Documente Academic
Documente Profesional
Documente Cultură
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/254155477
CITATIONS READS
2 88
5 authors, including:
Some of the authors of this publication are also working on these related projects:
Predicting of viscoelastic damaged behavior of a multilayer composite tubular structure View project
All content following this page was uploaded by Bezazi Abderrezak on 12 February 2015.
where Ur, U , and Uz are radial, circumferential, and axial As for the equilibrium equations in cylindrical
displacements respectively. coordinates, they take the following form [12]
r 1 r zr r (k ) (k )
(k ) (k ) (k )
+ + + =0 (4a)
r r z r
r 1 z 2 r
(k ) (k ) (k ) (k )
+ + + =0 (4b)
r r z r
zr 1 z z zr ( k )
(k ) (k ) (k )
+ + + =0 (4c)
r r z r
where r , , z
(k ) (k ) (k )
are radial, circumferential, and
axial stress vectors of the k th layer respectively; z
(k )
,
zr , r are shear stress in the plane z-, z-r, and -r
(k ) (k )
U r ( ) ( k ) 1 U ( ) U r ( ) ( k ) U z ( )
k k k k
dU ( ) U ( ) dU ( )
k k k
r( k ) = ; = + ; z = (3a) zr ( k ) = 0; r ( k ) = ; z( ) =
k
= 0 r (6b)
r r r z dr r dz
(k ) (k )
1 U z U
z ( k ) = + (3b) where 0 has physical interpretation - twist of the pipe per
r z
unit length.
U z
(k )
U r
(k ) These relations are based on the hypothesis that
zr ( k ) = + (3c) axial strains of all layers are equal to a constant 0 , i.e.
r z
z = 0 and that rotation of the cylinder z is independ-
U
(k ) (k )
1 U r ent of z [10].
r ( k ) = +r (3d)
r r r
The two equations (5b) and (5c) can be solved
according to the following form
2.2. Behaviour of the metal liner ties are equivalent in the plane y-v
'ij = T ij
(11a)
'ij = T ij
where 'ij and 'ij are stress and strain vectors expressed
in the base of Qi(r, , z); ij and ij are stress and strain
vectors expressed in the base of Qi (x, y, z); T and T are
matrixes of the base change of the stress and the strain
respectively [14]. Since vectorial form of the stresses and
strains, the matrices of basic changes are expressed, where
Fig. 2 Coordinate relation between cylindrical reference w is the winding angle about z .
and the reference mark of the fibre T and T are expressed respectively as follows
cos 2 sin 2 0 0 0
sin cos
sin 2 cos 2 0 0 0 sin cos
0 0 1 0 0 0
T = (11c)
0 0 0 cos sin 0
0 0 0 sin cos 0
2 sin cos 2 sin cos cos sin
2 2
0 0 0
Rigidity matrix of the composite is written in the the continuity and volume conservation and the conditions
following form [14] of loading on the other hand. It is supposed that there are
no slips in the interfaces and that there are continuity of
C11 C12 C13 0 0 C16 stresses and displacements. These boundary conditions
C C 22 C 23 0 0 C 26 allow determining the integration constants.
12 The number of unknown factors, or integration
C C 23 C 33 0 0 C 36 (12)
C c = 13 constant of the system, to be solved is 2(N+1) for N layers
0 0 0 C 44 C 45 0
of the tube composite/liner; where D(k), J(k), 0 , and 0
0 0 0 C 45 C 55 0
for k [1, N ] .
C16 C 26 C 36 0 0 C 66
The radii Rint ( ) and Rext
k
(k )
are introduced for
2.4. Problem position each layer k and it is noted that
(k ) (k ) (k ) (k )
C12 C13 0 C 26 2 C36
= (k )
+ 0 (13)
C33( ) The condition of continuity of radial stress results in
k
C33 r
tion (13) takes the form: The equilibrium condition of axial force due internal pres-
if ( k ) = 1 , then sure with end loading effect
N rk
U r ( ) = D( ) r + J ( ) / r 2 z ( r ) r dr = r02 p0 + F
k k k (k )
(14a) (19)
k =1 rk 1
if ( k ) 1
where F is the applied axial load.
(k ) (k )
Torque balance is
U r ( ) = D ( ) r + J ( ) r + 1( ) 0 r + 2( ) 0 r 2 (14b)
k k k k k
N rk
2 (r ) r
2
dr = M (20)
where k =1 rk 1
z
(k ) (k ) (k ) (k )
C c12 C c13 C 2C c 36 where M is the applied torque.
1( k ) = (k ) (k )
, 2 = c 26 ( k )
(k )
(k )
(15)
C c 33 C c 22 4C c 33 C c 22
The hypothesis of this study neglects torque and axial
loads, where F = 0 and M = 0. Thus one has 2(N+1) equa-
here D(k) and J(k) are integration constants. tions to identify the whole integration constants.
Boundary conditions on one hand are imposed by Analytical procedure of elastic behaviour resolu-
geometry conditions of the structure, since one supposes tion of a multi-layer tube reinforced by a metal liner under
15
pressure is solved using MATLAB software. This analysis The first type sequence [n/902n]. The variation
of a comparative study enters both types of laminate: the of axial stress z from the internal wall to the external one
quasi-isotropic [n/902n], and anti-symmetric [2n], is obtained on three stages for the laminates Seq1 to Seq3
where n=1. The two types of laminates are coated on a (Fig. 3): these three stages are limited from the ratio R, the
metal liner. The first type of stacking sequences [n/902n] first from 0 and 0.4, the second from 0.4 to 0.8, and the
is marked as: Seq1, Seq2, and Seq3. The second type third from 0.8 to 1.0. No noticeable difference is in the
[2n] is marked as: Seq4, Seq5, and Seq6. Table 1 pre- first as well as in the third stage and the values of the stress
sents the stacking sequences of the different studied lami- z in these stages reaches 545 and 90 MPa respectively.
nates, where the order of stacking angle of each laminate is The second stage is characterized by big difference of z
taken from interior to external. The tube is characterised values. Indeed, the stress z decreses to reach 420 MPa for
by an internal radius of 50 mm, 0.5 mm thickness of the the laminate Seq3, however it increases to reach 930 and
liner and as well as each layer of the composite. The prop- 710 MPa for the Seq1 and Seq2 respectively. Moreover,
erties of two materials are presented in Table 2. the second stage is characterised by the maximum stress
for Seq1 and Seq3 by the minimum stress.
Table 1
Stacking sequences of the tube under pressure
Sequence types Angle of wrap
Seq1 Liner/[ 30/(90)2]
Seq2 Liner/[ 45/(90)2]
Seq3 Liner/[ 60/(90)2]
Seq4 Liner/[ 30]2
Seq5 Liner/[ 45]2
Seq6 Liner/[ 60]2
Table 2
Elastic characteristics
Properties Carbon/epoxy Liner alu-
(T300/934) minium
Ex , GPa 141.6 69.5
Ey , GPa 10.7 69.5
Fig. 3 Axial distribution of stress z versus ratio R
Gxy, GPa 3.88 26.7
yx 0.268 0.3
The second type sequence [2n]. For the lami-
zy 0.495 -
nates Seq4, Seq5, and Seq6 axial stress z versus ratio R is
Internal wall of the composite tube is subjected to internal characterised by the stages and it is significantly lower
pressure of 10 MPa. All results are represented as a func- comparatively to the first type of laminates. In the first
tion of non-dimensional ratio R, which is expressed as stage (R from 0 to 0.4) the laminate Seq4 has the lowest z
fallowed by Seq5 and finally Seq6 having the values 207,
r r0 295 and 414 MPa respectively. However, the phenomena
R= (21) adverse in the stage 3 (R from 0.6 to 0.8) and the values of
ra r0 z for the Seq4, Seq5, and Seq6 are 440, 360 and 220 MPa
respectively. There is no noticeable difference of z in the
where ra is the external radius of the multi-layer tube. stage 2 (R from 0.4 to 0.6) and stage 3 (R from 0.6 to 0.8).
4. Results and discussions 4.1.2. Circumferential stress
4.1. Stresses
The distribution of circumferential stress for the
The distribution of the stresses z, , r and z two types of stacking sequences versus non-dimensional
along thickness is important in order to locate the critical ratio R is shown in Fig. 4. This distribution is in three
zones. The obtained results show the presence of compres- stages for the first type of the staking and four stages in the
sion stresses on the internal wall and tension stress on the second one. Moreover, the same remark of the discontinu-
external wall of the composite tube. ous variation in Fig. 3 is recorded for this stress. The val-
ues of the stress for the first type of stacking are much
4.1.1. Axial stress lower than for the second type in the internal wall (R=0).
The first type sequence [n/902n]. The variations
The variation of axial stress z for the two types of of are characterised by an increase from the internal
stacking sequences versus non-dimensional ratio R is wall to the external one and occur in three stages. The
shown in Fig. 3. Both types of sequences present a discon- laminate Seq1 is characterised by the largest followed
tinuous variation of z along thickness and the change of by Seq2 and finally Seq3 in the first stage (R from 0 to 0.4)
the fibres orientation from one layer to another influences as well as in the third one (R from 0.8 to 1) reaching the
the stress behaviour. The obtained results of stress z show maximum. However, in the second stage the phenomena
a large difference for the two types of stacking sequence. are in adverse with the laminate Seq3 which has the largest
16
stress values followed by Seq2 and finally Seq1. In addi- thickness of entire composite [15] is totally different com-
tion, the minimal stress (8 MPa) is recorded for the lami- pared to the results obtained for liner/composite tube.
nate Seq3 in the first stage and the maximal one (405 Mpa) Fig. 6 shows the distribution of radial stress r through the
for the Seq1 in the third stage. thickness of composite/liner tubes. The behaviour of radial
The second type sequence [2n]. The laminate stress through the thickness indicates the presence of
Seq4 has the maximum stress (240 MPa) in the internal maximal compression of -10 MPa in the internal wall and
wall, e.g. in the first stage (R from 0 to 0.4) and decreases minimal (equal to zero) in the external wall for all the stud-
in the second and the third but increase slightly in the forth ied laminates. In addition, this distribution of stresses from
stage. However, the laminates Seq5 and Seq6 have the the internal wall to the external one has a linear behaviour
minimum stresses (300 and 270 MPa respectively) and with an increase of the slope. The curves of radial stress r
increase to reach the maximum in the second stage. More- for the laminates of the first type [n/902n] (Seq1, Seq2,
over, the minimum stress of 60 MPa is obtained for the and Seq3) are below than the laminates of the second type
laminate Seq6 in the first stage and the maximum of 300 [2n] (Seq4, Seq5 and Seq6), e.g. the stress r is more
MPa in the second stage (R from 0.4o 0.6) of the same important for the laminates of the first type than for the
laminate. second ones. Indeed the Seq4 has the lowest stress r and
the Seq1 has the highest one.
nate Seq4 has the lowest values of the stress z in the nate Seq4 has the greatest values of the circumferential
thickness for all the stages. strain and Seq3 has the lowest one. In addition Seq2 and
Seq6 are really too close.
deformations are discussed there. The variation of all the 10. Xia, M., Takayanagi, H., Kemmochi, K. Analysis of
discussed parameters are represented through the thickness multi-layered filament -wound composite pipes under
of the composite tube and according to a non-dimensional internal pressure. -Composites Structures, 2001, v.53,
ratio R. The analysis of the obtained results makes it possi- p.483-491.
ble to release that: 11. Xia, M., Takayanagi, H., Kemmochi, K. Analysis of
- the laminate Seq3 liner/[60/(90)2] for stacking filament-wound reinforced sandwich pipe under com-
sequence [n/902n] has the best results, in terms of bined internal pressure and thermomechanical loading.
stresses and displacements comparatively to the two other Composite Structures, 2001, v.51, p.273-283.
laminates Seq1 and Seq2. The same remark is recorded for 12. Germain, P, Muller, P. Introduction la mcanique
Seq6 liner/[(60)2] comparatively to Seq6 and Seq5 for the des milieux continus. -Paris, Milan, Barcelone: Mas-
staking sequence [2n]; son, 1995.-467p.
- the stress ratio /z is constant and it varies 13. Timoshenko S., Woinowsky-Krieger S. Theory of
from a sequence to another according to the angle of orien- Plates and Shells.-New York: McGraw-Hill Book
tation of each layer. In this investigation the stress ratio Company, Inc., 1959.-580p.
indicates that axial stresses are much more important than 14. Berthelot, J.-M. Composite Materials. Mechanical
circumferential ones; Behavior and Structural Analysis.-New York:
- the metal liner produces a remarkable fall of Springer, 1999.-676p.
stresses and displacements, although it accounts for 1/5 of 15. Hocine, A., Boubakar, L., Bezazi, A., Benamar, A.
the total thickness of the tank; Analyse dun rservoir multicouche de stockage
- the assumption of heating effects responsibility dhydrogne.-Workshop sur lhydrogne WIH22005 -
is essential. This point will be the future object of the CDER, Alger 2005, p.[1-8].
works. A thermo mechanical comparative study enters
quasi-isotropic liner/[(60)/(90)2] and the anti-symmetrical
liner/[(60)2] will be established. A. Hocine, A. Bezazi, L. Boubakar, A. Benamar,
A. Kondratas
6. References
DAUGIASLUOKSNIS VAMZDINIS KOMPOZITAS SU
1. Bezazi, A., El Mahi, A., Berthelot, J.-M., Bezzazi, B. KLOTU: ELGSENOS, VEIKIANT SLGIUI,
Flexural fatigue behaviour of cross-ply laminates -An MODELIAVIMAS
experimental approach. -Strength of Materials, 2003,
v.35, No2, p.149-161. Rezium
2. Bezazi, A., El Mahi, A., Berthelot, J.-M., Bezzazi, B.
Analyse de lendommagement des stratifis en flexion Darbe pateikta daugiasluoksnio vamzdio (t. y.
3-points. Influence de la squence dempilement. cilindrinio bako vidurins dalies), armuoto metaliniu klo-
XVme Congrs Franais de Mcanique, Nancy, 2001, tu, tempim ir deformacij, veikiant vidiniam ir ioriniam
p.[1-6] slgiui, analiz. Tamprios struktros elgsenai analizuoti
3. Liang, C.C., Chen, H.W. and Wang, C.H. Optimum buvo taikomas analitinis modeliavimo metodas. Darbo
design of dome contour for filament-wound composite tikslas buvo itirti atitinkam kompozito sluoksni suda-
pressure vessels based on a shape factor. -Composite rymo eilikum, kad bt galima padidinti bako atsparum
Structure, 2002, v.58, p.469-482. didesniems slgiams. Buvo modeliuoti du kompozito
4. Verijenco, V.E., Adali, S., Tabakov, Y.P. Stress dis- sluoksni sudarymo ant kloto eilikumo tipai: antisimetri-
tribution in continuously heterogeneous thick lami- nis ir kvaziizotropinis. Itirti cilindrinio bako vidurins
nated pressure vessels. -Composite Structures, 2001, dalies ainiai, apskritiminiai, radialiniai ir lyties tempiai
v.54, p.371-377. bei deformacijos tomis paiomis kryptimis ir tempi san-
5. Vasiliev, VV., Krinakov, A.A., Razin, A.F. New gen- tykis bako sienels storio atvilgiu. Be to, buvo tirtas pa-
eration of filament-wound composite pressure vessels rametr kitimas bako sienels storio, ireikto santykine
for commercial applications. -Composite Structure, radialine koordinate, atvilgiu. Gauti rezultatai parod, kad
2003, v.62, p.449-459. anksiau minti parametrai kinta priklausomai nuo kompo-
6. Wild, P.M. Analysis of filament-wound cylindrical zit sluoksni sudarymo ant kloto eilikumo ir nuo
shells under combined centrifugal, pressure and axial sluoksni pluot orientacijos.
loading. -Composites, Part A, 1997, v.28 A, p.47-55.
7. Lifshitz, J.M., Dayan, H. Filamentwound pressure
vessel with thick metal liner. -Composite Structures, A. Hocine, A. Bezazi, L. Boubakar, A. Benamar,
1995, v.32, p.313-323. A. Kondratas
8. Hocine, A., Bezazi, A., Benamar, A. Analysis of the
influence of stacking sequence on the mechanical be- MULTILAYER TUBULAR COMPOSITE
haviour of composite fuel tanks.-JNC14 -2005, Com- REINFORCED BY A LINER: BEHAVIOUR UNDER
pigne, Published by Benzaggah M.L. and Lamon J., PRESSURE LOADING SIMULATION
p.641-649.
9. Rosenow, M.W.K. Wind angle effects in glass fiber- Summary
reinforced polyster filament wound pipes. -Composites,
1984, v.15, p.144-52. This paper presents an analysis of stresses and
19