Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 51, 1, pp. 131-141, Warsaw 2013 ON THE FIRST BOUNDARY VALUE PROBLEM OF THE PHOTOELASTIC THIN RING Djelloul Rezini, Tawfik Tamine University of UST-MB Oran, Faculty of Mechanical Engineering, Laboratoire LCGE, Algeria e-mail: redjellou@yahoo.fr Analytical solutions to the plane problems in terms of stresses for thin annular domains under compression loading arewell known in several papers.Moreover, the largemajority of the two-dimensional problems in the theory of elasticity are reducible to the solution of their boundary value problems. The two-dimensional photoelasticity methods easily provide the stress tensor components on the boundary, fromone photograph only. TheBeltrami-Michell equations with the Dirichlet photoelastic data state a well-posed hybrid problem in stress terms. It has been shown that the results obtained from the hybrid method developed in this paper, are applicable to any irregular shaped photoelastic domain of interest. Successful results have been obtained for more complicated forms and loads. The correctness of the results for the circular ring is confirmed and will be discussed in details. Key words: photoelasticity, Beltrami-Michell equation, annulus 1. Introduction The analysis of an annular thin disk subjected to diametral compression is often regarded as a problem of reference for verification and validation of theoretical or experimental techniques for stress or strain investigation of a study case, as done in (NavneetKumar andKhobragade (2011) and Sciammarella and Gilbert (1973). Analytical solutions to the plane problems in terms of stresses for such annular domains under compression loading are well known in several papers (Nomura et al., 2008; Filon, 1924). However, the concern of this paper is to show another way that exists in addition to those reference works listed by Tokovyy et al. (2010) for evaluating the stress state acting in a thin ring subject to diametral compression. Although the method developed in this study is not limited exclusively to this shape of structure, successful results have been obtained for more complicated forms and loads. Moreover, the large majority of the two-dimensional problems in the theory of elasticity are reducible to the solution of their boundary value problems. We shall be concerned only with those solutions in which the components of the stress tensor are continuous and single-valued throughout the region Ω ∪ Γ (Ω is the volume and Γ is the envelope surface) together with their first and second partial derivatives with respect to field variables. Moreover, the harmonic state of the stress invariant satisfies the Laplace equation, and is usually used to determine the interior potential of the stress sum if boundary values are known. It is well-known that each stress component is governed by the Beltrami-Michell equation (Timoshenko andGoodier, 1970). In fact, the source function of each Beltrami-Michell equation is expressed in terms of partial derivatives of the second order of the stress sumharmonic function. Therefore, the plane stress resolution is reduced to solving two Dirichlet problems successively. In spite of some apparent disadvantages, photoelasticity often serves as a useful complement to numerical analyses, especially for validation of complex stressedmodels orwhen theboundary conditions are difficult tomodel.The two-dimensional photoelasticitymethods easily provide the 132 D. Rezini, T. Tamine stress tensor components on theboundary, fromonephotoonly (seeKuske, 1979). TheBeltrami- Michell equationswith theDirichlet photoelastic data state awell-posedhybridproblem in terms of stress, as done by Rezini (1984). The purpose of the present study is summarized as follows: the Beltrami-Michell equations are discretized in terms of finite difference equations to provide harmonic solutions of each stress component in any irregular shaped photoelastic domain. Torealize this objective, thepresentpaper is organized in sections. InSection2, the statement of the problem and the expressions of stresses in two dimensions are reviewed in a convenient form for the analysis purpose.Thephotoelasticity proceeding and its associated boundaryvalues will bebrieflydescribed in Section 3.Thenumerical resolving of the photoelastic boundaryvalue problem in any domain geometries is presented in Section 4. Before conclusion, the results will be discussed in details in Section 5. 2. Statement of the stress problem Thedifficulty in analysis of stresses and strains in structural engineering depends on the comple- xity of the formand load of the structureparts and of the given specific conditions. Furthermore, from the strength ofmaterial point of view, the state of stresses ismore useful for engineers. The equilibrium equation, the compatibility condition, and thematerial law are the three fundamen- tal concepts of the theory of structures (Boresi et al., 2011). However, the theory of elasticity proposes two strategies to minimize the analysis. The first one provides analytical solutions to plane strain and plane stress problems and can be obtained by using several techniques which use functions to reduce the unknown of governing equations. The stresses are written in terms of this new function (Ugural and Fenster, 2011). According to the plane stress assumption in the absence of body forces, this problem is governed by the necessary and sufficient equilibrium conditions (Hetnarski and Ignaczak, 2011). The planar version of the equilibrium conditions are conveniently written withGreek indices as follows σαβ,β =0 (2.1) The stress function formulation is based on the idea representing the stress fields that satisfy the equilibrium equations. It is clear that equations (2.1) represent the necessary and sufficient condition for the existence of the function ψ, such that in absence of body forces the components of stresses are its partial derivates of thesecond order σαβ =− ∂2ψ ∂xα∂xβ + δαβ ∂2ψ ∂xκ∂xκ (2.2) where α,β,κ =1,2 and the Kronecker delta δij define the components of the identity matrix. Note that the summation over the repeatedGreek indices should be done. TheAiry stress func- tion ensures that the stresses according to (2.2) satisfy equilibrium conditions (2.1). This is a necessary condition but not sufficient. The function ψ should also ensure that the compatibi- lity and boundary conditions are satisfied. The introduced harmonic stress function obeys the Laplace potential. We use the notation Iσ =∇ 2ψ, where Iσ is evidently the first invariant or the trace of the stress tensor σ, equal to the stress sum. It is known that the linear invariant of stress (or strain) is a harmonic function in the absence of body forces Iσ = tr(σ)=∇ 2 ψ (2.3) From a formal contribution of the 2D Laplace-Operator ∇2 in equation (2.3), we obtain the Laplace potential equation of the stress sum ∇ 2 Iσ =∇ 2(σζ +σξ)= 0 (2.4) On the first boundary value problem of the photoelastic thin ring 133 where ζ-ξ is any orthonormal basis, By the contribution of the operator ∇2 in (2.2) with the help of equation (2.3) and rearrangement, we arrive at the Beltrami-Michell equations σαβ,κκ =−Iσαβ + δαβIσκκ (2.5) inwhich thecomma in the subscript indicates partial derivativewith respect to thecorresponding variables following further on; this notation will be applied in this paper. This was the second strategy of the theory of elasticity which provides solutions in terms of stress only if the boundary values are known. Note that Beltrami-Michell equations (2.5) are independent of the elastic constants. By examining the three-dimensional kinematic relations, it is found that the exact condition of the state stress in the two-dimensional problems of the elasticity for thin plates exists only if either Poisson’s ratio is set equal to zero (ν = 0) or the surface dilatation is constant and, does not exist except under theses conditions. This is an important assessment for the photoelasticity especially since it provides that there exists geometric similarity,material isotropy, linearity and similar applied loading of photoelastic specimen and its original. Then the stress distribution per unit load will be identical for each other. In the work by Muskhelishvili (1975) it is emphasized that for the elastic plane on the assumption that the holes contours are free from loading, the stress state does not depend on the elastic characteristics of the material. From equation (2.5), the following Poisson equations are available to determine the compo- nents of the Cartesian stress tensor individually ∇ 2σx =−Iσ,xx ∇ 2σy =−Iσ,yy ∇ 2τxy =−Iσ,xy (2.6) 3. Photoelasticity proceeding and associated boundary values The major task of the numerical treatment of the photoelastic data is the determination of stresses in all interior points of the specimen. However, the photoelastic experiment provides an incomplete solution; only two experimental results (σ1−σ2) are proportional to the isochromatic order N, and θ for eachpoint,while three stress componentparameters areneeded to completely define the stress state, sketched in as follows σ= [ σx τxy sym. σy ] ⇔ { N : isochromatic order θ : slope; isoclinic parameter Theprincipal stress element (Fig. 1) is used in the prediction of failure surfaces. Potential failure surfaces are the planes on which the maximum normal or maximum shear stress acts; in other words, in the principal planes and the plane of maximum shear. Fig. 1. Principal stress element configuration 134 D. Rezini, T. Tamine The equilibrium of the principal element (Fig. 1) provides the following expressions σm = σ1+σ2 2 τmax = σ1−σ2 2 (3.1) The principal stress σ1 is by convention the major and σ2 the minor one (σ1 > σ2). The following fundamental equation, (3.2), of photoelasticity, called the stress-optic law (Ramesh, 2000), which stipulates that the relative retardation N, isochromatic fringe order, at each point in themodel is directly proportional to the difference of principal stresses (σ1−σ2) at the point σ1−σ2 = Nfσ t = κN (3.2) where κ is an experimental constant, quotient of the stress-optical coefficient fσ, a constant that depends upon the model material and the wavelength of light employed (Ramesh, 2000), and t is the model thickness.We can rewrite the tensor as τ = 1 2 [ Iσ κN κN Iσ ] (3.3) Thismeans that the stress tensor τ relates the centre σm and the radius τmax of theMohr circle. Also in this stress state configuration, the first invariant of the stress tensor of equation (3.3) is verified Iσ =2σm. The stress tensor in relationship (3.3) is often used in order to separate the principal stresses, when its trace is avaible as experimental or numerical data (Kuske, 1979). To solve the set of partial differential equations (2.4) and (2.6), different kinds of stresses must be available on the boundaries. On the free boundaries, the stress state is generally easy to determine, while difficulties can arise in the loaded ones. The load types are various and each casemust be studied individually.When a concentrated load acts on the straight edge of a semi-infinite plate, the stress distribution is purely radial. To havemore details on several types of boundary loads, we can refer to Frocht (1948). By definition, the boundaries which are free from external shear and normal stresses will be referred to as free boundaries. The directions of the principal stresses are therefore either tangent or normal to the boundary with the normal stress assumed to be zero. It follows that at the points lying on the free boundaries, there is only one principal stress which is tangent to the boundary. Since the stress pattern provides the necessary data to calculate (σ1−σ2) at each point of the free boundary,where either σ1 or σ2 vanishes, the stress pattern gives directly the numerical value of the remaining principal stress. The sign of the boundary stress can be easily determined by considering the external loading or by pressure nail control on the boundary (seeWolf, 1976). The requiredboundarypotentials canbeeasily obtained fromphotoelastic studyof the specimen in which the principal stress sums can be found directly, because they are identical with the principal stress differences on the free boundaries (Doyle, 2004) (σ1+σ2) ∣ ∣ ∣ ∂Ω =(σ1−σ2) ∣ ∣ ∣ ∂Ω (3.4) The equilibrium of forces on an element in plain stress gives a relationship among the Cartesian stress components and the two principal stresses. The boundary equilibrium conditions and the knowledge of photoelasticity of the stress sum σx + σy can be coupled to yield the following system σxnx+ τxyny = tx τxynx+σyny = ty (σx+σy) ∣ ∣ ∣ ∂Ω = κN(x,y) ∣ ∣ ∣ ∂Ω (3.5) where nx, ny are the Cartesian components of the outward normal vector on the surface boun- dary and tx, ty are the Cartesian components of the specified boundary traction. The traction On the first boundary value problem of the photoelastic thin ring 135 wouldbe zero for the free boundary, or specifiedat regions of load application.Knowing the incli- nationangle θ (or the slope) at agivenpointon theedgeandthe tangential stress σB = κN(x,y) at the same point (see Fig. 2), the components of the Cartesian stresses are computed using separately the equilibrium conditions as follows σx = σB cos 2θ σy = σB sin 2θ τxy =−σB sinθcosθ (3.6) Fig. 2. Tensile state on the free boundary and associated Cartesian stresses 4. Numerical processing on an irregular domain The motivation of numerical processes to solve partial differential equations is to transform them into algebraic equations by means of finite differences or finite elements, or others. The simplest way is to replace the partial derivatives by discrete finite differences. FDM imposes usually a regular grid on the region of interest where the domain edges coincide with the axis of the coordinate system.This is one of themain disadvantages of FDMwhere complex geometries cannot be directly resolved by fitting the mesh to the object boundary. The crucial feature of the formulation is FDM discretisation of the homogeneous andmixed derivatives at irregularly shaped boundaries. For this task, many procedures exist; e.g. McKenney et al. (1996) proposed a fast Poisson-solver for complex geometries, Zhang (1998) developed a multigrid solution for Poisson’s equation. The Poisson-solvers list is long; a simple method to program is proposed for this task. The finite difference approximation is based on a uniform mesh size, and the resulting system of linear equations is solved by a residual method or multigrid method. A generalized approach is proposed in this paper in order to redefine the finite difference formula taking into account the irregular regions.Avarying5-point stencil is systematically used for thepurposeof discretization of the homogeneous and the mixed partial derivatives occurring in both the Laplace equation and in the right-hand-side of the Poisson equation. To approximate the solution to (2.4) and (2.6), a spatial discretization mostly in irregular geometries, using centred, second order finite-differences as discussed below is used. Let Φ be a generic grid variable used for the approximation.Using the 5-point stencil formula, the following compact algebraic equation at the stencil centres can be written as 4 ∑ k=0 α CkΦk =      0 Laplace f ∣ ∣ ∣ (k=0) Poisson (4.1) 136 D. Rezini, T. Tamine Fig. 3. Irregular 5-point stencils; LHS for homogeneous, RHS for mixed derivatives In this equation, the coefficients α Ck regulate the spacing to the boundary ∂Ω, and f represents each right-hand-side of stress component equations (2.6). For grid points far away or close to the boundary, the irregular 5-point discretization operator, introduced early by Shortley and Weller (1938) and still by Sangdong Kim et al. (2011), is used. The coefficients of the finite difference equations depend on the spacing to the boundaries. The Shortley-Weller coefficients will be distinctly expressed by α C1 = 1 h2α1(α1+α3) α C2 = 1 h2α2(α2+α4) α C3 = 1 h2α3(α3+α1) α C4 = 1 h2α4(α4+α2) α C0 =− 4 ∑ k=1 α Ck (4.2) Crucial undertaking is to formulate the discretization of the right-hand-side of the Poisson equation of shear stress, the last of equations (2.6) at the irregular boundaries. Themotivation of the following approach is to produce an accurate method for treating mixed derivatives. Near the boundary, the approximation of the second order mixed derivatives requires a 9-point stencil. A transformation by a rotation of 45◦ of the grid (stencil) to obtain a 5-point formula is also recommended. Notice that the 2-dimensional Taylor series approach, according to the approximation of mixed derivatives, leads to the similar form of Shortley-Weller’s coefficients by turning over 45◦ the stencil as shown in Fig. 3b. The finite difference coefficients for mixed derivates can be expressed in individual terms as follows β C1 = 1 2h2β1(β1+β3) β C2 = −1 2h2β2(β2+β4) β C3 = 1 2h2β3(β3+β1) β C4 = −1 2h2β4(β4+β2) β C0 =− 4 ∑ k=1 β Ck (4.3) The above symbols α, β are used respectively for the normal and inclined stencil. By definition, the irregular 5-point difference star is present when at least one of its star boughs becomes smaller than the mesh size h. The irregular difference star is the more general case and for the regular star one, αk = βk =1. Direct or iterativemethods are two different approaches that can beused for solving systems linear equations. In this study, the direct method which produces an exact solution in a finite number of operations is used. The advantages of this method are its numerical robustness and On the first boundary value problem of the photoelastic thin ring 137 the fact that it always guarantees a solution for various complex problemswithout the necessity of a convergence criterion. When solving successively the Laplace and the Poisson systems of equations, the errors in the solutions arise only from roundingoff the errors in the computational calculation. This method is tested on several applications examples and is found to produce accurate results. 5. Proceeding in annulus application Methods for automatic determination of parameters in photoelasticity were validated for the problem of a ring under diametrical load, as for example inGao (2010), ZhangDongsheng et al. (2002, 2007). Concerning the choice of the results for validation, a particular work is selected as the reference because of its more accurate results. Differentmethods were developed byGerlach (1968), later refined by Nurse and Allison (1972); and Redner (1974), who used a standard polariscope with fixed filters. All relevant data for the photoelastic measurement are measured electronically and are the input in a computer via an analogic-digital converter. A program was developed by Gerlach (1968) which enabled complete evaluation of the stresses along the line that had been measured on the model. The most important part of this program is the determination of an isochromatic fringe order from the light intensity distribution (skeletonizing of the fringe). For this, algorithms have been developed to allow exact determination of the fringe order by using two measurements with different wavelengths of light. Other parts of the program have to carry out the curve fit using themethod of least squares and to determine the stresses using the Shear-Stress-Differencemethod. The apparatus built byGerlach (1968) along with the programmade quasi-automatic evaluation of photoelastic data possible. Although this electrical measurementmethod is themost accurately method used, but it wasmore expensive. The photoelastic stress field fringe patterns are shown in Fig. 4. Isochromatic parameters [order] have been superimposed for convenience over the isoclinic parameters in the field along the line that had beenmeasured. Fig. 4. Photoelastic stress field fringe patterns of the annulus subjected to compressive forces; straight line 3-3 indicates the inclined cross section by 15◦) The stress integration can be started from the initial values, and the stress separation will be accomplished according to numerical integration of one of both equilibrium equations σx ∣ ∣ i = σx ∣ ∣ 0 − i ∑ 0 ∆τxy (∆x ∆y ) σy = σx± √ (σ1−σ2)2−4τ2xy (5.1) The initial values of stress are obtained from the boundary values at the point from conditions introduced by (3.6). For stress separation along the straight line, the last relation (5.1)2 is 138 D. Rezini, T. Tamine used. In this case, it is important to precise the difference between both procedures: in the SSD-process, the stress component σy is determined one from another by means of the Mohr circle statement; while the isoclinic parameter θ is used for the integration. It be noted that determination of the isoclinic parameter θ in its physical range is still a difficult problem. A number of full-field approaches have been proposed to determine it; see details in Ramesh et al. (2011), Fernández (2011), Pinit and Umezaki (2005, 2007), and Pinit (2007). The cumulative error in the integration process has also an effect on the stress separation. In the B-M BVP, each stress component occurs from the value on the boundary as BVP separately. In addition, it is noted that the annulus edge direction (slope) is the principal stress direction itself in the case of a free boundary. Fig. 5. Radial and tangential stress distributions along section 1 (Gerlach, 1968) Fig. 6. Radial and tangential stress distributions along section 2 (Gerlach, 1968) Fig. 7. Radial and tangential stress distributions along section 3 (Gerlach, 1968) On the first boundary value problem of the photoelastic thin ring 139 Separating in resulting data in individual components usually provides stresses at only di- screte locations. Techniques such as photoelasicity or thermoelasticity also suffer from this con- sequence of the point-by-point or line-by-line approach.However, discrete data can be converted into appropriate contours. Aswell as by FEM, post-processor concepts offer a suitable alternate approach to differentiate or help to convert discrete data into maps, particularly for irregularly geometrical shapes. In this purpose, the field results are presented in Figs. 8a,b. On account of symmetry, the different stresses are given as half fields. Fig. 8. Non-smoothed; (a) LHS: (σ1+σ2)-maps, RHS: τxy-maps; (b) LHS: σx-maps, RHS: σy-maps Generally, in the experimental analysis, the circular disk under diametral compression is used as the standard model for verification of the performance of any developed method, since the theoretical formula exists for the reconstruction, e.g. of the isochromatic parameters. It has been shown that the results obtained from the hybridmethod developed here, provides accurate results for a circular ring. 6. Conclusion The aim of this paper was to show an efficient additional method for stress separation (see a review by Fernández, 2010). With aminimum cost, the method is faster in proceeding, compa- red to those suggested in different works by Mahfuz et al. (1990), Jacob (1978) and Segerlind (1971). Based on numerical solving of the B-MBVP of Dirichlet type, this method reduces the photoelastic data to minimum, in order to separate the prevailing stresses. It is important to notice that thedirectmethod to solve the systemof finite difference equations, despite the sparse matrix, is more qualified to the nature of the experiment in general. The results of the investigation have shown that the method is easy to use and resolve engineering problems.Thismethod is suitable for automation of the hybridprocess of generating stress componentmaps and enables building thewhole stress tensor field.Thismethod is simple enough, so that it can be used with a few equipment elements. References 1. Boresi A.P., Chong A.P., Lee J.D., 2011,Elasticity in Engineering Mechanics, 3rd Ed., John Wiley and Sons, Inc. 2. Doyle J.F., 2004,Modern experimental stress Analysis, JohnWiley and Sons Ltd. 3. FernándezM.S-B., 2011,Towardsuncertaintyevaluation inphotoelasticmeasurements,J. Strain Analysis, 45, 275-285 140 D. Rezini, T. Tamine 4. FernándezM.S-B.,AlegreCalderón J.M.,BravoDiezP.M.,Cuesta Segura I.I., 2010, Stress-separation techniques in photoelasticity: a review, Strain, 1, 1-17 5. Filon L.N.G., 1924, The stresses in a circular ring, Inst. Civil Eng. London: Selected Engineer. Papers, 1, 12, 4-26 6. Frocht M.M., 1948,Photoelasticity, Vol. II, JohnWiley and Sons, NewYork 7. Gao S., 2010, In-Plane Stress Analysis using Tensor Field Photoelasticity, Thesis, University of British Columbia, Vancouver 8. GerlachH.-D., 1968,Eleklronische Hilfsmittel zur Automalisierung spannungsoptischerMessun- gen (ElectronicAids for theAutomation of PhotoelasticMeasurements),Doctoral thesis,University of Karlsruhe, FRGermany 9. Hetnarski R.B., Ignaczak J., 2011,The Mathematical Theory of Elasticity, Taylor Francis 10. Jacob K.A., 1978, Experimental numerical hybrid technique for stress analysis, VDI-Berichte, 335, 335-341 11. KuskeA., 1979, Separation of principal stresses in photoelasticity bymeans of a computer,Strain, 2, 43-49 12. Mahfuz H., Caser O., WongT.-L., 1990,Hybrid stress analysis by digitized photoelastic data and numerical methods,Experimental Mechanics, 2, 90-194 13. McKenney A., Greengard L., Mayo A., 1996, A fast Poisson solver for complex geometries, Journal of Computation Physics, 2, 348-355 14. Muskhelishvili N.I., 1975, Some Basic Problems of the Mathematical Theory of Elasticity, 4th Ed. English translation, Noordhoff International Publishing, Leyden 15. Navneet Kumar L., Khobragade N.W., 2011, Analysis of thermal stresses and displacement in a thick annular disc, Int. J. of Latest Trends in Mathematics, 2, 1, 39-46 16. Nomura Y., Pinit P., Umezaki E., 2008, Digital simulation of a circular ring loaded by a diametric compression for photoelastic analysis, Journal of JSEM, 8, Special Issue 83-87 17. NurseP.,Allison I.M., 1972,Automatic acquisitionofphotoelasticdata,J.B.C.S.A.Conference 18. Pinit P., 2007, Automated detection of singularities from orientation map of isoclinics in digital photoelasticity, 21st Conf.of Mech. Eng., Chonburi, Thailand 19. Pinit P., Umezaki E., 2005, Full-field determination of principal-stress directions using photo- elasticity with plane polarized RGB lights,Optical Review, 12, 3, 228-232 20. PinitP.,UmezakiE., 2007,Digitallywhole field analysis of isoclinic parameter in photoelasticity by four-step color phase-shifting technique,Optics and Lasers in Engineering, 45, 7, 795-807 21. Ramesh K., 2000,Digital Photoelasticity: Advanced Techniques and Applications, Springer 22. RameshK.,KasimayanT., SimonB.N., 2011,Digitalphotoelasticity–Acomprehensive review, Journal of Strain Analysis for Engineering Design, 46, 245-266 23. Redner S., 1974, New automatic polariscope system,Experimental Mechanics, 14, 12, 486-491 24. Rezini D., 1984, Randisochromaten als ausreichende Information zur Spannungstrennung und Spannungsermittlung, Doctoral thesis, Tech. University of Clausthal (TUC), FRGermany 25. Sangdong Kim, Soyoung Ahn, Philsu Kim, 2011, Local boundary element based a new finite difference representation for Poisson equations, Applied Mathematics and Computation, 217, 12, 5186-5198 26. Sciammarella C.A., Gilbert J.A., 1973, Strain analysis of a disk subjected to diametral com- pression bymeans of holographic interferometry,Applied Optics, 8, 12, 1951-1956 27. Segerlind L.J., 1971, Stress-difference elasticity and its application to photomechanics,Experi- mental Mechanics, 11, 440-445 On the first boundary value problem of the photoelastic thin ring 141 28. Shortley G.H., Weller R., 1938, The numerical solution of Laplace’s equation,Appl. Phys. 29. Timoshenko S.P., Goodier J.N., 1970,Theory of Elasticity, 3-rdEd., NewYork,Mc-GrawHill 30. Tokovyy Yu.V., Hung K.-M., Ma C.-C., 2010, Determination of stresses and displacements in a thin annular disk subjected to diametral compression, J. of Math. Sciences, 3, 342-354 31. Ugural A.C., Fenster S.K., 2011, Advanced Mechanics of Materials and Applied Elasticity, 5-th Ed., Prentice Hall 32. Wolf H., 1976, Spannungsoptik, Band 1 Grundlagen. Zweite Auflage, Springer 33. ZhangDongsheng,MinMaa, DwayneD.Arola., 2002, Fringe skeletonizing using an impro- ved derivative sign binarymethod,Optics and Lasers in Eng., 37, 51-62 34. Zhang Dongsheng, Yongsheng Han, Bao Zhang, Dwayne Arola, 2007, Automatic deter- mination of parameters in photoelasticity,Optics and Lasers in Eng., 45, 860-867 35. Zhang J., 1998, Fast and high accuracymultigrid solution of the three dimensional Poisson equ- ation, Journal of Computation Physics, 2, 449-461 Wyznaczanie pierwszej wartości brzegowej w zagadnieniu fotoelastyczności cienkiego pierścienia Streszczenie Analityczne rozważania płaskiego stanu naprężeńw cienkich strukturach pierścieniowych poddanych obciążeniu ściskającemu są szeroko znane w literaturze. Co więcej, większość przypadków tego typu za- gadnień teorii sprężystości daje się zredukować do problemuwyznaczania wartości własnych zagadnienia brzegowego.Metodydwuwymiarowej fotoelastycznościumożliwiają łatwą identyfikację składowychtenso- ra naprężeń na brzegu elementuw oparciu o pojedynczą fotografię.RównaniaBeltramiego-Michellawarz z parametrami fotoelastyczności Dirichleta pozwalają poprawnie sformułować hybrydowy model bada- nego układu w kontekście poszukiwanych naprężeń. W pracy wykazano, że taki model jest stosowalny do dowolnie nieregularnego kształtu próbki. Otrzymano bardzo dobre wyniki dla elementów o skom- plikowanym obrysie i poddanych złożonemu stanowi obciążenia. W szczegółowej analizie potwierdzono dokładność rezultatów dla pierścienia kołowego. Manuscript received February 29, 2012; accepted for print April 11, 2012