A fast forecast method based on high and low fidelity surrogate models for strength and stability of stiffened cylindrical shell with variable ribs

Abstract

In this work, a surrogate model based on high and low fidelity models for strength and stability analyses of a stiffened cylindrical shell with variable ribs is built. The design variables are the radius and thickness of the shell, the parameters of ribs and the space of ring-ribs. The outputs are the mid-plane circumferential stress of the shell, the longitudinal stress of the inner-face of the shell at ribs, rib stress, the local buckling eigenvalue and the global buckling eigenvalue. The Co-Kriging is used to establish the high and low fidelity surrogate model for predicting the response values within the design space. The high-fidelity model is established using ANSYS with fine meshes and is computationally expensive. In order to explore the influence to prediction accuracy by low fidelity models, two levels low fidelity models are established. One is constructed by ANSYS whose grids is coarse and the other is constructed by the classic formula. The Co-Kriging model is explored to find a relative better ratio of high fidelity samples and low fidelity samples for accurate forecasting the strength and stability indexes of the stiffened cylindrical shell with variable ribs. The results show that the accuracy of Co-Kriging surrogate model is better than traditional Kriging model with the same computational expense, and the samples and the accuracy of low fidelity model also have great influence to the Co-Kriging model.