Structural diagnoses by synergy of experimental, computational and material mechanics.

Mechanics of continua and its computational methods (e.g. by finite element discretization) make at present possible assessments of the safety margins of structures under extreme loads. Reliability of such results, concerning structures possibly deteriorated in service, clearly requires reliable input data concerning present damages, primarily concerning material properties and pre-existing stress states. Experiments, in laboratory or in situ, can provide the above data, but should be designed in order to reduce consequent damages, costs and times. Transition from results provided by optimized tests to parameters to employ for overall safety analyses requires solution of an inverse analysis problem based on test simulations and centred on the minimization of a “discrepancy function” which quantifies the difference between measurements and their counterparts provided by test simulations, with searched parameters as unknown variables. The above minimization often concerns non-convex objective function and non-linear constraints and, therefore, should be carried out by a suitably selected algorithm (“trust region” mathematical programming, genetic algorithm or artificial neural network). Kalman filters can be employed when probability density distributions of the resulting parameters estimates may be computed as consequences of uncertainties in measurements and in other input data. The present communication purpose is to provide a brief survey of the above methodology, by reference to some recent research results achieved by our team. The following subjects have been selected. (a) A typical procedure is superficially outlined step-by-step by an example concerning brittle fracture: indentation test in situ, finite element simulation, model reduction by “proper orthogonal decomposition”, parameters identification. (b) Back-analysis estimation of both residual stresses and inelastic material properties by standard “small punch” tests on metallic structural components of thermo- or hydro-electrical power plants. (c) Double purposes similar to those in (b), but achievable by a novel “hole drilling” method either in concrete dams or in geological layers crossed for hydrocarbons extraction. The latter procedure might be considered in the current large research project motivated by the Gulf of Mexico disaster and promoted and coordinated jointly by the National Academies, of Engineering and of Science, in the USA.