Jtam.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 48, 1, pp. 45-70, Warsaw 2010 NUMERICAL AND EXPERIMENTAL ANALYSIS OF POST-CRITICAL DEFORMATION STATES IN A TENSIONED PLATE WEAKENED BY A CRACK Tomasz Kopecki Rzeszów University of Technology, Faculty of Mechanical Engineering and Aeronautics, Rzeszów, Poland; e-mail: tkopecki@prz.edu.pl The paper presents the methodology of determination of the stress di- stribution in post-critical state of deformation of the rectangular plate weakenedwith the crack subjected to tension.The problemwas formula- ted as physically and geometrically non-linear. Using the finite elements method, numerical analyseswere performed.While solving the nonlinear issue, the progressive change of geometry of the structure in successive incremental steps were compared with results of experimental studies, performed simultaneously. The obtained results made the base for the assessment of reliability of effects of nonlinear numerical analysis, condi- tioned by the presence of imperfections of the plate in the neutral state. Two kinds of imperfections were considered: geometric - based on the assumption of the preliminary deflection of the plate in the zone of we- akness and the second one – in form of load perturbation, normal to the middle surface of the plate. Key words: plate, tension, crack, buckling, experiments, finite elements 1. Introduction Rational approach to load-bearing structure design suggests necessity of fo- cusing special attention on zones characterised with presence of high stress levels and gradients that are crucial for the structure durability and reliability. Determination of stress states in such zones in the design phase, when appro- priate amendments can be introduced to the structure before time-consuming and expensiveworkshop prototype solution is realised, requires analysis of the structure performancewithin the full range of operationally permissible defor- mations. The above applies, in particular, to thin-walled structures in which post-critical deformation states occur within admissible load ranges. Solution 46 T. Kopecki to the problem of stress distribution within these ranges requires application of nonlinear analysis based on numerical methods, mainly the finite element method. Despite the existence of a wide spectrum and algorithms useful in solving nonlinear problems, in the case of thin-walled structures with complex geometry, geometrical configurations corresponding to consecutive equilibrium states and obtained in the course of solving of a nonlinear problem can show significant divergence with respect to the current deformation state observed in the experiment. Among that category of structures, one can count thin plates weakened by presence of cracks or cutouts. When such structures are subject to tensile stress in conditions of reaching or exceeding the critical load value, displace- ments in the direction normal to the plate plane known as wrinkling build up in the weakened zone appear and develop with the increasing load. The occurredpost-critical deformation state results in fundamental change of plate geometry, a consequence of which redistribution of stress is significant. As the cause of loss of stability of a plate weakened by e.g. a crack, the stress state arising as a consequence of tension can be considered. The state is characterised with presence of strong gradients, especially in the area adjacent to the crack front.That can be seen in an isochromatic fringe pattern observed in the plate model made of the optically active material (Fig.1). Fig. 1. Isochromatic fringe patterns in a tensioned plate with a crack High orders of isochromatic fringes prove occurrence of high values of dif- ferences between main normal stresses with opposite signs. In the vicinity of the crack front, tensioning stresses occur oriented in the load direction toge- ther with high values of compression stress oriented transversally which, after reaching the critical load value, create favourable conditions for local buckling of the plate. Numerical and experimental analysis of post-critical... 47 Problems concerning stability of rectangular plates subject to uniaxial ten- sile stress andweakenedbypresenceof centrally located crackswere considered bymany authors (Brighenti, 2005a,b;Dyshel, 2002;MarkströmandStoräkers, 1980; Riks et al., 1992; Shaw and Huang, 1990; Sih and Lee, 1986). Those studies, carried out on the grounds of linearized theories, included mainly problems concerning determination of critical load values and plate buckling forms, depending on the orientation angle of the crack. Striving to extend the scope of research of the wrinkling effect typical for thin-walled structures, in this paper one considered the problem consisting in determination of stress fields in advanced post-critical deformation states in a plate weakened by a crack. It was assumed that the problemwas nonlinear in both physical and geometrical sense. One assumed as the postulate that for solving such a problem it is necessary to compare results of numerical analysis with appropriate results of experimental research. To this end, a plate of material showing elastic and inelastic properties was made, instantaneous characteristic ofwhichwasdetermined in anuniaxial tensile test. In the course of the experiment, the plate was subject to a variable load with a pulsating tensile force, as a result of which length of the crack increased, preserving its ”natural” character of the defect. Numerical analysis was carried out for three selected crack lengths obtained in the course of fatigue tests, for which deformation distribution in the direction normal to the plate was recorded by means of the shadowmoiré method (Patorski and Kujawińska, 1993). Numerical calculations were carried out by means of the finite elements method. Emphasis was put on effectiveness of used methods and procedures ensuring conformity of calculation results with the experiment. The degree of deformation conformity in the post-critical state was adopted as a criterion of reliability of the stress state determined on the grounds of nonlinear numerical analysis (Crisfield, 1997; Felippa, 1976; Felippa et al., 1994; Theory..., 2006). 2. Experimental research The subject of the experiment consisted in a plate with dimensions 300mm×250mm, 0.7mm thick (Fig.2) made of polycarbonate with physical characteristics presented in Fig.3. In the central part of the plate, a crack was cut perpendicular to the load direction, with the initial length close to L=30mm. The full size of 30mm was obtained as a result of propagation of the crack caused by a pulsating tensile load, varying from zero to Pmax =1500N (Kopecki and Zacharzewski, 48 T. Kopecki Fig. 2. Plate geometry Fig. 3. Tensile stress-strain plot for the plate material 2006). Theplatewasmounted in the experimental set-up providing a constant pressure force of the clamps. This kind of mount provided homogeneity of the displacement field with the dominating component corresponding to the tensile force direction, outside theweakness zone, located in the vicinity of the crack. This homogeneity was monitored during the experiment by continuous observation in polarized light. Figure 4 shows the experimental set-up with the plated mounted. Asa result ofvariable load, lengthof the crack increasedwith the increasing number of load cycles. The experimentwas carried out until the crack reached length of 70mm. Optical polarization properties of the plate material made possible simul- taneous recording of optical phenomena occurring in circularly polarized light. The registration was carried out using the reflected lightmethod. To this end, the inside surface of the platewas coveredwith a reflective layer. Figure 5 pre- sents distribution of optical effects for three deformation states corresponding Numerical and experimental analysis of post-critical... 49 Fig. 4. The set-up for examination of crack propagation, isochromatic fringe patterns and transverse deformations bymeans of the moiré method Fig. 5. Isochromatic fringe patterns in the crack zone at maximum load conditions; (a)L=30mm, (b) L=50mm, (c)L=70mm to selected crack lengths. Recording of resultswasmade at the same load level P =1500N. The obtained patterns did notmake a base for quantitative interpretation of the results because due to the presence of bending state in the zones of high effective stress levels, they cannot be interpreted as isochromatic lines (Aben, 1979; Laermann, 1982). Those patterns are very useful as qualitative experimental results for the assessment of the effective stress layouts obtained numerically and also for determination of load levels corresponding to first, local permanent deformations. 50 T. Kopecki In each of the phases considered, deflection (displacement in the direc- tion normal to the plate plane) was registered. A photograph depicting the post-critical deformation state corresponding to crack length L = 70mm is presented in Fig.6. Fig. 6. Picture of plate wrinkling in the vicinity of crack with length L=70mm The strain field recording was carried out by means of the shadow moiré method (Laermann, 1982). As a result, images of level contours were obtained (Fig.7) corresponding to three selected plate weakening states. Fig. 7. Images of moiré fringes representing qualitative and quantitative nature of transverse deformations: (a)L=30mm, f max =3.12mm, (b)L=50mm, f max =4.42mm, (c)L=70mm, f max =5.72mm Numerical and experimental analysis of post-critical... 51 To each of the contours, a constant deflection corresponded. From selected raster parameters and image recording conditions, the distance between neigh- bouring contours was determined as equaling 0.26mm. Thus, the maximum plate deflections in selected deformation phases were calculated (Fig.7). 3. Numerical analysis In nonlinear analysis of load-bearing structures, relations between a set of static parameters and corresponding set of geometric parameters can be pre- sented in the form of a matrix equation (Aben, 1979; Felippa, 1976; Felippa et al., 1994; Rakowski and Kacprzyk, 1993) g=K−1(g)f (3.1) where g is a set of geometric parameters describing the system deformation state caused by the load, f is a set of static parameters, and K is the stiffness matrix depending on the set of geometric parameters determining the current deformation state and a nonlinear constitutive relation. In view of permanent deformations observed in the course of experiments, the plate material physical characteristics determined in the uniaxial tensile test (Fig.3) was approximated by means of the ideally elastic-plastic body (Fig.8). Fig. 8. Themodel of the material Indescriptionof the constitutive equation related to the linear-elastic range σ=Dε (3.2) an assumption on the invariance of the normal segment (εz = 0) was kept in force. Therefore, the stress state in the plate is represented by vector σ= [σx,σy,τxy,τyz,τzx] ⊤ 52 T. Kopecki D= E 1−ν2        1 ν 0 0 0 ν 1 0 0 0 0 0 1−ν 2 0 0 0 0 0 1−ν 2k 0 0 0 0 0 1−ν 2k        (3.3) is the material constants matrix in which the effect of non-dilatational stra- in on the plate elastic energy was accounted for by introducing a correction coefficient k=1.2 (Theory..., 2006), while ε=        εx εy γxy γyz γzx        =                   ∂u ∂x + 1 2 [(∂u ∂x )2 + (∂v ∂x )2 + (∂w ∂x )2] ∂v ∂y + 1 2 [(∂u ∂y )2 + (∂v ∂y )2 + (∂w ∂y )2] ∂u ∂y + ∂v ∂x + ∂u ∂x ∂u ∂y + ∂v ∂x ∂v ∂y + ∂w ∂x ∂w ∂y ∂u ∂z + ∂w ∂y + ∂u ∂y ∂u ∂z + ∂v ∂y ∂v ∂z + ∂w ∂y ∂w ∂z ∂u ∂z + ∂w ∂x + ∂u ∂x ∂u ∂z + ∂v ∂x ∂v ∂z + ∂w ∂x ∂w ∂z                   (3.4) is a vector containing deformation state components corresponding to the Green-Saint-Venant deformation tensor (Theory..., 2006), and u, v, w are displacement vector components in local system of coordinates x, y, z. Numerical representations of the system nonlinear deformations are based on the assumption that at any solution stage with the corresponding load, the deformed system retains always a static equilibrium state. Thus, for a defined discrete system it is possible to formulate a system of equilibrium equations that, with respect to nonlinear structural analysis in its displacement-based representation, can be expressed in formof the residual forcematrix equation: r(u,Λ)=0 (3.5) where u is the state vector containing displacement components of the struc- ture nodes corresponding to the current geometrical configuration, Λ is the matrix composed of control parameters corresponding to the current load sta- te, and r is the residual vector containing uncompensated components of forces related to the current systemdeformation state (Crisfield, 1997; Felippa et al., 1976). In numerical algorithms, the components of matrix Λ are expressed as functions of the parameter λ defined as the state control parameter. It is Numerical and experimental analysis of post-critical... 53 a measure of increase of load related indirectly or directly with the pseudo- time parameter t. Thus, the system of equilibrium equations (3.5) can be also presented in the form r(u,λ)=0 (3.6) The above equation is knownas amonoparametric residual force equation. Its solution includes a finite number of consecutive structure deformation sta- tes, while each state corresponds to a combination of varying control parame- ters related to the system load, expressedbyasingle state control parameter λ. Transition from the current state to the subsequent one, representing the in- cremental step, is initialized by change of the control parameter related to new structure geometry determined by the new state vector (Felippa, 1976; Ramm, 1982). Development of numerical methods reflected in contemporary algorithms used in professional commercial programs resulted in constitution of their two fundamental types. The first one includes purely incremental methods known also as prediction ones, while the other type encompasses correctionmethods, called also prediction-correction or incremental-iteration methods.The first of thementioned groups is characterised with limited, often unsatisfactory accu- racy of obtained results. Moreover, they exclude continuation of calculations after crossing critical points on the equilibriumpath. Introduction of the itera- tion phase is therefore aimed at reduction of the solution error and possibility to determine critical points. That makes possible to analyse the structure in advanced deformation states. A feature common for bothmethod types consists in presence of the incre- mental phase.With respect to arbitrary increment, at transition from n-th to (n+1)-th state, the undetermined quantities are ∆un =un+1−un ∆λn =λn+1−λn (3.7) In order to determine them, an additional incremental control equation is formulated, known as the equation of constraints, expressed in form of the condition c(∆un,∆λn)= 0 (3.8) The fundamental componentof the incremental phase is its prediction step, determining a point in the state hyperspace corresponding to the subsequent state configuration defined by determination of the increment ∆u for the as- sumed ∆λwith equation (3.8) satisfied as the same time. Error of the solution at each increment step depends on the increment control equation and the ad- opted extrapolation formula. In each consecutive increment step, the value of 54 T. Kopecki total errormay increase, resulting in occurrence of the so-called drift error. Its minimization is ensured by the iteration phase. The fundamental method used in solution of structural mechanics nonli- near problems is the Newton-Raphson method with numerous program reali- sations and variations constituting awhole family ofmethods (Crisfield, 1997; Felippa et al., 1994; Rakowski and Kacprzyk, 1993). The core idea of these methods consists in expansion of the residual forces equation r= 0, and the increment control equation c=0, into Fourier series. Assuming that, as a result of k-th correction iteration step, values uk and λk are obtained and the equations take the following forms r k+1 = rk+ ∂r ∂u d+ ∂r ∂λ η+β=0 (3.9) ck+1 = ck+ ∂c ∂u d+ ∂c ∂λ η+β=0 where d=uk+1−uk η=λk+1−λk (3.10) Terms β inbothequations includeneglected residualvalues ofhigherorders. In the iteration process, consecutive values d and η are determinedwith respect towhich the solution convergence condition is checked at the assumed toleran- ce. The resulting set constituting a solution to nonlinear algebraic equations with respect to unknown nodal displacements, creates a base for determina- tion of the equilibrium path. The path, representing a relation between the static parameters corresponding to the structure and geometrical parameters related to displacements of its individual points represents a hyper-surface in amultidimensional spacewith the number of dimensions corresponding to the number of degrees of freedom of the system taken into account. In practice, representative relations between the two parameters are usually developed. As a representative relation for determination of the equilibrium path in the problem considered herein, a functional dependence between the maxi- mum deflection value at the crack half-length and the plate tensioning force is assumed. As a base in search for the solution, the incremental Newton-Raphsonme- thod and correction strategy based on the arc length control concept in the Riks-Wempner approach was adopted (Ramm, 1982). Attempts to use only the Newton-Raphsonmethods led to excessive divergence between the results of numerical calculations and the experiment as for the geometry of occur- ring wrinkling. The reliability of obtained results was assessed by comparing both equilibrium path shapes and deformation geometries. The two elements Numerical and experimental analysis of post-critical... 55 created a base for repeated corrections of the numerical model (Kopecki and Dębski, 2007). The plate was modelled using a curved four-node thin-shell element. It is an isoparametric double curved element described by means of bicubic inter- polation functions, based on the Koiter-Sanders shell theory(Theory..., 2006). At each node, the element has the following 12 degrees of freedom u, ∂u ∂θ1 , ∂u ∂θ2 , v, ∂v ∂θ1 , ∂v ∂θ2 , w, ∂w ∂θ1 , ∂w ∂θ2 , ∂2u ∂θ1∂θ2 , ∂2v ∂θ1∂θ2 , ∂2w ∂θ1∂θ2 where (θ1,θ2) are Gaussian coordinates related to the shell central plane, u, v,w are displacement components defined in global Cartesian system of coor- dinates x, y, z. Analysis of large deformations is ensured by use of Lagrange approximation functions. That approximation includes also the variant with finite deflections and torsions, with the assumption on small deformation ten- sor remaining in force. Use of such an element is recommended in nonlinear analysis of thin-walled structures (Theory..., 2006). Anumber of conditions related to the approximate nature of finite element method, distinguishing the resulting numerical solutions from the exact ones based on continuous medium models result in the fact that the obtained re- sults are, by their nature, chargedwith a solution error or,moreprecisely, with a discretization error. The discretized solution tends to the strict one, when the characteristic length of the used finite element tends to zero. That means in fact that numerical solving of a problem should be performedmany times, ”improving” consecutive results. That common practice, in case of nonline- ar analysis, sometimes encounters difficulties of both software and hardware nature. They are especially evident in analysis of thin-walled structures. For instance, a plate with a crack in its area but having no geometric imperfection in the direction normal to its plane, when subject to nonlinear numerical analysis should not, by assumption, lead to any deformation in the normal direction. Complexity of the numerical procedure, aswell as hardware- related conditions,may sometimes lead to results suggesting correctness of the obtained solution. 3.1. Plate without imperfection Realmechanical systems are characterised by the presence of various inac- curacies, which can be connected with not ideal geometry of objects or their material and load faults. This inaccuracies are widely called imperfections. A significance of kinds and values of imperfectionswill be considered in detail, using an example of the considered plate. 56 T. Kopecki The numerical model of the plate containing a crack in its area, without any imperfection, analysedusinganonlinearprocedure, according toprinciples should not have any displacements in the direction normal to the plain of the plate. However, the complexity of nonlinear procedures and numerical factors can lead to results suggesting the correctness of the incorrect solution. In order to provide a quantitative example for illustration of the issue of divergence of results with respect to the obtained ones, a numerical analysis was carried out based on the plate geometry according to Fig.2with the crack of length L=30mm.An ideal elastic-plastic material physical characteristics were adopted (Fig.3): Re= 50MPa, E =3000MPa, ν =0.36, and the load in form of evenly distributed tensile forces applied to the plate shorter edges. Along the longer edges, the degree of freedom was blocked corresponding to translation in the direction perpendicular to the plate plane. Moreover, di- splacements along the load directionwere constrained in two nodes located on the plate longer edges, on its transverse symmetry axis. The nonlinear analy- sis was carried out by means of MSC-Marc-7 software and MSC-Patran pre- and postprocessor. As a result, in the zone adjacent to the crack, a symme- tric form of transversal deformation was obtained shown in form of contours in Fig.9. Fig. 9. Constant deflection value (in cm) contours in the crack zone for the plate without imperfection. Amodel based on regular mesh (12700 elements) Despite geometrical similarity of constant deflection contour patterns, the results, from the quantitative point of view, do not reflect the deformation state identified in the experiment.What attracts attention is negligibly small absolute maximum value of the deflection equaling 1.43 · 10−17cm. Similar results were obtained no matter what parameters controlling the course of analysis, incrementmethod or correction strategywas adopted. In Fig.10, the distribution of effective stress according to H-M-H hypothesis is presented. Numerical and experimental analysis of post-critical... 57 Fig. 10. Contours representing equivalent stress distribution according to H-M-H hypothesis. The plate without imperfection σ max =36.8MPa The maximum effective stress value amounted to 36.8MPa which means that under the load of tensile force Pmax =1500N, nonon-elastic deformation occurred. The result is inconsistent with the outcome of the experiment that showed presence of permanent deformation after load release. It can be thus concluded that attempts to solve the problem of the flat plate without imper- fection weakened by presence of a crack as a geometrically nonlinear problem leads to incorrect results with an understated equivalent stress level. 3.2. Plate with geometric imperfection In the light of results presented above, in the next step of discussion a geometrical imperfection of the plate was taken into account. It consisted in placement of a centrally located point A (Fig.2) outside the plate plane in the normal direction. To determine the order of magnitude of the introduced de- fect in neutral state, a numberof preliminaryanalyseswas carried out inwhich a number of its different values was assumed. Finally, numerical analysis was carried outwith an initial deflection value at point A equaling 1mm.Anum- ber of mesh variants was considered. As a basic one, an irregular mesh was adopted, concentrated in the zone adjacent to the crack. The regular mesh option was also taken into account. In both cases, the number of elements did not exceed 13000. The presence of a crack was modelled by introducing doubled nodes with the same coordinates, assigned symmetrically to two ad- jacent areas on both sides of the crack edge. Identical model fastening and loading scheme was used as in the case of the plate without imperfection. Calculations were carried out bymeans of MSC-Marc-7 program, offering possibility to choose parameters of nonlinear analysis and select among diffe- rent incremental methods and correction strategies. 58 T. Kopecki In the course of preliminary analyses, a number of cases were encountered resulting, as it can be assumed, partly from geometrical conditions and partly from (not very fortunate) the initial selection of control parameters for cal- culation procedures, in which in the course of one of consecutive incremental steps the rapid change of sign of the control parameter into negative occurred. One such a case was observed in the course of analysis based on a regular mesh andusing theNewton-Raphsonmethod (Rakowski andKacprzyk, 1993) with a correction strategy corresponding to theRiks-Rammarc length control method (Ramm, 1982). As a result of the above-mentioned control parameter sign change, a ”turn-up” of shorter loaded edges of the plate beyond its plane occurred. Causes of similar incorrect results can be sought in too large system stiff- ness increase in consecutive steps and the resulting possibility of occurrence of a turningpoint on the equilibriumpath. Itwas found that effectivemethods of elimination of the obtained irregularity are: change of the increment method and/or correction strategy. In the problem considered herein, a fundamental solution quality improvement (at minimum correction of control parameters) was obtained by replacing the Newton-Raphson algorithm with the secant method, and switching at the same time from the arc length correction to the hyperspherical Crisfield correctionmethod (Crisfield, 1997; Ramm, 1982). Consequently, a solution was obtained in which the deflection distribution in the vicinity of the crack (Fig.13) turned out to be consistent in its character with that obtained experimentally (Fig.7a). As a result, in the vicinity of the crack, a deflection distribution presented in Fig.11 was obtained. Fig. 11. Plate deflection contours in the vicinity of the crack with length L=30mm.Model based on regular mesh with geometric imperfection (in cm) Numerical and experimental analysis of post-critical... 59 In the case of irregular mesh being used, with the same numerical proce- dures and identical set of control parameters, a similar result was obtained, shown in Fig.12. Fig. 12. Deflection contours in the vicinity of the crack with length L=30mm. Model with geometric imperfection based on irregularmesh Restricting ourselves to comparative assessment of the presented calcula- tion results with results of the experiment (Fig.7a), we can conclude that for both mesh variants the obtained results show satisfactory conformance as far as deflection distributions are concerned. However, quantitative comparison reveals divergence between the calculation and experimental results. A nu- merical manifestation of that divergence are maximum values of calculated deflections amounting to about 80% of actual values. Calculations performed for the plate with a crack 50mm long demonstrated slightly better conver- gence of the maximum deflection with the experiment exceeding 80% by a little. A fundamental change was obtained for a crack with length of 70mm. Here, conformity was as high as 95%. As formality, it must bementioned that numerical values of deflection obtained as a result of numerical analysis repre- sent deflection increments above the initial value resulting from the assumed imperfection, while the results of the experiment carried out by means of the moiré method represent actual deflection magnitudes. InFig.13, plate deflectionpatterns for caseswhere the crack reached length of 50mm and 70mm, respectively, are presented in form of level contours. In both cases, the same geometric imperfection was assumed consisting in placement of point A (Fig.2) outside the plate plane at a distance of 1mm. InFigs. 14, 15 and 16, equivalent stress distribution patterns are presented according to H-M-H hypothesis corresponding to calculated deformations. By comparing the results, it can be concluded that differences in the maximum 60 T. Kopecki Fig. 13. Deflection contours in the vicinity of the crack inmodels with irregularmesh Fig. 14. Equivalent stress distribution according to H-M-H hypothesis in the vicinity of the crack with length L=30mm in the plate model with irregularmesh and geometric imperfection Fig. 15. Equivalent stress distribution according to H-M-H hypothesis in the vicinity of the crack with length L=50mm in the plate model with irregularmesh and geometric imperfection values of reduced stresses, accounting for the flexural and membrane stress state, in the crack front zone, in both middle and external layer, are larger in the case of cracks with smaller lengths. As the crack length increases, the plasticization zone becomes virtually homogeneous along the plate thickness. In the most advanced deformation phase, deflection gradients also decreased, Numerical and experimental analysis of post-critical... 61 Fig. 16. Equivalent stress distribution according to H-M-H hypothesis in the vicinity of the crack with length L=70mm in the plate model with irregularmesh and geometric imperfection which shows that as crack length increases, the membrane state becomes the dominant one. In order to determine to what extent the conformance of deformations obtained fromnumerical calculations and from the experiment depends on the value of imposed imperfection, a number of numerical tests was carried out. They have shown that with increasing preliminary deflection above 1mm, the divergence of results in thewrinkling increases with respect to both values and deformation character. The effect is confirmedby calculation results presented in Fig.17. Fig. 17. Displacement distribution in the model with irregularmesh and geometric imperfection increased to 1.5mm; L=30mm In that case, calculations were based on themodel with an irregular mesh and geometric imperfection of 1.5mm, representing 48% of the maximum deflection value obtained experimentally. The maximum value of calculated deflection amounting to 2.78mm turned out to be 89% of the value observed in the experiment.Moreover, the distribution of contours (Fig.17) shows that the case appears the most divergent one. 62 T. Kopecki The performed tests and analyses allow one to conclude that the precon- dition of rational determination of the stress field in a weakened flat structure using exclusively nonlinear numerical analysis consists in introduction of such an imperfection thatwould ensure representation of the object geometry in the post-critical state remaining in as good conformity with results of experiment as possible. 3.3. Imperfection imposed by means of load normal to plate plane As an alternative to geometric imperfection, an idea was considered con- sisting in introduction, for the ideally flat structure, of a small transversal perturbation, non-negligible in nonlinear static numerical analysis. The sim- plest method of realisation of a defect of that kind consisted in loading the plate with a discrete normal force, localised at the central point A at the crack half-length. Preliminary analyses revealed that such a method led to significant variance of calculation results with respect to those observed in the experiment. What proved to be effective instead was the application of con- tinuous transverse load with constant intensity along both edges of the crack and less by several orders of magnitude with respect to the plate load force (Fig.18). Fig. 18. Method of introduction of a perturbation initializing transversal deformation of the plate Themethod was used in nonlinear analyses of identical models that were consideredwhen a geometric imperfectionwas used. Load values were selected in away ensuring the best possible conformance of plate deflection fields obta- ined numerically and experimentally. As a criterion, the maximum deflection valueswere adopted and similarity of contours representing constant deflection values. The best convergence of results was obtained using the secant incre- ment method, combined with the Riks-Ramm arc length control or Crisfield hyperspherical control correction strategy. The plot presented in Fig.19 shows the relation between crack length and magnitude of total transverse load FP (Fig.18) ensuring a satisfactory con- formance of the calculation results with the experiment. Numerical and experimental analysis of post-critical... 63 Fig. 19. Transverse load applied to the crack edge vs. crack length Fig. 20. Comparison of deflection distributions in the post-critical state. Models with continuous normal load 64 T. Kopecki Figure 20 shows deflection distributions in the post-critical state in the plate subject to the test for three crack lengths values: 30mm, 50mm, 70mm, for the identical loadwith a tensile force of 1500N.The factworth of attention consists in almost perfect conformity of deflection values obtained as a result of numerical analysis fmax(num), and in the experiment fmax(exp). Results presented in Fig. 20 prove the existence of a satisfactory simila- rity of contours representing constant deflection values obtained by means of numerical calculations and observed in the experiment. Theymake a base for a conclusion that, according to the solution uniqueness rule, stress fields cor- responding to the considered deformation states and determined bymeans of numerical analysis remain in conformitywith the state occurring in the actual structure. Figure 21 presents equivalent stress distributions according to H-M-H hy- pothesis in themiddleandexternal layer, correspondingtodeformations shown in Fig.20. Equivalent stress distributions show thatwith increasing crack length, dif- ferences in the stress distributions in the central and external surface disap- pear. In view of dominance of the membrane state in conditions of advanced deformations, the observed optical effects can be interpreted as ”pure” iso- chromatic, not disturbed by the flexural state. In Fig. 23, representative equilibrium paths are presented for numerical modelswithgeometric imperfectionsandmodelswithan imperfection imposed by means of the initial normal load. They represent relations between the maximum deflection value (at the crack half-length) and the resultant tensile force. The presented equilibrium paths for both forms of the imposed imperfec- tion, although similar as for the nature of the phenomenon, are quantitatively different. The results based on imposition of geometric imperfection led to overestimated plate stiffness. 4. Conclusions Nonlinear numerical analyses and experimental test carried out suggest anum- ber of conclusions that seem to be interesting from the research point of view. Anobservation of fundamental importance consists in the fact that –accor- ding to the experimental research – transition of the structure under examina- tion to the post-critical deformation state occurs smoothly, simply difficult to Numerical and experimental analysis of post-critical... 65 Fig. 21. Equivalent stress distribution according to H-M-H hypothesis in the crack front zone. Models with continuous normal load in the neutral state 66 T. Kopecki Fig. 22. Similarity of isochromatic fringe patterns with the equivalent stress distribution according to τ max hypothesis in the central surface (L=70mm) Fig. 23. Equilibrium paths showing load vs. maximum deflection; (a) models with geometric imperfection, (b) models with imperfection imposed by a load notice, despite the use of the deflectionmeasurementmethodwith sufficiently high sensibility for that kind of a task. Relations: the cause (tensile force) vs. the effect (maximum deflection va- lue), constituting the parameters of representative equilibrium path (Fig.26) reveal the nonlinear nature of phenomena occurring for both types of imposed imperfections. They are similar as for the nature of nonlinearity, but differ with respect to their values. Results of calculations based on the geometric imperfection imposition concept lead to overstated stiffness of the structure, as a result of which they do not assure satisfactory conformity with the experiment. Numerical and experimental analysis of post-critical... 67 An important element among the discerned relations is the conformity of the equilibriumpath obtained as a result of numerical calculation and, presen- ted in Fig.25, from the experiment. The experimental path was developed on the grounds of deflection level contours recorded in the course of examination of the crack with length of 30mm at three load levels: 500N, 1000N and 1500N (Fig.24). Fig. 24. Deflection level contours for a constant crack length L=30mm Fig. 25. Equilibrium paths for a crack with length L=30mm It can be concluded on the grounds of the above analysis that introduction of a perturbation in form of a continuous normal load in the weakening zone is a definitely more rational method of numerical representation of the effect of wrinkling of plates with cracks, assuring both qualitative and quantitative conformity with actual deformation of the structure. However, the method requires a proper selection of the load value, resulting in advisability, and sometimes necessity to carry out an experiment of local nature. 68 T. Kopecki The presented methodology seems to be applicable with satisfactory re- sults to flat structures with discontinuities of other kinds. Such a suggestion follows frompreliminary studies on crack propagation initiated in a platewith a circular cutout (Fig.26). Fig. 26. Pictures of the wrinkling effect in a plate with a circular cutout subject to fatigue tests In view of increasing possibilities in computer coding of nonlinear cha- racteristics of material models, including optically active ones, realisation of an adequate experiment can be significantly supportedby theoretical research, serving as the base for determination of efficiency of both developed numerical model and adopted procedure. Transfer of results onto real objects made of other materials with physical characteristics that could be approximated by means of the perfect elastic- plastic body can be reduced to determination of material constants entered into the calculation code or it becomes possible on the grounds of model simi- larity. References 1. Aben H., 1979, Integrated Photoelasticity, McGraw-Hill Book Co., London 2. Brighenti R., 2005a, Buckling of cracked thin-plates under tension and com- pression,Thin-Walled Structures, 43, 209-224 3. Brighenti R., 2005b,Numerical buckling analysis of compressed or tensioned cracked thin plates,Eng. Struct., 27, 265-276 Numerical and experimental analysis of post-critical... 69 4. Crisfield M.A., 1997, Non-Linear Finite Element Analysis of Solids and Structures, J.Wiley & Sons, NewYork 5. DyshelM.S., 2002, Stability and fracture of plates with a central and an edge crack under tension, Int. Appl. Mech., 38, 472-476 6. Felippa C.A., 1976, Procedures for Computer Analysis of Large Nonlinear Structural System in Large Engineering Systems, A.Wexler (Edit.), Pergamon Press, London 7. Felippa C.A., Crivelli L.A., Haugen B., 1994, A survey of the core- congruential formulation for nonlinear finite element, Arch. of Comput. Meth. in Enging., 1 8. Kopecki T., Dębski H., 2007, Buckling and post-buckling study of open sec- tion cylindrical shells subjected to constrained torsion,Arch. ofMech. Enging., LIV, 4 9. Kopecki T., Zacharzewski J., 2006, Fatigue life and stress state analysis of cracked thin-walled plate under cycles axial tension,Maintenance and Reliabi- lity, 31, 3, 19-26 10. Laermann K.H., 1982, The principle of integrated photoelasticity applied to experimental analysis of plates with nonlinear deformations,Proc. 7-th Intern. Conf. of Experim. Stress Analysis, Haifa 11. Markström K., Storäkers B., 1980, Buckling of cracked members under tension, Int. J. Solid and Struct., 16, 217-229 12. Patorski K., Kujawińska M., 1993,Handbook of the Moiré Fringe Techni- que, Elsevier, Amsterdam-London-NewYork-Tokyo 13. RakowskiG.,KacprzykZ., 1993,Metoda elementów skończonych wmecha- nice konstrukcji, OficynaWydawnicza PolitechnikiWarszawskiej,Warszawa 14. Ramm E., 1982,The Riks/Wempner Approach – An Extension of the Displa- cement Control Method in Nonlinear Analysis, Pineridge Press, Swensea 15. Riks E., Rankin C.C., Brogan F.A., 1992, The buckling behavior of a central crack in a plate under tension,Eng. Fract. Mech., 43, 529-547 16. Shaw D., Huang Y.H., 1990, Buckling behavior of a central cracked thin plate under tension,Eng. Fract. Mech., 35, 1019-1027 17. Sih G.C., Lee Y.D., 1986, Tensile and compressive buckling of plates weake- ned by cracks,Theor. Appl. Fract. Mech., 6, 129-138 18. Theory and user information, MSCMarc 2006 70 T. Kopecki Numeryczno-eksperymentalna analiza stanów zakrytycznej deformacji płyty rozciąganej osłabioną szczeliną Streszczenie Wopracowaniu zaprezentowanometodykę określania rozkładunaprężeńw zakry- tycznych stanach deformacji prostokątnej płyty osłabionej szczeliną, poddanej rozcią- ganiu. Problem został sformułowany jako geometrycznie i fizycznie nieliniowy. Prze- prowadzono nieliniową analizę numeryczną w ujęciumetody elementów skończonych. Stanydeformacjimodelunumerycznegootrzymywanewposzczególnychkrokachprzy- rostowychporównywano z deformacjami określonymi podczas eksperymentu.Uzyski- wanewyniki stanowiły podstawę ocenywpływu rodzaju i wielkości imperfekcji zasto- sowanychwmodelu numerycznympłyty nawiarygodność rezultatówanalizy. Rozwa- żano dwa rodzaje imperfekcji: pierwszą – zadanąw formie wstępnej deformacji płyty w strefie szczeliny oraz drugą – polegającą na zastosowaniu zaburzenia w kierunku normalnym do powierzchni środkowej płyty. Manuscript received February 19, 2009; accepted for print April 16, 2009