Development of structural design and assessment tools for innovative components and joints

This study presents the development of advanced assessment and optimization tools for threaded joint systems used across diverse engineering applications. The tools were established through an integrated methodological approach combining classical mechanics, international design standards, genetic algorithms (GA), finite element analysis (FEA), and machine learning (ML), with the aim of automating the bolt selection process and improving the efficiency, reliability, and accuracy of joint design and assessment. Initially, a computational tool was developed in Excel for simplified bolted geometries based on internationally recognized standards, including VDI 2230, EN 13001, and Eurocode 3, providing accurate estimates of bolt size and preload for specified loading conditions, geometries, and constraints. Building upon this tool, the methodology was translated into Python to enable automation and GA-based optimization, resulting in optimal bolt layout, reduced bolt size and number, and lower overall cost and material usage without compromising structural performance. In the next phase, FEA was employed to capture the nonlinear behaviour of joints, addressing limitations of the earlier simplified models. Moreover, to reduce the high computational cost of FEA, a hybrid FEA–ML framework was developed, producing surrogate models that rapidly and accurately predict joint performance while substantially reducing simulation time. Finally, an analytical elastic interaction model was formulated for a large-scale wind turbine blade-to-hub connection, integrating hybrid constitutive modelling with GA-based optimization and validated experimentally via ultrasonic preload measurements. Collectively, these assessment and optimization tools provide a comprehensive, computationally efficient methods for the realistic design, analysis, and optimization of threaded joint systems in advanced industrial applications.