Reliability-based optimization of trusses with random parameters under dynamic loads

Abstract

In this work, a reliability-based optimization technique is addressed to obtain the minimum mean value of random mass of the structures with random parameters under stationary stochastic process excitation. The challenge of the problem lies in randomness involved from both structural parameters and dynamic load, which renders the structural reliability becoming the random dynamic reliability of the first passage problem. In order to obtain minimum mean value of random gross mass, element and system dynamic reliability constraints are constructed, respectively, and the structural sizing and shape optimization models based on the dynamic reliability are then presented. Moreover, among two optimal strategies proposed for optimization models, the second one can effectively reduce the workload by avoiding the computation of the variance of the dynamic response during the iterative process. Finally, the implementation of three examples is discussed to display the feasibility and validity of optimization technique given.

Design optimization under uncertainties of a mesoscale implant in biological tissues using a probabilistic learning algorithm

“…When these works were published, design optimization with stochastic computational model was not still really possible for large scale multiscale computational models. The applications that have been published (see for instance [1,5,11,12,16,22,29] and also [35,54,57,68,70,72,80,99,96]) were devoted to optimization problems under uncertainties for which the computational models had a reasonable number of degrees of freedom, for which the optimizers were based on the use of relatively classical optimization algorithms and/or the introduction of approximations such as surface responses and surrogate models.…”

On the improvement of design space subsets formulation for iteratively optimizing structural systems

“…The stochastic design problems involve the multi-dimensional integrals that may represent the adopted objective function or the constraint. In this view, RBDO corresponds to the case where the constraints are formulated as multi-dimensional integrals to characterize the admissible design space [8,[16][17][18][19][20][21][22][23][24][25][26][27]. Unlikely, in the case of RDO, minimization of multi-objective stochastic objective functions problem is involved [14,[28][29][30].…”

Stochastic homogenized effective properties of three-dimensional composite material with full randomness and correlation in the microstructure

“…Herein, the Random Factor Method (RFM) proposed in [25,26] is extended to the computation of the random homogenized effective properties of a unidirectional fiber reinforced polymer (FRP) composite with orthotropic behavior at the macroscale which is a classic model in composite material. ) of FRP composites are formulated by combining the analytical homogenization approach with the RFM, whereby the randomness of the material properties and morphology parameters of the two constituents as well as the correlation among these random variables are simultaneously taken into account.…”