This action might not be possible to undo. Are you sure you want to continue?

Compurer~ & Struc~uresVol60. No 6.pp 1003-1012.1996 CopyrIght0 1996 Elsevm Swam Ltd Pnnted m Great Bntam Allnghts reserved 0045-7949196 $lSOO+OOO

FINITE

PURE BENDING

OF CURVED

PIPES

Dj. Boussaa,f,$

K. Dang Van,? P. Labbh# and H. T. Tang7

fLaboratoire de MBcanique et d’Acoustique, CNRS, 31, chemin Joseph Aiguler, 13402 Marseille cedex 20, France @ervice etudes et Projets Thermiques et NuclBaires, Electricitt de France, Villeurbanne, France T(Nuclear Power Division, Electric Power Research Institute, Palo Alto, CA, U.S.A

(Received 2 March 1995)

Abstract-We present an original treatment for the finite bending of curved pipes with arbitrary cross sections. The curved pipe is successively regarded as a three-dimensional continuum and a shell, and a formulation is proposed for each model. We show that, from a numerical point of view, the finite bending problem is reducible to an axisymmetric analysis augmented with 1 d.f. We also show how to take advantage of this analogy to solve the bending problem using standard axisymmetric FE routines. Copyright 0 1996 Elsevier Science Ltd

1. INTRODUCTION

pipes are among the most vulnerable components of piping systems. When compared to straight pipes of the same cross section and length, real-life curved pipes prove much more flexible and suffer much more severe stresses under the same bending moment. This contrast is due to the ovaling of the curved pipe cross section during bending, a phenomenon that thus cannot be accounted for by the usual beam theory. Even though the experimental discovery of the high flexibility of curved pipes is generally credited to Bantlin (1910), the above basic facts were actually known by the end of the 19th century. Piping systems of steamers were designed using beam theory, the extra flexibility of the curved parts being acounted for by multiplying their beam-stiffness by an empirical coefficient of 6/10, well-suited to the curved pipe geometries used at that time in naval engineering [l]. As regards modeling, the earliest conclusive contribution, in the small displacement setting, is due to von KinnAn (1911). Many attempts however preceded this work. In 1901, as is reported in its Curved Comptes Rendus de I’AcadPmie des Sciences (pp.1057-1058), the French Academy of Sciences

and Reissner [2] showed that a, for a wide range of geometries, is given by

(1)

where W is the radius of curvature of the curved pipe, b the radius of its cross section assumed to be circular, h its thickness assumed to be constant and v its Poisson’s ratio. In the large displacement setting, many researchers have paid attention to the problem ([4-71 and references therein). Except for the last reference, the treatment is purely analytical, involves many assumptions, and is based on Fourier series expansions which impose serious limitations (e.g. perfect geometries required) and suggests that more flexible solution techniques, such as the FE method, could prove better suited. Guidelines are given in Ref. [8] to treat the curvature controlled case using a standard axisymmetric FE analysis. A full formulation restricted to the small displacements case is developed in Ref. [9]. Moreover, it is shown in this reference that the FE dicretized pure bending problem may be regarded as an axisymmetric analysis augmented with 1 d.f. In some respects, our work may be inscribed as an extension of these two references, although we followed a different approach from those developed therein. The paper is organized as follows. The second section is devoted to the statment of the problem in the full three-dimensional framework. The similarities between this problem and an axisymmetric analysis are emphasized by a systematic decomposition of the kinematics of the bending problem into an axisymmetric part and additional terms. In the

acknowledged the contribution of naval engineer Marbec to justify the flexibility coefficient of 6/10. Always in the small displacement setting, Clark et al. [2, 31 proposed general formulae which are commonly used in the design of piping systems. An illustration is provided by the rigidity factor r]. Let l/EZ denote the beam flexibility of the curved pipe and l/@Z its effective flexibility, respectively. Clark $ To whom correspondence should be addressed.

CAS @II&K

1003

. with constant or varying thickness. respectively. As an illustration. followed by a brief conclusion. These modes are closer to those observed experimentally. and a normalized natural basis (ER. Remarks (1) No assumption was made on the cross section: the wall may be thin or thick.(O) + ZE. Accordingly. Fi Fig. In the large displacement setting. shown in Fig.. 9’p and Y$ the inner and outer skins. z) with a normalized natural basis (e. releasing a(t) adds only 1 d. denote the spatial derivative referred to the initial configuration. the assumption that cross sections remain plane is accounted for by the dependence of 0 on 0 solely. Notations Throughout the paper. (2) The kinematics.. the existence and uniqueness of the solution will ensure that the exhibited solution is the solution. y)~x. the curved pipe under study. (5) 1 where a(t) is a scalar function of time. imposing eqn (5) beforehand means that only rotationally symmetric instability modes can be accounted for. we present some numerical comparisons with existing results. the symbol @ will denote the standard tensor product of two vectors tensors ((x9 y)++x@y) or two second order ((X. . includes the usual assumptions for the in-plane bending problem. z = z(R.. (3) 2. e.2. The similarities between the present problem and an axisymmetric analysis are accounted for and the discretized formulation is shown to be reducible to an axisymmetric problem augmented with 1 d. the assumption that all the cross sections deform likewise in their plane by the independent of r and z of 8. Boussaa et al. (3) Fixing a(t) to zero in eqn (5) yields an axisymmetric evolution. 2. (5) Rigid body displacements compatibile with eqn (5) are translations along the Z-axis. . 9P0the radius (4) The pure in-plane bending of the curved pipe can be described conveniently with the following relations: r = r(R. we derive a matrix statement of the problem. by resorting to a FE discretization. the body deforms and a particle with initial position M occupies the new position m(r. 0. Z). Y)wX@Y). e. of curvature defined by third section we reduce this three-dimensional formulation to a shell formulation. 0.(B)+ze. THE CURVED PIPE AS A THREEDIMENSIONAL CONTINUUM Under external loads. z. In the fourth section.f. (4) In the small deformation setting. occupied by the matter and the hole. t). and M the position of a typical particle with cylindrical coordinates (R. 1 with the main notations. to those of this companion axisymmetric evolution. 2. Kinematics In its initial configuration assumed stress free. etc. E. eqn (5). 8 = (1 + a(t))@. of an axisymmetric body.1004 Dj.f. Let V. Ez): M = 0 + RE. is supposed to be a three-dimensional body occupying a sector Q. Finally.1. Y)i-+X *Y = trace(XYT)). t). A simple means to get rid of them is to fix the displacements along the Z-direction of one parallel. 1. The deformation gradient.): m=O+re. Cr’ and Cp the parts of Z. the dot the Euclidean inner product of two vectors ((x. the assumption of uniformity of the stretching of the parallels by the linearity between 0 and 0. General view and typical cross section of the curved pipe. the above assumptions will make it possible to exhibit one solution to the problem. & the bend angle of the curved pipe. y) or two second order tensors ((X. . Let Co denote a typical cross section. z.

Qe. They are related by S = JTF-lT = FP. (13) Differentiating eqn (19) with respect to time and choosing the current configuration as the reference . + rcWe.. the VVF associated with the axisymmetic companion transformation.3.. Note that the right-hand side of eqn (11) can be written as 6m = am.. (19) 6r = 6r(R. the gradient of a WF may be written as @E.. and Sa an arbitrary constant. Then the following relation holds: 6D = SD. To simplify the notations. that is If.OE..m@E. respectively. + f a. (8) The deformation follows: gradient F may be decomposed as 6L = V. denote the symmetric part of 6 L and 6L. we introduce 115) QE. Let Fz be the deformation gradient associated with the companion axisymmetric evolution.+&QE.. (20) where 6r and 6z are arbitrary scalar functions of R and Z. (17) F = (e. S and P.Qed. External loading The Cauchy.Smz. (6) V. i 62 = 6z(R. + r6Oe. (7) +6O~eB@E. all the kinematical quantities will be decomposed into axisymmetric and additional terms. eqn (14) is simplified to + 68@. (12) If the current configuration is considered as the reference configuration.Finite pure bending of curved pipes denoted by F. + &zD. respectively. it is the current configuration that is taken as the reference configuration. the determinant of F: J=(l +a)&. If the initial configuration is considered as the reference configuration. can also be D. + (1 + abe @Ee + e. first and second Piola-Kirchhoff stress tensors will be denoted by T.. (10) (18) By substituting the real velocity field rir for am. From now on.+~e. D. with the rotational symmetry conditions: (11) 2. (9) where A similar multiplicative decomposition applied to J. on the contrary.@e. = Vn.BER. we define in the same manner L.+~e.. where Jr is the determinant of Fx. The virtual velocity fields (VVF) compatible with the kinematics [eqn (5)] are of the form 6m = he.Sm = 6F Direct differentiation of eqn (4) with respect to its arguments and insertion of the results into eqn (6) yields 1005 with Sm. Z).+&JBE. (14) The expression of F-’ can be obtained from that of F by substituting in eqn (7) upper case letters to lower case letters and vice versa. @&)Fr .=e.e..m 0% + &m@E. 68 = aa@. + he. the relations.@E. and 6L.. reduce to T=S=P.. is by definition: F = V4m = d. Z). eqn (19). (16) Let 6 D and 6D.6m. La and Dx.

+ TooeoBe. Inserting eqns (21) and (28) into eqn (32) and applying the decomposition. In the case of curvature controlled loading. + &e. s4 (26) Equations (26).. dQt (32) OF. (19X (7) and the remark on F-’ that follows eqn (7) yield that at each point in the body: S = he.e. (Do GhW. - 6a sa D. and a prescribed moment in the case of moment controlled loading..5. Discussion of other types of constitutive equations and the corresponding implementations can be found in Ref. Oe. Weak form of equilibrium The local equilibrium equation written in the current configuration with zero body forces reads div.. to D and taking into account the equality between P and S . .e. Constitutive equations + S.dLdR.e. gives - J% c - S.6FdD+GaA$t=0. eqns (24) and (25). BE. BE.. + So.e. Inserting the boundary conditions. the moment is an unknown variable of the problem which needs to be calculated.6m d8R = 0. The forms of T and S are needed in the sequel.. + T. + Tzzez Oe. The rotational symmetry and the vanishing shear at both ends of the pipe imply that at each point in the body.@Fz 2.. which. T = T.) dR = 0. under the present context.)and C’“‘(0). .. and those (less common) prescribed on average [lo]. which are easily computable using standard FE routines. defined on the current configuration by where C is the standard constant elastic modulus tensor of the material which is assumed to be homogeneous and isotropic.4. Oez + T. = T.+ Tn.s. (28) the current configuration ation [ 121: as the reference configur- configuration The applied loads are of two types: those (typical in three-dimensional continuum descriptions) given pointwise. (31) Differentiating this equation with respect to time and then choosing the current configuration as the reference configuration yields the following rate equation: S. 2. into eqn (30) the vanishing of the normal force and the value of the moment [eqn (22)].aLrddn.. + &e.OJ&.1006 Dj. are the bending moment A’ and the normal force N. which can be proved by a direct application of the principle of virtual work with the (noncompatible) virtual velocity field re2e.. We assumed the following constitutive equation with In obtaining eqn (32) we made use of the following relation: KY. (29) The first type of boundary conditions in our problem is the vanishing of the stress vector on the outer and inner skins of the curved pipe and the shear force at the end cross sections: Tn t = 0 T. and writing the virtual work of the internal forces on the initial configuration. (24) (25) The boundary condition given on average corresponding to a pure bending is the vanishing normal force. (21) P=CD. [l 11. (27) The form of the constitutive equations is crucial for the quality of the predictions of the mode1 as well as for the numerical strategy to be implemented. We deliberately choose a simple form which has the advantage of involving only the standard axisymmetric small displacement stiffness matrix and the initial stress matrix. + S. (30) on Ynt and YeXr t t > on C’“‘(II/. leads to S=P+LP. Oe. T = 0. s an. Boussaa et al. = 0 Multiplying this equation by an arbitrary virtual velocity field and using the usual arguments leads to T.. eqn (17).

the outward normal to Y by n. Note that. Some notations.&bt Box I.1 o. are fields defined over the reference surface 9. A* =0. . (33) 3 -\ Fig.=hC and C. the spatial derivative defined over 9 by Vy and l as the third coordinate measured from 9’. (36) +6a. 0.(. On the other hand.-8 sB 6D. it is convenient to introduce C.OWD. Pnnciple of wrtual work. [13]. Moreover. which we have adopted.CD. consists in approximating C by the plane-stress elastic modulus tensor which is denoted hereafter by the same symbol. A last approximation. a 3. are treated first. by doing so. 2. The integrals defined over R are reduced to surface integrals by the following approximation on c. Kinematical fields D.pn.P*6L. where (.dn.). Lx and their virtual counterparts 6D. since the current configuration is chosen as a reference. CD. THE CURVED PIPE AS A SHELL (71 i. OP)D. The equality 6. dQ. This contrasts with the usual formulations in which Sm is entirely taken as a shell velocity field.. The notations to be used are defined in Fig. 2. is approximated {4) -2d by j-$)dQ. The derivation of a two-dimensional model from a three-dimensional one is achieved by reducing the three-dimensional fields and integrals defined over 0 to surface fields and integrals. -6a -26a s s n. n. and 6L. we define the tensor c as: p=l+SV. etc.) [ 141. yields the expression of the principle of virtual work given in Box 1. This approximation is compatible with many models (Koiter. In the shell framework. However. (37) 3.2. and not 6m taken as a whole is a shell velocity field (say of Kirchhoff-Love type. ) is any suitable argument.Finite pure bending of curved pipes 1007 SS- int n ext S- shall denote the reference surface by 9.. =j=~~. Portion of the cross section. This can be done in several ways.~(D. respectively. sB Dz. det(p) = 1. {3} where y and x.=gC. Donnel. D=u+<x (34) o= - s s 4 4 6 D.e. 3. We have adopted the Kirchhoff-Love approximation.fh:!2 ( . Assumptions We will now develop a shell model for our problem. (35) D.1.2 E 0. d!Z& {II L. One approach consists of developing a formulation ab initio stated in a two-dimensional framework.CD0dQ. respectively the membrane strain rate and the curvature rate tensors. these fields are usually reduced to affine functions of l. 6D. An alternative approach.dQt (D.dQ. the form of eqn (34) is common to many shell models starting from a three-dimensional description. this amounts to assuming that am.)dQ~=I.I Wp) dt dY. We the membrane and flexural stiffnesses.~2()dt d9’. classical in shell theory. consists in starting from the three-dimensional formulation in Box 1 and reducing it to a shell formulation. and owing to eqn (13) and its consequences. This latter assumption gives rise to coupling terms between the longitudinal and hoop directions in the strain rate tensor which are usually neglected without justification as discussed in Ref.oP 0 2 4 6 0 ___I 10 Fig.

C.2h (s 9. Releasing a adds 1 d. da. dY (1 (38) -6 fY h. that is. (45) With the hypothesis det(p) = 1. dY. -26a YX.. (YO @~)Yo dY. . (~0 63&o yr. ~2. are the components of the velocity vector of the ith node. the restriction of D. Equation (34) will make it possible to write SD.. be the force vector . (D. .d.. This reduces Term (3) to (3) = -6a s 9. = f4 + s yt where N= s 11. hr .2. the problem is reduced to an axisymetrical analysis. . A discretized version of the shell formulation in Box 2 can be obtained in the same manner. dR. GYO dY. 9 (40) 4.yOdY-ci I 9. ~YX GY. and develop a numerical solution procedure based on FE techniques for spatial discretization. it is easy to prove that D. (~0@N)YodY. If we designate by y. We give in the sequel the discretized version of the full three-dimensional formulation in Box 1. (YO B@ro dY. /’ (43) where ii and i. one has to approximate volume integrals in Terms (3. = YZ. 4. be the virtual counterpart of V. the integral of the right-hand side of eqn (43) is the volume Y=‘.f. in terms of surface fields. 4 and 6} by surface integrals.44). (41) The second term of (4) is likewise computable. FINITE ELEMENT SOLUTION PROCEDURE h/r and R= I -h/2 5Pdt. Spatial discretization We have mentioned that. (YO @@Y. the cross section is discretized into finite elements. We once again assume that we can solve the companion axisymmetric problem and will focus our attention on the additional terms 3-7 in Box 1. Term (7) is to be kept unchanged. f&y. (YO @N)YodY. . To make them shell terms. > (42) Vi={ . I Yl Box 2. }. i. GYO @t o= “axisymmetrical” terms -6a f 9. The following notations are needed. Finally. Let 6V. dY.7 Pdl -h/2 YZ. If moreover.1. the set of unknowns reduce to the set of mesh nodes velocities. the set of unknowns of the problem is V. WI Term {5) is to be kept in its original form in Box 1..i . formu- Putting Terms {3}-{6} together yields the formulation in Box 2. .. b}. and F. dY. CD. to the reference surface Y’. Boussaa er al. (~0 @~)Yo dYt > ~YZ.. Accordingly.. .. if ri and 6a are assumed to be zero. augmented by d. xx . represented as usual by a vector + I yt xz. + f 9. and then the kinematical unknowns are the velocities of the particles at any cross section. @P)D. So we have (4) = -26a (s 91 YZ .. Replacing C by its value in the term {6} yields (5) = -6ah (A ly...1008 3. . VT = (Vi. i o1 s y. Shell version of formulation m Box I. A shell version of the three-dimensional lation Dj. and D. (39) Resorting to eqn (34) once again yields D.

‘X.K. Term {6} reads {lo} = -G&(1 + 2jl)“yWo’.. (48) We consider curvature controlled loading. Physically. . -K. be defined as follows: P. (54) (52) Term (5) is to be kept in its present form: (5) = Gad”&. 161.=ciu. Solving eqn (57) in this case produces the foIlowing solution: These vectors will be used to restate Terms (l}-{7} in Box 1. be a fictitious displacement associated with E. This scheme yields the displacement increment AU resulting from a force increment AQ. Then (53) Thus (4) = -2F...2.SV. + h6Vx). 4.42=cif. (47) (58) (59) (60) X=F.. (50) Y -XTK. is well known and is given by K.(&zV. 1009 Putting Terms {l}-(7) together. Term {1) now reads (1) = -V. Term (3) reads (3) = -F. By analogy with this equation. AU = [Kq-‘(R’ + Term (4) can be evaluated in the same way as F.)D.: &=K. Y = (a + 2/l)Y”o1..f@aV. as follows [17]: step 1: step 2: R’=Q’-KU’. (56) t55) where 1 and k are the Lame constants.. s n. in a purely axisymmetric evolution: the vector F. (62) (63) stiff- where I<! is the standard axisymmetric small displacement stiffness matrix.dR. where B.SV. Term (2) can be rewritten as {2} = -V$K.+2F. + (i6Vx). and taking into account the arbitrariness of 6V yields a system of the form where The expression for F.‘F. Let P.‘X...=K’+K”. 4 appears to be the instantaneous ness to bending of the curved pipe..=(P. Time dkcretization where Kg is the standard initial stress matrix [15.D..Finite pure bending of curved pipes associated with the “initial stress” CD.. is the strain-displacement matrix for a purely axisymmetric transformation Jd.&.= x= v. FO= gCD. 01) A one-iteration incremental scheme with very small increments is implemented for time-discretization. Let U. . (49) 1 with U. is such that for every velocity field 6 Vx .

(73) is. + e. and E. which is multiplied by this value (1 + a). ’ 0. Aa = 3 = constant IL. since these not rotate in a purely axisymmetric evoleqn (70) in eqn (69). E. I* = 0. i.d a Fig. .@e. T.e..OE. The moment-rotation Reissner. Boussaa et al. + E. (68) 0 2 4 6 6 Using eqns (9) and (IO). About the numerical implementation In this section. corresponds to the axisymmetrical push-forward.1. NUMERICAL RESULTS 5. . then the value of a.. is expressed in the base generated by product of E.). according I-k(p*+a)(p*+2a) > . reads 5. (74) The term (1/Jz Fx PF:). the global push-forward of the second Piola-Kirchhoff stress tensor amounts to an axisymmetric push-forward followed by the division by (1 + a) of all the terms. + a)Tc@esBe. we obtain T$YeZ6e.+ (1 +a)e. 4.@Ez The tensor the tensor vectors do ution.+ + Tge. after n steps of calculations. 4. = > x (&BE.Oe. The second point is a comment on the time discretization of Li. a. = *o IF PF.e. + T. Inserting + TgE.. we restrict ourselves to the discussion of two points related to the grafting of the numerical procedure described above onto standard FE codes. except for the component e. A* = 1.e.OE. denoted by T. = (1 + Aa)n$. (70) m=- Jq/qiT) 7ch2b E ’ (75) to T. The first is concerned with the “pushforward” of the second Piola-Kirchhoff stress tensor between two consecutive configurations..0 - 0 2 4 ’ 6 8 a (72) Fig. and the extra terms to opening or closing of the curved pipe. (69) 9.OE. This operation is to be achieved through T = f FPFT.BE. Let m be a nondimensional moment defined by 6E.. + (1 + a)E. ’ ( J. i. (76) T= & (T.5.E..1010 Dj. In terms of components.e. If the increment Aa is chosen to be a constant between two consecutive steps. given by m=a curve is then.+ T.3. referred to the initial configuration. given by a. 5. this relation may be rewritten as T=1 lfa x (e.Oe.@e. Existing results Reissner’s formula for finite bending [6] may be summed up as follows.Oe.@e.)+(l (71) In short.

I* = 5. Using a thin shell theory developed by Reissner. Boyle [7] proposed a combined analytical-numerical formulation. When the push-forward procedure is skipped. we have also added the results that were obtained by skipping the push-forward procedure of the second Piola-Kirchhoff stress tensor. In these figures are also reported the results obtained by Boyle.1)) through T = CE. Accordingly.8 1. 9.2. in which case the second Piola-Kirchhoff and the Cauchy stress tensors are identical.6 E 0.0 r 0. I* = 10. Fig. Fig. the limit points tend to occur for lower values of the parameter a. I* = 20. if need be. It seems to us that in this reference. 7. However. Comparison with existing results Figures 3-9 show the relations between m and a we obtained by the shell version of our formulation.Finite pure bending of curved pipes 0. The motivation for the introduciton of this approximation is the way the behavior is formulated in [6]. the constitutive equation amounts to relating directly Cauchy stress the tensor T to the Green-Lagrange strain tensor (E = 1/2(FTF . one can note weak formulation against strong formulation. A* = 2.2 I- 1011 0. In this respect.6 E 0. our formulation could be better compared to the existing ones by adopting the same constitutive equations. 6. in addition to the strong dissimilarity between the two approaches to establish the governing equations. In the latter. . equilibrium was written in a strong form: the system of partial differential equations which express this equilibrium was reduced to a system of first order nonlinear differential equations which is solved by means of the multi-point shooting method. and those corresponding to eqn (76). and. 1.0 0 1 2 a 3 4 I 1 5 0.8 0. Among the differences. 8. our approach is notably different. a more realistic modeling of the bending problem should include material nonlinearities. where (77) provided b c<W and A* = Wh/b* > d-12. On these figures. Accordingly. the various approaches give very similar results. with the complete version of our algorithm.0 ii 0 1 2 a 3 4 Fig. nonlinear effects due to internal pressure (follower forces). Note that the coordinates of the calculated limit points are outside the range of elasticity of real life curved pipes. 5. and FE against multi-point shooting method.4 J 1 2 a 3 4 J 3 a Fig.6 1 0.3 0.

(1977). magement sous disme. Trans. Mech. Struct. Maidenhead (1977). L. Congress on Applied Mechanics (1956). 17. L. APronaut. K. Paris (1990). PhD thesis. Appl.1012 6. Une Prksentation de la MPthode afes l?lkments Finis. 33. Technical Report UCSESM 74-4. Mech. 5. Flexion plane du tuyau coude et endom12. Marit. Tech. from a numerical point of view. Numer. 93-122 (1951). Mech. J. P. Dhatt and G. Reissner. A. 2nd edn (1984). Bull.459-463 (1973). Boussaa. L. Trans.Reissner. Ericksen. 891-897 (1984). Wilson. Axisymmetric finite element analysis for pure moment loading of curved beams and pipe bends. Greenbaum. Bending of curved tubes Adu. Bull. Mod&lisation des Coques Minces &astiques. R. 643-680 (1928). Elastic tubes. Stress and deformations of toroidal shell problems. REFERENCES 11. In: Proc. an original formulation for the in-plane bending of curved tubes including geometrical nonlinearities has been proposed. A. 22. CA (1974). E. Comput. Thuloup. M. Ozdemir and E. Cook. 1. Dahl. 2. CONCLUSION Dj. Appl. C. 0: C. Englewood Cliffs. Destuynder. Bovle. Adv. H. 13. Finite Element ‘Pro&dares in Engineering Analysis. 2. J. Meth. Reissner.8. . Rev. 17. A. 386-392 (1959). Flexibilite des tubes. Special topics in elastostatics. McGraw Hill. Int. 19. 9th Int. appl. A. A. Hofmeister and D. Bathe. gl. 3. The Finite Element Method. ASME J. The influence of pressure on the bending of curved tubes. Solirls Skct. 37. AUDI. Marit. The finite bending of curved pines. 441-457 (1911). Marbec. J. Evensen. Masson. R. I. 7. Axelrad and F. Berkeley. i%ole National des Ponts et Chat&es (1992). 10. 515-529 (1981). J. Dj. Prentice-Hall. 483-487 (1989). . J. NJ (1982). . J. It has been shown that. 9. 37-48 (1952). Zienkiewicz. In this paper. Clark and E. Emmerling. Mech. H. G. Essai sur la fatigue des tuyaux minces a fibre moyenne plane ou gauche. Gilroy and E. 16. T. 4. Clark. L. Bathe. Engng 5. 32. ht. Touzot. R. Static and dynamic-geometric and material nonlinear analysis. the bending problem could be reduced to an axisymmetric problem augmented with 1 d. S. 3rd edn. Boussaa et al. Pure moment loading of axisymmetric finite element models. T. Ass. K. D. 15. University of Cahfornia. Appl. This similarity has been turned into account for the numerical treatment of the problem of bending using standard axisymmetric FE routines. Ass. Crandall and N. 14. On finite bending of pressurised tubes.f. Mech. G. A. Tech. 6. ASME J. E.

Sign up to vote on this title

UsefulNot useful- Principal Stresses
- Rotations in Computational Solid Mechanics
- me185_midterm2_spring04
- Top School in Noida
- Cauchy Stress Modeling
- Free Video Lectures for Mba
- Solid Mechanics-9.pdf
- Simple Stresses
- Mirzaei-ElasticityLecture
- Convection Chap1
- Web Class 4b Final
- Meric Tensor and Strains Notes
- A Self-Consistent Mechanics of Composite Materials
- Damage Effective Str
- Numerical Computation Slides3
- Stress
- Stress (Mechanics) - Wikipedia, The Free Encyclopediaq
- Solid Mech
- Lecture 27
- 735512191 (3)
- H.L. Duan, Y. Jiao, X. Yi, Z.P. Huang, J. Wang.pdf
- LectureslidesBlock1 Start Up and Solution
- 10.1.1.6.7968
- honoursthesis_jamesmbewu
- Solid Mechanic
- Unified Explicit Basis-free Expressions for Time Rate and Conjugate Stress of an Arbitrary Hill's Strain
- composite materials
- Public
- Index_2009_Elasticity-Second-Edition-.pdf
- Mechanics of Materials and composite_ UTP.docx
- Boussaa DangVan Computor Structure 1996