Beruflich Dokumente
Kultur Dokumente
net/publication/309428026
CITATIONS READS
0 241
3 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Kaustav Bakshi on 26 October 2016.
Abstract: Composite cylindrical shells are stiff surfaces with simple curved geometry and are extensively used to build large column
free areas in shopping malls, airports and car parking lots with reasonably less material consumption. Laminated composites gained
popularity in civil engineering structures as use of these materials results in reduced mass and mass induced forces like seismic
forces. Failure study of these materials is necessary, which includes the load value (first ply failure load) at which failure initiates, the
mode of failure and the failure propagation location. The present article aims to study first ply failure of uniformly loaded simply
supported cylindrical shells using finite element method. An eight noded curved quadratic isoparametric shell element is used to
develop the finite element program and validated through solution of benchmark problems. Well accepted failure criteria are used to
evaluate the failure loads and failure modes. .
Key Words: Failure modes, Finite element method, First ply failure loads, Laminated composite cylindrical shell.
where ‘v’ represents shell volume and ‘A’ shell area. Ex- K B D B dxdy
e
T
and
q 0 0 qz 0 0
T
(4)
Q N q dxdy
e
T
2.2 Finite element formulation The element stiffness matrix [Ke] and load vector {Qe}
are transformed to isoparametric coordinates ξ and h for
An eight noded curved quadratic isoparametric element numerical integration by 2×2 Gauss quadrature rule.
with C0 continuity is considered in the present study to Global stiffness matrix and load vector are obtained by
formulate the bending stiffness of the cylindrical shell. assembling the element matrices with proper transforma-
Five degrees of freedom that are considered include three tions due to the curved geometry of the shell and they are
displacements and two transverse rotations. expressed as,
The strain displacement matrix [B], laminate elasticity K d Q (9)
matrix [D] and the cubical shape functions used in the
present study are those as were reported by Chakravorty where
et al. (1995). ne
K K e
and
The strain energy of the shell is expressed as ne
1 Q Q e
d B D B d dxdy
T T
U1 e e (5) i 1
2 A
and the work done is expressed as, 2.4 Lamina stress calculation
V d e N q dxdy
T
(6) Generalized laminate midplane strains are evaluated using
A the strain displacement relationship. Inplane strain com-
To minimize the total potential energy of the shell with ponents for a lamina situated at a distance ‘z’ from the
respect to its deformations, the shell has to satisfy the lamina midplane are evaluated in global axes as,
following condition,
x x zk x
0
y y zk y
0
0 (7) (10)
d e
xy xy zk xy
0
Figure 1. Cylindrical shell. Figure 2. General doubly curved laminated composite shell element.
128
Table 2
Comparison of nondimensional downward
1 m 2mn x
2 2
n displacements of composite cylindrical shell
2
2 n 2mn y
4
m (11) Static displacement Static displacement
Lamination Exact
reported by Qatu and from present
mn mn n 2 (Degree) solution
6 xy Algothani (1994) formulation
129
Table 5
Uniformly distributed failure load of simply supported cylindrical shells of radius 1000 mm for different laminates
Location (x,y)
Lamination (degree) Failure theory Failure load (N/mm2) First failed ply Failure mode/failure tendency
(m,m)
Maximum stress 0.2725 (0,0) 2 Shear failure of the matrix
Maximum strain 0.2725 (0,0) 2 Shear failure of the matrix
L
0/90 Hoffman 0.2718 (0,0) 2 Shear failure of the matrix
Tsai-Hill 0.2725 (0,0) 2 Shear failure of the matrix
L
Tsai-Wu 0.2718 (0,0) 2 Shear failure of the matrix
Maximum stress 0.2961 (0,0) 3 Shear failure of the matrix
Maximum strain 0.2961 (0,0) 3 Shear failure of the matrix
0/90/0 Hoffman 0.2942L (0,0) 3 Shear failure of the matrix
Tsai-Hill 0.2961 (0,0) 3 Shear failure of the matrix
L
Tsai-Wu 0.2942 (0,0) 3 Shear failure of the matrix
Maximum stress 0.2897 (0,0) 4 Shear failure of the matrix
Maximum strain 0.2897 (0,0) 4 Shear failure of the matrix
0/90/0/90 Hoffman 0.2888L (0,0) 4 Shear failure of the matrix
Tsai-Hill 0.2897 (0,0) 4 Shear failure of the matrix
L
Tsai-Wu 0.2888 (0,0) 4 Shear failure of the matrix
Maximum stress 0.2999 (1,0) 4 Shear failure of the matrix
Maximum strain 0.2999 (1,0) 4 Shear failure of the matrix
L
0/90/90/0 Hoffman 0.2979 (1,0) 4 Shear failure of the matrix
Tsai-Hill 0.2999 (1,0) 4 Shear failure of the matrix
L
Tsai-Wu 0.2979 (1,0) 4 Shear failure of the matrix
Maximum stress 0.2280 (1,0) 2 Transverse matrix cracking
Maximum strain 0.1659L (1,0) 2 Transverse matrix cracking
45/-45 Hoffman 0.2154 (0,1) 2 Transverse matrix cracking
Tsai-Hill 0.2173 (0,1) 2 Transverse matrix cracking
Tsai-Wu 0.1973 (0,0) 2 Transverse matrix cracking
Maximum stress 0.2981 (0,0) 3 Transverse matrix cracking
L
Maximum strain 0.2159 (0,0) 3 Transverse matrix cracking
45/-45/45 Hoffman 0.2808 (0,0) 3 Transverse matrix cracking
Tsai-Hill 0.2836 (0,0) 3 Transverse matrix cracking
Tsai-Wu 0.2570 (0,0) 3 Transverse matrix cracking
Maximum stress 0.2857 (1,0) 4 Transverse matrix cracking
L
Maximum strain 0.2070 (1,0) 4 Transverse matrix cracking
45/-45/45/-45 Hoffman 0.2692 (1,0) 4 Transverse matrix cracking
Tsai-Hill 0.2718 (1,0) 4 Transverse matrix cracking
Tsai-Wu 0.2464 (1,0) 4 Transverse matrix cracking
Maximum stress 0.3256 (0,0) 4 Transverse matrix cracking
L
Maximum strain 0.2353 (0,0) 4 Transverse matrix cracking
45/-45/-45/45 Hoffman 0.3064 (0,0) 4 Transverse matrix cracking
Tsai-Hill 0.3095 (0,0) 4 Transverse matrix cracking
Tsai-Wu 0.2803 (0,0) 3 Transverse matrix cracking
Inplane degree of freedoms along the boundaries of the ment is assigned a high value (1030) to make them effec-
plate was released to model the partially clamped bound- tively zero to model a plate with no curvature.
ary condition. The material properties of the plate are Apart from solving the bench-mark problem for verify-
presented in Tables 3 and geometric properties are pre- ing the finite element code proposed here, authors solve a
sented as the footnote of the Tables 4. The radius of the number of cylindrical laminated composite shells under
principle curvature and cross curvature of the present ele- uniformly distributed pressure, with different laminations,
130
Table 6 supported boundary conditions under uniformly distrib-
Geometric dimensions of the cylindrical shell uted load. Infact the cross ply lamination showing the
Cylindrical shell dimensions Values least failure load (0/90) can support a superimposed
Length (a) 1000 mm pressure, more than what is obtained for the stiffest angle
ply shells (45/-45/45/-45) by about 13.4%.
Width (b) 1000 mm
Although the failure load values for cross and angle ply
Thickness (h) 10 mm cylindrical shell are markedly different, but one trend is
Radius(along X axis) Infinite common for both types of laminations. This is the fact
Radius(along Y axis) 1000 mm that the symmetric laminations always perform better than
the antisymmetric ones.
Among the cross-ply laminates, the 0/90/90/0
stacking orders and curvatures. These practical parametric
stacking sequence yields the maximum failure pressure.
variations include both angle and cross ply lamination of
For a cylindrical shell, symmetrically supported along all
both antisymmetric and symmetric stacking orders. The
the four edges, the loads and moments are transferred
curvature is varied so that the shell configuration is al-
mainly along the two plan directions. For cross-ply lami-
ways shallow (Rise/Span ratio less than 0.2). The results
nates, the on axis stiffness of the individual lamina play
of failure pressures obtained from the numerical experi-
an important role in resisting the load because they too are
mentation are presented in Table 5. Plies are started to be
aligned parallel to the direction of load transfer. This is
numbered from the top of the laminate i.e. the topmost ply
why the cross-ply shells are significantly better in per-
is numbered one and bottommost ply has the last ply
formance than the angle ply ones and for four layered
number. Material properties of the graphite-epoxy com-
cross-ply stacking sequence, (0/90/90/0) equal thick-
posite to fabricate the cylindrical shell are presented in
Table 3 and its geometric dimensions are furnished in ness of 0 and 90 laminae along the shell cross-sections
Table 6. cause a balanced load transfer mechanism along the plan
Table 1 shows good agreement of the present results direction and the failure load shows an improved value. A
with the established ones and this validate the cylindrical 90/0/0/90 shell, on the other hand gives a failure load
shell formulation. Table 3 also exhibits a very good agree- of 0.2571 N/mm2 which is 86.3% of the failure load for a
ment between present results and published values which 0/90/90/0 shell. The cylindrical shell configuration
validates the present first ply failure formulation. considered in the present study has curvature in y-
The results furnished in Table 5 shows that in all the direction and it is singly ruled in x-direction with no cur-
cases of cross ply laminate, the Hoffman and Tsai-Wu vature. This particular shape by virtue of its geometry
criteria yield the lowest value of failure pressure. It may only, has an enhanced stiffness along the arch or y-
be noted that the Hoffman and Tsai-Wu criteria converge direction where the bending and axial stiffness work to-
to the same condition when the transverse stress matrix gether to resist the load. Along the x-axis on the other
vanishes. For the cross-ply laminates under uniformly hand there is no such coupling and the load transfer is
distributed load, the plan direction of major load transfer almost like that of a plate. A 0/90/90/0 lamination has
and the orientations of the fibers are identical and hence the 0 fibers stiffening the x-direction and being away
hardly any transverse stress develops in the matrix. This is from the mid plane, renders adequate bending inertia to
why the two above mentioned criteria converge to give the shell. This brings about a more coMPatible balance of
the same result. Interestingly, on the other hand, in all the stiffness along the two plan directions and the failure load
cases of angle ply laminates the maximum strain criterion reaches the peak. In contrast to this, for 90/0/0/90
indicates the design failure loads. On these failure pres- shell the fiber stiffening the beam direction are more to-
sures coming from different criteria for cross and angle wards the midplane and contribute less significantly to the
ply laminates, the factor of safety should be applied, to bending inertia and the failure occurs for a much lower
obtain the working pressure value. value of the superimposed load indicating a failure along
The five failure criteria taken up here give comparable the beam direction.
results for cross ply shells although the Hoffman and Tsai
-Wu criteria consistently yield the minimum value. Con- 5. Conclusion
trary to this, the angle ply laminates the failure load val-
ues obtained from the criteria a part from maximum strain In all the cases of cross ply laminate, the Hoffman and
criterion are often quite high, when compared with the Tsai-Wu criteria yield the lowest value of failure pressure.
minimum pressure value. This also indicates that for cross Interestingly, on the other hand, in all the cases of angle
ply laminates, all the stresses and strains contributing in ply laminates , the maximum strain criterion indicates the
the failure criteria, increase in magnitude simultaneously design failure loads. This also indicates that for cross ply
as the superimposed load is increased. This indicates a laminates, all the stresses and strains contributing in the
more efficient utilization of material strength and this is failure criteria, increase in magnitude simultaneously as
why the cross ply surfaces are stiffer. the superimposed load is increased. This indicates a more
Another point which strikes designers’ attention is that, efficient utilization of material strength and this is why
the failure loads for angle ply laminations are remarkably the cross ply surfaces are stiffer. Although the failure load
less than that what are observed for cross ply ones. This values for cross and angle ply cylindrical shell are mark-
leads to the natural inference that for a given quantity of edly different, but one trend is common for both types of
material consumption, cross ply cylindrical shell should laminations. This is the fact that the symmetric lamina-
always be preferred than the angle ply ones for simply tions always perform better than the antisymmetric ones.
131
A 0/90/90/0 lamination has the 0 fibers stiffening the 7. Acknowledgement
x-direction and being away from the mid plane, renders
adequate bending inertia to the shell. This brings about a The second author gratefully acknowledges the financial
more compatible balance of stiffness along the two plan assistance of Council of Scientific and Industrial Research
directions and the failure load reaches the peak. The fail- (India) through the Senior Research Fellowship vide
ure modes/tendencies of the shell taken up here, it is ob- Grant no. 09/096 (0686) 2k11-EMR-I.
served in all the cases, the failure occurs at the corner of
the shell where shear is critical and at the bottommost References
lamina.
Akhras, G. and Li, W.C. (2007). Progressive failure analysis of thick
composite plates using the spline finite strip method. Composite
6. Notations Structures, Vol. 79, pp. 34-43.
Chakravorty, D., Sinha, P.K. and Bandyopadhyay, J.N. (1995). Finite
A Area of the shell. element free vibration analysis of point supported laminated compos-
a and b Length and width of shell in plan respectively. ite cylindrical shells. Journal of Sound and Vibration, Vol. 181,
D Flexural rigidity matrix of the laminate. No 1, pp. 43-52.
{d} Nodal displacements of the shell. Ganesan, R. and Liu, D.Y. (2008). Progressive failure and post buckling
E11, E22, Elastic moduli along the directions 1 and 2 of a lamina response of tapered composite plates under uni-axial compression.
respectively. Composite Structures, Vol. 82, pp. 159-176.
1,2 and 3 Local co-ordinates of a lamina respectively. Kam, T.Y. and Jan, T.B. (1995). First ply failure analysis of laminated
G12 Shear moduli of a lamina in 1-2planeof a lamina. composite plates based on the layerwise linear displacement theory.
ne Number of finite elements in the shell domain. Composite Structures, Vol. 32, pp. 583-591.
T Shear strength of a lamina in its 1-2 plane. Kam, T.Y. and Sher, H.F. (1995). Nonlinear and first-ply failure analy-
Tε Allowable shear strain of a lamina in its 1-2 plane. sis of laminated composite cross-ply plates. Journal of Composite
{u} Generalized displacement vector. Materials, Vol. 29, pp. 463-482.
w Nondimensional transverse displacement of shell. Kam, T.Y., Sher, H.F. and Chao, T.N. (1996). Predictions of deflection
=[wE22h3/(qa4)] and first-ply failure load of thin laminated composite plates via the
XT and XC Normal strengths of a lamina along the fiber direction in finite element approach. International journal of Solids and Struc-
tension and compression respectively. tures, Vol. 33, No 3, pp. 375-398.
XεT and XεC Allowable normal strains of a lamina along the fiber Prusty, B.G., Ray, C. and Satsangi, S.K. (2001). First ply failure analysis
direction in tension and compression respectively. of stiffened panels-a finite element approach. Composite Structures,
YT and YC Normal strengths of the matrix in tension and compres- Vol. 51, pp. 73-81.
sion respectively. Prusty, B.G., Satsangi, S.K. and Ray, C. (2001). Firstply failure analysis
YεT and YεC Allowable normal strains of the matrix in tension and of laminated panels under transverse loading. Journal of Reinforced
compression respectively. Plastics and Composites, Vol. 20, No 8, pp. 671-684.
x, y and z Global Cartesian co-ordinates of the shell. Qatu, M.S. and Algothani. (1994). A bending analysis of laminated
ex,ey Inplane normal strains along x and y axes respectively. plates and shells by different methods. Computers and Structures,
e1,e2 Inplane normal strains along 1 and 2 axes of a lamina Vol. 52, No 3, pp. 529-539.
respectively. Reddy, J.N. and Pandey, A.K. (1987). A first ply failure analysis of
ε6 Inplane shear strain in 1-2 plane of a lamina. composite laminates. Computers and Structures, Vol. 25, No 3, pp.
gxy Inplane shear strain in x-y plane of the shell. 371-393.
nij Poisson’s ratio which characterizes compressive strain Singh, S.B. and Kumar, A. (1998). Postbuckling response and failure of
along xj direction produced by a tensile strain applied in symmetric laminates under inplane shear. Composites Science and
xi direction. Technology, Vol. 58, pp. 1949-1960.
s1, s2 Inplane normal stresses along 1 and 2 axes of a lamina Turvey, G.J. (1980). An initial flexural failure analysis of symmetrically
respectively. laminated cross-ply rectangular plates. International Journal of Sol-
s6 Inplane shear stress in 1-2 plane of a lamina. ids Structures, Vol. 16, pp. 451-463.
κx, κy, κxy Curvature changes of the shell due to loading. Vinson, J.R. and Sierakowski, R.L. (2002). The Behavior of Structures
Composed of Composite Materials. 2nd ed., New York: Kluwer
Academic Publishers.
132