2nd WSEAS Int. Conf. On NON-LINEAR ANALYSIS, NON-LINEAR SYSTEMS and CHAOS (WSEAS NOLASC 2003) Vouliagmeni, Athens, Greece, December 29-31, 2003 Hybrid Finite Element in Non-Linear Structural Dynamics of Anisotropic Tubes Conveying Axial Flow M.H. Toorani1 and A.A. Lakis2 1. Nuclear Engineering Department, Babcock & Wilcox Canada P.O. Box 310, Cambridge, Ontario, Canada N1R 5V3 2. Mechanical Engineering Department, Ecole Polytechnique of Montreal C.P. 6079, Station Centre-ville, Montreal (QC), Canada H3C 3A7 Tel.:(514) 340-4711, x4906, Fax: (514) 340-4176 Abstract: A semi-analytical approach has been developed in the present theory to determine the geometrical non-linearity effects on the natural frequencies of anisotropic cylindrical tubes conveying axial flow. Particular important in this study is to obtain the natural frequencies of the coupled system of the fluidstructure, taking the geometrical non-linearity of the structure into consideration and estimating the critical flow velocity at which the structure loses its stability. The displacement functions, mass and stiffness matrices, linear and non-linear ones, of the structure are obtained by exact analytical integration over a hybrid element developed in this work. Linear potential flow theory is applied to describe the fluid effect that leads to the inertial, centrifugal and Coriolis forces. Stability of tubes subjected to the flowing fluid is also discussed. Numerical results are given and compared with those of experiment and other theories to demonstrate the practical application of the present method. Key-Words: Non-Linear, Dynamics, Flowing Fluid, Anisotropic, Cylindrical Tubes 1 instability can be illustrated as a feedback mechanism between structural motion and the resulting fluid forces. A small structural displacement due to fluid forces or whatever alerts the flow pattern, inducing a change in the fluid forces; this in turn leads to further displacement, and so on. When the flow velocity becomes higher, the vibration amplitude becomes larger and the impact phenomenon occurs that can lead to unacceptable tube damage due to fatigue and /or fretting wear in critical process equipment. Therefore, the evaluation of complex vibrational behavior of these structures is highly desirable in nuclear industry to avoid such problems. Introduction Component failures due to excessive flowinduced vibrations continue to affect the performance and reliability of nuclear components, piping system and tube heat exchangers. Fluid-elastic vibrations have been recognized as a major cause of failure in shelland-tube-type heat exchangers. Fluid-elastic vibrations result from coupling between fluidinduced dynamic forces and the motion of structure. Depending on the boundary conditions, static (buckling) and dynamic (flutter) instabilities are possible in the structures at sufficiently high flow velocities. The nature of fluid-elastic 2 Corresponding author 1 2.1 While several studies have been conducted on the dynamic stability of circular cylindrical shells conveying fluid, but in contrast, non-linear studies of shells subjected to either internal or external axial flow are few. Particularly interesting, in case of internal axial flow, are the studies of Païdoussis &Dennis [1], Lakis & Païdoussis [2, 3], Weaver & Unny [4], Pettigrew [5], Selmane &Lakis [6] and Amabili et al. [7]. These works are performed based on the classical shell theory by neglecting the shear deformation effect while this later plays a very important role in reducing the effective flexural stiffness of composite shells and also the moderately thick structures. The present work addresses the question of stability of anisotropic cylindrical shells, based on a shearable shell theory, subjected to internal and external axial flow. The non-linearities due to large amplitude shell motion are taken into account, by using the modal coefficient approach, while the amplitude of shell displacement remains within the linear range from the fluid point of view. Structure Model The equations of motion of anisotropic cylindrical shells in terms of U, V and W (axial, tangential and radial displacements of the mean surface of the shell), x and (the rotations of the normal about the coordinates of the reference surface), see Fig. 1, and in terms of Pij’s elements are written as follows: Lm U , V , W , x , , Pij 0. i, j 1,2,3,...,10 (1) where anisotropic elasticity matrix (Pij’s elements), and five linear differential operators (Lm), are fully given in [8]. The finite element developed is shown in Fig. 1. It is a cylindrical panel segment defined by two nodal lines i and j. Each node has five degrees of freedom, three displacements and two rotations. 1 1 2 m 1,..., 5 2 m Mathematical Model The analytical solution involves the following steps: a) The strain-displacement relations expressed in an arbitrary orthogonal curvilinear coordinate are inserted into the equations of motion, obtained based on shearable shell theory, of anisotropic cylindrical shells. The mass and linear stiffness matrices are determined for an empty finite element, Fig. 1, and assemble the matrices for the complete shell. (A) i j N Wmi 1mi Vmi i 2mi Umi m j (B) Figure 1: (A) Finite element discretization. (B) Nodal displacements at node i. The displacement functions associated with the axial wave number are assumed to be: b) The coefficients of the modal equations are derived using the non-linear part of the kinematics relations. m m xe ; x ( x, ) DCos xe L L m m (2) V ( x, ) BSin xe ; ( x, ) ESin xe L L m W ( x, ) CSin xe L U ( x, ) ACos c) A finite fluid element bounded by two nodal lines, Fig. 1-B, is considered to account the effect of the fluid on the structure. d) The linear and non-linear natural vibration frequencies are then obtained and compared with the available results. where m is the axial mode, and is a complex number. A system of five homogeneous linear functions is obtained by substituting (2) into equations of motion (1). For the solution to be non-trivial, the determinant of this system must be 2 equation (6). The global matrices [Ms] and [Ks(L)] may be obtained, respectively, by superimposing the mass and stiffness matrices for each individual panel finite element [9]. equal to zero. This brings us to the following characteristic equation (see References [8, 9] for more detail): Det H f 10 10 f 8 8 f 6 6 f 4 4 f 2 2 f 0 (3) 2.1.2 Non-Linear Stiffness Matrices of Structure The exact Green strain-displacement relations are used in order to describe the non-linear behavior, including large displacements and large rotations, of anisotropic cylindrical shells. In common with linear theory, it is based on refined shell theory in which the shear deformations and rotary inertia effects are taken into account. The approach developed by Radwan and Genin [10] is used with particular attention to geometric non-linearities. Each roots of this equation yields a solution to the linear equations of motion (1). The complete solution is obtained by adding the ten solutions independently. After carrying out the some intermediate manipulations, that are not displayed here (see Reference [9]), the following equations are obtained : U ( x, ) V ( x, ) i W ( x, ) N j ( x, ) x ( x, ) (4) The coefficients of the modal equations are obtained through the Lagrange method. Thus, the non-linear stiffness matrices of second and third order are then calculated by precise analytical integration and superimposed on the linear part of equations to establish the non-linear modal equations. The main steps of this method are as follow: where is the displacement vector at the boundaries and [N] represents the displacement function matrix. The constitutive relation between the stress and deformation vector of cylindrical shells is given as [9]: N xx N x Qxx N Nx Q M xx M x M M x P B i j 2.1.2.a Shell displacements are expressed as generalized product of coordinate sums and spatial functions: (5) The matrices [P], as a function of geometrical and mechanical parameters of anisotropic cylindrical shells, and [B] are given in [9]. u qi (t ) U i x, x qi (t ) xi x, i i v qi (t ) Vi x, qi (t ) i x, i i w qi (t ) Wi x, i 2.1.1 Mass and Linear Stiffness Matrices Using the procedure of the classical finite element, the mass and stiffness matrices are then calculated. For one finite element, they may be written as follows: where the qi (t)’s functions are the generalized coordinates and the spatial functions U, V, W, x L and are given by equation (2). m s h N T N dA 00 L k L B T PB dA (7) (6) 2.1.2.b The deformation vector is written as a function of the generalised coordinates by separating the linear part from non-linear one: 00 where s is the density of the shell, h its L NL T thickness, dA a surface element, [P] the elasticity matrix and the [N] and [B] are derived from equations (4, 5). The matrices [ms] and [ks(L)] are obtained analytically by carrying out the necessary matrix operations over x and in (8) This vector is given in [11]. The subscripts “L” and “NL” mean “linear” and “non-linear”, 3 respectively. In general, these terms can be expressed in the following form: carrying out a large number of the intermediate mathematical operations, while are not given here due to the complexity of the manipulations, the following non-linear modal equations are obtained. These non-linear modal equations are used to study the dynamic behavior of an empty anisotropic cylindrical shell. xo a j q j AA jk q j q k j j k xo b j q j BB jk q j q k j j k c j q j CC jk q j q k o x j j (9-a) k mijj kij j kijk ( L) o d j q j DD jk q j q k j j j k kijks o e j q j EE jk q j q k j j j j k x n j q j NN jk q j q k j j k x p j q j PPjk q j q k j j (9-b) k s j q j SS jk q j q k j j j k 2.1.2.c Lagrange’s equations of motion in the generalized coordinates qi (t) is defined as: T V Qi qi qi j k (12) i 1,2,... and P11 AAijks P22 BBijks P33CCijks (13-b) P44 DDijks P55 EEijks P66GGijks P77 NNijks P88 PPijks dA kijks( NL3) P99 SSijks P1010TTijks Pmn AUX ijks I AUX ijks J 57 58 P36 AUX ijks AUX ijks (10) Where T is the total kinetic energy, V the total elastic strain energy of deformation and the Qi’s are the generalized forces. Assuming NL 1 , 2 ,..., 10 T , the strain energy V can Where dA=R dx d and: m =1,2,...,9 I=1,3,5,...,55 and m, n3,6 n=m+1 to 10 J=I+1 The Pij’s are the terms of the elasticity matrix [P] and the terms AAijk, BBijk, ..., AUXijk58 and AAijks, BBijks,..., AUXijks58 represent the coefficients of the modal equations in step (2.1.2.d). Follow are the expressions for the coefficients ai, AAij, AAijk and AAjkrs, the others coefficients are obtained in the same way, details are given in [11]. be defined as follow: V j k s Qi ( NL 3) ( NL 2 ) P11 AAijk P22 BBijk P33CCijk P44 DDijk P55 EEijk (13-a) P GG P NN P PP 66 ijk 77 ijk 88 ijk dA kijk ( NL2) P99 SS ijk P1010TTijk Pmn AUX ijk I AUX ijk J 57 58 P AUX AUX 36 ijk ijk Note: AAij=AAji, BBij=BBji and etc. d T dt q i s k k t j q j TT jk q j q k j k j Where mij, kij(L) are the terms of mass and linear stiffness matrices given by equation (6). The terms of kijk(NL2) and kijks(NL3) , which represent the second and third-order non-linear stiffness matrices, are given by the following integrals in the case of anisotropic laminated cylindrical shell based on the refined shell theory in which the shear deformation and rotary inertia effects are considered: k o g j q j GG jk q j q k j j a L Pij i j Pkl k l R d x d 2 00 (11) Where: a =1 if i = j or k = l (i, j =1,2,...,10), (k,l=3,6) and i , j 3,6 a =2 if i j or k l (i ,j=1,2…,10) (k,l=3,6) 2.1.2.d After developing the total kinetic energy and strain energy, using definitions (9), and then substituting into the Lagrange equation (10) and 4 ai U i x AAij 2.2.1 Dynamic Pressure Based on the previous hypothesis, the potential function must satisfy the Laplace equation. This relation is expressed in the cylindrical coordinate system by: 1 U i U j Vi V j Wi W j 2 x x x x x x AAijk ai AA jk a j AAki ak AAij AAijks 2 AAis AA jk 2 (14) 1 r ,r ,r 12 , , xx 0. (16) r r is the potential function that represents the velocity potential. The components of the flow velocity are given by: where U,V and W are spatial functions determined by equations (2). In equation (14), the subscript “i,j”, “i,j,k” and “i,j,k,s” represent the coupling between two; three and four mode, respectively. Substituting equation (2) into equations (14), we obtain Vx U xu , x ; V 1 , ; Vr , r R (17) where Vx ,V and Vr are respectively the axial, tangential and radial components of the fluid velocity; Uxu is the velocity of the liquid through the shell section. The Bernoulli equation is given by: ai C i i mSin mx e i m 2 Sin 2 mx i j ( i j ) (15) AAij C i C je 2 2 (1 )m Cos mx i j 2 , t V P 2 f m m / L where i ( i =1,...,10) are the roots of characteristic equation (3) and m is the axial mode number. The same definitions, as relation (15), are obtained for other parameters and given in [11]. The constants Ci(i=1,...10) can be determined using ten boundary conditions for each element. The axial, tangential and radial displacements as well as the rotations have to be specified for each node. 0. (18) r A full definition of the flow requires that a condition be applied to the structure-fluid interface. The impermeability condition ensures contact between the shell and the fluid. This should be: Vr rR , r rR W,t U xW, x r R (19) From the theory of shells, we have: Substituting these definitions into equation (13) and then integrating over x and , the two expressions for the second- and third-order nonlinear matrices are obtained, as given in equation (12). W x, , t C j exp j it sin 10 j 1 m x (20) L Assuming then, x, , r , t R j (r ) S j x, , t 10 2.2 Fluid Model j 1 Linear potential flow theory is applied to describe the fluid effects that lead to the inertial, centrifugal and Coriolis forces. The mathematical model is based on the following hypothesis: i) the fluid flow is potential; ii) the fluid is irrotational, incompressible and non-viscous. (21) The function S j ( x, , t ) is explicitly determined by applying the impermeability condition (19) and using the radial displacement (20). Substituting the assumed function into equation (16) leads to the following differential Bessel equation: 5 2 r2 d R j (r ) dr 2 R j (r ) i 2 r m k2 r 2 dR j ( r ) dr Substituting relation (24) into (18), we obtain the equation for the pressure on the shell wall. (22) 0. i j j 1 where “i” is the complex number, i2=-1 and j is im r B Yi j L r (23) where J i j and Yi j are, respectively, the Bessel functions of the first and second kind of complex order “ i j ”. For inside flow, the solution (23) must be finite on the axis of shell (r=0); this means we have to set the constant “B” equal to zero. For outside flow ( r ); this means that the constant “A” is equal to zero. When the shell is simultaneously subjected to internal and external flow, we have to take the complete solution (23). We carry the Bessel equation solution back into (21) to obtain the final expression of velocity potential evaluated at the shell wall: 10 imRu (24) (r , x, , t ) Z uj ( )W j , t U xuW j , x j 1 3 mRu ) L and Z uj ( L ( L) (M s M f ) ( K s K f ) ( NL 2 ) f ( NL 3) 2 s (28) 3 s where { } is the displacement vector and [Ms], [KsL], [KsNL2] and [KsNL3] are, respectively the mass, linear and second- and third-order nonlinear stiffness matrices of the structure, respectively, and [Mf], [Kf] and [Cf] are the inertial, centrifugal and Coriolis forces, respectively, due to the fluid effect. Ru (26) if u e imRu Yi j 1 (imRu / L) i j L Yi j (imRu / L) where j ( j 1,...,10) Differential K 0 C K Ru if u i (25) J imRu i j 1 (imRu / L) i j L J i j (imRu / L) mRu ) L Non-Linear Relations The structural and fluid mass and stiffness matrices, either linear or non-linear, as well as the fluid damping matrix, obtained in the previous sections, are only determined for one element. The global mass, stiffness and damping matrices are obtained by assembling the matrices for each element. Assembling is done in such way that all the equations of motion and the continuity of displacements at each node are satisfied. These matrices are designated as [M], [K] and [C], respectively. where Z uj ( By introducing the displacement function (20) into pressure expression (27), performing the matrix operation and thereafter integration over the fluid element required by the finite element method, the linear matrices (mass [mf], damping [cf] and stiffness [kf]) of moving fluid are obtained. Finally, the global linear matrices [Mf], [Cf] and [Kf] may be obtained, respectively, by superimposing the different matrices for each individual fluid finite element. the complex solution of the characteristic equation for the empty shell. The general solution of equation (22) is given by: im R j r A J i j L 10 2 Pu f u Z uj W j ,tt 2U xuW j , xt U xu W j , xx (27) 2 are the roots of the characteristic equation of the empty shell; J i j and Yi j are, respectively, the Bessel functions of the first and second kind of order “ i j ”; “m” is the axial mode number; “R” is the Setting: q mean radius of the shell; “L” its length; the subscript “u” is equal to “i” for internal flow and is equal to “e” for external flow. (r ) q j (t ) j j (t ) j (0) 1 and j (0) 0. 6 (29) where in this work is a hybrid finite element based on a combination of refined shell theory, modal expansion approach and potential flow theory. This method is capable of obtaining the high as well as low frequencies with high accuracy. The values of shear correction factors used in the calculation have been taken 2 / 12 . represents the square matrix for eigenvectors of the linear system and qis a time related vector. Numerical solution of the coupled system (28) is difficult and costly. Here, we limit ourselves to solving the uncoupled system. In this case, equation (28) is reduced to the following equation: i i i i i (i / h) 2 i Non-Dimensional Frequency 90 2 i (30) i (i / h 2 ) i3 0. where i i and C f ii mii k ; i2 NL 3 sii mii NL 2 sii k k ii ; i h mii mii h ; mii msii m f ii 2 (31) y o Fibre 0 60 Sanders' Theory 50 Present 40 R/h=50 L/R=5 m=4 30 m=3 20 10 0 0 Results 2 3 4 5 6 7 8 9 10 Figure 2: Variation of non-dimensional natural frequencies in conjunction with variation of m. a) Linear Vibrations of empty and liquid filled isotropic and anisotropic cylindrical shells- It should be noted that in the two first examples, the natural frequencies of the structures are also obtained using Sanders’ theory (non-shearable shell theory), by authors. In the first example, the different longitudinal vibration modes ( o R 2 ( / E2 ) / h ) as a function of the circumferential wave number are drawn in Fig. 2. This figure shows the results for four symmetric layers cross-ply (0o/90o/90o/0o) laminated shell whose mechanical properties are given as: and This research work is focused on the shear deformation and geometrically non-linear effects on the dynamic behavior of anisotropic cylindrical shells conveying fluid. Non-linearity effects produce either hardening or softening behavior in circular cylindrical shells. Considering the shear deformation effects leads to reducing the flexural stiffness of the structures. The developed method 1 Circumferential Wave Number (n) E1=25E2; G23=0.2E2; G13=G12=0.5E2, ν12=0.25; ρ=1 3 ensional Freque/ncy Numerical Discussions x 70 m=1 k ii k k f ii L sii where “h” represents the shell thickness. The square root of coefficient k ii / mii represents the ith linear vibration frequency of system. The solution i (t ) of the non-linear differential equations (30), which satisfies the conditions in (29), is calculated by a fourth order Runge-Kuta numerical method. The linear and non-linear natural frequencies are evaluated by a systematic search for the i (t ) roots as a function of time. The NL / L ratio of linear and non-linear frequency is expressed as a function of non-dimensional ratio i / h where i is the vibration amplitude. 4 o Fibre 90 80 Present 2.5 Sanders' Theory 2 7 1.5 m=5 Figure 4: Natural frequencies of a simply supported cylindrical shell. Non-Dimensional Frequency 0.7 Figure 3: Variation of non-dimensional natural frequencies in terms of m & n. In the next example (Fig. 3), the effect of axial mode number on the non-dimensional natural frequencies ( o R( (1 2 ) / E )1 / 2 ) of an isotropic cylindrical shell is studied and the results are compared with the obtained corresponding values based on the Sanders’ theory. 0.3 200 0 140 0 Present R/h=20 m=1 n=1 0.3 0.2 R/h=100 0.1 1 2 3 4 5 6 7 The fluid depth effect is also studied for the halffilled cylindrical shell in Fig. 4. Fig. 5 is carried out for a simply supported, isotropic circular cylindrical shell completely filled with liquid. The frequency parameter, ( o R( (1 2 ) / E )1 / 2 ), is shown for different values of R/h and L/R and is compared with provided results in Ref. [14]. b) Stability of the shells subjected to flowing fluidThe influence of the flow velocity on the frequency parameter of cylindrical shells is studied through Figs 6 and 7 for different values of R/h, L/R, axial and circumferential wave numbers. The obtained results, in Fig. 6, are compared with those of theory [4]. 120 0 100 0 25 80 0 60 0 m=2 20 0 8 0 0 2 4 6 8 10 Circumferential Wave Number (n) 9 Figure 5: Frequency distribution of a fluid-filled cylindrical shell. 7680(kg / m3 ) 40 0 8 Length-to-Radius Ratio L/R 12 nsional Frequency Natural Frequency (Hz) 160 0 [Ref. 14] 0.4 R/h=300 Present, Empty Present, Full Experiment [13], Empty Present, Half-Filled 180 0 0.5 0 Fig. 4 shows the natural frequencies computed for closed simply supported, circular cylindrical shell for m=2 and compared with the experimental results, given in [13]. To determine natural frequencies with the developed program, based on the present theory, only 10 elements are required to provide acceptable accuracy. As can be seen, there is good agreement between the present theoretical results and those of experimental. Dimensions and material properties are given as follow: R 0.175(m) L 0.664(m) t 1(mm) E 206(GPa) 0.6 Present m=2 [Ref.4] 20 L/R=2 R/h=100 15 f/s=0.128 n=5 10 10 Figure 7: Stability of a cylindrical shell as a function of flow velocity. In Fig. 6, the first frequency becomes negative imaginary at U=2.96, indicating static divergence instability in the first axial mode, and reappeared and coalesced at U=3.36 with that of the second axial mode to produce mode flutter. Fig. 7 shows the divergence instability phenomenon for an isotropic simply supported cylindrical shell. Figure 6: Stability of a simply supported cylindrical shell as a function of internal flow velocity. K 1/ 2 Eh 3 uo 2 ( ) ;K L sh 12(1 2 ) c) Linear vibration of submerged cylindrical shells- Fig. 8 shows the non-dimensional, ( o R( s (1 2 ) / E )1 / 2 ) ,frequency variation as a function of circumferential wave number for three different cases, shell in air, fluid-filled shell and shell immersed in fluid and are compared with those of theory [15]. The two theories give nearly identical results. The u and ω are, respectively, the velocity of the flowing fluid and the natural frequency. As the flow velocity increases, Fig. 6, the two theories generate significantly different results. This might be attributed to i) not considering the influence of transverse shear deformation in Ref.[4] and ii) limitations of the theory (Ref.[4]) associated with the use of too few terms in the application of Galerkin’s method. d) Non-linear vibrations of empty and submerged cylindrical shells- The influence of non-linearities on the frequencies of a simply supported cylindrical shell, along with corresponding values given in References [16 and 17] is shown in Fig. 9. The given results in Ref. [16] were obtained based on Donnell’s simplified non-linear method. Raju and Rao [17] used the finite element method based on an energy formulation. In these two figures, the following parameters are defined: U u / u o ; / o ; o u o / L 2 L/R=2 R/h=100 f/s=.128 n=3 1 25 d n=4 20 n=5 m=1 - - - -m=2 15 n=3 10 n=4 n=5 5 9 0 0 1 2 3 Dimensionless Velocity 4 Non-Dimensional Frequency Non-Dimensional Frequency 30 b Present [Ref.15] 0.1 Shell in Air Fluid Filled m=1 L / R=4 R / h=400 b /d=1.0 f /s=.132 0.01 Shell Immersed in Fluid 5 0.001 0 2 4 6 8 10 12 Figure 10: Non-linearity effect on the frequency ratio of a submerged cylindrical shell. Figure 8: Frequency variation of empty, fluidfilled and immersed in fluid shell with respect to (n). 5 This paper deals with some of the problems that arise when considering geometric non linearities, shear deformation, rotary inertia and flowing fluid effects in the study of dynamic and stability behavior of elastic, anisotropic and isotropic cylindrical shells. An efficient hybrid finite element method, modal expansion approach, shearable shell theory and linear potential flow one have been used to develop the non-linear dynamic equations of the coupled fluid-structure system. The shell equations are used in full for the determination of the displacement functions. It is believed that the refined shear deformation theory and effects of geometric non-linearities of the structures presented here are essential for predicting an accurate response for anisotropic shell structures. A full implementation of the nonlinear dynamic equations is conducted to show the reliability and effectiveness of the present formulation that gives a very good description of geometrical non-linear and shear deformation effects on the dynamic and stability behavior of the cylindrical shells subjected to flowing fluid. Fig. 10 shows the non-linearity effect on the frequency ratio of a steel open (φ=100o) cylindrical shell totally submerged in fluid. The shell is simply supported. The following data are considered into calculations: R=450mm; h=1.5mm; L=1350mm, ρf/ρs=.128 1.2 Present Raju and Rao [17] Nowinski [16] m=1 n=4 1.1 1 0.9 0 0.5 1 1.5 2 2.5 3 Amplitude to Thickness Ratio j/h The natural frequencies of the coupled fluidstructure are lower than the corresponding values of empty shells due to increased kinetic energy without a corresponding increase in the strain energy. In the case of flowing fluid, the centrifugal and Coriolis terms generate complex eigenvalue problems, non-self-adjoint differential equations. Therefore, the system may experience static (buckling) and dynamic (flutter) instabilities. As long as the effective stiffness of the system remains positive as flow velocity increases, the system will oscillate asymptotically about its neutral equilibrium position; otherwise it will diverge to a new equilibrium position, different from neutral (buckling). As long as the Figure 9: Relative frequency of simply supported cylindrical shell versus relative amplitude. 2.1 1.9 n=1 n=2 1.7 Relative Frequency ωNl/ωL Relative Frequency NL/L 1.3 E=200GPa =0.3 =7800kg / m3 R=2.54 L=40 cm h=0.0254 cm cm 1.5 1.3 1.1 0.9 0.7 10 0.5 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 Amplitude to Thickness Ratio Γi/h 2 Conclusion effective fluid damping of the system remains positive as flow velocity is increased, vibrations will be damped; otherwise they will be amplified (flutter). Quiescent Fluid, Journal of Fluid and Structures, No. 12, 1998, pp. 883-918. [8] M.H. Toorani and A.A. Lakis, General Equations of Anisotropic Plates and Shells Including Transverse Shear Deformations, Rotary Inertia and Initial Curvature Effects, Journal of Sound and Vibration, 237 (4), 2000, pp. 561-615. It is shown that the non-linearities associated with fluid, under no-flow conditions, have no or very little effect on the natural frequency of a cylindrical shell for amplitudes up to two times the shell thickness [6]. Under flow conditions, non-linear effects were found to increase with flow rate increasing but the importance of the contribution of flow-non-linearities to this overall trend has yet to be determined. Attempting to this work is left to future investigations. [9] M.H. Toorani and A.A. Lakis, Shear Deformation Theory in Dynamic Analysis of Anisotropic Laminated Open Cylindrical Shells Filled With or Subjected to a Flowing Fluid, Computer Methods in Applied Mechanics and Engineering, 190, 2001, pp. 4929-4966. [10] H. Radwan and J. Genin J., Non-Linear Modal Equations for Thin Elastic Shells, International Journal of Non-Linear Mechanics, 10, 1975, pp. 15-29. References: [1] M. P. Païdoussis and J.P. Denis, Flutter of Thin Cylindrical Shells Conveying Fluid, Journal of Sound and Vibration, No. 20, 1972, pp. 9-26. [11] M.H. Toorani and A.A. Lakis, Geometrically Non-Linear Dynamics of Anisotropic Open Cylindrical Shells with a Refined Shell Theory, Technical Report, Polytechnique of Montreal, EPM-RT-01-07, 2002. [2] A.A. Lakis and M.P. Païdoussis, Free Vibration of Cylindrical Shells Partially Filled with Liquid, Journal of Sound and Vibration, No. 19, 1971, pp. 1-15. [12] J.L. Sanders J.L., Nonlinear Theories for Thin Shells, Quarterly of Applied Mathematics, 21, 1963, pp.21-36. [3] A.A. Lakis and M.P. Païdoussis, Shell Natural Frequencies of the Pickering Steam Generator, Atomic Energy of Canada Ltd., AECL, Report No. 4362, 1973. [4] D.S. Weaver and T.E. Unny, On the Dynamic Stability of Fluid-Conveying Pipes, Journal of Applied Mechanics, No. 40, 1973, pp. 48-52. [13] M. Amabili and G. Dalpiaz, Breathing Vibrations of a Horizontal Circular Cylindrical Tank Shell, Partially Filled with Liquid, Journal of Vibration and Acoustics, 117, 1995, pp. 187191. [5] M.J. Pettigrew, Flow-Induced Vibration Technology: Application to Steam Generators, Lecture Series Presented at Babcock & Wilcox Canada, November 2000. [14] A.A. Lakis and M. Sinno, Free Vibration of Axisymmetric and Beam-Like Cylindrical Shells Partially Filled with Liquid, Int. J. of Num. Meth. In Eng., 33, 1992, pp. 235-268. [6] A. Selmane and A.A. Lakis, Non-Linear Dynamic Analysis of Orthotropic Open Cylindrical Shells Subjected to a Flowing Fluid, Journal of Sound and Vibration, No. 202, 1997, pp. 67-93. [15] P.B. Gonçalves and R.C. Batista, Frequency Response of Cylindrical Shells Partially Submerged or Filled with Liquid, Journal of Sound and Vibration, 113(1), 1987, pp. 59-70. [16] J.L. Nowinski J.L., Non-Linear Transverse Vibrations of Orthotropic Cylindrical Shells, AIAA Journal, 1, 1963, pp. 617-620. [7] M. Amabili M., F. Pellicano and M.P. Païdoussis, Non-Linear Vibrations of Simply Supported Circular Cylindrical Shells, Coupled to 11 [17] K.K. Raju and G.V. Rao, Large Amplitude Asymmetric Vibrations of Some Thin Elastic Shells of Revolution, Journal of Sound and Vibration, 44, 1976, pp. 327-333. 12