SOME THEOREMS IN THE BENDING THEORY OF POROUS THERMOELASTIC PLATES

Transcription

1 ANALELE ŞTIINŢIFICE ALE UNIVERSITĂŢII AL.I.CUZA IAŞI Tomul L, s.i a, Matematică, 2004, f.2. SOME THEOREMS IN THE BENDING THEORY OF POROUS THERMOELASTIC PLATES BY MIRCEA BÎRSAN Abstract. In the context of the linear theory of thermoelastic materials with voids, we study the dynamic bending of Mindlin type plates. We formulate the corresponding boundary initial value problem and establish a reciprocal theorem. Then we present two variational theorems of Gurtin type. Mathematics Subject Classification 2000: 74K20. Key words: materials with voids, thermoelastic, plate, bending, porous. 1. Introduction. The Nunziato Cowin theory for the treatment of porous elastic solids in which the interstices (or pores) are void of material was established in [1,2]. In this theory, the bulk mass density is written as the product of two fields, the matrix material density field and the volume fraction field. Capriz and Podio Guidugli [3] remarked that the theory proposed by Nunziato and Cowin can be regarded as a particular case of the theory of materials with microstructure. The linear theory of thermoelastic materials with voids was investigated by Ieşan [4]. Due to its intended applications in the mechanics of granular materials and manufactured porous bodies, the theory of elastic solids with voids has been intensively studied in the last two decades. The present paper is concerned with the bending of thermoelastic thin plates made from an isotropic and homogeneous material with voids. We employ a plate theory of Mindlin type, taking into account the transverse shear deformation in the flexural motions of plates. In [5] Mindlin introduced a plate model with transverse shear deformation. Since then, a large number of papers have been devoted to the study of the bending theory

2 306 M. BÎRSAN 2 for Mindlin type plates. Several articles have investigated the behaviour of plates made from materials with microstructure (see, e.g., [6]). The theory of plates is presented in detail in the work of Naghdi [7]. The bending theory of thermoelastic plates made from a material with voids was derived by Bîrsan [8]. Some results concerning the study of porous plates and shells were presented in [9,10]. In the first part of the present article we remind the fundamental equations for the bending of thermoelastic porous plates established in [8]. Then, the corresponding boundary initial value problem is expressed in a form in which the initial conditions are incorporated into the field equations. Using this formulation of the problem, a reciprocal theorem is derived. In the final section we prove two theorems which provide a variational characterization of the solution of the boundary initial value problem. 2. Basic equations. In this section we summarize the fundamental equations that govern the bending of porous thermoelastic plates of Mindlin type. For a detailed derivation of these equations we refer to [8]. We denote by B the interior of a right cylinder of length h 0 with open cross-section and smooth lateral boundary Π. Let C be the boundary of. We assume that the region B is occupied by a linearly thermoelastic material with voids. The length h 0 is supposed to be small enough so that the body represents a thin plate of uniform thickness h 0. The rectangular Cartesian coordinate frame is chosen such that x 1 Ox 2 is the middle plane of the plate. Let u be the displacement vector field, ϕ the change in volume fraction field and θ the temperature field measured from the constant absolute temperature T 0 of the reference state. Following [5,6,7], we study the states of bending characterized by (1) u α = x 3 v α (x 1, x 2, t), u 3 = w(x 1, x 2, t), ϕ = x 3 ψ(x 1, x 2, t), θ = x 3 T (x 1, x 2, t) on B T, where T = [0, t 1 ) is a given interval of time (which may also be [0, )). Throughout this paper, Latin subscripts range over the integers 1, 2, 3, whereas Greek subscripts are confined to the range 1, 2. The usual summation and differentiation conventions are also employed. We say that (v α, w, ψ, T ) is an admissible process on T provided: (i) v α, w and ψ are of class C 2 on T ; (ii) T is of class C 2,1 on T ;

3 3 THEORY OF POROUS THERMOELASTIC PLATES 307 (iii) v α, w, ψ and T are of class C 1 on T. Let us denote by D the set of all admissible processes. The linear strain measures are (2) ε αβ = 1 2 (v α,β + v β,α ), γ α = v α + w,α. The quantities γ α represent the angles of rotation of the cross sections x α = constant about the middle surface (see, e.g., [11]). For thermoelastic porous plates made from an isotropic and homogeneous material, we have the following constitutive equations (3) M αν = I[λε ρρ δ αν + 2µε αν + bψδ αν βt δ αν ], N α = µγ α, H ν = αiψ,ν, G = I( bε ρρ ξψ + mt ), σ = I(βε ρρ + mψ + at ), Q α = kit,α, Γ = αψ, R = kt, where λ, µ, b, β, α, ξ, m, a and k are constitutive constants and I = h 3 0 /12. For the significance of the functions M αν, N α, H ν, G, σ, Q α, Γ and R which appear in relations (3), we refer to [8]. By virtue of the second law of thermodynamics, the coefficient k is nonnegative. We denote by ρ the reference mass density. The balance equations for the bending of porous plates are (4) M βα,β h 0 N α + f α = ρi v α, N α,α + f = ρẅ, H α,α + G h 0 Γ + l = ρκi ψ, Q α,α h 0 R + S = T 0 σ, where f α, f, l and S are prescribed functions representing resultant external body loads and heat supply (see [8]). We consider the following mixed boundary conditions (5) v α = ṽ α on C 1 T, M βα n β = M α on C 2 T, w = w on C 3 T, N α n α = Ñ on C 4 T, ψ = ψ on C 5 T, H ν n ν = H on C 6 T, T = T on C 7 T, Q α n α = Q on C 8 T,

4 308 M. BÎRSAN 4 where n designate the outward unit normal of and C i (i = 1, 2,.., 8) are subsets of such that C 1 C 2 = C 3 C 4 = C 5 C 6 = C 7 C 8 = C and C 1 C 2 = C 3 C 4 = C 5 C 6 = C 7 C 8 =. We assume that the prescribed functions ṽ α, w, ψ and T are continuous on their domains of definition, whereas the given functions M α, Ñ, H and Q are piecewise regular and continuous in time. The initial conditions are (6) v α (x γ, 0) = vα(x 0 γ ), v α (x γ, 0) = vα(x 1 γ ), w(x γ, 0) = w 0 (x γ ), ẇ(x γ, 0) = w 1 (x γ ), ψ(x γ, 0) = ψ 0 (x γ ), ψ(x γ, 0) = ψ 1 (x γ ), σ(x γ, 0) = σ 0 (x γ ), (x γ ), where the functions appearing in the right hand sides are given continuous functions. The boundary initial value problem associated with the bending of thermoelastic porous plates consists of equations (2) (6). We mention that the field equations (2) (4) can be expressed in terms of the functions v α, w, ψ and T in the following form (7) I[µ v α + (λ + µ)v ν,να + bψ,α βt,α ] µh 0 (v α + w,α ) + f α = ρi v α, µ w + µv α,α + f = ρẅ, I[α ψ bv ν,ν + mt ] (ξi + αh 0 )ψ + l = ρκi ψ, I[k T βt 0 v α,α mt 0 ψ] kh 0 T + S = ait 0 T, where is the Laplace operator. In [8] we have studied the solutions of the above boundary initial value problem and we have proved some uniqueness and continuous dependence results. 3. Reciprocity. In this section we derive a reciprocal theorem using the method presented in [12]. For two arbitrary scalar fields u and v on T that are continuous in time, we denote by u v the convolution of u and v [u v](x, t) = t 0 u(x, t τ)v(x, τ)dτ

5 5 THEORY OF POROUS THERMOELASTIC PLATES 309 and by ū the function defined on T by We introduce the functions ū(x, t) = t 0 u(x, τ)dτ. e(t) = 1, j(t) = t, t T. Taking the convolution of equations (4) 1,2,3 with j and of equation (4) 4 with e, we find (8) j (M βα,β h 0 N α ) + F α = ρiv α, j N α,α + F = ρw, j (H β,β + G h 0 Γ) + L = ρκiψ, e (Q α,α h 0 R) + W = T 0 σ, where the functions F α, F, L and W are defined by (9) F α = j f α + ρi(tv 1 α + v 0 α), F = j f + ρ(tw 1 + w 0 ), L = j l + ρκi(tψ 1 + ψ 0 ), W = e S + T 0 σ 0. We remark that equations (8) are equivalent to the balance equations (4) together with the initial conditions (6). Let us consider two external data systems L (α) = {f γ (α), f (α), l (α), S (α), ṽ γ (α) (α), M γ, w (α), Ñ (α), ψ (α), H (α), T (α), Q (α), vγ 0(α), vγ 1(α), w 0(α), w 1(α), ψ 0(α), ψ 1(α), σ 0(α) } and let p (α) = {v γ (α), w (α), ψ (α), T (α), M (α) βγ, N (α) β, H(α) β, G(α), σ (α), Q (α) β, Γ(α), R (α) } be solutions of the boundary initial value problem corresponding to the data systems L (α) (α = 1, 2), respectively. We denote by M (α) γ = M (α) βγ n β, N (α) = N (α) β n β, H (α) = H (α) β n β, Q (α) = Q (α) β and let F γ (α), F (α), L (α), W (α) be the quantities defined in (9) corresponding to the two data systems L (α) (α = 1, 2). We introduce the notation (10) E αβ (r, s) = M (α) γσ (r)ε (β) γσ (s) + h 0 N σ (α) n β (r)γ σ (β) (s) G (α) (r)ψ (β) (s)+ +H γ (α) (r)ψ,γ (β) (s) + h 0 Γ (α) (r)ψ (β) (s) σ (α) (r)t (β) (s),

6 310 M. BÎRSAN 6 for all r, s T, and α, β {1, 2}, where (11) ε (α) γσ = 1 2 (v(α) γ,σ + v σ,γ (α) ), γ σ (α) = v σ (α) + w,σ (α). From the constitutive equations (3) one can easily prove that (12) E αβ (r, s) = E βα (s, r). On the other hand, from the relations (10), (11) and the balance equations (4) 1,2,3 and (8) 4 we obtain (13) E αβ (r, s)da= [M γ (α) (r)v γ (β) (s)+h 0 N (α) (r)w (β) (s)+ C +H (α) (r)ψ (β) (s) 1 T 0Q (α) (r)t (β) (s)]dl+ [f γ (α) (r)v γ (β) (s)+ +h 0 f (α) (r)w (β) (s)+l (α) (r)ψ (β) (s) 1 T 0W (α) (r)t (β) (s)]da ρ[i v γ (α) (r)v γ (β) (s)+h 0 ẅ (α) (r)w (β) (s)+κi ψ (α) (r)ψ (β) (s)]da + + k [h 0 T (α) (r)t (β) (s)+it (α),γ (r)t,γ (β) (s)]da. T 0 If we put r = τ and s = t τ, relations (12) and (13) yield (M γ (1) v γ (2) + h 0 N (1) w (2) + H (1) ψ (2) e 1 C T 0Q (1) T (2) )dl+ + [f γ (1) v γ (2) + h 0 f (1) w (2) + l (1) ψ (2) 1 T 0W (1) T (2) ]da ρ[i v γ (1) v γ (2) + h 0 ẅ (1) w (2) + κi ψ (1) ψ (2) ]da = (14) = (M γ (2) v γ (1) + h 0 N (2) w (1) + H (2) ψ (1) e 1 C T 0Q (2) T (1) )dl+ + [f γ (2) v γ (1) + h 0 f (2) w (1) + l (2) ψ (1) 1 T 0W (2) T (1) ]da ρ[i v γ (2) v γ (1) + h 0 ẅ (2) w (1) + κi ψ (2) ψ (1) ]da. Taking the convolution of equation (14) with j and making use of relations of the type j ẅ = w(t) (tw 1 + w 0 ),

7 7 THEORY OF POROUS THERMOELASTIC PLATES 311 we are led to the following reciprocal theorem. Theorem 1. Let p (α) be a solution of the boundary initial value problem corresponding to the external data system L (α) (α = 1, 2). Then the following relation holds (15) j [M γ (1) v γ (2) + h 0 N (1) w (2) + H (1) ψ (2) e 1 C T 0Q (1) T (2) ]dl+ + [F γ (1) v γ (2) +h 0 F (1) w (2) +L (1) ψ (2) j 1 T 0W (1) T (2) ]da = = j [M γ (2) v γ (1) +h 0 N (2) w (1) +H (2) ψ (1) e 1 C T 0Q (2) T (1) ]dl+ + [F γ (2) v γ (1) +h 0 F (2) w (1) +L (2) ψ (1) j 1 T 0W (2) T (1) ]da. 4. Variational theorems. With a view toward establishing variational theorems of Gurtin type, we follow the method described in [13]. The equation (8) can be put in the form of a vector equation (16) AU = φ, where we have introduced the vectors (17) U = (v 1, v 2, w, ψ, T ), φ = (F 1, F 2, h 0 F, L, j 1 T 0W ) and the operator A defined on D by (18) A = (A 1, A 2, A 3, A 4, A 5 ), A α U=ρIv α j [µi v α +(λ+µ)iv γ,γα µh 0 (v α +w,α )+biψ,α βit,α ], A 3 U = ρh 0 w j µh 0 ( w + v α,α ), A 4 U = ρκiψ j [αi ψ biv γ,γ (ξi + αh 0 )ψ + mit ], A 5 U = j [ I(βv γ,γ + mψ + at ) + e k T 0 (I T h 0 T )]. In view of the equations (7), we see that U represents a solution of the boundary initial value problem if and only if the relation (16) holds and U satisfy the boundary conditions (5).

8 312 M. BÎRSAN 8 Let us consider first the case of homogeneous boundary conditions. For U = (v α (1), w (1), ψ (1), T (1) ) and V = (v α (2), w (2), ψ (2), T (2) ) satisfying the homogeneous boundary conditions corresponding to (5), the reciprocal theorem (15) yields (19) (AU V ) = (U AV ), where we have denoted by the convolution scalar product defined by (u v) = u i v i da, for any vectors u, v continuous on T. Relation (19) means that the linear operator A is symmetric in convolution. Then we may apply a theorem presented in [13] which asserts that U is a solution of equation (16) if and only if δf(u) = 0, where (20) F(U) = 1 (AU U) (U φ). 2 Substituting (17) and (18) into (20) we find the expression of the functional F in our case (21) F(U)= 1 j [λiv α,α v β,β +2µIε αβ ε αβ +µh 0 γ α γ α +2bIv γ,γ ψ 2 2βIv γ,γ T +αiψ,γ ψ,γ +(ξi + αh 0 )ψ ψ 2mIψ T ait T ]da+ + 1 [ρ(iv α v α +h 0 w w+κiψ ψ) j e k (h 0 T T +IT,α T,α )]da 2 T 0 α v α + h 0 F w + L ψ j (F 1 W T )da. T 0 Thus, we have obtained the following result. Theorem 2. Let M be the set of all admissible processes U=(v α, w, ψ, T ) which satisfy homogeneous boundary conditions. Then δf(u) = 0 on M, if and only if U M satisfy the field equations (7) and the initial conditions (6).

9 9 THEORY OF POROUS THERMOELASTIC PLATES 313 In the case of inhomogeneous boundary conditions (5),we define G(U) = F(U) { j Mα v α dl + h 0 Ñ w dl + H ψ dl e 1 C 2 C 4 C 6 C 8 T 0 The following variational theorem holds. } Q T dl. Theorem 3. Let K be the set of all admissible processes U=(v α, w, ψ, T ) that satisfy the boundary conditions Then if and only if v α = ṽ α on C 1 T, w = w on C 3 T, ψ = ψ on C 5 T, T = T on C 7 T. δg(u) = 0 on K, U is a solution of the boundary initial value problem. Proof. In order to prove the theorem we can use arguments similar to those presented in [13]. Alternatively, we can verify the assertion of the theorem directly, with the help of the relation (22) [ ] d dy G(U + yu ) = {[j (M βα,β h 0 N α ) + F α ρiv α ] vα+ y=0 +h 0 (j N α,α + F ρw) w +[j (H β,β +G h 0 Γ)+L ρκiψ] ψ + +j 1 (Q α,α h 0 R) + W T 0 σ] T T 0[e }da+ { +j (M βα n β M α ) vαdl + h 0 (N α n α Ñ) w dl+ C 2 C 4 + (H β n β H) ψ dl e 1 } (Q α n α C 6 C 8 T Q) T dl, 0 for every U = (vα, w, ψ, T ) such that U +yu K, for all real numbers y. Using (22) and taking into account equations (8), we complete the proof in the classical manner (see [14]). Theorems 2 and 3 provide a variational characterization of the solution of the boundary initial value problem for the bending of thermoelastic porous plates.

10 314 M. BÎRSAN 10 REFERENCES 1. Nunziato, J.W.; Cowin, S.C. A non linear theory of elastic materials with voids, Arch. Rational Mech. Anal., 72(1979), pp Cowin, S.C.; Nunziato, J.W. Linear elastic materials with voids, J. Elasticity, 13(1983), pp Capriz, G.; Podio Guidugli, P. Materials with spherical structure, Arch. Rational Mech. Anal., 75(1981), pp Ieşan, D. A theory of thermoelastic materials with voids, Acta Mech., 60(1986), pp Mindlin, R.D. Influence of rotary inertia and shear on flexural motions of isotropic, elastic plates, J. Appl. Mech., 18(1951), pp Eringen, A.C. Theory of micropolar plates, ZAMP, 18(1967), pp Naghdi, P.M. The Theory of Shells and Plates, in: W. Flügge, Handbuch der Physik, vol. VI a/2 (edited by C. Truesdell), Springer Verlag, Berlin Heidelberg New York, 1972, pp Bîrsan, M. A bending theory of porous thermoelastic plates, J. Thermal Stresses, 26(2003), pp Bîrsan, M. On the bending of plates in the theory of elastic materials with voids, An. Şt. Univ. Al.I. Cuza Iaşi, s. Matematica, 49(2003), pp Bîrsan, M. On a theory of porous thermoelastic shells, An. Şt. Univ. Al.I. Cuza Iaşi, s. Matematica, 46(2000), pp Lagnese, J.E.; Lions, J.L. Modelling, Analysis and Control of Thin Plates, Collection RMA, vol. 6, Masson, Paris, Ieşan, D. On some theorems in thermoelastodynamics, Rev. Roum. Sci. Tech., série Mécanique Appliquée, 34(1989), pp Ieşan, D. On some reciprocity theorems and variational theorems in linear dynamic theories of continuum mechanics, Memorie dell Accad. Sci. Torino, Cl.Sci.Fis.Mat.Nat., Serie 4, 17(1974). 14. Gurtin, M.E. The Linear Theory of Elasticity, in: W. Flügge, Handbuch der Physik, vol. VI a/2 (edited by C. Truesdell), Springer Verlag, Berlin Heidelberg New York, 1972, pp Received: 3.XI.2003 Al.I. Cuza University, Faculty of Mathematics, Bd. Carol I, nr. 11, , Iaşi, ROMÂNIA, bmircea@uaic.ro