Indexed by:Journal Papers

Date of Publication:2019-03-01

Journal:COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING

Included Journals:EI、SCIE

Volume:345

Page Number:476-501

ISSN No.:0045-7825

Key Words:Reliability-based design optimization; Isogeometric analysis; Enhanced step length adjustment; Second-order reliability method; Sequential optimization and reliability assessment approach

Abstract:Reliability-based design optimization (RBDO) is a powerful tool to handle the influence of various uncertainties during optimization. However, unbearable computation cost is one of the largest barriers for its application, especially for the finite element method (FEM)-based RBDO. In this paper, an efficient and accurate RBDO framework is established based on isogeometric analysis (IGA) for complex engineering problems. Furthermore, an enhanced step length adjustment (ESLA) iterative algorithm and a second-order reliability method-based stepped-up sequential optimization and reliability assessment approach (SSORA-SORM) are proposed to boost the efficiency of RBDO. According to the situation of iteration process, the step length of search the most probable target point can be adaptively updated by the proposed criterion to improve the robustness in ESLA. In the proposed framework, the analytical first-order sensitivity is derived based on IGA in the optimization process to substitute the time-consuming finite difference method. The robustness, accuracy and efficiency of proposed methods are verified via several numerical benchmarks. Besides, three complex IGA-based examples demonstrate that the proposed method is able to save much computational cost without losing accuracy, which is inherently suitable for the RBDO of complex engineering problems. (C) 2018 Elsevier B.V. All rights reserved.