Development of a rapid strength assessment method for dented pressure vessels using ai: a case study of dataset formation for rectangular plates under pressure

Pressure vessels are key components in the chemical and oil and gas industries. Ensuring their reliability is a current challenge. With an increasing number of damages due to prolonged operation and, notably, ongoing military activities in Ukraine, defects such as dents on cylindrical shells have become particularly prevalent. Dents can pose a significant hazard, necessitating the development of rapid and accurate tools for strength assessment. This paper proposes a methodology for creating a training dataset for artificial intelligence (AI) models aimed at evaluating the strength and stress-strain state of vessels with local dent-type defects. To test the approach and validate the methodology, a simplified model of a thin-walled rectangular plate, rigidly clamped at its edges and subjected to uniform pressure, is considered. The dataset creation procedure using the Python programming language is described in detail. This includes: generating a wide range of input parameters (geometry, material, load); automated finite element method (FEM) modeling of each configuration in Ansys to obtain maximum deflections and equivalent von Mises stresses; and analytical calculation of deflection for FEM verification. All generated and calculated data (input, FEM, analytical) are consolidated into a single file. A thorough analysis of the generated dataset was conducted, including: visualization of parameter distributions using histograms; calculation and analysis of the Pearson correlation matrix to identify interdependencies; and validation of FEM results by comparison with analytical calculations. The prospects of using the created dataset and methodology for training AI models capable of performing rapid structural diagnostics and solving inverse problems in mechanics, particularly for assessing the strength of cylindrical shells with real defects, are substantiated. This will contribute to the development of effective tools to support engineering decision-making.