Jtam.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 48, 2, pp. 435-451, Warsaw 2010 ON THE PROBLEM OF A TRANSVERSELY ISOTROPIC HALF-SPACE WEAKENED BY A PENNY-SHAPED CRACK FILLED WITH A GAS Bohdan Monastyrskyy Pidstryhach Institute for Applied Problems of Mechanics and Mathematics, NASU, Lviv, Ukraine e-mail: bmonast @ gmail.com Andrzej Kaczyński Warsaw University of Technology, Faculty of Mathematics and Information Science, Warsaw, Poland; e-mail: akacz@mini.pw.edu.pl An axially-symmetric problem of a penny-shaped crack situated in a position parallel to the boundary of a semi-infinite transversely isotropic solid is formulated with due regard for the presence of an ideal gas in the crack.Themethod of theHankel integral transforms is used to solve this problem. The dual integral equations obtained are reduced to a set of integral equationswhich are solvednumerically.The graphs presented illustrate the influence of the gas on the stress intensity factors ofMode I andMode II. Key words: transversely isotropic half-space, gas-filled penny-shaped crack, integral equations, stress intensity factors 1. Introduction The class of problems for solids with cracks and cavities, provided that the defects are filled with some substance, has wide applications in many areas, namely, in geomechanics, the petroleum industry, gas-producing industry,mi- ning geotechnical engineering and others. Such problems are arisen during in- vestigation of hydraulic fracture of rocks, gas-filtration into cavities appearing during coal excavation, spalling of concrete at high temperature, etc. There is a considerable amount of literature on the topic. A great deal of interest is focused on themodelling of liquid or gas-filled cracks (see, for example, Abe et al., 1976; Zazovskii, 1979; Bui and Parnes, 1982; Advani et al., 1997; Feraille- Fresnet andEhrlacher, 2000; Savitski andDetournay, 2002; Feraille-Fresnet et 436 On the problem of a transversely isotropic half-space... al., 2003). Comprehensive accounts of developments pertaining to fluid-filled crack problems can be found in Chapter 8 of Bui’s book (2006). A simplified analysis of these problems, which retains the description of fracture phenome- na, can be carried out using the concepts of linear elastic fracture mechanics. This approach was oriented mainly to the construction of appropriate strain- stress solutions for idealized situations with some assumptions involving fluid crack interaction. In this respect,we canmention the research results obtained by Evtushenko and Sulim (1981) for a plane problem involving a crack filled with a compressible barotropic liquid, by Balueva and Dashevskii (1995) and Dashevskii (2007) who studied the growth of gas-filled cracks, by Kit et al. (2003) andMachyshyn andNagórko (2003) for systemswith gas-filled contact gaps.Attention should also be paid to an article byMatczyński et al. (1999) in which the combined thermal andmechanical influence of the heat-conducting ideal gas filling the crack on stresses is analysed in the case of plane strain. In the present paper, we continue the investigations originated by the au- thors (Kaczyński andMonastyrskyy, 2004, 2007), and study an axisymmetric elastostatic problemofdeterminationof disturbancesof stressesdueto thepre- sence of a gas-filled penny-shaped crack positioned parallel to the boundary of the semi-infinity body treated as a transversely isotropic medium (modelling, for example, a rock layered horizontally (Gil, 1991) or stratified rockmass (Sa- lamon, 1968)). Our goal is to examine the integrated effect of the crack filler on the stress distribution around the crack, essentially on the stress intensity factors. 2. Statement of the problem 2.1. Formulation Suppose that a penny-shaped crack of radius a is embedded in a trans- versely isotropic half-space with the axis of elastic symmetry normal to the crack plane as shown in Fig.1. We refer to a system of cylindrical coor- dinates (r,θ,z) with the origin placed at the centre of the crack and the z-axis in the transverse direction such that the crack occupies the region S = {(r,θ,z = 0) : 0 ¬ r ¬ a ∧ 0 < θ ¬ 2π} and z = h defines the boundary of the solid. Here, h is the distance of the crack away from the half-space surface. Keeping in mind the presence of a gas in this crack, we touch upon a problemwhichhas two interrelated components regarding the external loading B. Monastyrskyy, A. Kaczyński 437 Fig. 1. A half-space with a gas-filled penny-shaped crack and thebehaviour of thefiller.For formulation of this problem, some simplified assumptionswill bemade. It allows effective examination of the stressed-strain state of the half space and determination of pressure of the gas in the crack. In what follows, we assume that the body is subjected at infinity and the boundary to a constant tensile or compressive load p. The crack is filled with an ideal and compressible gas whose state is described by thewell-known Boyle-Clayperon-Mendeleyev equation, written in the simple form PgasV = g0 = const (2.1) where Pgas and V stand for the pressure and volume of the gas, and g0 is a constant on the assumption that themass of the gas and temperature remain constant. The mechanical action of the gas filled the crack is simulated by the internal pressure Pgas, so only uniformly distributed normal forces −Pgas act on the surfaces of the crack. It is noteworthy that Pgas is unknown and according to Eq. (2.1) is a function of gas properties and the volume V which is equal to the volume of the crack depending, in turn, on the external load p. Hence, the gas pressure is an additional unknown parameter of the problem to be determined in the course of its solution. Thus the problem under study lies in the determination of the stress-and- strain state of the body, payingmuch attention to the distribution of stresses in the neighbourhood of the crack. In particular, the stress intensity factors as the local important parameters controlling the fracture instability are of prime interest. 438 On the problem of a transversely isotropic half-space... 2.2. Governing equations The above-mentioned problemmay be treated as axially-symmetric (inde- pendentof theangle θ)with theonlynon-vanishingdisplacements in the radial and axial directions ur(r,z), uz(r,z) and components of the stress tensor σzz, σrz, σrr, σθθ. In this case, the linear constitutive relations of a transversely isotropicmedium characterised by the five elastic moduli c33, c13, c44, c11, c12 are (Lekhnitskii, 1963) σzz(r,z) = c33 ∂uz(r,z) ∂z + c13 1 r ∂[rur(r,z)] ∂r σrz(r,z) = c44 (∂ur(r,z) ∂z + ∂uz(r,z) ∂r ) (2.2) σrr(r,z) = c11 ∂ur(r,z) ∂r + c12 ur(r,z) r + c13 ∂uz(r,z) ∂z σθθ(r,z) = c12 ∂ur(r,z) ∂r + c11 ur(r,z) r + c13 ∂uz(r,z) ∂z The equilibrium equations for the unknown displacements ur(r,z) and uz(r,z), in the absence of body forces, are given by c11 ∂ ∂r (1 r ∂[rur(r,z)] ∂r ) +(c13+ c44) ∂2uz(r,z) ∂r∂z + c44 ∂2ur(r,z) ∂z2 =0 (2.3) c44 1 r ∂ ∂r ( r ∂uz(r,z) ∂r ) +(c13+ c44) 1 r ∂2[rur(r,z)] ∂r∂z + c33 ∂2uz(r,z) ∂z2 =0 3. The boundary-value problem and its solution Following the classical approach in crack problems based on the superposition principle,we separate theproblemunder study into twoparts: thefirst relating to thebodywithnocrack subjected to thegivenexterior loadpand thesecond, corrective part involving local perturbations caused by the gas-filling penny- shaped crack. Since the first part is trivial, we will draw attention to finding the corrective solution. To formulate the perturbed problem with the crack located on the plane z=0, it is convenient to treat the considered solid as a layer of thickness h, described by 0 ¬ z ¬ h joined to the half-space z ¬ 0 of the same transversely isotropic material (see Fig.1). Bearing in mind all aforementioned assumptions, we arrive at the following boundary conditions: B. Monastyrskyy, A. Kaczyński 439 – on the crack σ (1) zz (r,0)=σ (2) zz (r,0)=−Pgas −p 0¬ r¬ a σ (1) rz (r,0)=σ (2) rz (r,0)= 0 0¬ r¬ a (3.1) – outside the crack (continuity of stresses and displacements) σ (1) zz (r,0)=σ (2) zz (r,0) σ (1) rz (r,0)=σ (2) rz (r,0) a< r<∞ u (1) z (r,0)=u (2) z (r,0) u (1) r (r,0)=u (2) r (r,0) a¬ r <∞ (3.2) – on the boundary σ(2)zz (r,h) =σ (2) rz (r,h) = 0 (3.3) – at infinity (regularity conditions) σ(1)zz (r,−∞) =σ(1)rz (r,−∞) = 0 (3.4) where superscripts (1) and (2) refer to quantities associated with the region z¬ 0 and 0¬ z¬h, respectively. Besides, recall that the unknown gas pressure Pgas is found during solving the problem by using Eq. (2.1), in which V = ∫∫ S (u(2)z −u(1)z ) dS=2π a∫ 0 r[u(2)z (r,0)−u(1)z (r,0)] dr (3.5) The solution to the non-trivial perturbedproblem is grounded on the jump displacementmethod. It lies in reducing this problem to a set of simultaneous integral equations the solution to which can be obtained only in a numerical fashion. The use of the Hankel integral transforms (see Sneddon and Lowen- grub, 1969) is a mathematical tool. At the first stage, we solve an auxiliary problem described by Eqs. (3.3) and (3.4), and the following boundary conditions on the whole plane z=0 σ (1) zz (r,0)=σ (2) zz (r,0) σ (1) rz (r,0)=σ (2) rz (r,0) u (2) z (r,0)−u(1)z (r,0)=∆uz(r) u (2) r (r,0)−u(1)r (r,0)=∆ur(r) (3.6) where ∆uz(r) and ∆ur(r) stand for the jumps of normal and tangential di- splacements, respectively, unknown beforehand. A commonly employed method of obtaining the solution is based on the theory of Hankel’s transforms. Only some relevant results will be given. We 440 On the problem of a transversely isotropic half-space... apply Hankel’s transforms of the first order for Eq. (2.3)1 and the zero order for Eq. (2.3)2, with the aid of the definitions f̃n(ξ,z)≡Hn[f(r,z);r→ ξ]≡ ∞∫ 0 rf(r,z)Jn(rξ) dξ n=0,1 (3.7) where Jn stands for the Bessel function of the first kind of the order n, and ξ is the transformparameter. Using some properties ofHankel’s transform,we arrive at two-coupled ordinary differential equations for theHankel transforms of displacements (ũz)0 and (ũr)1. Solving the set of these equations separately for the half-space z¬ 0 (bearing Eq. (3.4) in mind) and the layer 0¬ z¬h, we get the following expressions: — for z¬ 0 (ũ(1)z )0 =S1(ξ)exp(k1ξz)+S2(ξ)exp(k2ξz) (3.8) (ũ(1)r )1 = c44−k21c33 (c13+ c44)k1 S1(ξ)exp(k1ξz)+ c44−k22c33 (c13+ c44)k2 S2(ξ)exp(k2ξz) — for 0¬ z¬h (ũ(1)z )0 =X1(ξ)cosh(k1ξz)+X2(ξ)sinh(k1ξz)+X3(ξ)cosh(k2ξz)+ +X4(ξ)sinh(k2ξz) (3.9) (ũ(1)r )1 = c44−k21c33 (c13+ c44)k1 [X2(ξ)cosh(k1ξz)+X1(ξ)sinh(k1ξz)]+ + c44−k22c33 (c13+ c44)k1 [X4(ξ)cosh(k2ξz)+X3(ξ)sinh(k2ξz)] in which the functions S1, S2 and X1, X2, X3, X4 are unknown and remain to be found.Moreover, k2i (i=1,2) are the roots of the equation c33c44k 4+(c213+2c13c44− c11c33)k2+ c44c11 =0 (3.10) Here we have confined to the case of distinct positive roots given explicitly by Ding et al. (2006) k1 = √ ( √ c11c33− c13)( √ c11c33+ c13+2c44) 4c33c44 + − √ ( √ c11c33+ c13)( √ c11c33− c13−2c44) 4c33c44 (3.11) B. Monastyrskyy, A. Kaczyński 441 k2 = √ ( √ c11c33− c13)( √ c11c33+ c13+2c44) 4c33c44 + + √ ( √ c11c33+ c13)( √ c11c33− c13−2c44) 4c33c44 Note that there are no principal difficulties to solve the problem in the special case of equal roots, but in the present paper it is omitted. With the help of Eqs. (3.8) and (3.9), the Hankel transforms of stresses σzz and σrz are found from Eqs. (2.2)1,2 to be: — for z¬ 0 (σ̃(1)zz )0 = c44 (c13+ c44) ξ · · [c13+ c33k21 k1 S1(ξ)exp(k1ξz)+ c13+ c33k 2 2 k2 S2(ξ)exp(k2ξz) ] (3.12) (σ̃(1)rz )1 =− c44 c13+ c44 ξ · · [ (c13+ c33k 2 1)S1(ξ)exp(k1ξz)+(c13+ c33k 2 2)S2(ξ)exp(k2ξz) ] — for 0¬ z¬h (σ̃2zz)0 = c44 (c13+ c44) ξ (c13+ c33k21 k1 [X2(ξ)cosh(k1ξz)+X1(ξ)sinh(k1ξz)]+ + c13+ c33k 2 2 k2 [X4(ξ)cosh(k2ξz)+X3(ξ)sinh(k2ξz)] ) (3.13) (σ̃(1)rz )1 =− c44 c13+ c44 ξ ( (c13+ c33k 2 1)[X1(ξ)cosh(k1ξz)+X2(ξ)sinh(k1ξz)]+ +(c13+ c33k 2 2)[X3(ξ)cosh(k2ξz)+X4(ξ)sinh(k2ξz)] ) Thenext step in the solution is to express the above transforms in terms of theHankel transformsof two functions given in the regionof the crack, namely, the jumps of normal and tangential displacements: ∆uz(r) and ∆ur(r). To this end, we apply the Hankel transformation to boundary conditions (3.3) and (3.6), and with the aid of Eqs. (3.8), (3.9), (3.12), (3.13), we see that these conditions are satisfied if the six unknown functions S1,S2 and X1,X2, X3,X4 fulfil the following set of linear equations 442 On the problem of a transversely isotropic half-space... c13+ c33k 2 1 k1 [X2(ξ)cosh(k1ξh)+X1(ξ)sinh(k1ξh)]+ + c13+ c33k 2 2 k2 [X4(ξ)cosh(k2ξh)+X3(ξ)sinh(k2ξh)] = 0 (c13+ c33k 2 1)[X1(ξ)cosh(k1ξh)+X2(ξ)sinh(k1ξh)]+ +(c13+ c33k 2 2)[X3(ξ)cosh(k2ξh)+X4 sinh(k2ξh)] = 0 (c13+ c33k 2 1) k1 [X2(ξ)−S1(ξ)]+ (c13+ c33k 2 2) k2 [X4(ξ)−S2(ξ)] = 0 (3.14) (c13+ c33k 2 1)[X1(ξ)−S1(ξ)]+(c13+ c33k22)[X3(ξ)−S2(ξ)] = 0 X1(ξ)−S1(ξ)+X3(ξ)−S2(ξ)= ( ∆̃uz(ξ) ) 0 c44−c33k21 (c13+ c44)k1 [X2(ξ)−S1(ξ)]+ c44− c33k22 (c13+ c44)k2 [X4(ξ)−S2(ξ)] = ( ∆̃ur(ξ) ) 1 Its solution is written in the form X1(ξ)=S1(ξ)− c13+ c33k 2 2 c33(k 2 1 −k22) ( ∆̃uz(ξ) ) 0 X2(ξ)=S1(ξ)− (c13+ c33k 2 2)k1 c33(k 2 1 −k22) ( ∆̃ur(ξ) ) 1 (3.15) X3(ξ)=S2(ξ)+ c13+ c33k 2 1 c33(k 2 1 −k22) ( ∆̃uz(ξ) ) 0 X4(ξ)=S2(ξ)+ (c13+ c33k 2 1)k2 c33(k 2 1 −k22) ( ∆̃ur(ξ) ) 1 provided that S1(ξ)= c13+ c33k 2 2 c33(k 2 1 −k22) [D1z(ξ) ( ∆̃uz(ξ) ) 0 +D1r(ξ) ( ∆̃ur(ξ) ) 1 ] (3.16) S2(ξ)= c13+ c33k 2 1 c33(k 2 1 −k22) [D2z(ξ) ( ∆̃uz(ξ) ) 0 +D2r(ξ) ( ∆̃ur(ξ) ) 1 ] where (i=1,2) Diz(ξ)= 1 2 ( (−1)3−i+ k1+k2 k1−k2 exp(−2kiξh)− 2ki k1−k2 exp[−(k1+k2)ξh] ) (3.17) Dir(ξ)= ki 2 ( (−1)3−i− k1+k2 k1−k2 exp(−2kiξh)+ 2k3−i k1−k2 exp[−(k1+k2)ξh] ) B. Monastyrskyy, A. Kaczyński 443 Thus, we can establish the representations of the Hankel transforms of di- splacements (3.8), (3.9) and stresses (3.12), (3.13) through the transforms( ∆̃uz(ξ) ) 0 and ( ∆̃ur(ξ) ) 1 with the help of Eqs. (3.15)-(3.17). It can be obse- rved that these representations satisfy all theboundaryconditions of theposed principal problem, except conditions (3.1) and (3.2)2. Applying theHankel in- version to Eqs. (3.8), (3.9) and (3.12), (3.13), we find that conditions (3.1) and (3.2)2 yield the system of simultaneous dual integral equations for the functions ( ∆̃uz(ξ) ) 0 and ( ∆̃ur(ξ) ) 1 : — for 0¬ r¬ a ∞∫ 0 [( 1 k1k2(k1+k2) −U1z(ξh) )( ∆̃uz(ξ) ) 0 −U1r(ξh) ( ∆̃ur(ξ) ) 1 ] ξ2J0(ξr) dξ= = 2c33(c13+ c44)(Pgas +p) (c13+ c33k 2 1)(c13+ c33k 2 2)c44 (3.18) ∞∫ 0 [( 1 k1+k2) −U2r(ξh) )( ∆̃ur(ξ) ) 1 −U2z(ξh) ( ∆̃uz(ξ) ) 0 ] ξ2J1(ξr) dξ=0 — for r >a ∞∫ 0 ξ ( ∆̃uz(ξ) ) 0 J0(ξr) dξ=0 (3.19) ∞∫ 0 ξ ( ∆̃ur(ξ) ) 1 J1(ξr) dξ=0 In the above, the quantities Uiz(ξh), Uir(ξh), i=1,2 stand for U1z(ξh)= 1 (k1−k2)2 2∑ i,j=1 (−1)i+j2exp[−(ki+kj)ξh] ki+kj U1r(ξh)=U2r(ξh)=− 1 (k1−k2)2 [exp(−k2ξh)− exp(−k1ξh)]2 (3.20) U2z(ξh)= 1 (k1−k2)2 2∑ i,j=1 (−1)i+j 2kikj exp[−(ki+kj)ξh] ki+kj Following Sneddon (1996), if the unknown transforms are taken to be of the form 444 On the problem of a transversely isotropic half-space... ( ∆̃uz(ξ) ) 0 = ξ−1 a∫ 0 ϕz(t)sin(ξt) dt (3.21) ( ∆̃ur(ξ) ) 1 = ξ−1 a∫ 0 ϕr(t) (sin(ξt) ξ − cos(ξt) ) dt then Eqs. (3.19) are identically satisfied, and the inserting of Eqs. (3.21) into (3.19) yields two equations for the new auxiliary functions ϕz and ϕr defined in [0,a] 1 k1k2(k1+k2)r ∂ ∂r r∫ 0 tϕz(t)dt√ r2− t2 − a∫ 0 ϕz(t) dt ∞∫ 0 ξU1z(ξh)sin(ξt)J0(ξr) dξ+ − a∫ 0 ϕr(t) dt ∞∫ 0 ξU1r(ξh) (sin(ξt) ξ − cos(ξt) ) J0(ξr) dξ= = 2c33(c13+ c44)(Pgas +p) (c13+ c33k 2 1)(c13+ c33k 2 2)c44 (3.22) 1 k1+k2 ( ∂ ∂r + 1 r ) r∫ 0 ϕr(t)dt√ r2− t2 − a∫ 0 ϕz(t) dt ∞∫ 0 ξU2z(ξh)sin(ξt)J1(ξr) dξ+ − a∫ 0 ϕr(t) dt ∞∫ 0 ξU2r(ξh) (sin(ξt) ξ − cos(ξt) ) J1(ξr) dξ=0 Finally, Eqs. (3.22) may be inverted to give π 2 1 k1k2(k1+k2) ϕz(r)− a∫ 0 ϕz(t)K1z(r,t) dt− a∫ 0 ϕr(t)K1r(r,t) dt= = 2c33(c13+ c44)(Pgas +p) (c13+c33k 2 1)(c13+ c33k 2 2)c44 (3.23) π 2 1 k1+k2 ( ϕr(r)+ r∫ 0 ϕr(t)dt t ) − a∫ 0 ϕz(t)K2z(r,t) dt+ − a∫ 0 ϕr(t)K2r(r,t) dt=0 B. Monastyrskyy, A. Kaczyński 445 Here, the following notations for the kernels have been employed K1z(r,t)= 1 (k1−k2)2 2∑ i,j=1 (−1)i+j 2 ki+kj I1(h,ki+kj,r,t) K1r(r,t)=− 1 (k1−k2)2 2∑ i,j=1 (−1)i+j[I5(h,ki+kj,r,t)− I2(h,ki+kj,r,t)] (3.24) K2z(r,t)=− 1 (k1−k2)2 2∑ i,j=1 (−1)i+jI3(h,ki+kj,r,t) K2r(r,t)= 1 (k1−k2)2 2∑ i,j=1 (−1)i+j 2kikj ki+kj · ·[I6(h,ki+kj,r,t)− I4((h,ki+kj,r,t)] in which I1(h,k,r,t) = ∞∫ 0 exp(−hξk)sin(ξt)sin(ξr) dξ= = 2hkrt [h2k2+(r− t)2][h2k2+(r+ t)2] I2(h,k,r,t) = ∞∫ 0 exp(−hξk)cos(ξt)sin(ξr) dξ= = r(h2k2+r2− t2) [h2k2+(r− t)2][h2k2+(r+ t)2] I3(h,k,r,t) = ∞∫ 0 exp(−hξk)sin(ξt)[1− cos(ξr)] dξ= = r2t(3h2k2+r2− t2) [h2k2+(r− t)2](h2k2+ t2)[h2k2+(r+ t)2] (3.25) I4(h,k,r,t) = ∞∫ 0 exp(−hξk)cos(ξt)[1− cos(ξr)] dξ= = r2hk(h2k2+r2−3t2) [h2k2+(r− t)2](h2k2+ t2)[h2k2+(r+ t)2] I5(h,k,r,t) = ∞∫ 0 exp(−hξk) sin(ξt) ξt sin(ξr) dξ= 1 4t ln h2k2+(r+ t)2 h2k2+(r− t)2 446 On the problem of a transversely isotropic half-space... I6(h,k,r,t) = ∞∫ 0 exp(−hξk) sin(ξt) ξt [1− cos(ξr)] dξ= = 1 2t ( arctan r− t hk +2arctan t hk −arctan r+ t hk ) Note that Eqs. (3.23) contain the unknown pressure of the gas Pgas. To determine this quantity, we use governing state equation (2.1) and, in view of Eqs. (3.5) and (3.21)1, we can obtain the following formula Pgas = g0 2π a∫ 0 rϕz(r) dr (3.26) Thus, the original problem is reduced to the solution of the system of simultaneous integral equations (3.23) supplemented with Eq. (3.26). Once the functions ϕz(r) and ϕr(r) are known, the displacements and stresses in the half-space can be find by applying Hankel’s inversion theorem to Eqs. (3.8), (3.9) and (3.12), (3.13) with the known functions given by Eqs. (3.15), (3.16) and (3.21). From the viewpoint of linear fracturemechanics (seeKassir andSih, 1975), it is of great importance to investigate highly intensified normal and tangen- tial stresses around the crack edge resulting in fracture initiation under the environment of a given external load and the presence of the gas in the crack. The most widely used fracture criterions in Mode I and II of crack propaga- tion are based on the knowledge of the stress intensity factors (SIFs) KI,KII describing the asymptotic behaviour of stresses in the immediate vicinity of the crack border r >a, i.e. σzz(r,θ,0)= KI√ 2π(r−a) +O(1) σrz(r,θ,0)= KII√ 2π(r−a) +O(1) (3.27) It turns out that in our case these factors are given in terms of the solution to governing integral equations (3.23) as KI = √ π a c44(c13+ c33k 2 1)(c13+ c33k 2 2) 2(c13+ c44)c33k1k2(k1+k2) ϕz(a) (3.28) KII = √ π a c44(c13+ c33k 2 1)(c13+ c33k 2 2) 2(c13+ c44)c33(k1+k2) ϕr(a) B. Monastyrskyy, A. Kaczyński 447 4. Numerical analysis 4.1. Numerical procedure The complicated system of coupled integral equations (3.23) can be so- lved only by recourse to numerical techniques. A certain numerical procedure, briefly outlined below, was used. We shall proceed on the well-known fact that any continuous function in a bounded domain can be uniformly approximated up to any accuracy by a polynomial. By some arguments resulting from the structure of kernels and geometry of the problem in hand, we represent the approximated solutions to governing equations (3.23) as ϕz(r)≈ϕzN(r)= cz1r+ cz2r3+ . . .+czNr2N−1 (4.1) ϕr(r)≈ϕrM(r)= cr1r2+ cr2r4+ . . .+ crMr2M where czn (n=1,2, . . . ,N) and crm (m=1,2, . . . ,M) stand for the unknown coefficients to bedetermined. Substituting the above-assumed expressions into Eqs. (3.23) and (3.26), and then satisfying them in the set of chosen collocation points (Eq. (4.1)1 at the points rn = na/N and Eq. (4.1)2 at the points rm =ma/M, we arrive at a set of non-linear algebraic equations (the discrete analogue of Equations (3.23), (3.26)) for the unknown coefficients czn, crm and parameter Pgas. Its solution is found by Newton’s method. The desired accuracy is achieved by increasing the power of approximating polynomials in Eqs. (4.1). 4.2. Numerical results The numerical analysis was carried out for the following dimensional pa- rameters Pgas =10 3Pgas c44 p=103 p c44 c11 = c11 c44 =3.88 c33 = c33 c44 =3.15 c13 = c13 c44 =1.31 h= h a g0 = g0 c44a3 =10−5 KI =10 3KI √ a c44 KII =10 4KII √ a c44 (4.2) InFig.2 a graph of the internal gas pressure is plotted against the external load. As the load increases, the pressure of the gas decreases. The high slope 448 On the problem of a transversely isotropic half-space... of the curve is observed for the compressive load. While the external load becomes tensile, the slope decreases and Pgas tends to zero. Fig. 2. Dependence of the gas pressure on the external load Fig. 3. Variations of SIF of Mode I (a) andMode II (b) with the external load A graphical representation of the SIF of Mode I and II is given in Fig.3a andFig.3b, respectively. As in the case presented inFig.2, the dependences of theSIFs on the external load arenonlinear. It shouldbe remarkedhere thatwe obtain physically reasonable values of theSIFswithin the range of compressive external load. This directly indicates the effect of the crack filler. Moreover, it can be seen that KI and |KII| increase in magnitude as the boundary is approached, i.e. for decreasing values of h=h/a. This tendency has also been observed in Fig.4 for p=0. Unlike that, Fig.5 shows that the pressure of the gas in this case slowly decreases as the crack surface approaches the half-space boundary. 5. Conclusions Thepresented researchwas carried outwith the aim of demonstrating the role of the gas filling a penny-shaped crack situated parallel to the boundary of a semi-infinite transversely isotropic space on the limiting equilibrium stress B. Monastyrskyy, A. Kaczyński 449 Fig. 4. SIF ofMode I (a) andMode II (b) versus h=h/a Fig. 5. Boundary effect on the gas pressure state.As canbe seen fromthenumerical solution to complex integral equations (3.23), a change in the mechanical behaviour has been noted in comparison to the case of the non-filled crack. The main result of the paper is the non- linearity of the relations between the external load and the internal pressure of the gas as well as the stress intensity factors. This phenomenon is due to the non-linear response of the filler on the change of its volume, governed by the state equation. References 1. Abe H., Mura T., Keer L.M., 1976, Growth of penny-shaped crack in hy- draulic fracturing of rocks, Journal of Geophysical Research, 81, 29, 5335-5340 2. Advani S.H., Lee T.S., Dean R.H., Pak C.K., Avasthi J.M., 1997, Con- sequences of fluid lag in three-dimensional hydraulic fractures, International Journal for Numerical and Analytical Methods in Geomechanics, 21, 229-240 3. Balueva A.V., Dashevskii I.N., 1995, Qualitative estimates of growth of gas-filled cracks, Izvestiya AN SSSR:Mekhanika Tverdogo Tela, 30, 6, 122-128 4. Bui H.D., 2006, Fracture Mechanics. Inverse Problems and Solutions, Solid Mechanics and its Applications 139, Springer, The Netherlands, Dordrecht 450 On the problem of a transversely isotropic half-space... 5. Bui H.D., Parnes R., 1982, A reexamination of the pressure at the tip of a fluid-filled crack, International Journal of Engineering Science, 20, 11, 1215-1220 6. Dashevskii I.N., 2007,To the kinetics of diffusion cracks,Mechanics of Solids, 42, 1, 149-156 7. DingH., ChenW., ZhangL., 2006,Elasticity of Transversely IsotropicMa- terials, Solid Mechanics and its Applications 126, Springer, The Netherlands, Dordrecht 8. Evtushenko A.A., Sulim G.T., 1981, Stress concentration near a cavity filled with a liquid,Materials Science, 16, 6, 546-549 9. Feraille-Fresnet A., Bui H.D., Ehrlacher A., 2003, Hydrostatic inte- raction of a wetting fluid and a circular crack in an elasticmaterial,Mechanics of Materials, 35, 581-586 10. Feraille-Fresnet A., Ehrlacher A., 2000, Behaviour of a wet crack sub- mitted to heating up to high temperature,Mechanics ofMaterials, 32, 471-484 11. Gil H., 1991,The Theory of Strata Mechanics, Developments in Geotechnical Engineering 63, Elsevier-PWN,Amsterdam-Warsaw 12. Kaczyński A., Monastyrskyy B., 2004, On the problem of some interface defect filled with a compressible fluid in a periodic stratified medium, Journal of Theoretical and Applied Mechanics, 42, 1, 41-57 13. Kaczyńki A., Monastyrskyy B., 2007, The elasticity problem for a stra- tified semi-infinite medium containing a penny-shaped crack filled with a gas, Acta Mechanica et Automatica, 1, 1, 63-66 14. KassirM.K., SihG.C., 1975,Three-Dimensional CrackProblems.Mechanics of Fracture 2, Noordhoff International Publishing, The Netherlands, Leyden 15. Kit G.S, Martynyak R.M., Machishin I.M., 2003, The effect of a fluid in the contact gap on the stress state of conjugate bodies, International Applied Mechanics, 39, 3, 292-298 16. Lekhnitskii S.G., 1963,Theory of Elasticity of an Anisotropic Elastic Body, Holden-Day, San Francisco 17. Machyshyn I., Nagórko W., 2003, Interaction between a stratified elastic half-space and an irregular base allowing for the intercontact gas, Journal of Theoretical and Applied Mechanics, 41, 2, 289-304 18. Matczyński M., Martynyak R., Kryshtafovych A., 1999, Contact pro- blemof a crackfilledwith heat-conducting gas,Proceedings of 3rd International Congress on Thermal Stresses, Cracow, Poland, 127-130 19. SalamonM.D.G., 1968,Elasticmoduli of a stratified rockmass, International Journal of Rock Mechanics and Mining Sciences and Geomechanics Abstracts, 5, 6, 519-527 B. Monastyrskyy, A. Kaczyński 451 20. Savitski A.A., Detournay E., 2002, Propagation of a penny-shaped fluid- driven fracture in an impermeable rock: asymptotic solutions, International Journal of Solids and Structures, 39, 6311-6337 21. Sneddon I.N., 1966, Mixed Boundary Value Problems in Potential Theory, North-Holland Publishing Company, The Netherlands, Amsterdam 22. Sneddon I.N., LowengrubM., 1969,CrackProblems in theClassical Theory of Elasticity, Wiley, NewYork 23. Zazovskii A.F., 1979, Growth of coin-shaped crack produced by hydraulic fracture inan impermeable rock, IzvestiyaANSSSR:MekhanikaTverdogoTela, 14, 2,103-109 [in Russian] O zagadnieniu poprzecznie izotropowej półprzestrzeni osłabionej szczeliną kołową wypełnioną gazem Streszczenie W pracy rozważono osiowo-symetryczne zagadnienie kołowej szczeliny wypełnio- nej idealnym gazem i położonej równolegle do brzegu półprzestrzeni sprężystej po- przecznie izotropowej. Używając techniki transformacji całkowej Hankela, rozpatry- wany problem został sprowadzony do złożonego układu równań całkowych. Na pod- stawie procedury numerycznej zbadano i zilustrowano wpływ gazu na współczynniki intensywności naprężeń. Manuscript received September 14, 2009; accepted for print October 22, 2009