Inverse and crack identification problems in engineering mechanics (Q5926478)

The book is devoted to the study of inverse problems arising in unilateral contact problems, summarizing some research works of the author. The subject is presented without a detailed description of mechanical problems, in a concise mathematical form, containing relevant citations and significant survey on the title topics. The book contains: part I. Introduction. Problem description; part II. Theoretical computational tools, and part III. Application to inverse problems. In chapter 2, variational inequalities, the potential energy optimization problem and the complementary energy optimization problem are outlined within linear and nonlinear elastic material laws, in elastostatics, under the hypothesis of small displacements. The link with convex optimization is pointed out, and the discretizations of the related mechanical problems are obtained by means of finite elements. Unilateral contact problems are introduced in section 2.1.2. The pointwise interface law is described in a general subdifferential form, via the convex potential energy function, as a monotone interface law (decomposed normally and tangentially to the interface). The author considers the particular cases of frictionless unilateral contact and of frictional contact which introduce nondifferentiable terms in the potential energy. The corresponding elastostatic problems lead to variational inequalities for frictionless and frictional unilateral contact laws with a given normal contact traction, and to quasivariational inequalities when the general frictional effects are involved. In section 2.1.4 the author describes the boundary element reformulation of unilateral problems into a linear complementarity problem in standard form, by using slack variables. Algorithms for inequalities containing quadratic programming, algorithms for complementarity problems, and algorithms for two-dimensional unilateral contact and frictional problems are discussed in section 2.2, since these problems arise in applications studied in the book. A short derivation of reduced elastodynamic equations, when the elastic quantities are time-harmonic, and the corresponding boundary element solution methods are presented in section 2.3. The direct dynamic boundary element formulation is given for transient elastodynamics based on an appropriate reciprocal theorem that extends elastostatic Betti's theorem. About 80 titles are cited in this chapter. Chapter 3, devoted to computational and structural optimization, contains a short summary of results, with 70 references, related to optimality conditions for smooth and for nondifferentiable convex problem restricted to inequalities. The mathematical optimization theory used for the modelling and effective solutions in computational mechanics is presented in connection with finite element approximations in unilateral contact problems. Chapter 4, selected computational tools, presents neural network and genetic algorithms, and fuzzy and filtering techniques used for the solution of identification and inverse problems. These methods are appreciated by the author to have the potential to overcome certain problems due to ill-conditioning or scaling, nonconvexity and nondifferentiability, as they arise in the applications considered in the book. This chapter contains applied computing methods together with appropriate citations, about 50, for further details. Part III, devoted to the applications to inverse problems, contains static problems (chapter 5), steady-state dynamics (chapter 6), and transient dynamics (chapter 7). Each chapter starts with introduction and literature survey and ends with references, accumulating in total 130 titles. Here the author develops and tests a solution technique for nondestructive crack identification problems. The identification (inverse) problems are formulated as appropriate optimization problems for differences between the computed and desired response, within the space of variables which define different possible configurations of the structure. In the course of solving inverse problems, some direct problems have to be solved several times. For a two-dimensional specimen which contains an unknown crack idealized by a number of parameters, the direct problem is numerically solved by the methods outlined in chapter 2. For the inverse problems, different numerical methods described mainly in chapter 4, are tested and compared with classical optimization algorithms in order to determine the parameters which define the unknown crack. As the author notices, there is still no enough experience in inverse problems with nonlinear behaviour, and in this respect the reported results should be considered as a first attempt in this direction.