Prediction of probabilistic settlements via spectral stochastic meshless local Petrov-Galerkin method

摘要

This study introduces the prediction of probabilistic settlements with the uncertainty in the spatial variability of Young's modulus to illustrate the preliminary development of a spectral stochastic meshless local Petrov-Galerkin (SSMLPG) method. Generalized polynomial chaos expansions of Young's moduli and a two-dimensional meshfree weak-strong formulation in elasticity are combined to derive the SSMLPG formulation. Because of the local and truly meshless nature, the SSMLPG method is more computationally efficient than available stochastic numerical methods. Two examples further show that SSMLPG-based predictions remain sufficiently accurate even in case of scattered nodes. Therefore, the SSMLPG method can be a valuable alternative for solving stochastic boundary-value problems.