Deep neural networks for maximum stress prediction in piping design

Abstract

Piping design mainly consists of design, modeling, and analysis steps. Once all processes of the design and modeling steps are completed, the maximum stress values obtained in the analysis step are compared with those prescribed by the regulations to complete the piping design. If these values do not satisfy those provided by the regulations, the entire design must be modified. In the analysis step, bottlenecks occur because both design and modeling must be re-performed. This requires considerable time and effort from the designer, and it is a major factor lowering designer productivity. To achieve efficiency, the required maximum stress value should be considered in the initial step itself. In this study, a deep neural network was used to predict the maximum stress. Based on the accuracy of the predicted analysis results, it was possible to shorten the design time while improving the piping design.