Stochastic analysis of concrete dams with alkali aggregate reaction

Abstract Following a comprehensive literature survey, this paper will first address the theoretical underpinnings of stochastic modeling. An algorithm to model arch dam inhomogeneity in terms of characteristic length is presented next and is applied to assess the impact of concrete's elastic modulus and volumetric AAR expansion. Results are contrasted with both a single deterministic analysis, and a series of classical Monte Carlo simulations in which non-Gaussian stochastically varying properties are used. The impact of randomness in material properties on displacements, joint openings, and stresses are investigated. It is found that whereas mean values of these responses are for the most part little impacted, standard deviations exhibit a greater variation vis a vis simple Monte Carlo simulations. Furthermore, the safety is assessed through fragility surfaces, and meta-modeling. This study determined that whereas randomness may affect local results e.g. stresses, their impact may be neglected for globally averaged responses e.g. displacements.