Co py rig hte dM ate ria l

CHAPTER 12 BENDING OF PLATES UNDER THE COMBINED ACTION OF LATERAL LOADS AND FORCES IN THE MIDDLE PLANE OF THE PLATE 90. Differential Equation of the Deflection Surface. In our previous discussion it has always been assumed that the plate is bent by lateral loads only. If in addition to lateral loads there are forces acting in the middle plane of the plate, these latter forces may have a considerable effect on the bending of the plate and must be considered in deriving the corresponding differential equation of the deflection surface. Proceeding as in the case of lateral loading (see Art. 21, page 79), we consider the equilibrium of a small element cut from the plate by two pairs of planes parallel to the xz and yz coordinate planes (Fig. 191). In addition to the forces discussed in Art. 21 we now have forces acting in the middle plane of the plate. We denote the magnitude of these forces per unit length by Nx, Nyj and Nxy = Nyx, as shown in the figure. Projecting these forces on the x and y axes and assuming that there are no body forces or tangential forces acting in those directions at the faces of the plate, we obtain the following equations of equilibrium:
dNx
dx

dNxy _
dy

rciftt

eNxv

dNy _

(216)

These equations are entirely independent of the three equations of equilibrium considered in Art. 21 and can be treated separately, as FIG. 191 will be shown in Art. 92. In considering the projection of the forces shown in Fig. 191 on the z axis, we must take into account the bending of the plate and the resulting small angles between the forces Nx and Ny that act on the opposite sides of the element. As a result of this bending the projection

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

of the normal forces Nx on the z axis gives

Co py rig hte dM ate ria l

After simplification, if the small quantities of higher than the second order are neglected, this projection becomes In the same way the projection of the normal forces Ny on the z axis gives Regarding the projection of the shearing forces Nxy on the z axis, we observe that the slope of the deflection surface in the y direction on the two opposite sides of the element is dw/dy and dw/dy + (d2w/dx By) dx. Hence the projection of the shearing forces on the z axis is equal to An analogous expression can be obtained for the projection of the shearing forces Nyx Nxy on the z axis. The final expression for the projection of all the shearing forces on the z axis then can be written as Adding expressions (a), (6), and (c) to the load q dx dy acting on the element and using Eqs. (216), we obtain, instead of Eq. (100) (page 81), the following equation of equilibrium: Substituting expressions (101) and (102) for Mx, My, and Mxyy we obtain (217) This equation should be used instead of Eq. (103) in determining the deflection of a plate if in addition to lateral loads there are forces in the middle plane of the plate.
Copyright 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

If there are body forces1 acting in the middle plane of the plate or tangential forces distributed over the surfaces of the plate, the differential equations of equilibrium of the element shown in Fig. 191 become

Co py rig hte dM ate ria l

(218) Here X and Y denote the two components of the body forces or of the tangential forces per unit area of the middle plane of the plate. Using Eqs. (218), instead of Eqs. (216), we obtain the following differential equation2 for the deflection surface: (219) Equation (217) or Eq. (219) together with the conditions at the boundary (see Art. 22, page 83) defines the deflection of a plate loaded laterally and submitted to the action of forces in the middle plane of the plate. 91. Rectangular Plate with Simply Supported Edges under the Combined Action of Uniform Lateral Load and Uniform Tension. Assume that the plate is under uniform tension in the FIG. 192 x direction, as shown in Fig. 192. The uniform lateral load q can be represented by the trigonometric series (see page 109). Equation (217) thus becomes This equation and the boundary conditions at the simply supported edges
1

An example of a body force acting in the middle plane of the plate is the gravity force in the case of a vertical position of a plate. 2 This differential equation has been derived by Saint Venant (seefinalnote 73) in his translation of Clebsch, "Theorie de Telasticite des corps solides," p. 704, 1883.

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

will be satisfied if we take the deflection w in the form of the series

Co py rig hte dM ate ria l

Substituting this series in Eq. (6), we find the following values for the coefficients amn: in which m and n are odd numbers 1, 3, 5, . . . , and amn = 0 if m or n or both are even numbers. Hence the deflection surface of the plate is Comparing this result with solution (131) (page 110), we conclude from the presence of the term Nxm2/ir2Da2 in the brackets of the denominator that the deflection of the plate is somewhat diminished by the action of the tensile forces Nx. This is as would be expected. By using M. Levy's method (see Art. 30) a solution in simple series may be obtained which is equivalent to expression (e) but more convenient for numerical calculation. The maximum values of deflection and bending moments obtained in this way1 for v = 0.3 can be represented in the form The constants a, /3, and /3i depend upon the ratio a/b and a parameter and are plotted in Figs. 193, 194, and 195. If, instead of tension, we have compression, the force Nx becomes
1 H. D. Conway, J. Appl. Mechanics, vol. 16, p. 301, 1949, where graphs in the case of compression are also given; the case Nx = Ny has been discussed by R. F. Morse and H. D. Conway, J. Appl. Mechanics, vol. 18, p. 209, 1951, and the case of a plate clamped all around by C. C. Chang and H. D. Conway, J. Appl. Mechanics, vol. 19, p. 179, 1952. For combined bending and compression, see also J. Lockwood Taylor, The Shipbuilder and Marine Engine Builder, no. 494, p. 15, 1950.

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

Co py rig hte dM ate ria l

FIG.

negative, and the deflections (e) become larger than those of the plate bent by lateral load only. It may be seen also in this case that at certain values of the compressive force Nx the denominator of one of the terms in series (e) may vanish. This indicates that at such values of Nx the plate may buckle laterally without any lateral loading. 92. Application of the Energy Method. The energy method, which was previously used in discussing bending of plates by lateral loading (see Art. 80, page 342), can be applied also to the cases in which the

193

lateral load is combined with forces acting in the middle plane of the plate. To establish the expression for the strain energy corresponding to the latter forces let us assume that these forces are applied first to the unbent plate. In this way we obtain a two-dimensional problem which can be treated by the methods of the theory of elasticity.1 Assuming that this problem is solved and that the forces Nx, Ny, and Nxy are known at each point of the plate, the components of strain of the middle plane of the plate are obtained from the known formulas representing Hooke's
1

SeC, for example, S. Timoshenko and J. N. Goodier, "Theory of Elasticity," 2d ed., p. 11, 1951.

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

law, viz.,

Co py rig hte dM ate ria l

The strain energy, due to stretching of the middle plane of the plate, is then
(220)

where the integration is extended over the entire plate. Let us now apply the lateral load. This load will bend the plate and produce additional strain of the middle plane. In our previous discussion of bending of plates, this latter strain was always neglected. Here,

FIG.

194

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

Co py rig hte dM ate ria l

FIG.

195

however, we have to take it into consideration, since this small strain in combination with the finite forces Nx, Nv, Nxy may add to the expression for strain energy some terms of the same order as the strain energy of bending. The x, y, and z components of the small displacement that a point in the middle plane of the plate experiences during bending will be denoted by U1 v. and w, respectively. Considering a linear element AB of that plane in the x direction, it may be seen from Fig. 196 that the elongation of the element due to the displacement u is equal to (du/dx) dx. The elongation of the same element due to the displacement w is FIG. 196 %(dw/dx)2 dx, as may be seen from the comparison of the length of the element AiBi in Fig. 196 with the length of its projection on the x axis. Thus the total unit elongation in the x direction of an element taken in the middle plane of the plate is (221)

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

Similarly the strain in the y direction is

(222)

Co py rig hte dM ate ria l

Considering now the shearing strain in the middle plane due to bending, we conclude as before (see Fig. 23) that the shearing strain due to the displacements u and v is du/dy + dv/dx. To determine the shearing strain due to the displacement w we take two infinitely small linear elements OA and OB in the x and y directions, as shown in Fig. 197. Because of displacements in the z direction these elements come to the positions OiAi and 0\B\. The difference between the angle TT/2 and the angle AiOiBi is the shearing strain corresponding to the displacement w. To determine this difference we consider the right angle .B2OiAi, in which B2Oi is parallel to BO. Rotating the plane .B2OiAi about the axis OiAi by the angle dw/dy, we bring the plane B2OiAi into coincidence with the plane BiOiA 1* and the point B2 to position C. The displacement B2<7 is equal to (dw/dy) dy and is inclined to the vertical B2Bi by the angle dw/dx. Hence BiC is equal to (dw/dx) (dw/dy) dy, FIG. 197 and the angle C0iBh which represents the shearing strain corresponding to the displacement w} is (dw/dx) (dw/dy). Adding this shearing strain to the strain produced by the displacements u and v, we obtain
(223)

Formulas (221), (222), and (223) represent the components of the additional strain in the middle plane of the plate due to small deflections. Considering them as very small in comparison with the components ex, ey, and yxy used in the derivation of expression (220), we can assume that the forces Nx, Ny, Nxy remain unchanged during bending. With this assumption the additional strain energy of the plate, due to the strain produced in the middle plane by bending, is

Substituting expressions (221), (222), and (223) for e'XJ e'y, and y'xy, we

* The angles dw/dy and dw/dx correspond to small deflections of the plate and are regarded as small quantities.

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

finally obtain

Co py rig hte dM ate ria l

(224)

It can be shown, by integration by parts, that the first integral on the right-hand side of expression (224) is equal to the work done during bending by the forces acting in the middle plane of the plate. Taking, for example, a rectangular plate with the coordinate axes directed, as shown in Fig. 192, we obtain for the first term of the integral

Proceeding in the same manner with the other terms of the first integral in expression (224), we finally find

The first integral on the right-hand side of this expression is evidently equal to the work done during bending by the forces applied at the edges x = 0 and x = a of the plate. Similarly, the second integral is equal to the work done by the forces applied at the edges y = 0 and y = b. The last two integrals, by virtue of Eqs. (218), are equal to the work done during bending by the body forces acting in the middle plane. These integrals each vanish in the absence of such corresponding forces. Adding expressions (220) and (224) to the energy of bending [see Eq. (117), page 88], we obtain the total strain energy of a bent plate under the combined action of lateral loads and forces acting in the middle plane of the plate. This strain energy is equal to the work Tv done by the lateral load during bending of the plate plus the work Th done by the forces acting in the middle plane of the plate. Observing that this latter work is equal to the strain energy Vi plus the strain energy represented by the first integral of expression (224), we conclude that the work pro-

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

duced by the lateral forces is

Co py rig hte dM ate ria l

(225)

Applying the principle of virtual displacement, we now give a variation Bw to the deflection w and obtain, from Eq. (225),

(226)

The left-hand side in this equation represents the work done during the virtual displacement by the lateral load, and the right-hand side is the corresponding change in the strain energy of the plate. The application of this equation will be illustrated by several examples in the next article.

93. Simply Supported Rectangular Plates under the Combined Action of Lateral Loads and of Forces in the Middle Plane of the Plate. Let us

begin with the case of a rectangular plate uniformly stretched in the x direction (Fig. 192) and carrying a concentrated load P at a point with coordinates and i). The general expression for the deflection that satisfies the boundary conditions is

To obtain the coefficients amn in this series we use the general equation (226). Since Ny = Nxy = 0 in our case, the first integral on the righthand side of Eq. (225), after substitution of series (a) for w, is

The strain energy of bending representing the second integral in Eq. (225) is [see Eq. (d), page 343]

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

To obtain a virtual deflection 8w we give to a coefficient amix an increase 8amini. The corresponding deflection of the plate is

Co py rig hte dM ate ria l

The work done during this virtual displacement by the lateral load P is The corresponding change in the strain energy consists of the two terms which are Substituting expressions (d) and (e) in Eq. (226), we obtain from which Substituting these values of the coefficients amifll in expression (a), we find the deflection of the plate to be If, instead of the tensile forces Nx, there are compressive forces of the same magnitude, the deflection of the plate is obtained by substituting Nx in place of Afx in expression (g). This substitution gives
Copyright 1959 by The McGraw-Hill Companies Retrieved from: www.knovel.com

Co py rig hte dM ate ria l

(227) where m must be chosen so as to make expression (227) a minimum. Plotting the factor against the ratio a/b, for various integral values of m, we obtain a system of curves shown in Fig. 198. The portions of the curves that must be
FIG.

The smallest value of Nx at which the denominator of one of the terms in expression (Zi) becomes equal to zero is the critical value of the compressive force Nx. It is evident that this critical value is obtained by taking n = 1. Hence

198

used in determining k are indicated by heavy lines. I t is seen that the factor k is equal to 4 for a square plate as well as for any plate that can be subdivided into an integral number of squares with the side 6. I t can also be seen that for long plates k remains practically constant at a value of 4.* Since the value of m in Eq. (227) may be other than 1 for oblong plates, such plates, being submitted to a lateral load combined with compression, do not generally deflect1 in the form of a half wave in the direction of the longer side of the plate. If, for instance, a/b = 2, 4, . . . the respective elastic surface becomes markedly unsymmetrical with respect to the middle line x = a/2 (Fig. 192), especially so for values of Nx close to the critical value (Nx)cr. By using the deflection (g) produced by one concentrated load, the * A more detailed discussion of this problem is given in S. Timoshenko, "Theory of Elastic Stability," p. 327, 1936. 1 Several examples of such a deformation have been considered by K. Girkmann, Stahlbau, vol. 15, p. 57, 1942,

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

Co py rig hte dM ate ria l

The critical value of the forces Nx and Ny is obtained from the condition1 where m and n are chosen so as to make Nx and Ny a minimum for any given value of the ratio Nx/Ny. In the case of a square plate submitted to the action of a uniform pressure p in the middle plane we have a = b and Nx = Ny = p. Equation (j) then gives The critical value of p is obtained by taking m = n = 1, which gives (228) In the case of a plate in the form of an isosceles right triangle with simply supported edges (Fig. 161) the deflection surface of the buckled plate which satisfies all the boundary conditions is2 Thus the critical value of the compressive stress is obtained by substituting m 1, n = 2 or m 2} n = 1 into expression (/c). This gives (229)
1 A complete discussion of this problem is given in Timoshenko, "Elastic Stability," p. 333. 2 This is the form of natural vibration of a square plate having a diagonal as a nodal line.

deflection produced by any lateral load can be obtained by superposition. Assuming, for example, that the plate is uniformly loaded by a load of intensity q, we substitute q d% drj for P in expression (g) and integrate the expression over the entire area of the plate. In this way we obtain the same expression for the deflection of the plate under uniform load as has already been derived in another manner (see page 381). If the plate laterally loaded by the force P is compressed in the middle plane by uniformly distributed forces Nx and Ny, proceeding as before we obtain

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

Co py rig hte dM ate ria l

in which In the case of a circular plate without a hole1 the solution of Eq. (a) is of the form
FIG.

94. Circular Plates under Combined Action of Lateral Load and Tension or Compression. Consider a circular plate (Fig. 199) submitted to the simultaneous action of a symmetrical lateral load and a uniform compression Nr = Nt = N in the middle plane of the plate. Owing to the slope <p of the deformed plate (Fig. 27) the radial compression N gives a transverse component N d<p/dr which we have to add to the shearing force Q (Fig. 28) due to the lateral load. Hence the differential equation (54) becomes

199

where J\ is the Bessel function of the order one, ^ 0 a particular solution of Eq. (a) depending on Q, and Ci a constant defined by the boundary conditions of the plate. Let us take as an example a rigidly clamped2 plate carrying a uniform load of intensity q. Then, as a particular solution, we use

and therefore

It follows, by integration, that

where J0 is the Bessel function of the order zero and C2 a second constant. Having calculated Ci from the condition <p = 0 on r = a, and C2 from the condition w = 0 on r = a, we obtain the final solution3

The deflections (/) become infinite for Jx(Iz) = 0. Denoting the zeros of the function J1 in order of their magnitude by J1, j 2 , . . . we see that the condition k = jx

1 In the case of a concentric hole a term proportional to a Bessel function of second kind must be added to expression (c). The inner boundary must be submitted then to the same compression N, or else the problem becomes more complex because of the inconstancy of stresses Nr and Nt. 2 The case of an elastic restraint without transverse load has been discussed by H. Reismann, / . Appl. Mechanics, vol. 19, p. 167, 1952. 3 This result may be found in A. Nadai, "Elastische Platten," p. 255, Berlin, 1925.

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

defines the lowest critical value

Co py rig hte dM ate ria l

of the compressive stress iV. Now, for the function J\(k) we have the expression in which ji = 3.83171, J2 = 7.01559, . . . . k2/j2 beginning with the second parentheses. As k < ji we can neglect the terms Observing, furthermore, that by virtue of Eqs. (6) and (g) we have, approximately, where Making use of the expression (i), it can be shown that, approximately, 1 where wq is the deflection due to the load q alone. Cases with other boundary conditions and other laws of distribution of the lateral load may be handled in like manner. In the general case of a symmetrical lateral load combined with compression we can put, approximately, for the center of the plate (r = 0) and on the boundary (r = a) where wq relates to a plate carrying the given lateral load alone and a. = N/Ncr has the following meaning: Na2 For a simply supported plate: a. = For a clamped plate:
1

a =

Na2

14.DoJL'

See O. Pettersson, Acta Polytech., Stockholm, no. 138, 1954. The following results are taken from this paper, in which, more generally, an elastic restraint at the edge is assumed.

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

TABLE 81. VALUES OF CONSTANTS IN APPROXIMATE EXPRESSIONS (JC) AND (I) nn

Co py rig hte dM ate ria l

Case
1

Load distribution

Boundary conditions

Constants

Uniform edge couples

Simply supported

C0 = 0.305 c' = -0.270 c" = -1.219

Uniform load

Simply supported

C0 = 0.0480 c' = c" = -0.0327

Clamped

c"

C0 = 0.308 = -0.473
C

Central uniform load over area Simply supported of radius ea

, 2.153 i 1 + " i-1.31n. c' = c" = 0.205 1.308 . In e

Clamped

C0 =

c" = 0.0539

the former value being valid for v = 0.3. The values of the constants C0, c', and c" are given in Table 81. If the circular plate is subjected to a lateral load combined with a uniform tension JV, instead of compression, then we have, approximately,

where a is the absolute value of N/Ncr.

As for the curvatures, a factor

instead of the factor (1 + Ca)/(I a) must be used in expressions (Jc) and (I), the constant c having the meaning of C0, cr, and c", respectively.

95. Bending of Plates with a Small Initial Curvature.1 Assume that a plate has some initial warp of the middle surface so that at any point there is an initial deflection W o which is small in comparison with the thickness of the plate. If such a plate is submitted to the action of transverse loading, additional deflection w\ will be produced, and the total deflection at any point of the middle surface of the plate will be W o + Wi. In calculating the deflection Wi we use Eq. (103) derived for flat plates. This procedure is justifiable if the initial deflection W0 is 1 See S. Timoshenko's paper in Mera. Inst. Ways Commun., vol. 89, St. Petersburg 1915 (Russian).
Retrieved from: www.knovel.com

Copyright 1959 by The McGraw-Hill Companies

Co py rig hte dM ate ria l

(230)

small, since we may consider the initial deflection as produced by a fictitious load and apply the principle of superposition.1 If in addition to lateral loads there are forces acting in the middle plane of the plate, the effect of these forces on bending depends not only on Wi but also on WQ. To take this into account, in applying Eq. (217) we use the total deflection w = WQ + Wi on the right-hand side of the equation. It will be remembered that the left-hand side of the same equation was obtained from expressions for the bending moments in the plate. Since these moments depei.d not on the total curvature but only on the change in curvature of the plate, the deflection W1 should be used instead of w in applying that side of the equation to this problem. Hence, for the case of an initially curved plate, Eq. (217) becomes

It is seen that the effect of an initial curvature on the deflection is equivalent to the effect of a fictitious lateral load of an intensity

Thus a plate will experience bending under the action of forces in the xy plane alone provided there is an initial curvature. Take as an example the case of a rectangular plate (Fig. 192), and assume that the initial deflection of the plate is defined by the equation

If uniformly distributed compressive forces Nx are acting on the edges of this plate, Eq. (230) becomes

Let us take the solution of this equation in the form

Substituting this value of W\ into Eq. (b), we obtain

1 In the case of large deflections the magnitude of the deflection is no longer proportional to the load, and the principle of superposition is not applicable.

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com

Co py rig hte dM ate ria l

in which The maximum deflection will be at the center and will be This formula is analogous to that used for a bar with initial curvature.1 In a more general case we can take the initial deflection surface of the rectangular plate in the form of the following series: Substituting this series in Eq. (230), we find that the additional deflection at any point of the plate is in which It is seen that all the coefficients bmn increase with an increase of Nx. Thus when Nx approaches the critical value, the term in series (Zi) that corresponds to the laterally buckled shape of the plate [see Eq. (227)] becomes the predominating one. We have here a complete analogy with the case of bending of initially curved bars under compression. The problem can be handled in the same manner if, instead of compression, we have tension in the middle plane of the plate. In such a case it is necessary only to change the sign of Nx in the previous equations. Without any difficulty we can also obtain the deflection in the case when there are not only forces Nx but also forces Ny and Nxy uniformly distributed along the edges of the plate.
1

With this value of A expression (c) gives the deflection of the plate produced by the compressive forces Nx. Adding this deflection to the initial deflection (a), we obtain for the total deflection of the plate the following expression:

See S. Timoshenko, "Strength of Materials/' part II, 3d ed., p. 56, 1956.

Copyright 1959 by The McGraw-Hill Companies

Retrieved from: www.knovel.com