38 Bulletin of TICMI AN EXTENSION OF THE VEKUA-BITSADZE METHOD FOR SOLVING EQUATIONS OF THE SHELLS Meunargia T. I. Vekua Institute of Applied Mathematics of Iv. Javakhishvili Tbilisi State University Key words: Shell, Vekua-Bitsadze method parameter, nonlinear theory. MSC 2000 : 74K25. I. Vekua has constructed several versions of the refined linear theory of thin and shallow shells, containing the regular process by means of the method of reduction of three-dimensional problems of elasticity to two-dimensional ones. In the present paper by means of the I. Vekua method the system of differential equations for the nonlinear theory of non-shallow shells is obtained. Using the method I. Vekua and the method of a small parameter 2-D system of equations for the nonlinear and non-shallow shells is obtained. For any approximations of order N the complex representations of Vekua-Bitsadze type [2] of the general solutions are obtained. Under thin and shallow shells I. Vekua meant three-dimensional shell-type elastic bodies satisfying the conditions aβα − x3 bβα ∼ = aβα , −h(x1 , x2 ) ≤ x3 ≤ h(x1 , x2 ) (α, β = 1, 2), (∗) where aβα and bβα are mixed components of the metric tensor and the curvature tensor of the shell’s midsurface, x3 is the thickness coordinate and h is the semi-thickness, depending on curvilinear coordinates x1 , x2 . In the sequel, under by non-shallow shells we mean elastic bodies not subject to assumption (*), i. e., such that ¯ ¯ aβα − x3 bβα aβα ⇒ ¯x3 bβα ¯ ≤ q < 1. 1. To construct the theory of shells we use the coordinate system which is normally connected with the midsurface S. This means that the radius-vector of any point of the domain Ω can be represented in the form [1] R(x1 , x2 , x3 ) = r(x1 , x2 ) + x3 n(x1 , x2 ) (x3 = x3 ), where r and n are radius-vector and the unit vector of the normal of the midsurface S(x3 = 0); x1 , x2 are the Gaussian parameters of S. Covariant and contravariant basis vectors Ri and Ri of the surface Ŝ(x3 = const) and the corresponding basis vectors r i and r i of the midsurface S(x3 = 0) are connected by the following relations [1]: Ri = A.ji. r j = Aij r j , Ri = Ai..j r j = Aij r j (i, j = 1, 2, 3), Volume 13, 2009 39 where β β α. −1 α α i 3 3 A.β α. = aα − x3 bα , A.β = ϑ [(1 − 2Hx3 )aβ + x3 bβ ], A3 = Ai = δi , ϑ = 1 − 2Hx3 + Kx23 , R3 = R3 = r 3 = r 3 = n (α, β = 1, 2). (1) Here (aαβ , aαβ , aαβ ) and (bαβ , bαβ , bαβ ) are the components (co, contra, mixed) of the metric tensor and curvature tensor of the midsurface S. By H and K we denote a middle and Gaussian curvature of the surface S, where 2H = bαα = b11 + b22 , K = b11 b22 − b12 b21 . The main quadratic forms of the midsurface S have the form I = ds2 = aαβ dxα dxβ , II = ks ds2 = bαβ dxα dxβ , (2) where ks is the normal curvature of the surface S, and aαβ = r α r β , bαβ = −r α nβ , ks = bαβ sα sβ , sα = dxα . ds Here and in the sequel, under a repeated indices we mean summation; note that the Greek indices range over 1, 2, while Latin indices range over 1, 2, 3. To construct the theory of non- shallow shells, it is necessary to obtain formulas for a family of surfaces Ŝ(x3 = const), analogous to (2) of the midsurface S(x3 = 0) which have the form [1] I = dŝ2 = gαβ dxα dxβ , II = kŝ dŝ2 = b̂αβ dxα dxβ , (3) where gαβ = aαβ − 2x3 bαβ + x23 (2Hbαβ − Kaαβ ), b̂αβ = (1 − 2Hx3 )bαβ + x3 Kaαβ , and kŝ the normal curvature of the surface Ŝ. It is not now difficult to get the expression for the tangential normal l̂ of the surface Ŝ directed to ŝ [3]: l̂ = ŝ × n = [(1 − x3 ks )l − x3 τs s] ds , dŝ = dŝ q 1 − 2x3 ks + x23 (ks2 + τs2 )ds, where s and l are the unit vectors of the tangent and tangential normal on S, dŝ and ds are the linear elements of the surfaces Ŝ and S, and τs is the geodesic torsion of the surface S. 2. We write the equation of equilibrium of elastic shell-type bodies in a vector form √ 1 ∂ gσ i + Φ = 0, ⇒ ∇i σ i + Φ = 0, (4) √ g ∂xi 40 Bulletin of TICMI where g is the discriminant of the metric quadratic form of the three-dimensional domain Ω, ∇i are covariant derivatives with respect to the space coordinates xi , Φ is on external force, σ i are the contravariant constituents of the stress ∗ vector σ ∗ acting on the area with the normal l and representable as the (l) Cauchy formula as follows: µ ¶ ∗ ∗ ∗ i σ ∗ = σ l i , l i = l Ri , . (l) For the stress vector acting on the area with normal l̂, we obtain σ (l̂) = σ α (l̂Rα ) = ϑσ α (lr α ) ds . dŝ (5) The stress-strain relation for the geometrically nonlinear theory of elasticity has the form σ i = σ ij (Rj + ∂j U ) = E ijpq epq (Rj + ∂j U ), (6) where σ ij are contravariant components of the stress tensor, eij are covariant components of the strain tensor, U is the displacement vector, E ijpq and eij are defined by the formulas: 1 E ijpq = λg ij g pq + µ(g ip g jq + g iq g ip ), eij = (Ri ∂j U + Rj ∂i U + ∂i U ∂j U ). (7) 2 To reduce the three-dimensional problems of the theory of elasticity to the two-dimensional problems, it is necessary to rewrite the relation (4-7) in forms of the bases of the midsurface S of the shell Ω. The relation (4) can be written as √ 1 ∂ aϑσ α ∂ϑσ 3 √ + + ϑΦ = 0, (a = a11 a22 − a212 ). (8) ∂x3 a ∂xα From (1), (6), (7) we obtain 1 σ i = Aii1 App1 , M i1 j1 p1 q1 [(r q1 ∂p U ) + Aqq1 (∂p U ∂q U )](r j1 + Ajj1 ∂j U ), 2 (9) M i1 j1 p1 q1 = λai1 j1 ap1 q1 + µ(ai1 p1 aj1 q1 + ai1 q1 aj1 p1 ) (aij = r i r j ) 3. The isometric system of coordinates on the surface S is of special interest, for in this system we can obtain basic equations of the theory of shells in a complex form which in turn allows one to construct for a rather wide class of problems complex representations of general solutions by means of analytic functions of one variable z = x1 + ix2 . The main quadratic forms in the system of coordinates are of the type I = ds2 = Λ(z, z̄)dzdz̄, ¤ 1 £ II = ks ds2 = Λ Q̄dz 2 + 2Hdzdz̄ + Qdz̄ 2 , 2 Volume 13, 2009 Q = 0.5(b11 − b22 + 2ib12 ), 41 Λ(z, z̄) > 0. Introducing the well-known differential operators 2∂z = ∂1 − i∂2 , 2∂z̄ = ∂1 + i∂2 and the notations τ.ji. = ϑσ i r j , Xi = ϑΦr i , for the geometrically nonlinear theory of non-shallow shells from (8) and (9) we obtain the following complex form both for the system of equations of equilibrium and Hooke’s law: 3 ∂τ·+ 1 ∂Λτ + r + ∂ τ̄ + r + + ·+ + X+ = 0, + − Λ(Hτ3 + Qτ̄3 ) + Λ µ ∂z ∂ z̄ ¶ ∂x3 + + ¡ ¢ ∂τ 3 1 ∂Λτ3 ∂Λτ̄3 + H(τ11 + τ22 ) + Re Q̄τ + r + + 3 + X3 = 0, + Λ ∂z ∂ z̄ ∂x3 © £ ¡ ¢ ¤ª ¡ + ¢ τ + = ϑ λΘ+µ R+ ∂z U + R̄+ +2∂z U ∂ z̄ U R +2∂ z̄ U n£ ´ ¤³ + + z̄ z + µϑ R ∂z̄ U +(R+ +2∂z̄ U ) ∂ U R̄ +2∂ U £ ¤ ª + R+ ∂3 U +(n+∂3 U ) ∂ z̄ U (n+∂3 U) , ¸ ½· 1 3 3 3 τ = ϑ λΘ + 2µ(n∂ U + ∂3 U ∂ U ) (n + ∂3 U ) 2 · 1 +µ ( R̄+ ∂3 U + n∂z U + ∂z U ∂3 U )(R+ + 2∂ z̄ U ) 2 ¸¾ 1 + z + ( R+ ∂3 U + n∂z̄ U + ∂z̄ U ∂3 U )(R̄ + 2∂ U ) , 2 (10) (11) (∂ 3 U = ∂3 U ). Here ¡ ¢ ¡ ¢ τ + r + = τ 1 + iτ 2 (r 1 + ir 2), τ̄ + r + = τ 1 − iτ³2 (r 1 + ´ ir 2), τ.3+. = τ + n, ¡ ¢ + 3. τ.+ = τ 3 r + , τ33 = τ 3 n, 2∂ z̄ U = R+ R+ ∂z U + R+ R̄ ∂z̄ U, τ.βα. = τ α r β , £¡ ¤¢ £ Θ = 2Re ¤ R+ +∂ z̄ U ∂z U +∂3 U 3 +0.5 (∂3 U )2 , R+ = ϑ−1 (1−Hx3 ) r ++ +x3 Qr̄ + , R+ = (1 − Hx3 )r + −x3 Qr̄ + , R+ = R1 +iR2, R+ = R1 +iR2, + R+ R+ = 4x3 (Λϑ2 )−1 (1−Hx3 )Q, R+ R̄ = 2(Λϑ2 )−1 (ϑ+2x23 QQ̄), + R+ r + = 2ϑ−1 Qx3 , R̄ r + = 2ϑ−1 (1 − Hx3 ), r + r̄ + = 2Λ−1 , r + = r 1 + ir 2 . We have the formulas + + r + ∂z U = Λ−1 ∂z U+ − ¡ HU3 , r ∂z̄¢U =+∂z̄ U −+QU3 , X+ = X1 +iX2 , n∂z U = ∂z U3 + 0.5 Q̄U+ +H Ū+ (U = U r , U+ = U r + , U3 = U n) . 4. In the present paper the three-dimensional problems of the theory of elasticity are reduced to the two-dimensional ones by the n method ³ x ´osuggested by 3 is complete I. Vekua. Since the system of Legendre polynomials Pm h 42 Bulletin of TICMI in the interval [-h,h], for equation (8) we obtain the infinite system of twodimensional equations Zh · ¸ √ ¡ ¢ 1 ∂ aϑσ α ∂ϑσ 3 √ + + ϑΦ Pm xh̄3 dx3 = 0 (m = 0, 1, ...) α 3 ∂x a ∂x −h or in a form (m) α ∇α σ 2m + 1 − h µ (m−1) 3 σ ¶ (m) + ... + F = 0, (m−3) 3 + σ (12) where µ (m) i (m) ¶ σ ,Φ 2m + 1 = 2h Zh ¡ i ¢ ¡ ¢ 2m + 1 ϑσ , ϑΦ Pm xh̄3 dx3 = 2h −h (m) (m) 2m + 1 F = Φ+ 2h Zh ¡ ¢ τ i , X Pmdx3 , −h µ (+) (+) 3 m ϑ σ −(−1) (−) (−) ϑ σ ¶ µ ¶ ϑ = ϑ(±h) , (±) 3 ∇α are covariant derivatives on the midsurface S. The equation of state (9) may be written as "à !µ ¶ ∞ ∞ (m2 ) (m1 ) X X (m) (m) (m) i ip ijp i1 j1 p1 q1 σ =M r q1 ·Dp U A i1 p1r j1 + A i1 j1 p1Dj U (m1 ) (m1 ,m2 ) m =0 m =0 1 2 à !µ ¶# ∞ ∞ (m2 ) (m1 ) (m3 ) X (m) (m) 1 X ipq ij pq rj + Dj U Dp U Dq U + A A 2 m =0 (m1 ,m2i)1 p1 q1 1 m =0 (m1 ,m2 ,m3i)1 j1 p1 q1 2 (13) 3 where (m) Di U = δiβ ∂β (m) A (m1 ) ij i1 j1 (m) U +δi3 2m + 1 = 2h (m) ijpq A i1 j1 p1 q1 (m1 ,m2 ,m3 ) 0 0 Zh ϑAii1 , Ajj1 , Pm1 µ(m+1) (m+3) ¶ U + U +... , ³x ´ 3 h Pm ³x ´ 3 h (14) dx3 −h ijp A i1 j1 p1 (m1 ,m2 ) (m) (m) 2m + 1 U ; U = h (m) 2m + 1 = 2h 2m + 1 = 2h Zh ϑAii1 , Ajj1 , App1 Pm1 Pm2 Pm dx3 , −h Zh ϑAii1 , Ajj1 , App1 Aqq1 Pm1 Pm2 Pm3 Pm dx3 . −h The boundary conditions on the lateral contour ∂S take the form: (15) Volume 13, 2009 43 a) for the stresses σ 2m + 1 = σ (ll) l + σ (ls) s + σ (ln) n = 2h (m) (m) (l) (m) (m) Zh σ (l) ³x ´ dŝ 3 Pm dx3 , ds h (16) h b) for the displacements 2m + 1 U= 2h m Zh U Pm ³x ´ 3 h (m) dx3 = U (m) (l) l + U (m) (s) (s) + U 3 n. (17) h Thus we have constructed an infinite system of two-dimensional equations of geometrically non-linear and non-shallow shells (12-17), which is consistent (±) with the boundary conditions on the face surfaces, i.e. σ 3 = σ 3 (x1 , x2 , ±h). The passage to finite systems can be realized by various methods one of which consists in considering of a finite series, i.e. ¶ N µ ³x ´ X (m) i (m) (m) 3 i σ , U , Φ Pm (ϑσ , U , ϑΦ) = = (τ i , U , X), h m=0 where N is a fixed nonnegative number. In other words, it is assumed that (m) (m) U = 0, σ i = 0, if m > N. Approximation of this type will be called approximation of order N . The integrals of type (15) can be calculated [5], for example, (m) A (m1 ) αβ α1 β1 2m + 1 = 2h Zh ϑ −1 Bαα1 (x3 )Bββ1 (x3 )Pm1 ³x ´ 3 h Pm ³x ´ 3 h dx3 = −h · µ ¶¸y2 α β Lα Lβ m 2m + 1 α Pm1 (y)Qm (y), m1 ≤ m β √ Bα1 (hy)Bβ1 (hy) + 1 1 σm , 1 Qm1 (y)Pm (y), m1 ≤ m y1 K 2 Eh (18) m , if E = H 2 − K = 0; where Qm (y) is if E 6= 0 K 6= 0 and aαα1 aββ1 δm 1 the Legendre function of the second kind, E is the Euler difference, Bβα (x) = aαβ + xLαβ , Lαβ = bαβ − 2Haαβ . Under the square brackets we mean the following: √ [f (y)]yy21 = f (y2 ) − f (y1 ), y1,2 = [(H ∓ E)h]−1 . For the integrals containing the product of three (four) Legendre polynomials we have (m) A (m1 ,m2 ) α1 α2 α3 β1 β2 β3 2m + 1 = 2n Zh −h Bβα11 Bβα2 Bβα3 2m + 1 Pm1 Pm2 Pm dx3 = × 1 − 2Hx3 + Kx3 K 2 h4 44 Bulletin of TICMI " à !#y2 2 n P (y)Q (y), s ≤ m s m y ∂ α α α n αm1 m2 r , Cβ11β22β33 h ∂y ∂y y − y Q (y)P (y), s ≥ m 1 2 1 2 s m n=0 y min(m1 ,m2 ) × X r=0 3 X n 1 where s = m1 + m2 − 2r, apqr = Ap−r Ar Aq−r 2(p + q) − 4r + 1 1.3 · · · 2p − 1 , Ap = , Ap+q−r 2(p + q) − 2r + 1 p! Bβα11 (x)Bβα22 (x)Bβα33 (x) = 3 X n α α α n Cβ11β22β33 x , n=0 5. Three-dimensional shell-type bodies are characterized by inequalities |hbαβ | ≤ q < 1 (α, β = 1, 2) Therefore they can be represented as follows |εbαβ R| ≤ q < 1, where ε = hR−1 is a small parameter. Here h is semi-thickness of the shell, R is a certain characteristic radius of curvature of the midsurface S [4]. Now, following Signorini [3] we assume the validity of the expansions ! à ¶ X µ ∞ (m,n) (m,n) (m,n) (m) i (m) (m) εn . σi , U , F σ , U, F = n=1 Substituting the above expansions into the (12,13) and (10,11) than equalizing the coefficients of expansions for εn we obtain the following 2-D finite system of equilibrium equations with respect to components of displacement vector in the isometric coordinates a11 = a22 = Λ(x1 , x2 ), which has the form: ³ ´ (m,n) (m,n) (m,n) 4µ∂z Λ−1 ∂z u+ + 2(λ + µ)∂z θ + 2λ∂z u03 − (2m + 1)µ i (m,n) h ³(m−1,n) (m−3,n) ´ (m−1,n) (m−3,n) 2∂z u3 + u3 + · · · + u0+ + u0+ + · · · + F+ = 0, (19) ³ (m,n) (m,n)´ h ³(m−1,n) (m−3,n) ´ µ ∇2 u3 + θ0 − (2m + 1) λ θ + θ +··· + ´i (m,n) ³(m−1,n) (m−3,n) (λ + 2µ) u03 + u03 + · · · + F3 = 0, ³ ´ where u+ = u1 + iu2 , θ = Λ−1 ∂z u+ ∂z u+ , z = x1 + ix2 ,2∂z = ∂1 − i∂2 , ∇2 = 4 ∂2 . Λ ∂z∂z Volume 13, 2009 45 Obviously, in passing from the n-th step of approximation to the (n + 1)-th step only the right-hand side of equations are changed. Below we will omit upper index n. Consider now the cases: N = 0, 1, 2, 3 [6]. Case N = 0. From (3) we get ³ ´ (0) (0) 4µ∂z Λ−1 ∂z u+ + 2(λ + µ)∂z θ = 0, ³(m) (m) ´ (0) µ∇2 u3 = 0 F+ = F3 = 0, m = 0, 1, 2, 3 . (20) The complex representation of general solutions has the form ZZ 0 ZZ 0 (0) λ + 3µ 1 ϕ (ζ)dS 1 ϕ (ζ)dS u+ = − + − ψ(z), λ+µ π π S ζ −z S ζ −z ³ ´ (0) u3 = f (z) + f (z), dS = Λ(ζ, ζ)dζ dζ̄, ζ = ξ + iη , (21) where f (z), ϕ(z) and ψ(z) are holomorphic functions of z. We note that (0) for plane (i.e. Λ = 1) the expression of u+ coincides with the well-known representation of Kolosov-Muskhelishvili. (0) (1) (1) (0) Case N = 1. With respect to the components (u+ , u3 ) and (u1 , u3 ) we have two systems of equations: ³ −1 4µ∂z Λ (0) ∂z u+ ´ (0) (1) + 2(λ + µ)∂z θ + 2λ∂z u3 = 0, h (0) (1)i (1) µ∇2 u3 + 3 λ θ + (λ + 2µ) θ = 0 and ³ ´ ³ (0) (1) ´ (1) (1) 4µ∂z Λ−1 ∂z u+ + 2(λ + µ)∂z θ − 3µ 2∂z u3 + u+ = 0, (0) ∇ 2 u3 (1) (22) (23) + θ = 0. The complex representation of general solutions has the form: ZZ 0 ZZ 0 (0) 5λ + 6µ 1 ϕ (ζ)dS 1 ϕ (ζ)dS λ ∂w u+ = − + − ψ(z) − 3λ + 2µ π π 6(λ + µ) ∂z S ζ −z S ζ −z µ ¶ (0) 12(λ + µ) 2λ (ϕ0 + ϕ0 ), ∇2 w = w , (24) u3 = w − λ + 2µ λ + 2µ and ZZ Φ0 (ζ) + Φ0 (ζ) 4(λ + µ) 00 ∂χ dS + Φ (z) − 2Ψ0 (z) + i , 3µ ∂z ζ− S Z Zz ³ ´ (0) u3 = Ψ(z) + Ψ(z) − Φ0 (ζ) + Φ0 (ζ) ln |ζ − z|dS (∇2 χ = 3χ), (1) u+ 1 =− π S 46 Bulletin of TICMI where Φ(z) and Ψ(z) are holomorphic functions of z. Note that the systems (10) and (11) coincide with I. Vekua’s refined systems of equations for the stretch-strain and bending of plate, respectively. (0) (2) (1) Case N = 2. In this case with respect to the components (u1 , u+ , u3 ) and (1) (0) (2) (u+ , u3 , u3 ) we have two systems of equations: ³ ´ (0) (0) (1) 4µ∂z Λ−1 ∂z u+ + 2(λ + µ)∂z θ + 2λ∂z u3 = 0, ³ ´ ³ (1) ´ (2) (2) (2) 4µ∂z Λ−1 ∂z u+ + 2(λ + µ)∂z θ − 5µ 2∂z u3 + 3u+ = 0, ³ (1) h (0) i (2)´ (1) 2 µ ∇ u3 + 3 θ − 3 λ θ + (λ + 2µ)u3 = 0 and (25) ³ ´ ³ (0) (1) ´ (1) −1 (1) 4µ∂z Λ ∂z u + 2(λ + µ)∂z θ − 3µ 2∂z u3 + u+ = 0, (0) (1) ∇2 u3 + θ = 0, h (1) i (2) (2) µ∇2 u3 − 5 λ θ + 3(λ + 2µ)u3 = 0. (26) In this case the complex representations of the general solutions of the equations (12) and (13) take the forms ZZ 0 ZZ 0 (0) 5λ + 6µ 1 1 ϕ (ζ)dS ϕ (ζ)dS u+ = − + − ψ(z) 3λ + 2µ π π S ζ −z S ζ −z 2 2λ X 1 ∂vk − , λ + 2µ k=1 αk ∂z µ ¶ 2 X (2) 2 ∂ω α3−k ∂vk 2λ 00 u+ = i + ϕ (z) + , (27) 3 ∂z 3λ + 2µ αk ∂z k=1 ¢ (1) 2λ ¡ 0 ϕ (z) + ϕ0 (z) , u3 = v 1 + v 2 − λ + 2µ where ∇2 vk = αk vk , αk2 − and 12(λ + µ) 180µ(λ + µ) αk + = 0 (k = 1, 2); ∇2 ω = 15ω, λ + 2µ (λ + 2µ)2 ZZ 16(λ + µ) 00 Φ0 (ζ) + Φ0 (ζ) dS + Φ (z)− 3(λ + 2µ) ζ −z S (28) ∂χ λ ∂w 0 − , −2Ψ (z) + i ∂z 10(λ + µ) ∂z ZZ ³ ´ (0) 1 λ Φ0 (ζ) + Φ0 (ζ) ln |ζ − z|dS + w, u3 = Ψ(z) + Ψ(z) − π 20(λ + µ) S ´ (2) 2λ ³ 0 u3 = w − Φ (z) + Φ0 (z) , 3λ + 2µ (1) u+ 1 =− π Volume 13, 2009 where ∇2 w = 60(λ+µ) λ+2µ 47 w, ∇2 χ = 3χ. Case N = 3. For this case we have (0) u+ 5λ + 6µ 1 =− 3λ + 2µ π ZZ S 1 ϕ0 dS + ζ −z π ZZ 3 S (2) u+ µ ¶ 3 (2) X ∂v 2 ∂ω 2λ k i ϕ00 + Ak , = + 3 ∂z 3λ + 2µ ∂z k=1 (1) u3 3 (1) X = Ak vk − (3) u3 = k=1 3 X ³ vk , (1) 2λ X 1 + Ak ∂vk ϕ0 dS −ψ− λ + 2µ k=1 αk ∂z ζ −z (29) 2λ (ϕ0 + ϕ0 ), 3λ + 2µ 2 2 ´ ∇ vk = αk vk , ∇ ω = 15ω , k=1 where αk3 − 7 · 900(λ + µ) 180(λ + µ) 2 120(λ + µ)(7λ + 15µ) αk + αk + = 0, 2 λ + 2µ (λ + 2µ) λ + 2µ · ¸· ¸−1 3(9λ + 4µ) 180µ(λ + µ) 12(λ + µ) 180µ(λ + µ) 2 Ak = αk − αk − αk + , λ + 2µ (λ + 2µ)2 λ + 2µ (λ + 2µ)2 ¸· ¸−1 · (2) λ 12(λ + µ) 12(λ + µ) 180µ(λ + µ) 2 αk − αk − αk + , Ak = −10 λ + 2µ λ + 2µ λ + 2µ (λ + 2µ)2 (1) and (1) u+ (3) u+ (0) u3 (2) u3 ZZ Φ0 + Φ0 4 23λ + 24µ 00 dS + Φ − 2Ψ + 15 λ + 2µ S ζ −z ¶ 2 µ X κk − 3 ∂χk 6λ ∂wk + i − , 3 ∂z λ + 2µ ∂z k=1 µ 2 X ∂χk 2 γ3−k ∂wk ¶ 4 3λ + 2µ 00 + − = i Φ (z), ∂z 5 γ ∂z 15 λ + 2µ k k=1 1 = − π (30) (∇2 wk = γk wk , ∇2 χk = κk χk , ), ¶ ZZ 2 µ X γ3−k 1 3λ − wk − (Φ0 + Φ0 ) ln |ζ − z|dS + Ψ + Ψ, = λ + 2µ 5 γ k S k=1 = 2 X wk − k=1 µ γk2 2λ (Φ0 + Φ0 ) 3(λ + 2µ) ¶ 60(λ + µ) 35µ(λ + µ) 2 − γk + 60 = 0, κk − 45κk + 105 = 0 . 3(λ + 2µ) (λ + 2µ)2 48 Bulletin of TICMI The general solution of the homogeneous system (19) we can find the form ³ 1 Z Z ϕ0 (ζ) − æ ϕ0 (ζ)dS ´ 1 0 ζ 0 = ∂z V + + − ψ00 (z) δ0m − π ζ −z S ´ ³ 1 Z Z ϕ0 (ζ) + ϕ0 (ζ)dS ζ 1 1 + η1 ϕ001 (z) − 2ψ10 (z) δ1m π ζ −z (m) (m) u+ S (31) +æ2 ϕ000 (z)δ2m + η2 ϕ001 (z)δ3m , Z Z ³1 ´ (m) (m) u3 = V3 − (ϕ01 (ζ) + ϕ01 (ζ))ln|ζ − z|dSζ − ψ1 (z) − ψ1 (z) δ0m π S i 3 h 0 0 0 0 (m = 0, 1, ..., N ) − æ2 (ϕ0 (z) + ϕ0 (z))δ1m − (ϕ1 (z) + ϕ1 (z))δ2m 2 (0) (0) (0) V1 = V2 = 0, u3 = ψ1 (z) + ψ1 (z), if N = 0 (dSζ = Λ(ζ, ζ)dζdζ, ζ = ξ + iη). where ϕ00 (z),ϕ01 (z),ψ00 (z),ψ10 (z) are holomorphic functions of z and express the biharmonic solution of the system (19). Then æ1 , æ2 , η1 , η2 are known constants. (m) Substituting expressions (31) into (19) the matrix equations for Vi are obtained ∇2 V − AV = X, ∇2 Ω − BΩ = Y, (32) where V and Ω are column-matrices of the form (N ) (0) (1) (N )¢ ¡ (0) (1) T V = V1 , V1 , ..., V1 , V3 , V3 , ..., V3 , (N )¢ ¡ (0) (1) T Ω = V2 , V2 , ..., V2 , and A and B are block-matrices 2N +2×2N +2 and N +1×N +1 respectively. Using now the formulae Vekua-Bitsadze for the homogenous matrix equations (32) we obtain the following complex representation of the general solutions A V = 2Re{ϕ(z) + 4 Zz Zz Λ(t, t)R(z, z, t, t)ϕ(t)dtdt}, z0 z 0 B Ω = 2Re{f (z) + 4 Zz Zz Λ(t, t)r(z, z, t, t)f (t)dtdt}, z0 z 0 or Zz h i V (z, z) = 2Re αR(z, z, z0 , z 0 )ϕ(z) + Φ(t)R(z, z, t, z 0 )dt , z0 Volume 13, 2009 49 Zz h i Ω(z, z) = 2Re βr(z, z, z0 , z 0 ) + Ψ(t)r(z, z, t, z 0 )f (t)dt , z0 ³ α ϕ(z) = 2 Zz Φ(t)dt, V (z0 , z 0 ) = α, Φ(t) = ∂V (z0 , z 0 ) ´ , ∂z Ψ(t)dt, Ψ(t) = ∂Ω(z0 , z 0 ) ´ , ∂z z0 ³ β f (z) = 2 Zz Ω(z0 , z 0 ) = β, z0 where R and r are the Riemann’s matrix functions of the equations (32), ϕ(z) and f (z) are holomorphic column-matrices: ϕ(z) = (ϕ0 (z), · · ·, ϕN (z), ϕN +1 (z), · · ·ϕ2N (z))T , f (z) = (f0 (z), · · ·, fN (z))T . Then particular solutions of the matrix equations (32) have the form 1 V (z, z) = 4 ∧ Zz Zz Λ(t, t)R(z, z, t, t)X(t, t)dtdt, z0 z 0 1 Ω(z, z) = 4 ∧ Zz Zz Λ(t, t)r(z, z, t, t)Y (t, t)dtdt, z0 z 0 where A R(z, z, t, t) = E + 4 Zz Zz Λ(t1 , t1 )dt1 dt1 + t t ³ A ´2 Zz Zz ³ Zt1 Zt1 ´ Λ(t1 , t1 ) Λ(t2 , t2 )dt2 dt2 dt1 dt1 · · · 4 t t t r(z, z, t, t) = E + B 4 Zz t z Z Λ(t1 , t) dt1 dt1 + t t ³ B ´2 Zz Zz ³ Zt1 Zt1 ´ Λ(t1 , t1 ) Λ(t2 , t2 )dt2 dt2 dt1 dt1 + · · · 4 t t t t For the first boundary condition (in stress) we have ³ ∂ u dz ´ (m) (m) ∂ u+ ³ dz ´2 (m) 3 = a1 + i b1 , Im = c1 (on ∂S) (λ + µ) θ − 2µΛ ∂z ds ∂z ds (m) (m) (m) 50 Bulletin of TICMI The second boundary condition (in displacements) for any m takes the form (m) dz u+ ds (m) (m) (m) (m) = a2 + i b2 , u3 = c2 (on ∂S). The basic boundary conditions (N = 0) for any n have the form: a) for the first boundary problem (in displacements) (0,n) (0,n) (0,n) U (`) + i U (s) = i U + dz̄ (n) = d+ , ds (0,n) U 3 (n) = d 3 on ∂D (33) b) for the second boundary problem (in stresses) · µ ¶ ¸ (n) (0,n) 1 (0,n) + dz σ (``) + i σ (`s) = σ̄ r + − σ + r + = e +, dz̄ ¶ ¸ ·µ 2 (n) (0,n) (0,n) dz̄ σ (`n) = −Im σ +n = e3 on ∂D. ds (0,n) (0,n) (34) Here we present a general scheme of solution of boundary problems when the domain D is a circle of radius r0 . The first boundary problem for any n takes the form (on |z| = r0 ), (0,n) U + æ =− π ZZ (n) Λϕ0 (ζ)dξdη 1 Λdξdη 0 ϕ (z) − ψ(z) = G + + π ζ̄ − z̄ ζ̄ − z̄ ZZ D D (0,n) U (n) (35) 3 (n) = f (z) + f (z) = G 3 (z = reiϕ , ζ = ρeiψ ), (n) (0,1) (36) (0,n−1) where G + and G 3 are the known values containing solutions U i , · · · , U i of the previous approximations. (n) Let Λ(z, z̄) depend only on r = |z|, next ϕ0 (z), ψ(z) and G + are expanded in power series of the type 0 ϕ (z) = ∞ X k=0 k ak z , Ψ(z) = ∞ X k ∞ X (n) bk z , G + = k=0 Ak eikϑ . k=−∞ Substituting these expansions into (24), we obtain a0 = r0k+1 Ak+1 r0 æA1 + Ā1 , a = (k ≥ 1), k α0 æ2 − 1 æαk Āk α0 r0k+2 bk = − k − Ak+2 , (k ≥ 0), αk = 2 æαk+1 r0 Zr0 ρ2k+1 Λ(ρ)dρ. 0 Volume 13, 2009 (0,n) U 3 51 is representable in the form of the Poisson integral, 1 U 3 (r, ϑ) = 2π (0,n) Z2π (n) G 3 (ψ) 0 r02 − r2 dψ. r2 − 2r0 r cos(ψ − ϑ) + r02 (37) Thus for any n we can construct formal solutions of the problem (22), when N = 0. From the second boundary condition (23), we obtain (on ∂D) µ ¶ (n) (0,n) (n) (0,n) (0,n) 1 (0,n) dz̄ σ (``) + i σ (`s) = ` + ⇒ (λ + µ) Θ −2µ = P +, (38) U+ Λ dz (0,n) (n) (0,n) ∂ U 3 dz̄ (n) σ (`n) = `3 ⇒ Im (39) = P3 , ∂z ds (0,n) (0,n) 1 ∂ U + ∂ Ū + (0,n) + Θ = . Λ ∂z ∂ z̄ Consider the case of a spherical shell, whose midsurface is a spherical segment of radius R0 sin ϑ, where R0 is the radius of a sphere. Isometric coordinates on the sphere can be represented in the form ϑ z = x1 + ix2 = reiϕ , r = tg , Λ = 4R2 (1 + z z̄)−2 2 (0 ≤ ϑ ≤ ϑ0 ). Let the expressions µ ¶ ∞ ∞ X X (n) (n) k g+ , g3 = (ak , bk , ck )z , (Ak , Bk )eikϕ , (ϕ (z), Ψ (z), f (z)) = 0 0 k=0 (n) k=−∞ (0,1) (n) (0,n−1) be valid, where g+ and g3 are known values expressed by U i , · · · , U i of the previous approximations. Substituting these expansions into (27), (28) and taking into account that principal vector and moment of stresses are zero, we obtain 1 Ak , k 1 + 2æ(1 + r 2 )β 2µr0 k 0 · ¸ −1 (1 + r02 )−1 ((1 + r02 )k + 2r02 )Ak+1 bk = + Āk−1 (k ≥ 0). 1 + 2æ(1 + r02 )βk+1 r02 2µr0k−1 k + 2r02 Zz 1 2 R0 Bk (z − t)tk dt (k ≥ 1), B = 0, β (z) = ck = . 0 k µ 1 + r02 kr0k−1 z k+2 (1 + tz̄)3 ak = 0 52 Bulletin of TICMI From here we obtain the well-known Dini’s formula r0 U 3 (r0 , ϕ) = − π (0,n) Z2π (n) iϑ P3 (r0 , ϕ) ln |σ − z|dϑ + const (σ = r0 e ). 0 References [1] I.N. Vekua. Shell Theory: General Methods of Construction. Pitman Advanced Publishing Program, Boston-London-Melburne, 1985, 287p. [2] A. Bitsadze. Boundary Value Problems for Second Order Elliptic Equations. Moscow, Nauka, 1966, 203p (in Russian) [3] P.G. Ciarlet. Mathematical Elasticity. V.I three-dimensional elasticity. NorthHolland Publishing, Amsterdam-New-Oxford-York-Tokyo, 1998, 471p. [4] A.L. Goldenveizer. Theory of Elastic thin Shells. Moscow, Nauka, 1976, 512p. (in Russian) [5] T.V. Meunargia. A small parameter method for I. Vekua’s nonlinear and non-shallow shells. Proceedings of the IUTAM Symposium on Relation of Shell, Plate, Beam, and 3D Models Dedicated to Centenary of Ilia Vekua, 23-27 April, 2007, Tbilisi, Georgia. IUTAM Bookseries, 9 (2008), 155-166 [6] T.V. Meunargia. On the complex representations for the nonlinear and non-shallow shells. Rep. Enlarged Sess. Semin. I. Vekua Appl. Math., 23 (2009), 7-11 Received 30.11.2009; revised 23.12.2009; accepted 29.12.2009