layers introduces more interfaces and more unknowns for generalized displacement
components at such numerical interfaces leading to an increased computational cost. The strain-
displacement relations are given as [1]:

 xx  U k , x  k ,  
R
R0  R  z 
Vk ,  k  U k  k sin   Wk  k cos   ,

U k ,  k  Vk  k sin   ,
R
 zz  Wk  k , z ,  x  Vk , x  k 
R0  R  z 

Wk ,  k  Vk  k sin  
R
 xz  Wk , x  k  U k  k , z ,   z  Vk  k , z   6
R0  R  z 

It is worth mentioning that the in-plane normal and shear stresses (̅ , ̅ , and ̅ ) are not
continuous through the thickness while the out-of-plane normal and shear stresses (̅ , ̅ , and
̅ ) are continuous through the thickness of the laminated composite conical panels.

4. Equations of Motion
The equilibrium equations of a thick fiber-steered laminated composite conical panel with N
physical layers are obtained by using the principle of virtual displacement [1]:
t2

 U  V   K  dt  0
t1
7

where δU, δV, and δK are the virtual strain energy, virtual work done by external loads, and
virtual kinetic energy, respectively. Virtual strain energy δU is defined as [1]:
h /2 0 L
 z 
U     
 h /2  0
 xx       zz zz    z  z   xz xz   x  x  1 
xx
 R
 R 0dxd dz

8
0

The total virtual work done by the applied external loads δV is:

  z 
  1
 
 V    q  x, , t   w  Nˆ kxx  Nˆ xxk  HT  Wk , x Wk , x  2 Nˆ k  Nˆ k  HT  Wk , Wk , 1   R0d dx
A  R 
R0
 2 ˆk  z 
    
N x  Nˆ x  Wk ,  Wk , x 1   R0 d dx
k HT

A  R 
R0
 2M 0  t  2T  t   z 
    U k  0  U k sin    Vk cos    2 F0  t   U k 1   R0 d dx  9 
A
R0  z R0  z   R 
where F0 (t), T0 (t), and M0 (t) represent the time-dependant values of the distributed axial load,
torque, and bending moment, respectively, as shown in Fig. 1. The intensity of the applied

10
external transverse load is denoted by q (x, θ, t) (see Fig.1). Finally, the total virtual kinetic
energy δK defined as:
h /2
 z 
K     u u  v v  w w  1   R 0dxd dz
 R 
A
 h /2

  I U k  U j  Vk  V j  Wk  W j  R 0dxd
kj
10a 
A

where ρ (x, θ, z) represents the conical panel density, and the mass terms Ikj are defined:
h /2
 z 
Ikj     k  j 1   dz 10b 
 h /2  R 
Employing the fundamental lemma of calculus of variations, the equilibrium equations of
variable stiffness composite conical panels are obtained:
Qxk M xk N k 2T0
 U k : Qxk R0  Rxk cos   Nk sin    cos   x R0   R0  zk cos   sin 
 x x R0  zk
2M 0
  R0  zk cos    2 F0  R0  zk cos    R0 I kjU j 11a 
R0  zk
Rxk Q k N k
Vk : Qk R0  Rkz cos   Qkz sin   Qxk sin   cos   x R0  
x x 
2T0
  R0  zk cos   cos   R0 I kjV j 11b 
R0  zk
Rzk Q k Q k
 Wk : Nk cos   N zk R0  M zk cos   cos   z R0   z
x x 
 
  R0  zk cos   q  x, , t   k  N 1   R0  zk cos   Nˆ xx  Nˆ xx  Wk , xx
k k HT

1  zk  k  z 

R0  R 
 k  HT 
1   Nˆ   Nˆ    
Wk ,  2 1  k  Nˆ xk HT   Nˆ xk Wk , x   R0 I kjW j
 R 
11c 

 k  1, 2,..., N LWT
 1

where δkj is the Kronecker delta. Since the external transverse load is applied on the upper
surface of the conical panel, δk(N+1) is multiplied by q (x, θ, t) to reflect that. It should be
emphasized that Eqs. (11) can be reduced to the governing equation of cylindrical shell presented
in reference [56] for α = 0. The results are, in general, 3(NLWT+1) equilibrium equations
corresponding to 3(NLWT+1) unknown functions Uk, Vk, and Wk. The generalized stress and
moment resultants are defined as:

11
 h /2

N k
x , Nk , Qxk , Qkz , Qzk     xx ,   ,  x ,   z ,  xz   k dz 12a 
 h /2
h /2

M k
x
k
,R ,R
x
k
z     xx ,  x ,  xz  z k dz 12b 
 h /2
h /2

N k
z , Qxk , Qk     zz ,  xz ,   z   k , z dz 12c 
 h /2
h /2

M k
z , Rxk , Rkz     zz ,  xz ,   z  z k , z dz 12d 
 h /2

It should be mentioned that the superscript k in Eqs. (11) and (12) refers to the kth numerical
layer in the VS composite conical panel. The three-dimensional constitutive law within the kth
layer of the laminate can be written as [1, 57, and 58]:
k  k   k  k  
  xx  C11 C12 C13 0 0 C16      xx T   xx M 
      xx    
    C12 C22 C23 0 0 C26        T   M  
C      T   zz M  
  zz  C23 C33 0 0 C36    zz  
    13 
   z 
  zz 

13
 z  0 0 0 C44 C45 0   0 
0     
  xz  0 0 C45 C55 0     xz  0 
     
  x  C16 C66     x   2 T  2 M  
C26 C36 0 0
  x x  

where ̅ represents the stiffness components. ̅ is constant for a constant stiffness

composite conical panel, while it becomes a function of x and θ coordinates for a fiber-steered
composite conical panel defined in Appendix A. ̅ and ̅ are the transformed coefficients of
thermal and moisture expansions, respectively. A uniform temperature change ΔT and uniform
moisture concentration change ΔM are assumed for the analysis. The hygro-thermal stress
resultants are defined as:

 M   , M      C      ,      C      ,      C      ,      2C      ,      T,M  
h /2
k T k H k k k k k k k k k k k k
  12 xx xx 22   23 zz  26 x  k dz
 h /2

 N   , N      C      ,      C      ,      C      ,      2C      ,      T,M  
h /2
k T k H k k k k k k k k k k k k
x x 16 xx xx 26   36 zz  66 x  k dz
 h /2

 M   , M      C      ,      C      ,      C      ,      2C      ,      T,M  1  Rz  

h/ 2
k T k H k k k k k k k k k k k k
x x 11 xx xx 12   13 zz  16 x  k dz
h/ 2 

   C      ,      C      ,      C      ,      2C      ,      T,M  1  Rz  

h/ 2
M x  , M x 
k H

k T k k k k k k k k k k k k
16 xx xx 26   36 zz  66 x  k dz
h/ 2 

12
   C      ,      C      ,      C      ,      2C      ,      T,M  1  Rz  
h/ 2
Nz   , Nz  dz 14 
k H

k T k k k k k k k k k k k k
13 xx xx 23   33 zz  36 x  k,z
h/ 2 

The generalized stress and moment resultants in LWT are expressed in terms of the
displacement functions on substitution of Eqs. (6) in Eq. (13) and the subsequent results in Eqs.
(12). The final results are:
sin  kj kj kj
N k
x , Nk , Qxk  
R0
 D12 , D22 , D26 U j   D11kj , D12kj , D16kj U j , x   D16kj , D26kj , D66kj U j ,
1
R0
sin  kj kj kj

R0
 D16 , D26 , D66 V j   D16kj , D26kj , D66kj V j , x   D12kj , D22kj , D26kj V j ,
1
R0
 cos  kj cos  kj cos  kj 

 R0
D12  B13kj ,
R0
D22  B23kj ,
R0 
k HT k HT k HT

D26  B36kj  W j  M x   , M    , N x  15a  
 
Qkz , Qzk , Rzk    B45jk , B55jk , B55jk U j    sinR  D44kj  R0 B44jk ,  sinR  D45kj  R0 B45jk ,  sinR  D45kj  R0 B45kj V j
 0 0 0 
  D45kj , D55kj , D55kj W j , x 
1
R0
 
D44kj , D45kj , D45kj W j , 15b 

M k
x , Rxk  
sin  kj kj
R
 
D12 , D26 U j   D11kj , D16kj U j , x   D16kj , D66kj U j ,
1
R


sin  kj kj
R
 
D16 , D66 V j   D16kj , D66kj V j , x 
1
R
D12kj , D26kj V j ,  
 cos  kj cos  kj 

 R0
D12  B13kj ,
R0
D26  B36kj  W j  M x   , M x 

k HT k HT
  15c 

 N zk , M zk   sinR   B23jk , B23jk U j   B13jk , B13jk U j ,x  R1  B36jk , B36jk U j ,

0 0

sin 

R0
 B36jk , B36jk V j   B36jk , B36jk V j , x   B23jk , B23jk V j ,
1
R0
 cos  jk cos  jk 

 0R
B23  A33kj ,
R 0
B23  A33kj  W j  N z  

k HT
  15d 
 kj sin  jk kj sin  jk 
Q , Q    A
k
x
k
 , A45 U j   A45 
kj
55
kj
B45 , A44  B44 V j   B55 , B45jk W j , x   B45jk , B44jk W j , 15e 
jk 1
 R0 R0  R0
 
 Rxk , Rkz    A55kj , A45kj U j   A45kj  sinR  B45jk , A44kj  sinR  B44jk V j   B55jk , B45jk W j ,x  R1  B45jk , B44jk W j , 15 f 
    

where the laminate stiffnesses in Eqs. (15) are defined as:

 Apqkj , Bpqkj , Dpqkj      k ,z  j ,z , k  j ,z , k  j  dz

N zi1
 i
C pq
zi
i 1

13
 Apqkj , Bpqkj , Dpqkj      k ,z  j ,z , k  j ,z , k  j  zdz
N zi1

i
C pq
zi
i 1

 k , j  1, 2,..., N LWT  1
N
R
   C pq
i 
16 
zi1
kj
Dpq k  j dz
i 1
zi R  z
where z-direction is along the thickness direction as shown in Fig. 1. These laminate stiffnesses
are a function of the coordinates z and x, and are defined in Appendix B with the help of Eq. (4).
The governing equations are derived by substituting Eqs. (15) in Eqs. (11). We should remind
that in contrast to constant stiffness composites, the elements of the stiffness matrices of variable
stiffness composites are a function of the spatial coordinate. In other words, the derivatives of
these matrices with respect to x and θ coordinates are not zero. Therefore, the governing
equations are dissimilar from the conventional governing equations for a constant stiffness
composite.

5. Space Solution
In this section, we introduce a numerical methodology for solving the coupled governing
differential equations in the space coordinate system for alternative boundary conditions of the
fiber-steered composite conical panels. The boundary conditions can be simply-supported,
clamped, free, or a hybrid of any of those aforementioned. The Galerkin method is adopted here
for solving the governing differential equations to enable us predicting the performance of fiber-
steered composites of arbitrary boundary conditions. Mathematical expressions for boundary
conditions considered in this study are [1, 59]:
(a) Immovable simply-supported edges (S):
U k Vk  Wk  0  axial  at x  0, L
U k Vk  Wk  0  circumferential  at   0 17a 
(b) Clamped edges (C):
Wk Wk
U k  Vk  Wk   0  axial  at x  0, L
 x
Wk Wk
U k  Vk  Wk   0  circumferential  at   0 17b 
 x
(c) Free edges (F):
N xk  M xk  Qxk  Rxk  Rzk  0  axial  at x  0, L
Nk  Qkz  Qxk  Rxk  Rkz  0  circumferential  at   0 17c 
14
where k = 1, 2, …, NLWT+1. It should be emphasized that governing equations are 3(NLWT+1)
equations and each of the boundary conditions in Eqs. (17) provides minimum 3(NLWT+1)
equations. To implement the Galerkin method, the displacement field should be expressed in the
following form for arbitrary boundary conditions [1, 59]:
 
 m  x 
U k  x,  , t   U kmn  t   n  
m 1 n 1 x
 
 n  
Vk  x,  , t   Vkmn  t   m  x 
m 1 n 1 
 
Wk  x,  , t   Wkmn  t   m  x   n    k  1, 2,..., N  1 18
m 1 n 1
where , , and (m and n are integer numbers) are unknown coefficients determined
by satisfying governing differential equations and boundary conditions given in Eq. (17). In
addition, admissible functions and are defined as [1]:

 m  x   1 cosh  m x    2 cos  m x    m 3 sinh  m x    4 sin  m x  

 n    1 cosh  n    2 cos  n    n   3 sinh  n    4 sin  n   19 
where λm and λn are defined in Table 1 for alternative boundary conditions. As the panel could be
subjected to an external distributed out-of-plane load q, distributed axial load F0, torque T0, and
bending moment M0, they should be also expanded as:

Q  t  , F  t  , M  t  , T  t  
mn mn mn mn

0
    q  x, y, t  , F  t  , M t  , T t     x     dxd
2
 20
L

l0
0 0 0 m n
 0 0

By implementing the Galerkin method and the approximate displacement fields given in Eqs.
(18) and (20), we can solve the governing differential equations by [1]:
  p  x  
 R1 U k ,Vk ,Wk   l   
mn mn mn

 x 
L    
0 0  R2 U k ,Vk ,Wk   p  x    dxd  0
0
mn mn mn l
 21
 
 R3 U kmn ,Vkmn ,Wkmn   p  x   l   
 
 
 p  1, 2,..., and l  1, 2,...
where Ri (i = 1, 2, and 3) are the residuals of the governing differential equations, when the
admissible displacement fields are substituted into the governing differential equations.

15
Substituting Eq. (18) in the governing equations and then applying the Galerkin formulation (Eq.
(21)) lead to an expanded formulation, leading to a system of 3(NLWT+1) differential equations:
 M 11 [0] N 1 N 1 [0] N 1 N 1  U mn   K K12 K13  U kmn 
   k   11  
[0] N 1 N 1 M 22 [0] N 1 N 1   Vkmn    K 21 K 22 K 23   Vkmn 
  W mn    mn 
K 33  Nˆ  Wk 
 
k   K 31
  N 1 N 1
[0] [0] N 1 N 1 M 33 K 32

 2Tmn 2 M mn 
 R  z  R0  zk cos   sin   R  z  R0  zk cos    2 Fmn  R0  zk cos   
 0 k 0 k

 2Tmn 
  R0  zk cos   cos   22 
 R0  zk 
  R0  zk cos   Qmn k  N 1 
 
 
It should be mentioned that [M] and [K] are symmetric stiffness and mass matrices, respectively,
and their components can be obtained by the integration of [M'] and [K'] presented in Appendix
C for arbitrary boundary conditions, using Eq. (21). In addition, ̂ is a function of buckling loads
defined in Appendix C for alternative boundary conditions. For harmonic and free vibration
analyses, solutions are sought in the following form [1]:
Ukmn  t   Ukmn 0eimnt , Vkmn  t   Vkmn 0eimnt , Wkmn  t   Wkmn 0eimnt  23
Substitution of Eq. (23) in Eq. (22) yields an eigenvalue problem in the absence of mechanical
load, elevated temperature, moisture concentration change, and nonlinear in-plane force resultant
̂ . The fundamental frequency of the panels can be obtained by solving the following eigenvalue
problem:

 K   2  M   0  24
where ω is the fundamental frequency of the fiber steered composite conical panel and | |
represent determinant in Eq. (24). The smallest eigenvalue obtained from Eq. (24) is called the
fundamental vibration frequency. It should be mentioned that the nonlinear displacement term ̂
is omitted for static bending and forced vibration analysis, but it is preserved for the bifurcation
buckling study of fiber-steered composite conical panels. Admissible trigonometric functions for
different boundary conditions of fiber-steered composite conical panels are presented in Table 1,
in which the symbols S, C, and F stand for immovable simply-supported, clamped and free edge
boundary conditions, respectively. For example, CSCS refers to a fiber-steered composite
conical panel with clamped edges at x = 0 and L and simply-supported edges at θ = ±θ0. It should

16
be mentioned that the admissible functions must satisfy the essential boundary conditions (i.e.,
generalized displacement) of the problem. However, admissible functions, only for the simply-
supported case, satisfy both natural (i.e., generalized forces) and essential boundary conditions. It
is worthwhile mentioning that free edge boundary conditions are not reasonable to be considered
for critical buckling temperature or moisture concentration as no hygro-thermal reactions can be
applied on the free edges.

Table 1: Values of αj, λi, and ζi for S-S, C-C, F-F, C-S, C-F and F-S boundary conditions [1, 59].
Characteristic equations
Boundary αj or βj
and values of μλi (L λm or ζi (ζm or ζn)
conditions (j=1, 2, 3, and 4)
2θ0 λn)
α1=0, α2=0 sin(μλi)=0
S-S 1 or 1
α3=0, α4=-1 mπ or nπ

α1=1, α2=-1 cos(μλi)cosh(μλi)=1

C-C
α3=1, α4=-1 μλi=4.730, 7.853, …

α1=-1, α2=-1 cos(μλi)cosh(μλi)=1

F-F
α3=-1, α4=-1 μλi=4.730, 7.853, …

α1=1, α2=-1 tan(μλi)=tanh(μλi)

C-S α3=1, α4=-1 μλi=3.927, 7.069, …

α1=1, α2=-1 cos(μλi)cosh(μλi)=-1

C-F α3=1, α4=-1 μλi=1.875, 4.694, …

α1=1, α2=1 tan(μλi)=tanh(μλi)

F-S
α3=1, α4=1 μλi=3.927, 7.069, …

6. Results and Discussion

The hygro-thermo-mechanical analysis of fiber-steered composite conical panels is
investigated for thin to thick structures. We should mention that, in this paper, we limit our
analysis to uncoupled hygro-thermo-elasticity and steady state hygro-thermal load. In the
uncoupled hygro-thermo-elasticity, there is no effect of mechanical field on thermal and
hygroscopic field and the coupling of hygro-thermal fields is ignored. In addition, because of the
analogy between hygroscopic and thermal fields, most of the results in this paper is provided for
thermal responses. For the coupled multiphysics problems, including coupled hygro-thermo-
elasticity, the readers are referred to reference [60]. Two fiber-steering paths are considered,
fiber path with a linear angle variation and a fiber path with a constant curvature. Deflection and

17
resultant stresses are obtained for the fiber-steered composites subjected to a uniform transverse
load and/or elevated temperature or moisture concentration. All physical layers are assumed to
have equal thickness of 0.125 mm [33] and are modeled as being made up of p numerical layers.
In all the subsequent calculations, p is assumed to be 9. The material properties of Carbon/Peek
composites used for the analysis of constant and variable stiffness laminates are given in Table 2
[23, 41]. We consider a conical panel made up of a 40-ply balanced and symmetric laminate with
a variable stiffness design of [±<T0|T1>]10s with length L = 10.777 cm (L/h = 21.5), minor
reference radius R b = 6 cm, semi-vertex angle α = 21.80°, and span angle 2θ0 = 90° [33] (unless
otherwise mentioned). In addition, quasi-isotropic, and constant stiffness composite conical
panels with the lay-up sequences of [0/±45/90]5s, and [±θ]10s, respectively, are considered as a
baseline to compare their responses to those of fiber-steered composite conical panels.

Table 2: Material properties of Carbon/Peek composite laminate [23, 41].

Mechanical properties Value
E11 143 (GPa)
E22 =E33 9.1 (GPa)
G12 4.82 (GPa)
G13 = G23 4.9 (GPa)
ʋ12 = ʋ13 = υ23 0.3
ρ 1500 (kg/m3)
α1 -0.2×10-6 (°C-1)
α2 = α3 24×10-6 (°C-1)
β1 146×10-6
β2 = β3 4770×10-6

The numerical results described here focus on the effect of the following parameters on
deflection, stress distribution, critical buckling load, and fundamental frequency: fiber path, fiber
orientation, boundary conditions, and geometrical features such as length-to-thickness ratio,
semi-vertex, and span angles of variable stiffness conical panels. The out-of-plane deflection (w),
critical buckling load (Ncr), fundamental frequency (ω), and stress (σ) are provided in the
following non-dimensional forms:
E22h3 L2 L2  
w  w0 , N  N cr ,   ,   25
L4 q0 E22h 3 h E22 q0

6.1. Validation

18
 In order to validate the accuracy of the developed methodology, numerical results are
compared with the results of finite element method (FEM) and existing data presented in the
literature. For the validation purpose, the mechanical and geometrical properties of the constant
stiffness cylindrical panel are taken from reference [61], while for the rest of numerical studies;
material properties given in Table 2 are used. It should be mentioned that the formulations
derived here for conical panels can also be used for the analysis of cylindrical composite panels
when α = 0°. As a result, non-dimensional critical buckling loads obtained by the developed
model for cross-ply SSSS cylindrical panels are compared with those reported in reference [61]
based on the first-order shear deformation shell theory. In addition, the buckling analysis of the
constant stiffness composite conical panel with α = 20° is also conducted by the finite element
method using ANSYS 2017 to compare their results with those predicted by the proposed
method. The element used to perform the analysis is the layered solid element, SOLID 185.
SOLID 185 is a 3D element which is characterized by 8 nodes and each node has three degree of
freedoms. This element can be used to model both homogeneous and layered solids. For layered
solids such as composites, a section should be defined and associated with elements in order to
define the number of layers per element, layer’s thickness, material properties of each layer, and
layer’s orientation. The number of integration points (1, 3, 5, 7 or 9), located through the
thickness of each layer used for element calculations, can be designated. Two points are located
on the top and bottom surfaces; and the remaining points are distributed equally between the two
points. Here, three integration points are used through the thickness of each layer. In addition,
one element or more, depend on mesh number for each layer in the lay-up sequence, is defined in
FEM. Table 3 presents the non-dimensional critical buckling loads of cross-ply SSSS and CCCC
cylindrical and conical panels for two composite laminates. As shown in Table 3, the maximum
difference of 0.2% is found between the results confirming the reliability and accuracy of our
developed method.

Table 3: Comparison of non-dimensional critical buckling load ( ̅) of SSSS and CCCC cylindrical and conical
composite panels.
Cylindrical panel Conical panel
Lay-up
SSSS CCCC SSSS CCCC
sequence
Present Ref. [61] Present Ref. [61] Present FEMANSYS Present FEMANSYS
[0°/90°/0°] 0.2764 0.2765 0.4167 0.4168 0.2570 0.2568 0.3987 0.3985
[0°/90°] 0.1523 0.1525 0.2404 0.2406 0.1462 0.1460 0.2254 0.2251

19
6.2. Effects of length-to-thickness ratio
In this section, we present the effects of length-to-thickness ratio on the out-of-plane
deformation and stress distributions of fiber-steered composite conical panels. Figure 4 presents
the non-dimensional deflection ( ̅ , normal stress (̅ ), and shear stresses (̅ and ̅ ) of the
CCFF fiber-steered conical panels with [±<49.5|36.2>]ns symmetric lay-up sequence along the
circumferential direction for n = 1, 5, 10, and 20 [33]. It should be mentioned that the data are
obtained at z = 0 and x = L/2. It can be seen in Fig. 4 that by increasing the number of physical
layers of the composite conical panel from 4 (n = 1, or L/ h = 215.4) to 80 (n = 20, or L/ h =
10.77), the magnitude of free-edge stresses increases opposed to the out-of-plane deflections
which decrease. These interlaminar stresses are existed due to the mismatch in
hygroscopic/thermal expansion coefficients of adjacent plies. Therefore, the thicker the
composite laminates, the larger free-edge effects will be.

(a) (b)

20
 (c) (d)
Figure 4: (a) Non-dimensional deflection, (b) normal stress ( ̅ ), (c) shear stress ( ̅ ), and (d) shear stress ( ̅ ) of
the CCFF fiber-steered conical panels with the [±<49.5|36.2>]ns lay-up sequence along the circumferential direction.

The proposed model in this paper is able to capture the free-edge effects for thick constant and
variable stiffness composites in which free-edge stresses play an important role in delamination
and catastrophic failures. It is worth mentioning that LWT is capable of accurate prediction of
the non-zero transvers shear stresses in thick composite panels opposed to the ESL plate and
shell theories. In addition, we should mention that free-edge stresses are observed in the free
edges of composite structures under multiphysical loads no matter the composite structure is a
constant stiffness or a variable stiffness.

6.3. Effects of fiber paths on hygro-thermo-mechanical response

As explained in Section 2 for fiber-steered composite panels with a linear angle variation or a
constant curvature, the parameters T0 and T1 define the fiber path trajectory. This section
examines the impact of the fiber path parameters, T0 and T1, on the hygro-thermo-mechanical
responses of a thick variable stiffness composite conical panel with the [±<T0|T1>]ns lay-up
sequence. For the buckling analysis, buckling loads in the x-direction are specified (see Fig. 1).
Figures 5a and 5b present a design chart for a CCCC fiber-steered composite conical panel
with [±<T0|T1>]10s lay-up sequence for the non-dimensional critical buckling load ( ̅ versus the
non-dimensional fundamental frequency ( ̅ shown for the two types of fiber paths: linear angle

21
variation and constant curvature fiber paths. In Fig. 5, T0 (fiber angle at the minor radius) varies
from T0 = 0° to T0 = 90°, while T1 (fiber angle at the major radius) varies from T1 = 0° to T1 =
45° in Fig. 5a and from T1 = 45° to T1 = 90° in Fig. 5b.

(a) (b)
Figure 5: Design chart for fiber-steered composite conical panels [±<T0|T1>]10s for the non-dimensional critical
buckling load ( ̅ vs. the non-dimensional fundamental frequency ( ̅ .

As shown in Fig. 5a, For T0 = 0°, the critical buckling load and fundamental frequency decrease
when the fiber angle at the minor radius T1 varies between 0° and 45°; the trend is opposite for
T0 = 90° . The behavior is reversed for 45° < T1 < 90° as shown in Fig. 5b. Comparing the results
obtained for variable stiffness design shows that the constant curvature fiber path shifts the
response domain towards a higher buckling load and fundamental frequency. As opposed to a
quasi-isotropic and constant stiffness design whose response domains are represented by a point
and a line, respectively, a variable stiffness design yields to domains of larger extent providing
composite designers with a series of fiber path selection options to optimize the performance of
composite structures subjected to multiphysical loading scenarios. Furthermore, the buckling
loads and fundamental frequencies of composite conical panels could be improved by 57% and
44%, respectively, by using a constant curvature fiber-steered lay-up instead of a quasi-isotropic
design.
Figures 6a and 6b demonstrate a design chart for a CCCC fiber-steered composite conical
panel with [±<T0|T1>]10s lay-up sequence for the critical buckling temperature (ΔTcr) versus the

22
non-dimensional maximum deflection ( ̅ shown for two fiber paths including a path with a
linear angle variation, and a path with constant curvature. In Fig. 6, T0 varies from T0 = 0° to T0
= 90°, while T1 varies from T1 = 0° to T1 = 45° in Fig. 6a, and from T1 = 45° to T1 = 90° in Fig.
6b. It is noted that the maximum out-of-plane deflection occurs at the conical midpoint. As can
be seen in Fig. 6a, once T1 varies from 0° to 45°, the critical buckling temperature decreases for
T0 = 0° while the maximum deflection increases for T0 = 90°. These trends are reversed in Fig.
6b where 45° < T1 < 90°. As can be seen, the regions offering low maximum deflections
correspond to regions with a high critical buckling temperature. In addition, the critical buckling
temperature of constant curvature fiber-steered conical panels is greater than that of linearly
varying fiber-steered conical panels while the maximum deflection of constant curvature fiber-
steered conical panels is less than that of linearly varying fiber-steered conical panels. In
addition, the critical buckling temperatures of composite conical panels could improve up to
64%, by using a constant curvature fiber-steered lay-up instead of a quasi-isotropic lay-up, while
their maximum deflections decrease up to 67%. It should be mentioned that while high ΔTcr is
theoretically acceptable, in reality, composite materials, specifically polymers, could not bear
such a harsh environment since the composite material can show some chemical reaction in the
elevated temperature and moisture conditions.

(a) (b)
Figure 6: Design chart for fiber-steered composite conical panels [±<T0|T1>]10s for the critical buckling temperature
(ΔTcr) vs. the non-dimensional maximum deflection ( ̅ .

23
 (a) (b)
Figure 7: Design chart for fiber-steered composite conical panels [±<T0|T1>]10s for the non-dimensional
fundamental frequency buckling load ( ̅ ̅ vs. the non-dimensional free-edge stress ( ̅ ).

Figures 7a and 7b show a design chart for a CCFF fiber-steered composite conical panel with
[±<T0|T1>]10s lay-up sequence for the non-dimensional fundamental frequency buckling load
( ̅ ̅ versus the non-dimensional free-edge stress ( ̅ ) shown for the two types of fiber paths.
In Fig. 7, T0 varies from T0 = 0° to T0 = 90°, while T1 varies from T1 = 0° to T1 = 45° in Fig. 7a,
and from T1 = 45° to T1 = 90° in Fig. 7b. We observe that the fundamental frequency buckling
load-free edge stress domain for a variable stiffness conical panel for 45° < T1 < 90° is
significantly larger than the corresponding fundamental frequency buckling load-free edge
stress domain for 0° < T1 < 45°. As can be seen, the regions offering low free edge stresses
correspond to regions with high critical buckling loads and fundamental frequencies.

Cylindrical panel (α = 0°) Conical panel (α = 21.80°)

24
Conical panel (α = 45°) Conical panel (α = 75°)

Circular panel (α = 90°)

25
 Figure 8: Non-dimensional critical buckling load ( ̅ for constant curvature fiber-steered composite cylindrical,
conical, and circular panels [±<T0|T1>]10s.

Bearing in mind that locating fibers along the loading direction results in the highest in-plane
stiffness but not the lowest free edge normal stress, we observe here the opportunity to find
trade-off between the in-plane and out-of-plane performance under vibration frequency and
buckling load constraints. It should be noted that using a constant curvature fiber path shifts the
response domain towards a higher buckling load and fundamental frequency and a lower free
edge stress compared to using a linearly varying fiber path for the variable stiffness composite
conical panels. For example, the fundamental frequency buckling load amplitude increases by
45% and the free-edge stress decreases by 15% for a conical panel with a constant curvature path
compared to a linearly varying fiber path. It is worth mentioning that the free edge stresses of
composite conical panels could be decreased by 60%, when an appropriate constant curvature
fiber-steered lay-up is used instead of a quasi-isotropic lay-up.
Figure 8 presents the fiber path parameters (i.e., T0, and T1) versus non-dimensional critical
buckling loads of a constant curvature fiber steered composite cylindrical, conical, and circular
panels [±<T0|T1>]10s with the CCCC boundary conditions. As shown in Fig. 8, the buckling loads
of conical panels with α = 21.80° decrease with increasing the fiber angle at the major radius of
the conical panels from T1 = 0° to T1 = 45° while the buckling load increases from T1 = 45° to T1
= 90° while the fiber angle at the minor radius (T0) is kept constant. When the fiber angle at the
major radius has a low or high values (0° < T1 < 15° and 75° < T1 < 90°), the buckling loads
decrease with varying the fiber angle at the minor radius from T0 = 0° to T0 = 90° otherwise the
buckling loads increase from T0 = 0° to T0 = 90°. As can be seen from Fig. 8, there is a relation
between the high buckling bands, shown with the red color in Fig. 8, with semi-vertex angle α of
fiber-steered conical panels. The angle of the red band, for maximum buckling load with respect
to the axis of T0 is almost equal to the semi-vertex angle in Fig. 8.

6.4. Effects of geometrical parameters on hygro-thermo-mechanical response

In this section, we investigate the effects of boundary conditions, number of layers, and
geometrical parameters, such as length L , semi-vertex angle α, and span angle 2θ0 of conical
panels on the hygro-thermo-mechanical responses of constant curvature fiber-steered conical
panels with the lay-up sequences of [±<0|90>]10s.

26
 (a) (b)
S C S C C C S C
S C F F S F F S
S C F F S C S C
S C S C C F F S

Figure 9: Design chart for fiber-steered composite conical panels [±<0|90>]10s with a constant curvature path for
the non-dimensional fundamental frequency buckling load ( ̅ ̅ vs. the non-dimensional von Mises
stress maximum deflection ( ̅ ̅ .

Figure 9 illustrates a design chart for fiber-steered composite conical panels with the
[±<0|90>]10s lay-up sequence with a constant curvature path for the non-dimensional
fundamental frequency mechanical buckling load ( ̅ ̅ versus the non-dimensional von Mises
stress maximum deflection ( ̅ ̅ for different boundary conditions including CCCC,
SSSS, CCFF, and SSFF boundary conditions (Fig. 9a) and CSCS, CCSS, CFCF, and SFSF
boundary conditions (Fig. 9b). Semi-vertex angle increases from α = 0° to α = 90° and the length
increases from L = 0.5R b to L = 2.5R b . As seen in Fig. 9, the effectiveness of variable stiffness
design, compared to a constant stiffness design, depends on the boundary conditions.
Figure 10 shows a design chart for CCCC fiber-steered conical panels with [±<0|90>]ns lay-
up sequence for the non-dimensional fundamental frequency buckling load ( ̅ ̅ versus the
critical buckling moisture concentration (ΔMcr) shown for two different types of fiber paths. In
Fig. 10, the span angle increases from 2θ0 = 11.25° to 2θ0 = 360°, and the number of layers

27
increases from 4 (n = 1, or L/ h = 215.4) to 80 (n = 20, or L/ h = 10.77). Increasing the number
of layers (i.e., composite thickness) of a fiber-steered conical panel shifts the domain of
fundamental frequency buckling load versus critical buckling moisture concentration domain to
higher values. It is also seen in Fig. 10 that increasing the span angle from 2θ0 = 11.25° to 2θ0 =
360° leads to an increase in the fundamental frequency, buckling load, and critical buckling
moisture concentration. It is worth mentioning that increasing the span angle of a conical
composite provides a better opportunity for optimizing the overall performance of variable
stiffness composite conical panels by steering fibers in arbitrary paths.

Figure 10: Design chart for fiber-steered composite conical panels [±<0|90>]ns for the non-dimensional
fundamental frequency buckling load ( ̅ ̅ vs. the critical buckling moisture concentration (ΔMcr).

In order to investigate the effect of simultaneous hygro-thermal loading on the responses of

fiber-steered composite conical panels, a thick fiber-steered conical is selected and the variation
of hygro-thermal stresses in the form of von Mises stress through the thickness is shown. Figure
11 presents the distribution of the non-dimensional von Mises stress ̅ along the thickness
of the CCCC fiber-steered conical panels with the [±<0|90>]20s lay-up sequence (80 layers)
subjected to a temperature change (ΔT = 100 °C) and moisture concentration change (ΔM = 1
%). It should be mentioned that the von Mises stresses are obtained at θ = 0 and x = L/2. As seen

28
in Fig. 11, with the superposition of thermal and moisture concentration responses, stresses for a
fiber-steered composite subjected to the combined steady-state hygro-thermal loading can be
obtained. In addition, it should be noted that the magnitudes of the in-plane hygro-thermal
stresses are greater than those of the out-of-plane hygro-thermal stresses.

Figure 11: Distribution of non-dimensional von Mises stress ̅ along the thickness of CCCC fiber-steered
conical panels with the [±<0|90>]20s lay-up sequence due to temperature change (ΔT = 100 °C), moisture
concentration change (ΔM = 1 %), and simultaneous temperature and moisture concentration changes (ΔT = 100 °C
and ΔM = 1 %).

7. Conclusion
We develop a semi-analytical methodology to accurately predict uncoupled hygro-thermo-
mechanical responses of thick fiber-steered composite conical panels. The methodology enables
the analysis of fiber-steered panels in the form of conical panels, cylindrical panels, and circular
plates. Structural analysis for exploring the static bending, tension, torsion, hygro-thermo-
mechanical buckling, and free vibration has been formulated based on Layer-wise theory. The
governing differential equations have been solved by using a hybrid Fourier-Galerkin method.
The effects of length-to-thickness ratio, boundary conditions, number of layers, geometrical
parameters, and fiber paths on hygro-thermo-mechanical responses of thick VS composite
conical panels are examined to provide a case study for the optimization of the structural

29
performance of composite panels by tuning the fiber path. The significant feature of the proposed
methodology is its capability for capturing free-edge effects of thick VS composite panels. The
major findings of this study can be highlighted as:
 The buckling loads and fundamental frequencies of composite conical panels can be
respectively improve up to 57% and 44% compared to a quasi-isotopic composite with
the same weight and thickness, by employing a constant curvature fiber-steered
composite with the length-to-thickness ratio of L/h = 21.5.
 The critical buckling temperatures of composite conical panels can be increased up to
64% by using a constant curvature fiber-steered lay-up instead of a quasi-isotropic lay-up
while the maximum deflection of the panel subjected to the mechanical load also
decreases by 67%.
 Compared to constant stiffness composites, buckling can occur in a lower temperature
(ΔTcr) in VS composite laminates with a linearly varying fiber angle. This observation
implies the significance of a multi-criteria design (stiffness, strength, and hygro-thermal
performances) for advanced composite structures.
 The fundamental frequency buckling load amplitude increases by 45% and the free-edge
stress decreases by 15% for a composite conical panel with a constant curvature path
compared to a linearly varying fiber path.
 The free-edge stresses of composite conical panels can be decreased by 60% by
designing a constant curvature fiber-steered composite conical panel instead of a quasi-
isotropic composite.

Acknowledgment
A.H. Akbarzadeh acknowledges the financial support provided by McGill University and the
Natural Science and Engineering Research Council of Canada (NSERC) through NSERC
Discovery Grant RGPIN-2016-04716.

Appendices
Appendix A
The components of ̅ are constant in a constant stiffness conical panel while they become a
function of θ in a fiber-steered conical panel. Therefore, the following relations hold:

30
 C11 k   k   m4C11  2m2 n 2  C12  2C16   n 4C22 , C13 k   k   m2C13  n 2C23 ,
C12 k   k   m2 n 2  C11  C22  4C66    n 4  m4  C12 , C23 k   k   n 2C13  m 2C23
C16 k   k   mn  m 2  C11  C12  2C66   n 2  C12  C22  2C66   ,
C22 k   k   n 4C11  2m 2 n 2  C12  2C16   m4C22 , C33 k   k   C33
C26 k   k   mn  n2 C11  C12  2C66   m2 C12  C22  2C66   ,
C36 k   k   nm  C13  C23  , C44 k   k   m2C44  n 2C55 , C45 k   k   nm C55  C44 

C55 k   k   n2C44  m2C55 , C66 k   k   m2 n 2  C11  2C12  C22    n 2  m 2  C66  A.1

2

and the transformed coefficients of thermal and moisture expansions are:

 xx ,  xx    xx ,  xx  m2   ,   n2 ,  x ,  x    xx   ,  xx    nm

  ,     xx ,  xx  n2   ,   m2 ,  zz ,  zz    zz ,  zz   A.2 

where m = cos (φk), and n = sin (φk), in which φk is defined in Eqs. (1) and (2).

Appendix B
The laminate stiffnesses based on LWT are obtained by carrying out the integrations of Eq. (16):
 C pq
 k 1
C pq k 1
hk 1C pq  k 1 
  , ,  if j  k  1
 hk 1 2 6 

  k 1  k  k 1  k  k 1 k  
  C pq C pq C pq C pq hk 1C pq hk C pq
     if j  k
 Apqkj , Bpqkj , Dpqkj     hk 1 hk 2 2
, ,
3 3 
 k  k  k  
  C pq C pq hk C pq
  , ,  if j  k  1
 hk 2 6 
 
  0,0,0  if j  k  1 or j  k  1

A kj
pq , B pqkj , D pq
kj

  C pq  k 1
C 
k 1
h C 
k 1

 
 2hk 1
 zk  zk 1  ,  pq  2 zk  zk 1  , k 1 pq  zk  zk 1  

if j  k  1
  6 12 
  k 1 k   k 1 k   k 1 k 
 C pq C pq C pq C pq hk 1C pq hk C pq 

  2hk 1
 k k 1 
z  z 
2hk
 k 1 k 
z  z ,
6
 k k 1 
2 z  z 
6
 k k 1 
2 z  z ,
12
 hk 1  2  k k 1  
z  z 
12
 hk  2  zk 1  zk    if j  k


  k  k  k
  C pq C hC 
 
 2hk
 zk 1  zk  , pq  2 zk  zk 1  , k pq  zk 1  zk  

if j  k  1
  6 12 
  0,0,0  if j  k  1 or j  k  1

and

31
   k 1 
R C pq R z  R z  R  z 
  1  2  k 1  1    k 1  1 ln  k   if j  k  1
 2  hk 1   hk 1  R  zk 1  
  k 1 k 
 R C pq  R  zk 1  R  zk 1 R  zk   R C pq  R  zk 1  R  zk 1 R  zk 1  
 1  2  1 ln   1  2 1  ln  if j  k
D pq   2 
kj
hk 1  hk 1 R  zk 1   2  hk  hk R  zk  
 k  
 R C pq R z  R z  R  z 
  1  2  k 1  1    k 1  1 ln  k 1   if j  k  1
2  hk   hk  R  zk  

 if j  k  1 or j  k  1
 0

Appendix C
The components of [K] and [M] matrices in Eq. (22) are calculated by the integration of
components of [K'] and [M'] matrices using Eq. (21):
 1 
K11    D11kj cos   R0 D11kj  m2  m , x3  n 
1 kj 2
D66n  m , x  n , 2  2  D16kj  D16kj  mn  m , x2  n ,
R0  R 
 sin 2  kj 
 D22  A55kj cos   R0 A55kj   m , x  n
 R0 
K12    D16kj cos   R0 D16kj  m2  m , x2  n ,  D66n  m n , 3   D12kj  D66  mn  m,x n, 2
1 kj 2 kj

R0

  D16kj  D26
kj

 sin m m,x2  n,  D22kj  D66kj  sinR    
0
n m n , 2

 sin 2  kj 
 D26  A45kj cos   R0 A45
kj
 B45jk sin    m n ,
 R0 
 
K13    D12kj  B13kj  B55jk  cos   R0  B13kj  B55jk  m m, x  n
 cos  kj kj   cos  sin  kj kj  k  HT 
 D26  B45jk  B36  n  m  n ,   D22  sin  B23   m n  M sin 
 0R   R0 

  1
   R0 D66kj cos   R0 D66kj m2 m, x2  n ,  D22kj n2 m n , 3  2 D26kj mn  m , x n , 2
K 22
R0
 sin 2  kj 

 R0
 
D66  D44kj   B44kj  B44jk  sin   A44kj cos   R0 A44kj   m n ,


32
  
    D26kj  B36kj  B45jk  cos   R0  B36kj  B45jk   D45kj sin  m m, x  n
K 23
 cos  kj sin  kj   cos  sin  kj 
 D22  B44jk  D44  B23kj  n  m  n ,   D26  sin  B36kj   m n  N x  sin 
k HT

 R0 R0   R0 
 1 
  1
K33   D55kj cos   R0 D55kj m2  m, x2  n  D44kj n2 m  n , 2  2  D45kj  D45kj  m n  m, x  n ,
R0  R 
 cos 2  kj 
 D22   B23kj  B23jk  A33kj  cos   R0 A33kj   m n  M k  HT  cos   R0 N zk  HT 
 R0 
M11  R0 I kj m, x n ,   R0 I kj mn, ,
M 22   R0 I kj mn
M 33
1  z  k
 
Nˆ    R0  zk cos   Nˆ kxx  Nˆ xx  m2  m, x2  n  1  k  Nˆ 
k HT

R0  R 
 
 Nˆ   n2 m n, 2
k HT

 z 
 
2 1  k  Nˆ xk  Nˆ xk HT  m n  m, x  n ,
 R 

References
[1] Reddy JN. Mechanics of laminated composite plates and shells: theory and analysis. CRC
press 2004.
[2] Kisch RA. Automated fiber placement historical perspective. Proceedings of the SAMPE
2006 Conference, Long Beach, CA, USA, 2006.
[3] Martin JP, Langone RJ, Pasanen MJ, Mondo JA. Cost-effective, automated equipment for
advanced composite structure development and production. Evolving technologies for the
competitive edge 1997:48-55.
[4] Morey B. Automating composites fabrication. Manufacturing Engineering 2008:140(4).
[5] Hyer MW, Charette RF. Use of curvilinear fiber format in composite structure design. AIAA
Journal 1991;29(6):1011-5.
[6] Gürdal Z, Olmedo R. In-plane response of laminates with spatially varying fiber orientations:
variable stiffness concept. AIAA Journal 1993;31(4):751-8.
[7] Alhajahmad A, Abdalla MM, Gürdal Z. Optimal design of tow-placed fuselage panels for
maximum strength with buckling considerations. Journal of Aircraft 2010;47(3):775-82.
[8] Gürdal Z, Tatting BF, Wu CK. Variable stiffness composite panels: effects of stiffness
variation on the in-plane and buckling response. Composites Part A: Applied Science and
Manufacturing 2008;39(5):911-22.

33
[9] Setoodeh S, Abdalla MM, IJsselmuiden ST, Gürdal Z. Design of variable-stiffness composite
panels for maximum buckling load. Composite structures 2009;87(1):109-17.
[10] Akhavan H, Ribeiro P. Natural modes of vibration of variable stiffness composite laminates
with curvilinear fibers. Composite Structures 2011;93(11):3040-7.
[11] Honda S, Narita Y. Natural frequencies and vibration modes of laminated composite plates
reinforced with arbitrary curvilinear fiber shape paths. Journal of Sound and Vibration
2012;331(1):180-91.
[12] Akhavan H, Ribeiro P, De Moura MF. Large deflection and stresses in variable stiffness
composite laminates with curvilinear fibres. Int Journal of Mechanical Sciences 2013;73:14-
26.
[13] Khani A, IJsselmuiden ST, Abdalla MM, Gürdal Z. Design of variable stiffness panels for
maximum strength using lamination parameters. Composites Part B: Engineering
2011;42(3):546-52.
[14] Stanford BK, Jutte CV. Comparison of curvilinear stiffeners and tow steered composites for
aeroelastic tailoring of aircraft wings. Computers & Structures 2017;183:48-60.
[15] Khani A, Abdalla MM, Gürdal Z. Optimum tailoring of fibre-steered longitudinally
stiffened cylinders. Composite Structures 2015;122:343-51.
[16] Sabido A, Bahamonde L, Harik R, van Tooren MJ. Maturity assessment of the laminate
variable stiffness design process. Composite Structures 2017;160:804-12.
[17] Abdalla MM, Setoodeh S, Gürdal Z. Design of variable stiffness composite panels for
maximum fundamental frequency using lamination parameters. Composite structures
2007;81(2):283-91.
[18] Kameyama M, Fukunaga H. Optimum design of composite plate wings for aeroelastic
characteristics using lamination parameters. Computers & structures 2007;85(3):213-24.
[19] Olmedo R, Gurdal Z. Buckling response of laminates with spatially varying fiber
orientations. 34th Structures, Structural Dynamics and Materials Conference 1993:1567.
[20] Ribeiro P, Akhavan H. Non-linear vibrations of variable stiffness composite laminated
plates. Composite Structures 2012;94(8):2424-32.
[21] Lopes CS, Camanho PP, Gürdal Z, Tatting BF. Progressive failure analysis of tow-placed,
variable-stiffness composite panels. Int. Journal of Solids and Structures 2007;44(25):8493-
516.

34
[22] IJsselmuiden ST, Abdalla MM, Gürdal Z. Thermomechanical design optimization of
variable stiffness composite panels for buckling. Journal of Thermal Stresses
2010;33(10):977-92.
[23] Duran AV, Fasanella NA, Sundararaghavan V, Waas AM. Thermal buckling of composite
plates with spatial varying fiber orientations. Composite Structures 2015;124:228-35.
[24] Blom AW, Stickler PB, Gürdal Z. Optimization of a composite cylinder under bending by
tailoring stiffness properties in circumferential direction. Composites Part B: Engineering
2010;41(2):157-65.
[25] Rouhi M, Ghayoor H, Hoa SV, Hojjati M. Effect of structural parameters on design of
variable-stiffness composite cylinders made by fiber steering. Composite Structures
2014;118:472-81.
[26] Rouhi M, Ghayoor H, Hoa SV, Hojjati M. Multi-objective design optimization of variable
stiffness composite cylinders. Composites Part B: Engineering 2015;69:249-55.
[27] Ghayoor H, Rouhi M, Hoa SV, Hojjati M. Use of curvilinear fibers for improved bending-
induced buckling capacity of elliptical composite cylinders. Int. Journal of Solids and
Structures 2017;109:112–122.
[28] Goldfeld Y, Arbocz J. Buckling of laminated conical shells given the variations of the
stiffness coefficients. AIAA Journal 2004;42(3):642-9.
[29] Sun M, Hyer MW. Use of material tailoring to improve buckling capacity of elliptical
composite cylinders. AIAA Journal 2008;46(3):770-82.
[30] Khani A, Abdalla MM, Gürdal Z. Circumferential stiffness tailoring of general cross section
cylinders for maximum buckling load with strength constraints. Composite Structures
2012;94(9):2851-60.
[31] Rouhi M, Ghayoor H, Hoa SV, Hojjati M, Weaver PM. Stiffness tailoring of elliptical
composite cylinders for axial buckling performance. Composite Structures 2016;150:115-23.
[32] Wu CP, Lee CY. Differential quadrature solution for the free vibration analysis of laminated
conical shells with variable stiffness. Int. Journal of Mechanical Sciences 2001;43(8):1853-
69.
[33] Blom AW, Setoodeh S, Hol JM, Gürdal Z. Design of variable-stiffness conical shells for
maximum fundamental eigenfrequency. Computers & Structures 2008;86(9):870-8.

35
[34] Haddadpour H, Zamani Z. Curvilinear fiber optimization tools for aeroelastic design of
composite wings. Journal of Fluids and Structures 2012;33:180-90.
[35] Wu K, Gurdal Z. Thermal testing of tow-placed, variable stiffness panels. 19th AIAA
Applied Aerodynamics Conference 2001:1190.
[36] Abdalla MM, Gürdal Z, Abdelal GF. Thermomechanical response of variable stiffness
composite panels. Journal of Thermal Stresses 2008;32(1-2):187-208.
[37] White SC, Weaver PM, Wu KC. Post-buckling analyses of variable-stiffness composite
cylinders in axial compression. Composite Structures 2015;123:190-203.
[38] Ungwattanapanit T, Kamath S, Baier H. Gradient-Based Optimization of Postbuckled,
Steered-Fiber Aircraft Shell Using Equivalent Static Loads. 57th AIAA/ASCE/AHS/ASC
Structures, Structural Dynamics, and Materials Conference 2016:1967.
[39] Nik MA, Fayazbakhsh K, Pasini D, Lessard L. Surrogate-based multi-objective
optimization of a composite laminate with curvilinear fibers. Composite Structures
2012;94(8):2306-13.
[40] Blom AW, Lopes CS, Kromwijk PJ, Gurdal Z, Camanho PP. A theoretical model to study
the influence of tow-drop areas on the stiffness and strength of variable-stiffness laminates.
Journal of Composite Materials 2009;43(5):403-425.
[41] Akbarzadeh AH, Nik MA, Pasini D. The role of shear deformation in laminated plates with
curvilinear fiber paths and embedded defects. Composite Structures 2014;118:217-27.
[42] Akbarzadeh AH, Nik MA, Pasini D. Vibration responses and suppression of variable
stiffness laminates with optimally steered fibers and magnetostrictive layers. Composites Part
B: Engineering 2016;91:315-26.
[43] Tullu A, Kang BS. Elastic deformation of fiber-reinforced multi-layered composite
cylindrical shells of variable stiffness. Composites Part B: Engineering 2016;100:44-55.
[44] Tullu A, Ku TW, Kang BS. Elastic deformation of fiber-reinforced multi-layered composite
conical shell of variable stiffness. Composite Structures 2016;154:634-45.
[45] Sarvestani HY, Sarvestani MY. Interlaminar stress analysis of general composite laminates.
Int. Journal of Mechanical Sciences 2011;53(11):958-67.
[46] Sarvestani HY, Sarvestani MY. Free-edge stress analysis of general composite laminates
under extension, torsion and bending. Applied Mathematical Modelling 2012;36(4):1570-88.

36
[47] Yazdani Sarvestani H, Hoa SV, Hojjati M. Stress analysis of thick orthotropic cantilever
tubes under transverse loading. Advanced Composite Materials 2016:1-28.
[48] Yazdani Sarvestani H. Effects of Lay-up Sequences on Stresses of Thick Composite
Cantilever Tubes. Advanced Composite Materials 2017;26(3):273-294.
[49] Sarvestani HY, Hojjati M. Failure analysis of thick composite curved tubes. Composite
Structures 2017;160:1027-41.
[50] Sarvestani HY, Hojjati M. Three-dimensional stress analysis of orthotropic curved tubes-
part 2: laminate solution. European Journal of Mechanics-A/Solids 2016;60:339-58.
[51] Yazdani S, Ribeiro P, Rodrigues JD. A p-version layerwise model for large deflection of
composite plates with curvilinear fibres. Composite Structures 2014;108:181-90.
[52] Houmat A. Nonlinear free vibration of laminated composite rectangular plates with
curvilinear fibers. Composite Structures 2013;106:211-24.
[53] Tornabene F, Fantuzzi N, Bacciocchi M. Higher-order structural theories for the static
analysis of doubly-curved laminated composite panels reinforced by curvilinear fibers. Thin-
Walled Structures 2016;102:222-45.
[54] Blom AW. Structural performance of fiber-placed, variable-stiffness composite conical and
cylindrical shells. Delft University of Technology 2010.
[55] Fayazbakhsh K, Nik MA, Pasini D, Lessard L. Defect layer method to capture effect of gaps
and overlaps in variable stiffness laminates made by automated fiber placement. Composite
Structures 2013;97:245-51.
[56] Nosier A, Miri AK. Boundary-layer hygrothermal stresses in laminated, composite, circular,
cylindrical shell panels. Archive of Applied Mechanics 2010;80(4):413-40.
[57] Akbarzadeh AH, Chen ZT. Hygrothermal stresses in one-dimensional functionally graded
piezoelectric media in constant magnetic field. Composite Structures 2013;97:317-31.
[58] Akbarzadeh AH, Chen ZT. Magnetoelectroelastic behavior of rotating cylinders resting on
an elastic foundation under hygrothermal loading. Smart Materials and Structures
2012;21(12):125013.
[59] Loy CT, Lam KY. Vibration of cylindrical shells with ring support. Int. Journal of
Mechanical Sciences 1997;39(4):455-71.

37
 [60] Akbarzadeh AH, Pasini D. Multiphysics of multilayered and functionally graded cylinders
under prescribed hygrothermomagnetoelectromechanical loading. Journal of Applied
Mechanics 2014;81(4):041018.
[61] Shadmehri F, Hoa SV, Hojjati M. Buckling of conical composite shells. Composite
Structures 2012;94(2):787-92.

38

View publication stats