Cluster Computing-Aided Open-Source Programming Framework for Model Updating of Civil Structures

Abstract

The finite element model updating (FEMU) and structural optimization of high-fidelity numerical models for large civil structures require significant computational resources and efficient optimization algorithms. However, prior research has predominantly relied on commercial software, which has more restrictions compared to open-source ones. A cluster computing-aided programming framework for the FEMU of large civil structures was developed based on the open-source platforms OpenSees and Python. The high-performance computing (HPC) cluster was built to connect the cloud/local computing resources. Then, the cluster computing-aided particle swarm optimization (PSO) algorithm, suitable for scientific computing on HPC cluster, was developed. The software interfaces were programmed to connect OpenSees with HPC cluster to achieve high-performance FEMU and structural optimization. The advantages of the framework include (1) an open-source cluster computing platform suitable for FEMU and structural design optimization is developed utilizing dispy; (2) the framework is convenient to use, highly efficient in computation, and is capable of fully utilizing both local and cloud computational resources to improve computational efficiency; and (3) it has strong compatibility and is flexible to be customized for various engineering problems by embedding objective functions. Four examples were used to illustrate the applications of this framework in different fields.