A two variable refined plate theory for orthotropic plate analysis

A two variable refined plate theory for orthotropic plate analysis R.P. Shimpi *, H.G. Patel Department of Aerospace Engineering, Indian Institute of Technology Bombay, Powai, Mumbai 400076, Maharashtra, India Abstract A new theory, which involves only two unknown functions and yet takes into account shear deformations, is presented for orthotropic plate analysis. Unlike any other theory, the theory presented gives rise to only two governing equations, which are completely uncoupled for static analysis, and are only inertially coupled (i.e., no elastic coupling at all) for dynamic analysis. Number of unknown functions involved is only two, as against three in case of simple shear deformation theories of Mindlin and Reissner. The theory presented is variationally consistent, has strong similarity with classical plate theory in many aspects, does not require shear correction factor, gives rise to transverse shear stress variation such that the transverse shear stresses vary parabolically across the thickness satisfying shear stress free surface conditions. Well studied examples, available in literature, are solved to validate the theory. The results obtained for plate with various thickness ratios using the theory are not only substantially more accurate than those obtained using the classical plate theory, but are almost comparable to those obtained using higher order theories having more number of unknown functions. Keywords: Refined plate theory; Shear deformable plate theory; Orthotropic plate theory; Inertial coupling; Elastic coupling; Free vibrations 1. Introduction The purpose of this paper is to present a new theory for orthotropic plate analysis, which involves only two unknown functions and yet takes into account shear deformations. With the increasing use of composite materials, the need for advanced methods of analysis became obvious. In case of composite materials, transverse stresses and strains strongly influence the behavior. In particular, the transverse shear stress effects are more pronounced. The classical plate theory, which is not formulated to account for the effect of these stresses, is not satisfactorily applicable to orthotropic plate analysis. Therefore, over the years, researchers developed many theories, which took into account transverse shear effects.

6784 It is worthwhile to note some developments in the plate theory. Reissner (1944, 1945) was the first to develop a theory which incorporates the effect of shear. Reissner used stress based approach. At the same level of approximation as the one utilized by Reissner, Mindlin (1951) employed displacement based approach. As per Mindlin s theory, transverse shear stress is assumed to be constant through the thickness of the plate, but this assumption violates the shear stress free surface conditions. Mindlin s theory satisfies constitutive relations for transverse shear stresses and shear strains in an approximate manner by way of using shear correction factor. A good discussion about Reissner s and Mindlin s theories is available in the reference of Wang et al. (001). A refined plate theory for orthotropic plate, based on stress formulation, was proposed by Medwadowski (1958). In this theory, a nonlinear system of equations was derived from the corresponding equations of the three-dimensional theory of elasticity. The system of equations was linearized and use of stress function was made. Approach of Librescu (1975) makes the use of weighted transverse displacement. Constitutive relations between shear stress and shear strain are satisfied. Reissner s formulation comes out as a special case of Librescu s approach. Approach of Donnell (1976) is to make correction to the classical plate deflections. Donnell assumes uniform distribution of shear force across the thickness of the plate, and, to rectify the effects of the assumption, introduces a numerical factor, which needs to be adjusted. Formulation by Levinson (1980) is based on displacement approach and his theory does not require shear correction factor. The governing equations for the motion of a plate obtained by Levinson s approach are same as those obtained by Mindlin s theory, provided that the shear correction factor associated with the Mindlin s theory is taken as 5/6. Many higher order theories are available in the literature e.g., theories by Nelson and Lorch (1974) with nine unknowns, Krishna Murty (1977) with 5,7,9,...unknowns, Lo et al. (1977a,b) with 11 unknowns, Kant (198) with six unknowns, Bhimaraddi and Stevens (1984) with five unknowns, Reddy (1984) with five unknowns, Soldatos (1988) with three unknowns, Reddy (1990) with eight unknowns, Hanna and Leissa (1994) with four unknowns. It is important to note that Srinivas et al. (1970) and Srinivas and Rao (1970) have presented a three-dimensional linear, small deformation theory of elasticity solution for static as well as dynamic analysis of isotropic, orthotropic and laminated simply-supported rectangular plates. It may be worthwhile to note that a critical review of plate theories was carried out by Vasil ev (199). Whereas, Liew et al. (1995) surveyed plate theories particularly applied to thick plate vibration problems. A recent review paper is by Ghugal and Shimpi (00). Plate theories can be developed by expanding the displacements in power series of the coordinate normal to the middle plane. In principle, theories developed by this means can be made as accurate as desired simply by including a sufficient number of terms in the series. These higher-order theories are cumbersome and computationally more demanding, because each additional power of the thickness coordinate, an additional dependent is introduced into the theory. It has been noted by Lo et al. (1977a) that due to the higher order of terms included in their theory, the theory is not convenient to use. This observation is more or less true for many other higher order theories as well. And, thus there is a scope to develop simple to use higher order plate theory. It is to be noted that Shimpi (00) presented a theory for isotropic plates. He applied the theory to flexure of shear-deformable isotropic plates. The theory, using only two unknown functions, gave rise to two governing equations, which are uncoupled in case of static analysis. Also, unlike any other theory, the theory has strong similarities with the classical plate theory in some aspects. In this paper, using similar approach, a new orthotropic plate theory is presented.. Orthotropic plate under consideration Consider a plate (of length a, width b, and thickness h). The plate occupies (in 0 x y z right-handed Cartesian coordinate system) a region 0 6 x 6 a; 0 6 y 6 b; h= 6 z 6 h= ð1þ

6785 o q (x,y,t) h/ h/ x,u z,w b y,v a Fig. 1. Geometry of a plate. Geometry of plate under consideration is shown in Fig. 1. The plate can have any meaningful boundary conditions at edges x = 0,x = a and y = 0,y = b. The orthotropic plate has following material properties: E 1, E are elastic moduli, G 1, G 3, G 31 are shear moduli and l 1, l 1 are Poisson s ratios. Here subscripts 1,, 3 correspond to x, y, z directions of Cartesian coordinate system, respectively (this notation is the same as that used by Jones (1999, Chapter ) and Reddy (1984)). For isotropic plate these above material properties reduce to E 1 = E = E, G 1 = G 3 = G 31 = G, l 1 = l 1 = l. 3. Refined plate theory (RPT) for orthotropic plate 3.1. Assumptions of RPT Assumptions of RPT would be as follows: 1. The displacements (u in x-direction, v in y-direction, w in z-direction) are small in comparison with the plate thickness and, therefore, strains involved are infinitesimal. As a result, normal strains x, y, z and shear strains c xy, c yz, c zx can be expressed in terms of displacements u, v, y using strain-displacement relations: x ¼ ou ox ; c xy ¼ ov ox þ ou oy ; y ¼ ov oy ; z ¼ ow oz c yz ¼ ow oy þ ov oz ; 9 >= c zx ¼ ou oz þ ow >; ox. The transverse displacement w has two components: bending component and shear component w s. Both the components are functions of coordinates x, y and time t only. wðx; y; tþ ¼ ðx; y; tþþw s ðx; y; tþ 3. (a) In general, transverse normal stress r z is negligible in comparison with inplane stresses r x and r y. Therefore, stresses r x and r y are related to strains x and y by the following constitutive relations: E 1 r x ¼ x þ l 9 1E y 1 l 1 l 1 1 l 1 l >= 1 r y ¼ E y þ l 1E 1 x 1 l 1 l 1 1 l 1 l 1 >; (b) The shear stresses s xy, s yz, s zx are related to shear strains c xy, c yz, c zx by the following constitutive relations: s xy ¼ G 1 c xy ; s yz ¼ G 3 c yz ; s zx ¼ G 31 c zx ð5þ ðþ ð3þ ð4þ

6786 4. The displacement u in x-direction consists of bending component u b and shear component u s. Similarly, the displacement v in y-direction consists of bending component v b and shear component v s. u ¼ u b þ u s ; v ¼ v b þ v s ð6þ (a) The bending component u b of displacement u and v b of displacement v are assumed to be analogous, respectively, to the displacements u and v given by classical plate theory (CPT). Therefore, the expressions for u b and v b can be given as u b ¼ z o ox v b ¼ z o oy ð7þ ð8þ It may be noted that the displacement components u b, v b, and together do not contribute toward shear strains c yz, c zx and therefore to shear stresses s zx and s yz. (b) The shear component u s of displacement u and the shear component v s of displacement v are such that (i) they give rise, in conjunction with w s, to the parabolic variations of shear strains c yz, c zx and therefore to shear stresses s zx and s yz across the cross section of the plate in such a way that shear stresses s zx and s yz are zero at z = h/ and at z = h/, and (ii) their contribution toward strains x, y,andc xy is such that in the moments M x, M y,andm xy there is no contribution from the components u s and v s. 3.. Displacements, moments, shear forces in RPT Based on the assumptions made in the preceding section, it is possible, with some efforts, to get the expressions for the shear component u s of displacement u, and shear component v s of displacement v; and these can be written as z z 3 ows ð9þ u s ¼ h 1 5 4 h 3 h v s ¼ h 1 z 5 z 3 4 h 3 h ox ows oy ð10þ Using expressions (3), and (6) (10), one can write expressions for displacements u, v, w as uðx; y; z; tþ ¼ z ow b ox þ h 1 z 5 z 3 ows 4 h 3 h ox vðx; y; z; tþ ¼ z ow b oy þ h 1 z 5 z 3 ows 4 h 3 h oy wðx; y; tþ ¼ ðx; y; tþþw s ðx; y; tþ ð11þ ð1þ ð13þ Using expressions for displacements (11) (13) in strain-displacement relations (), the expressions for strains can be obtained. x ¼ z o ox þ h 1 z 5 z 3 o w s ð14þ 4 h 3 h ox y ¼ z o oy þ h 1 z 5 z 3 o w s ð15þ 4 h 3 h oy z ¼ 0 ð16þ c xy ¼ z o oxoy þ h 1 z 5 z 3 o w s ð17þ 4 h 3 h oxoy

c yz ¼ 5 4 5 z ows h oy c zx ¼ 5 4 5 z ows h ox Expressions for stresses can be obtained using strain expressions from (14) (19) in constitutive relations (4) and (5). Constitutive relation can be written in matrix form as follows: 8 9 38 9 r x Q 11 Q 1 0 0 0 x >< r y >= Q 1 Q 0 0 0 >< y >= s xy ¼ 0 0 Q 66 0 0 c xy ð0þ 6 7 s yz 4 0 0 0 Q 44 0 5 c yz >: >; >: >; s zx 0 0 0 0 Q 55 c zx where, E 1 Q 11 ¼ ; Q 1 l 1 l 1 ¼ l 1E ¼ l 1E 1 1 1 l 1 l 1 1 l 1 l 1 E Q ¼ ; 1 l 1 l 1 Q 44 ¼ G 3 ; Q 55 ¼ G 31 ; Q 66 ¼ G 1 It is well known that the material property matrix in expression (0) is symmetric (e.g. reference of Jones, 1999). The moments and shear forces are defined as 8 9 8 9 r x z >< >: M x M y M xy Q x Q y >= ¼ >; Z z¼h= z¼ h= >< r y z >= s xy z dz >: s zx s yz >; Using expressions (14) (0) in (), expressions for moments M x, M y and M xy and shear forces Q x and Q y can be obtained. These expressions are: o M x ¼ D 11 ox þ D o 1 ð3þ oy o M y ¼ D oy þ D o 1 ð4þ ox o M xy ¼ D 66 ð5þ oxoy Q x ¼ A 55 ð6þ ox Q y ¼ A 44 ð7þ oy where, D 11 ¼ Q 11h 3 1 ; D ¼ Q h 3 A 44 ¼ 5Q 44h ; A 55 ¼ 5Q 55h 6 6 1 ; D 1 ¼ Q 1h 3 9 >= >; 1 ; D 66 ¼ Q 66h 3 1 ; 6787 ð18þ ð19þ ð1þ ðþ ð8þ

6788 It may be noted that expressions for moments M x, M y and M xy contain only as an unknown function. Also, the expressions for shear forces Q x and Q y contain only w s as an unknown function. 3.3. Expressions for kinetic and total potential energies It should be noted that displacement w, given by Eq. (13), is not a function of z. As a result of this, normal strain z comes out to be zero. Therefore, the expressions for kinetic energy T and total potential energy P of the plate can be written as T ¼ Z z¼ h Z y¼b Z x¼a z¼ h y¼0 x¼0 " þ ov þ ow # dxdy dz ot ot 1 q ou ot ð9þ P ¼ Z z¼ h Z y¼b Z x¼a z¼ h y¼0 x¼0 Z y¼b Z x¼a y¼0 x¼0 1 r x x þ r y y þ s xy c xy þ s yz c yz þ s zx c zx dxdy dz q½ þ w s Šdxdy ð30þ in which q is the intensity of externally applied transverse load (acting along the z-direction). Using expressions (11) (0) in Eqs. (9) and (30), expressions for kinetic energy and total potential energy can be written as Z y¼b Z x¼a ( o o þ o ) o ot ox ot oy Z y¼b Z x¼a ( o ow s ot ox T ¼ qh3 dxdy þ qh3 4 y¼0 x¼0 016 y¼0 x¼0 þ o ) dxdy þ qh Z y¼b Z x¼a o þ ow s dxdy ð31þ ot oy y¼0 x¼0 ot ot P ¼ 1 Z y¼b Z ( x¼a o o o o ) D 11 þ D y¼0 x¼0 ox þ D oy 1 ox oy þ 4D o 66 dxdy oxoy þ 1 Z y¼b Z ( " x¼a 1 y¼0 x¼0 84 D o w s o w s o w s o # w s 11 þ D ox þ D oy 1 ox oy þ 4D o w s 66 oxoy ) Z ow y¼b Z x¼a s þ A 44 þ A 55 dxdy q½ þ w s Šdxdy ð3þ oy ox y¼0 x¼0 3.4. Obtaining governing equations and boundary conditions in RPT by using Hamilton s principle Governing differential equations and boundary conditions can be obtained using well known Hamilton s principle Z t t 1 dðt PÞdt ¼ 0 ð33þ where d indicates a variation w.r.t. x and y only; t 1, t are values of time variable t at the start and at the end of time interval (in the context of Hamilton s Principle), respectively. Using expressions (31) and (3) in the preceding equation and integrating the equation by parts, taking into account the independent variations of and w s, one obtains the governing differential equations and boundary conditions, and these are as follows:

6789 3.4.1. Governing equations in RPT The governing differential equations for the plate are o 4 D 11 ox þ ðd 4 1 þ D 66 Þ o4 ox oy þ D o 4 qh3 o r o w oy 4 1 ot b þ qh þ o w s ¼ q ot ot o w s A 55 ox þ A o w s 44 þ 1 oy 84 D o 4 w s 11 ox þ ðd 4 1 þ D 66 Þ o4 w s ox oy þ D o 4 w s qh3 oy 4 84 1 þ qh o þ o w s ¼ q ot ot o ot r w s ð34þ ð35þ where, r ¼ o ox þ o oy ð36þ 3.4.. Boundary conditions in RPT The boundary conditions of the plate are given as follows: 1. At corners (x =0,y = 0), (x =0,y = b), (x = a, y = 0), and (x = a, y = b), the following conditions hold: (a) The condition involving (i.e., bending component of transverse displacement) ð D 66 Þ o ¼ 0 or is specified ð37þ oxoy (b) The condition involving w s (i.e., shear component of transverse displacement) 1 ð 84 D 66Þ o w s ¼ 0 or w s is specified ð38þ oxoy. On edges x = 0 and x = a, the following conditions hold: (a) The conditions involving (i.e., bending component of transverse displacement) o 3 D 11 ox þ ð D 3 1 þ 4D 66 Þ o3 þ qh3 o o ¼ 0 or w oxoy 1 ot b is specified ð39þ ox o D 11 ox þ D o o 1 ¼ 0 or is specified ð40þ oy ox (b) The conditions involving w s (i.e., shear component of transverse displacement) A 55 ox 1 84 D o 3 w s 11 ox þ ð D 3 1 þ 4D 66 Þ o3 w s oxoy þ qh3 o ¼ 0 or w 1 ot s is specified ð41þ ox 1 84 D o w s 11 ox þ D o w s 1 ¼ 0 or is specified ð4þ oy ox 3. On edges y = 0 and y = b, the following conditions hold: (a) The conditions involving (i.e., bending component of transverse displacement) o 3 D oy þ ð D 3 1 þ 4D 66 Þ o3 þ qh3 o o ¼ 0 or w ox oy 1 ot b is specified ð43þ oy o D oy þ D o o 1 ¼ 0 or is specified ð44þ ox oy

6790 (b) The conditions involving w s (i.e., shear component of transverse displacement) A 44 oy 1 84 D o 3 w s oy þ ð D 3 1 þ 4D 66 Þ o3 w s ox oy þ qh3 o ¼ 0 or w 1 ot s is specified ð45þ oy 1 84 D o w s oy þ D o w s 1 ¼ 0 or is specified ð46þ ox oy 4. Comments on RPT 1. With respect to governing equations, following can be noted: (a) In RPT, there are two governing equations (Eqs. (34) and (35)). Both the governing equations are fourth-order partial differential equations. (b) The governing equations involve only two unknown functions (i.e., bending component and shear component w s of transverse deflection). Even theories of Reissner (1944), Mindlin (1951), which are first order shear deformation theories and are considered to be simple ones, involve three unknown functions. Mindlin s theory does not exactly satisfy transverse shear stresses and shear strains constitutive relations. Whereas, in contrast, in RPT, these constitutive relations are exactly satisfied. (c) It is emphasized here that RPT is the only theory, to the best knowledge of the authors, wherein in the governing equations, there is only inertial coupling, and there is no elastic coupling at all.. With respect to boundary conditions, following can be noted: (a) There are two conditions per corner. (i) One condition is stated in terms of and its derivatives only (i.e., condition (37)). (ii) The second condition is stated in terms of w s and its derivatives only (i.e., condition (38)). (b) There are four boundary conditions per edge. (i) Two conditions are stated in terms of and its derivatives only (e.g., in the case of edge x = 0, conditions (39) and (40)). (ii) The remaining two conditions are stated in terms of w s and its derivatives only (e.g., in the case of edge x = 0, conditions (41) and (4)). 3. Some entities of RPT (e.g., a governing equation, moment expressions, boundary conditions) have strong similarity with those of CPT. The relevant equations, expressions in respect of CPT are given in Jones (1999, Chapter 5). (a) The following entities of RPT are identical, save for the appearance of the subscript, to the corresponding entities of the CPT: (i) Moment expressions for M x, M y, M xy (i.e., expressions (3) (5)). (ii) Corner boundary condition (i.e., condition (37)). (iii) Edge boundary conditions (i.e., conditions (39), (40), (43) and (44)). The bending component of transverse displacement figures in the just mentioned entities of RPT, whereas transverse displacement w figures in the corresponding equations of the CPT. (b) One of the two governing equations (i.e., Eq. (34)) has strong similarity with the governing equation of CPT. (If in Eq. (34) the entity o w s ot is ignored, and if is replaced by w, then the resulting equation is identical to the governing equation of CPT.) 4. It is seen from the displacements u, v, w expressions (11) (13) that, as the thickness becomes smaller and smaller, the theory converges in the limit to the classical plate theory and, therefore, it is equally applicable to thick as well as thin plate analysis. 5. Illustrative examples for static analysis of simply supported rectangular plates Well studied examples, available in literature, would be taken up to demonstrate the effectiveness of RPT. The examples have also been studied by Reddy (1984), Srinivas and Rao (1970). Consider a plate (of length a, width b, and thickness h) of orthotropic material. The plate occupies (in 0 x y z right-handed Cartesian coordinate system) a region defined by Eq. (1). The plate has simply supported

boundary conditions at all four edges x =0,x = a, y = 0 and y = b. The plate carries on surface z = h/, a uniformly distributed load of intensity q 0 acting in the z-direction. The properties of orthotropic material used for analysis are indicated in the concerned result Tables 1 7. 5.1. Governing equations for illustrative examples (static analysis) The governing equations for static analysis of plate can be obtained from Eqs. (34) and (35) by setting the kinetic energy terms to zero, as follows: o 4 D 11 ox þ ðd 4 1 þ D 66 Þ o4 ox oy þ D o 4 oy ¼ q 4 0 ð47þ o w s A 55 ox þ A o w s 44 þ 1 oy 84 D o 4 w s 11 ox þ ðd 4 1 þ D 66 Þ o4 w s ox oy þ D o 4 w s ¼ q oy 4 0 ð48þ 6791 5.. Boundary conditions for illustrative examples (static analysis) The boundary conditions of the plate are given as follows: 1. At corners (x =0,y = 0), (x =0,y = b), (x = a, y = 0), and (x = a, y = b) the following conditions hold: ¼ 0 w s ¼ 0. On edges x = 0 and x = a, the following conditions hold: ¼ 0 o D 11 ox þ D o 1 ¼ 0 oy w s ¼ 0 1 84 D 11 o w s ox þ D 1 o w s oy ¼ 0 3. On edges y = 0 and y = b, the following conditions hold: ¼ 0 o D oy þ D o 1 ¼ 0 ox w s ¼ 0 1 84 D o w s oy þ D 1 o w s ox ¼ 0 5.3. Solution of illustrative examples (static analysis) ð49þ ð50þ ð51þ ð5þ ð53þ ð54þ ð55þ ð56þ ð57þ ð58þ The following displacement functions and w s satisfy the boundary conditions (49) (58). ¼ X1 X 1 W bmn sin mpx sin npy a b w s ¼ m¼1;3;... n¼1;3;... X1 X 1 m¼1;3;... n¼1;3;... W smn sin mpx a sin where, W bmn, W smn are constant coefficients associated with and w s, respectively. npy b ð59þ ð60þ

679 Also externally applied uniformly distributed load of intensity q 0 can be represented in a double Fourier series as follows: q 0 ¼ X1 X 1 q mn sin mpx sin npy ð61þ a b where m¼1;3;... n¼1;3;... q mn ¼ 16q 0 p mn for uniformly distributed load of intensity q 0 Using expressions (59) (61) in the governing Eqs. (47) and (48), one obtains two completely uncoupled equations which can be written in matrix form, as follows: K 11mn 0 W bmn ¼ q mn ð6þ 0 K mn W smn q mn where, mp 4 K 11mn ¼ D 11 þ ð D1 þ D 66 Þ mp np np 4 þ D a a b b mp np K mn ¼ A 55 þ A44 a b þ 1 84 D mp 4 11 þ ð D1 þ D 66 Þ mp np np 4 þ D a a b b Coefficients W bmn ; W smn can be determined from Eq. (6), and then can be substituted in Eqs. (59) and (60) to get displacements. Here it is observed from the Eq. (6) that bending and shearing components are uncoupled. The following nondimensionalized deflections w, ^w and stresses r x, r y, s zx are tabulated in Tables 1 5. w ¼ wq 11 =hq 0 : for orthotropic plate ^w ¼ wg=hq 0 : for isotropic plate and r x ¼ r x =q 0 r y ¼ r y =q 0 s zx ¼ s zx =q 0 : for orthotropic plate. Results are discussed separately after Table 7. Table 1 Comparison of nondimensional central displacement w of simply-supported orthotropic rectangular plate under uniformly distributed transverse load a Plate dimensional parameters Nondimensional displacement w at x = a/, y = b/ by various theories a/b h/a h/b EXACT b Reddy c Reissner b CPT b RPT 0.5 0.05 0.05 1,54 1,54 1,54 1,01 1513.5 0.1 0.05 1408.5 1408.5 1408.4 135.1 140.4 0.14 0.07 387.3 387.5 387.7 344.93 384.0 1.0 0.05 0.05 10,443 10,450 10,44 10,46 10413.4 0.1 0.1 688.57 689.5 688.37 640.39 681.73 0.14 0.14 191.07 191.6 191.0 166.7 187.75.0 0.05 0.1 048.7 051.0 047.9 1988.1 04.74 0.1 0. 139.08 139.8 138.93 14.6 137.8 0.14 0.8 39.79 40.1 39.753 3.345 39.6 a w ¼ wq 11 =hq 0 (E /E 1 = 0.5500, G 1 /E 1 = 0.693, G 13 /E 1 = 0.15991, G 3 /E 1 = 0.6681, l 1 = 0.44046, l 1 = 0.314). b Taken from reference of Srinivas and Rao (1970). c Taken from reference of Reddy (1984).

6793 6. Illustrative examples for free vibrations of simply supported rectangular plates Well studied examples, available in literature, would be taken up to demonstrate the effectiveness of RPT. The examples have also been studied by Reddy (1984, 1985), Srinivas et al. (1970), Srinivas and Rao (1970). Consider a plate (of length a, width b, and thickness h) of orthotropic material. The plate occupies (in 0 x y z right-handed Cartesian coordinate system) a region defined by Eq. (1). The plate has simply supported boundary conditions at all four edges x = 0, x = a, y = 0 and y = b. The properties of orthotropic material used for analysis are indicated in the concerned result Tables 1 7. 6.1. Governing equations for illustrative examples (free vibration analysis) The governing equations for free vibration of plate can be obtained from Eqs. (34) and (35) by setting the external load (i.e., transverse load q) to zero, as follows: Table Comparison of nondimensional central displacement ^w of simply-supported isotropic rectangular plate under uniformly distributed transverse load a Plate dimensional parameters Nondimensional displacement ^w at x = a/, y = b/ by various theories a/b h/a h/b EXACT b Reissner b CPT b RPT 0.5 0.05 0.05 6855 685.9 6806.5 6861.10 0.1 0.05 437.5 437.0 45.4 439.06 0.14 0.07 116.91 116.66 110.74 117.70 1.0 0.05 0.05 761.3 760.00 79.9 765.7 0.1 0.1 178.45 178.13 170.6 179.46 0.14 0.14 48.40 48.47 44.414 48.9.0 0.05 0.1 437.5 437.0 45.4 439.06 0.1 0. 9.604 9.49 6.588 9.999 0.14 0.8 8.45 8.405 6.91 8.659 Note: Results are not available for Reddy s theory. a ^w ¼ wg=hq 0 (E = E 1 = E, G 1 = G 13 = G 3 = G =[E/(1 + l)], l 1 = l 1 = l = 0.3). b Taken from reference of Srinivas (1970). Table 3 Comparison of nondimensional stress r x of simply-supported orthotropic rectangular plate (under uniformly distributed transverse load) a Plate dimensional parameters Nondimensional stress r x at x = a/, y = b/, z = h/ by various theories a/b h/a h/b EXACT b Reddy c Reissner b CPT b RPT 0.5 0.05 0.05 6.67 6.6 6.07 6.6 6.78 0.1 0.05 65.975 65.95 65.379 65.564 66.07 0.14 0.07 33.86 33.84 33.65 33.451 33.96 1.0 0.05 0.05 144.31 144.3 143.87 144.39 144.68 0.1 0.1 36.01 36.01 35.578 36.098 36.36 0.14 0.14 18.346 18.34 17.906 18.417 18.68.0 0.05 0.1 40.657 40.67 40.477 40.86 40.98 0.1 0. 10.05 10.05 9.846 10.15 10.33 0.14 0.8 5.0364 5.068 4.8603 5.118 5.3 a At x = a/, y = b/ z = h/ r x ¼ r x =q 0 (E /E 1 = 0.5500, G 1 /E 1 = 0.693, G 13 /E 1 = 0.15991, G 3 /E 1 = 0.6681, l 1 = 0.44046, l 1 = 0.314). b Taken from reference of Srinivas and Rao (1970). c Taken from reference of Reddy (1984).

6794 o 4 D 11 ox þ ðd 4 1 þ D 66 Þ o4 ox oy þ D o 4 qh3 o r o w oy 4 1 ot b þ qh ð ot w b þ w s Þ ¼ 0 ð63þ o w s A 55 ox þ A o w s 44 þ 1 oy 84 D o 4 w s 11 ox þ ðd 4 1 þ D 66 Þ o4 w s ox oy þ D o 4 w s qh3 o r w oy 4 84 1 ot s þ qh o ð ot w b þ w s Þ ¼ 0 ð64þ 6.. Boundary conditions for illustrative examples (free vibration analysis) Boundary conditions are as those of illustrative example I (i.e., expressions (49) (58)). Table 4 Comparison of nondimensional stress r y of simply-supported orthotropic rectangular plate (under uniformly distributed transverse load) a Plate dimensional parameters Nondimensional stress r y at x = a/, y = b/, z = h/ by various theories a/b h/a h/b EXACT b Reissner b CPT b RPT 0.5 0.05 0.05 79.545 79.337 79.11 79.30 0.1 0.05 0.04 0.001 19.78 19.94 0.14 0.07 10.515 10.31 10.09 10.5 1.0 0.05 0.05 87.08 86.91 86.487 86.68 0.1 0.1.1.048 1.6 1.80 0.14 0.14 11.615 11.453 11.031 11.1.0 0.05 0.1 54.79 54.134 53.838 54.04 0.1 0. 13.888 13.743 13.46 13.65 0.14 0.8 7.794 7.1358 6.8671 7.06 Note: Results are not available for Reddy s theory. a At x = a/, y = b/, z = h/ r y ¼ r y =q 0 (E /E 1 = 0.5500, G 1 /E 1 = 0.693, G 13 /E 1 = 0.15991, G 3 /E 1 = 0.6681, l 1 = 0.44046, l 1 = 0.314). b Taken from reference of Srinivas and Rao (1970). Table 5 Comparison of nondimensional stress s zx of simply-supported orthotropic rectangular plate (under uniformly distributed transverse load) a Plate dimensional Nondimensional stress s zx at x =0,y = b/, z = 0 by various theories parameters a/b h/a h/b EXACT b Reddy c Reissner b CPT b RPT Direct Indirect Direct Indirect Direct Indirect 0.5 0.05 0.05 14.048 13.98 14.00 14.114 0.0 14.154 1.67 14.03 0.1 0.05 6.966 6.958 6.998 7.0611 0.0 7.138 6.31 6.78 0.14 0.07 4.878 4.944 4.997 5.0445 0.0 5.105 4.47 4.70 1.0 0.05 0.05 10.873 10.85 10.88 10.864 0.0 10.97 8.045 10.87 0.1 0.1 5.3411 5.38 5.4 5.467 0.0 5.564 4.0 5.31 0.14 0.14 3.7313 3.805 3.857 3.8741 0.0 3.986.839 3.585.0 0.05 0.1 6.434 6.163 6.184 6.191 0.0 6.4175 4.151 6.9 0.1 0..9573.885 3.044 3.054 0.0 3.40.03.868 0.14 0.8 1.9987.080.131.148 0.0.3175 1.417 1.9 Direct Shear stresses are calculated using the constitutive relations. Indirect Shear stresses are calculated using the equilibrium equations. a At x =0, y = b/, z =0 s zx ¼ s zx =q 0 (E /E 1 = 0.5500, G 1 /E 1 = 0.693, G 13 /E 1 = 0.15991, G 3 /E 1 = 0.6681, l 1 = 0.44046, l 1 = 0.314). b Taken from reference of Srinivas and Rao (1970). c Taken from reference of Reddy (1984).

6795 6.3. Solution of illustrative examples (free vibration analysis) The following displacement functions and w s satisfy the boundary conditions (49) (58): ¼ X1 X 1 W bmn sin mpx sin npy sinðx mn tþ a b m¼1;;... n¼1;;... w s ¼ X1 X 1 W smn sin mpx sin npy sinðx mn tþ a b m¼1;;... n¼1;;... where, W bmn, W smn are constant coefficients and x mn is the circular frequency of vibration associated with mth mode in x-direction and nth mode in y-direction. ð65þ ð66þ Table 6 Comparison of non-dimensional natural frequencies x mn of simply-supported orthotropic rectangular plate a m n Non-dimensional natural frequency x mn by various theories EXACT b Reddy c Reissner b CPT b RPT 1 1 0.0474 0.0474 0.0474 0.0497 0.0477 1 0.1033 0.1033 0.103 0.110 0.1040 1 0.1188 0.1189 0.1187 0.1354 0.1198 0.1694 0.1695 0.169 0.1987 0.17 1 3 0.1888 0.1888 0.1884 0.154 0.1898 3 1 0.180 0.184 0.178 0.779 0.197 3 0.475 0.477 0.469 0.309 0.50 3 0.64 0.69 0.619 0.3418 0.675 1 4 0.969 0.969 0.959 0.3599 0.980 4 1 0.3319 0.3330 0.3311 0.4773 0.3340 3 3 0.33 0.336 0.331 0.4470 0.3407 4 0.3476 0.3479 0.3463 0.4480 0.3534 4 0.3707 0.370 0.3696 0.5415 0.3774 pffiffiffiffiffiffiffiffiffiffiffiffi a x mn ¼ x mn h q=q 11 ; h/a = 0.1, b/a = 1.0; (E /E 1 = 0.5500, G 1 /E 1 = 0.693, G 13 /E 1 = 0.15991, G 3 /E 1 = 0.6681, l 1 = 0.44046, l 1 = 0.314). b Taken from reference of Srinivas and Rao (1970). c Taken from reference of Reddy (1984). Table 7 Comparison of non-dimensional natural frequencies ^x mn of simply-supported isotropic square plate a m n Non-dimensional natural frequency ^x mn by various theories EXACT b Reddy b Reissner b CPT b RPT 1 1 0.093 0.0931 0.0930 0.0955 0.0930 1 0.6 0. 0.19 0.360 0.0 0.341 0.3411 0.3406 0.373 0.3406 1 3 0.4171 0.4158 0.4149 0.469 0.4151 3 0.539 0.51 0.506 0.5951 0.508 1 4 0.6545 0.650 0.7668 0.655 3 3 0.6889 0.686 0.6834 0.8090 0.6840 4 0.7511 0.7481 0.7446 0.896 0.7454 3 4 0.8949 0.8896 1.0965 0.8908 1 5 0.968 0.930 0.9174 1.1365 0.9187 5 1.0053 0.9984 1.549 1.0001 4 4 1.0889 1.0847 1.0764 1.3716 1.0785 3 5 1.1361 1.168 1.4475 1.19 Against an entry p ffiffiffiffiffiffiffiffiffi indicates that results/data are not available. ^x mn ¼ x mn h q=g ; h=a ¼ 0:1, b/ a = 1.0 (E = E 1 = E, G 1 = G 13 = G 3 = G =[E/(1 + l)], l 1 = l 1 = l = 0.3). b Taken from reference of Reddy and Phan (1985).

6796 Using expressions (65) and (66) in the governing Eqs. (63) and (64), one obtains two equations, which can be written in matrix form, as follows: K 11mn 0 W bmn M x 11mn M 1mn W bmn mn ¼ 0 ð67þ 0 K mn W smn M 1mn M mn W smn 0 where, M 11mn ¼ qh3 mp np þ þ qh 1 a b M 1mn ¼ qh M mn ¼ qh3 mp np þ þ qh 1008 a b Eq. (67) is of standard eigen value form, and solving it one gets free vibration frequencies. Here it is observed from the Eq. (67) that bending and shearing components are elastically uncoupled, but are inertially coupled. The following nondimensionalized free vibration frequencies x mn, ^x mn are tabulated in Tables 6 and 7. pffiffiffiffiffiffiffiffiffiffiffiffi x mn ¼ x mn hp q=q ffiffiffiffiffiffiffiffiffi 11 : for orthotropic plate and ^x mn ¼ x mn h q=g : for isotropic plate. Results are discussed separately after Table 7. It should be noted that RPT can be adapted, apart from solving plates of rectangular planform, to solve plates of any other planform also. But, in such cases, analytical solutions would not be feasible. For solving such problems, it would be necessary to use some numerical techniques (e.g. devising and making use of finite element based on the present theory). 7. Discussion on results Static analysis results using RPT are tabulated in Tables 1 5 and free vibration (predominantly bending frequency) results are tabulated in Tables 6 and 7. The tables also present solutions obtained using exact theory (Srinivas et al., 1970), Reddy s theory {termed as Higher-order Shear Deformation Plate Theory (HSDPT) in references of Reddy (1984), Reddy and Phan (1985)}, Reissner s theory {termed as First-order Shear Deformation Plate Theory (FSDPT) in references of Reddy (1984), Reddy and Phan (1985)} and classical plate theory (CPT) taking into account rotary inertia {as reported in references of Reddy (1984), Reddy and Phan (1985)}. It is to be noted that results by Reddy (1984) are calculated for m,n =1,3,...,19. Whereas, present results are obtained using same values of m and n as those used for obtaining results using exact theory (Srinivas and Rao, 1970). Exact theory results available in the literature (Srinivas et al., 1970; Srinivas and Rao, 1970) are used as basis for comparison of results obtained by various theories. The % error is calculated as follows: value obtained by a theory % error ¼ corresponding value by exact theory 1 100 7.1. Discussion on static analysis results On results obtained for displacement Orthotropic plate: Table 1 shows the comparison of static deflection at the center of simply-supported orthotropic plate obtained by various theories. RPT shows very good accuracy in results e.g., maximum error for the present results is 1.74% for thick square plate case (a/b = 1, h/a = 0.14). Whereas CPT has maximum error of 18.7% for a/b =,h/a = 0.14 case. Reddy s and Reissner s theories give marginally

improved results (maximum error in Reddy s theory is 1.06%, maximum error in Reissner s theory is 0.01%) as compared to those obtained by RPT. Isotropic plate: Table shows the comparison of static deflection at the center of simply-supported isotropic plate obtained by various theories. CPT has 18.11% error for a thick rectangular plate (a/b =, h/a = 0.14). RPT shows very good accuracy in results e.g.,.46% error for the same plate Reissner s theory gives marginally improved results ( 0.58% error for the same plate) in comparison with RPT. Reddy s results are not available for displacement. On results obtained for stresses for orthotropic plate Stress r x : Table 3 shows comparison of stress r x obtained by various theories. CPT and Reissner s theory give almost same stress results. RPT gives accurate stresses than CPT and Reissner s theory for moderately thick plates (e.g., for a/b = 0.5, h/a = 0.05, 0.1, 0.14) but % error is more for thicker plate case (e.g., 5.67% error for rectangular plate with a/b =, a/h = 0.14, for the same case CPT and Reissner s theory both have 3.5% error). Reddy s theory shows good accuracy (i.e., maximum % error is about 0.63%). Stress r y : Table 4 shows comparison of stress r y obtained by various theories. Results using Reddy s theory are not available. For a rectangular plate (having a/b =.0, h/a = 0.14) result for r y for RPT have maximum error of.99%, whereas CPT has 5.66% and Reissner s theory has 1.97%. RPT results are more accurate than those of CPT. Reissner s theory shows marginally improved accuracy than those of RPT. Stress s zx : Table 5 shows comparison of shear stress s zx obtained by various theories. Direct method of calculating shear stresses involves use of constitutive relations. Indirect method makes use of equilibrium equations. For example, in the case of CPT, transverse shear stresses cannot be obtained using shear stress and shear strain constitutive relations, and these are required to be obtained by indirect method. In this method, first stresses r x, r y and s xy are obtained. Then, these stresses are substituted in the equilibrium equations of the three dimensional theory of elasticity, and then integrating the equations and finding the constants of integrations, one obtains the expressions for transverse shear stresses s zx and s yz. When the indirect method is used, results obtained by RPT are the best amongst the results presented. For a square plate case (a/b = 1.0, a/h = 0.14) results obtained by RPT have 3.9% error. It is interesting to note that even though Reissner s is a stress based approach, % error in shear stress results obtained using his approach is quiet high (e.g., 7.1% for rectangular plate having a/b =.0, a/ h = 0.14). Using CPT, shear stresses can be found out only using indirect method and % error increases for thicker plates. Unlike CPT, using RPT shear stresses can be found out using direct method, but the % error is quite high (for thick plate, a/b =.0, h/a = 0.14 case error is 9.1%). Using Reddy s theory, shear stresses obtained by direct as well as indirect methods shatisfactory results (around 5% error), but it is to be noted that stresses converge very slowly (Reddy, 1984) and results quoted for Reddy s theory are only for m, n = 1,3,......19. Use of more number of terms may lead to different conclusion. 7.. Discussion on free vibration results Results of free vibration frequencies for an orthotropic and also for isotropic square plate (b/a = 1,h/a = 0.1) are presented in Tables 6 and 7, respectively. Also, results quoted are in ascending order of frequencies. When m = 1 and n = 1, the frequency indicated is the fundamental frequency. The following observations can be made: Orthotropic plate: CPT results are not satisfactory for higher modes (e.g., error is 46%, when m = 4, n = ). RPT gives good accuracy (e.g., maximum % error for the RPT results presented is.51% when m = 3 and n = 3). Reddy s theory and Reissner s theory give slightly better accuracy than RPT (e.g., for the results presented in respect of Reddy s and Reissner s theories, maximum % error is about 0.36%). It is to be noted that RPT has only one variable for capturing shear effects unlike other theories. Isotropic plate: It is observed from Table 7 that CPT results are not satisfactory for higher modes (e.g., error is 6%, when m =4,n = 4). RPT gives good accuracy (e.g., error is 0.96%, when m =4,n = 4). Reddy s theory and Reissner s theory give marginally accurate results than RPT (e.g., Reddy has 0.39% error and Reissner has 1.15% error, when m =4,n = 4). 6797

6798 It should be noted that RPT involves only two unknown functions and two differential equations as against five unknown functions and five differential equations in case of Reddy s theory, and three unknown functions and three differential equations in case of Reissner s theory. Moreover, in RPT, both the differential equations are only inertially coupled and there is no elastic coupling; and, therefore, the equations are easier to solve. Whereas, in Reddy s theory as well as Reissner s theory, all the differential equations are inertially as well as elastically coupled, and therefore, these equations are comparatively more difficult to solve. Also unlike Reissner s theory, RPT satisfies shear stress free boundary conditions, and also constitutive relations in respect of transverse shear stresses and strains (and, therefore, does not require shear correction factor). CPT results are not satisfactory in respect of thick plates for static flexure as well as for free vibrations. 8. Concluding remarks In this paper, a new two variable refined plate theory (RPT) is presented for orthotropic plate analysis. The efficacy of the theory has been demonstrated for static and free vibration problems. The following points need to be noted in respect of RPT: 1. Use of the theory results in two fourth order governing differential equations. For static analysis case, both equations are fully uncoupled; whereas for dynamic analysis case, both equations are only inertially coupled and there is no elastic coupling at all. No other theory, to the best of the knowledge of the authors, has this feature.. Number of unknown functions involved in the theory is only two. Even in the Reissner s and Mindlin s theory (first order shear deformation theories), three unknown functions are involved. 3. The theory is variationally consistent. 4. The theory has strong similarity with the classical plate theory in many aspects (in respect of a governing equation, boundary conditions, moment expressions). 5. (a) The theory assumes displacements such that transverse shear stress variation is realistic (giving rise to shear stress free surfaces and parabolic variation of shear stress across the thickness). (b) Constitutive relations in respect of shear stresses and shear strains are satisfied (and, therefore, shear correction factor is not required). 6. The classical plate theory (CPT) comes out as a limiting case of RPT formulation (in situations wherein shear effects become ignorable, e.g., flexure of thin plate). Therefore, the finite elements based on RPT will be free from shear locking. 7. CPT involves use of only one unknown function and only one differential equation. Compared to CPT, use of RPT involves only one additional function and one additional number of differential equation. But, when RPT is used, gain in accuracy of results is not only substantial as compared to CPT, but almost comparable to higher order theories containing more number of unknown functions. 8. As mentioned in Section 4, RPT is equally applicable to thick as well as thin plate analysis. However, with increase in thickness of plate and increase in orthotropy of plate material, the results using RPT may show a trend (which is common to many other two-dimensional theories) of slight decrease in accuracy of results. In conclusion, it can be said that for orthotropic plate analysis, RPT can be successfully utilized as it is the simplest yet accurate shear deformable theory with only two variables. References Bhimaraddi, A., Stevens, L.K., 1984. A higher order theory for free vibration of orthotropic, homogeneous, and laminated rectangular plates. ASME Journal of Applied Mechanics 51 (1), 195 198. Donnell, L.H., 1976. Beams, Plates and Shells. McGraw-Hill Book Company, New York.

Ghugal, Y.M., Shimpi, R.P., 00. A review of refined shear deformation theories of isotropic and anisotropic laminated plates. Journal of Reinforced Plastics and Composites 1 (9), 775 813. Hanna, N.F., Leissa, A.W., 1994. A higher order shear deformation theory for the vibration of thick plates. Journal of Sound and Vibration 170 (4), 545 555. Jones, R.M., 1999. Mechanics of Composite Materials, second ed. Taylor and Francis, Inc., Philadelphia. Kant, T., 198. Numerical analysis of thick plates. Computer Methods in Applied Mechanics and Engineering 31 (1), 1 18. Krishna Murty, A.V., 1977. Higher order theory for vibrations of thick plates. AIAA Journal 15 (1), 183 184. Levinson, M., 1980. An accurate, simple theory of the statics and dynamics of elastic plates. Mechanics Research Communications 7 (6), 343 350. Librescu, L., 1975. Elastostatics and Kinetics of Anisotropic and Heterogeneous Shell-Type Structures. Noordhoff International, Leyden, The Netherlands. Liew, K.M., Xiang, Y., Kitipornchai, S., 1995. Research on thick plate vibration: a literature survey. Journal of Sound and Vibration 180 (1), 163 176. Lo, K.H., Christensen, R.M., Wu, E.M., 1977a. A high-order theory of plate deformation. Part 1: Homogeneous plates. ASME Journal of Applied Mechanics 44 (4), 663 668. Lo, K.H., Christensen, R.M., Wu, E.M., 1977b. A high-order theory of plate deformation. Part : Laminated plates. ASME Journal of Applied Mechanics 44 (4), 669 676. Medwadowski, S.J., 1958. A refined theory of elastic, orthotropic plates. ASME Journal of Applied Mechanics 5, 437 441. Mindlin, R.D., 1951. Influence of rotatory inertia and shear on flexural motions of isotropic, elastic plates. ASME Journal of Applied Mechanics 18 (1), 31 38, Transactions ASME 73. Nelson, R.B., Lorch, D.R., 1974. A refined theory for laminated orthotropic plates. Journal of Applied Mechanics 41, 177 183. Reddy, J.N., 1984. A refined nonlinear theory of plates with transverse shear deformation. International Journal of Solids and Structures 0 (9/10), 881 896. Reddy, J.N., 1990. A general non-linear third-order theory of plates with moderate thickness. International Journal of Non-Linear Mechanics 5 (6), 677 686. Reddy, J.N., Phan, N.D., 1985. Stability and vibration of isotropic, orthotropic and laminated plates according to a higher-order shear deformation theory. Journal of Sound and Vibration 98 (), 157 170. Reissner, E., 1944. On the theory of bending of elastic plates. Journal of Mathematics and Physics 3, 184 191. Reissner, E., 1945. The effect of transverse shear deformation on the bending of elastic plates. ASME Journal of Applied Mechanics 1 (), A69 A77, Transactions ASME 67. Shimpi, R.P., 00. Refined plate theory and its variants. AIAA Journal 40 (1), 137 146. Soldatos, K.P., 1988. On certain refined theories for plate bending. ASME Journal of Applied Mechanics 55 (4), 994 995. Srinivas, S., 1970. Three dimensional analysis of some plates and laminates and a study of thickness effects. Ph.D. Thesis, Dept. of Aeronautical Engineering, Indian Institute of Science, Bangalore, India. Srinivas, S., Rao, A.K., 1970. Bending, vibration and buckling of simply-supported thick orthotropic rectangular plates and laminates. International Journal of Solids and Structures 6 (11), 1463 1481. Srinivas, S., Joga Rao, C.V., Rao, A.K., 1970. An exact analysis for vibration of simply-supported homogeneous and laminated thick rectangular plates. Journal of Sound and Vibration 1 (), 187 199. Vasil ev, V.V., 199. The theory of thin plates. Mechanics of Solids (Mekhanika Tverdogo Tela) 7 (3), 4 (English translation from original Russian by Allerton Press, New York). Wang, C.M., Lim, G.T., Reddy, J.N., Lee, K.H., 001. Relationships between bending solutions of Reissner and Mindlin plate theories. Engineering Structures 3 (7), 838 849. 6799