Nonlinear First Ply Failure Response of Composite Cylindrical Shells Under Non-Uniform Transverse Pressure
Nonlinear First Ply Failure Response of Composite Cylindrical Shells Under Non-Uniform Transverse Pressure
Nonlinear First Ply Failure Response of Composite Cylindrical Shells Under Non-Uniform Transverse Pressure
Abstract - Use of laminated composite materials in place of conventional materials in weight sensitive disciplines of engineering
including structural engineering started from the second half of the twentieth century. Confident use of composite structural units
necessitates comprehensive understanding of the material behaviour including failure characteristics which the researchers are focusing
on in the recent papers. Failure in composite materials may initiate from the surface and from inner laminae as well which may remain
undetected. Unnoticed flaw may progress leading to overall failure of the structural unit. This paper intends to explore the first ply
failure characteristics of shells and takes up the industrially popular cylindrical configuration having wide industrial applications about
which only a very few papers report the failure related information. Failure of clamped graphite epoxy shells is studied using Sanders’
linear strains together with von-Kármán nonlinear strain components. The shells subjected to nonuniform sinusoidal is modeled using
isoparametric Serendipity element having five degrees of freedom at each node. The paper reports the gross failure behavior of a
number of shell options with varying lamination and curvature. The results are presented systematically for lucid understanding and
are interpreted from practical engineering standpoints. The paper concludes with pin pointed guidelines based on which the different
shell combinations may be taken up for relative performance study .
Keywords: Nonlinear analysis, first ply failure, cylindrical shells, non-uniform pressures, composite material
1. Introduction
The hunt in the search of advanced structural materials in the second half of the last century resulted in introduction of
laminated composites in different engineering branches including civil engineering. The use of these materials having high
specific strength and stiffness properties does on only help to make the building frames economical but also lead to
reduced foundation costs as both dead and seismic loads on the structures are reduced drastically. Cylindrical shell are
popular in the industry as these singly ruled units offer ease of fabrication. Though the laminated composites have high
specific strength and stiffness, high fatigue strength, capacity of being assembled fast, less susceptibility to thermal
expansion and less vulnerability to weathering action and moisture but the failure of laminated composite initiate from an
inner lamina. Such a failure may remain unnoticed and may cause propagation of internal damage leading to gross failure.
Hence the load at which the failure initiates (first ply failure) in the laminate has to be known to the end user.
Failure study of laminated composites is an active area of research. Progressive damage of composite plates due to
bending was studied by Ferreira et al. [1] .They used a FORTRAN based code and studied the variation of different
stress parameters considering sinusoidal surface pressure. Ghosh and Chakravorty [2] reported failure initiation of
laminated hypar shell roofs of antisymmetric cross and angle ply laminations considering different boundary conditions.
They used nonlinear strains and the recently proposed Puck’s criterion of failure together with serviceability failure in
terms of permissible deflection limits. Nonlinear first ply failure behaviour of shallow thin composite conoidal shells
subjected to central point load was reported by Bakshi and Chakravorty [3] . Kumar et al. [4] worked on the failure loads
of laminated composite and sandwich cylindrical shells using finite element model based on higher order zig-zag theory.
They considered different loading and boundary conditions . Well established failure theories were employed by Kumar
and Srivastava [5] studied first ply failure load of cross ply stiffened plates subjected to uniformly distributed and
ICSECT XXX-1
sinusoidal load . Prusty [6] reported the load of failure initiation for laminated stiffened and unstiffened panels under
central concentrated load, uniformly distributed and sinusoidal load considering maximum stress, maximum strain,
Hoffman, Tsai-Wu and Yeh-Stratton failure criteria. Bakshi and Chakravorty [7] developed a finite element code to predict
the first ply failure load of simply supported thin laminated composite conoidal shell subjected to uniformly distributed
loading.
From the literature survey it is evident that research on first ply failure of composite cylindrical shell under non
uniform load using the nonlinear strains has not received due attention.
Only one recent paper [1] reported failure of composites under non-uniform transverse loading but that too for plates.
This paper reports nonlinear first ply failure characteristics of cross ply clamped composite cylindrical shells subjected to
sinusoidal loading to partially fulfill the above mentioned lacuna. The paper of Kam et al. [8] is used for validating the
present approach with the published ones.
The present isoparametric finite element formulation uses eight noded doubly curved elements with C 0 continuity.
Five degrees of freedom u, v, w, αx and αy (Fig. 1) are set at each node of the element. The strain displacement relation of
cylindrical shell may be expressed as a combination of mid-surface strains and curvatures which is given below.
The above equation can be expressed in the three-dimensional field of strain problem as,
ICSECT XXX-2
{ }{ }
εinplane ε 'inplane
T
{ ε }={ ε x ε y γ xy κ x κ y κ xy γ xz γ yz } = ε bending + 0
εshear 0
[ [ ]]
¿ [ B0 ] + 0.5 B ( { de } ) { d e } = { εlineaar } + { ε nonlineaar } (2)
where, the in-plane, bending and shear strain components are expressed as,
{ ( )} { )}
T T
{ε inplane } =
∂u
∂x
∂v w
+
∂y R yy )( ∂u ∂v
+
∂y ∂x
,{ ε bending } =
∂αx
∂x
∂α y
∂y ( ∂ αx ∂ αy
∂y
+
∂x
,
{( )}
T
{ε shear } =
∂w
∂x
+ αx )( ∂w
∂y
+ αy
'
Finally, the nonlinear components of in-plane strains {ε inplane } are defined as,
{( )}
T
) ( ) ( )(
2 2
{ε '
inplane } = 12 ∂w
∂x
1 ∂w v
+
2 ∂y R yy
∂w
∂x
∂w v
+
∂y R yy
(3)
In Eq. (2), [ B0 ] is the linear part while [ B ] is the nonlinear part of the strain-displacement matrix and is a function of
nodal displacements {de}. Thus, for geometric nonlinear analysis resultant strain-displacement matrix [ B ] is given as,
[ ][ ]
T
∂w ∂w v ∂ Ni
0 + 8 0 0 0 0
∂x ∂y R yy ∂x
[ B ] = [ B0 ] + [ B ] = [ B0 ] + [ A ][ G ] = [ B0 ] +
∂w v ∂w
∑ Ni ∂ Ni
i=1
0 + 0 0 0
∂y R yy ∂x R yy ∂y
(4)
where, Ni and Ryy denote the shape function at ith node and radius of curvature along the y axis of the shell respectively.
To maintain the equilibrium condition, the virtual work done by all forces applied on the system must be zero. The internal
force over the domain of the system is
¿ (5)
Externally generated force {Pn} on the system by the applied force is expressed as ,
ICSECT XXX-3
8 ❑
{ Pn }=∑ ∫ [ N i ] { F } dA
T
i=1 A
(7)
T
where applied force { F }={ p x p y p z m x m y } in which p x , p y , p z are the applied non-uniform pressure along x , y and
z directions respectively and m x , m y are the applied moments per unit area along x , y axes respectively. Except p z, all other
components are assumed to be zero for the present investigation and p z denotes the applied non-uniform sinusoidal
transverse pressure expressed as,
πx πy
p z= p0 sin sin , (8)
a b
where p0 is the peak transverse surface pressure of laminated composite cylindrical shell surface. As per the theorem of
virtual work,
¿ (9)
Here { r }is the residual force in case of nonlinear equilibrium equation which is to be minimised to get improved
displacement. The detailed expression for tangent stiffness matrix [ k ]t is given as,
❑ ❑ ❑ ❑
[ k ]t =∫ [ B 0 ]T [ E ] [ B0 ] dA+∫ [ B 0 ]T [ E ][ B ] dA +∫ [ B ]T [ E ] [ B 0 ] dA+∫ [ B ] T [ E ][ B ] dA
A A A A
[]
∂Ni
[ ] [ ]
❑
∂ Ni ∂Ni N x N xy ∂ x dA
+∫ (10)
A ∂x ∂y i =1 ,2 … ,8 N xy N y ∂Ni
∂y
Here, Nx, Ny and Nxy represent the components of normal stress resultant vector and the expressions of which are the
same as it were reported earlier by Ghosh and Chakravorty [*]. The laminate elasticity matrix [E] and proper shear
correction factors are adopted from Ghosh and Chakravorty [*] while obtaining the shear stress resultants.
The tangent and secant stiffness matrices and external and internal force vectors are calculated by numerical
integration technique using 2×2 Gauss quadrature rule. Global stiffness matrices and force vectors are obtained by
assembling the element matrices with proper transformations due to the curved geometry of the shell [ghosh, Chakravorty
tws paper number]. The convergence of the Newton – Raphson iterative process is checked using the guidelines proposed
by Chatterjee et al [**]. The converged displacements of the cylindrical shell are used to obtain the lamina stresses and
strains [Chatterjee et al tws]. The first ply failure pressures are obtained by applying those stresses and strains in well-
known failure theories like maximum stress, maximum strain, Tsai-Hill, Tsai-Wu, Hoffman, Hashin and Puck failure
criteria. The failure theories are taken up here from Chatterjee et al. [**].
ICSECT XXX-4
of partially clamped composite square plate under concentrated loading are compared with the values reported by earlier
researchers Kam et al. [*] and the values obtained from present nonlinear finite element code in Table *. In this benchmark
problem, both the experimental and analytical failure load values are compared with the failure loads calculated from the
present analysis.
Once the accuracy of the proposed finite element code is confirmed, it is further used to study the first ply failure
response of laminated composite cylindrical shell panels under non-uniform sinusoidal transverse pressure with *******
boundary condition. In this present study, the authors adopted the graphite-epoxy composite material which was used by
Kam et al. [**] and the properties are furnished in Table 2. Several symmetric and antisymmetric stacking orders of
composite skew plates like ************************ are considered here. The developed stresses and strains under
applied sinusoidal pressures are continuously checked with different well-established failure criteria mentioned in the
previous section to get the first ply failure pressure values. Among all these first ply failure pressure values coming from
different failure criteria, the minimum values are considered as the governing first ply failure pressures for various
cylindrical shell options. The results of the present analysis are reported and discussed systematically by various tables and
figures in the following paragraphs.
Table 1: Nonlinear first ply failure load (kN) for graphite epoxy ( 0 ° 2 /90 ° )s square plate
Failure criteria Experimental first ply Analytical first ply failure load
failure load Kam et al. [**] Present
Kam et al. [**]
Maximum stress 0.1476 0.1400
Maximum strain 0.1853 0.1941
0.1573
Hoffman 0.1431 0.1382
Tsai-Wu 0.1578 0.1517
Tsai-Hill 0.1444 0.1521
ICSECT XXX-5
pressure with increasing curvature is an interesting point of study and it is found that introduction of curvature increases
the load resistance capacity no doubt, but when R/a becomes less than 0.75, the load capacity suffers a decline (Refer Fig.
##). Among the class of shells taken up here R/a = 0.75 may be recommended.
4. Conclusion
The following conclusions are evident from the present study.
References
The IEEE citation format is used. Books and book chapters should be referenced as [1] and [2] respectively. Patents
are referenced based on [3] and a thesis can be referenced as [4]. Finally, conference presentations/papers and journal
papers need to be reference based on [5] and [6] respectively.
With the increasing availability of useful information that can be found on the internet, website references must also
be reported based on [7]. Meanwhile, due to the dynamic nature of web pages and the fact that in most cases the
information is not peer-reviewed, the use of published resources are very much preferred and advised over online
references.
The reference section at the end of the paper should be edited based on the following:
[1] B. Klaus and P. Horn, Robot Vision. Cambridge, MA: MIT Press, 1986.
ICSECT XXX-6
[2] L. Stein, “Random patterns,” in Computers and You, J. S. Brake, Ed. New York: Wiley, 1994, pp. 55-70.
[3] J. P. Wilkinson, “Nonlinear resonant circuit devices,” U.S. Patent 3 624 125, July 16, 1990.
[4] J. O. Williams, “Narrow-band analyzer,” Ph.D. dissertation, Dept. Elect. Eng., Harvard Univ., Cambridge, MA, 2000.
[5] U. V. Koc and K. R. Liu, “Discrete-cosine/sine-transform based motion estimation,” in Proceedings of the IEEE
International Conference on Image Processing, Austin, TX, 1994, vol. 3, pp. 771-775.
[6] R. E. Kalman, “New results in linear filtering and prediction theory,” J. Basic Eng., vol. 83, no. 4, pp. 95-108, 1961.
[7] K. Author. (2015, May 10). Facility Greenhouse Gas Reporting (2nd ed.) [Online]. Available:
http://www.ec.gc.ca/ges-ghg/default.asp?lang=En&n=040E378D-1
ICSECT XXX-7