Applications of Artificial Intelligence Techniques for Optimization of Structural Steel Connections

EasyChair Preprint 6837

Abstract

Steel connections in building projects are traditionally designed in Canada by the fabricator’s engineer using connection design software or spreadsheets, often supplemented by hand calculations. Implicit in these processes is the desire to optimize the net fabrication and erection costs based on several parameters such as material, task and process repetition, machinery availability, fabrication and erection constraints, and available optimization tools. This study aims to explore the realm of structural optimization by employing evolutionary algorithms to automate and optimize the design of steel shear connections. Evolutionary algorithms are search algorithms used within artificial intelligence to determine an optimal solution from a population of possible solutions within user-defined optimization criteria. A design optimization tool built using parametric modelling and connection design software utilizing component-based finite element modelling is applied to a shear end plate connection. The results are tested and compared against an example end plate connection design from the Canadian Handbook of Steel Construction. The optimization results showcase the potential of using evolutionary algorithms for steel connection design, and the practicality of the optimization tool for use in a professional environment is presented.

Keyphrases: Artificial Intelligence, Evolutionary Algorithms, Genetic Algorithms, Steel Shear Connections, Structural Engineering, computational design, generative design, parametric design, steel connections

@booklet{EasyChair:6837,
  author    = {Eric Duong and A.J. Darras and Robert Driver and Maha Essa and Ali Imanpour},
  title     = {Applications of Artificial Intelligence Techniques for Optimization of Structural Steel Connections},
  howpublished = {EasyChair Preprint 6837},
  year      = {EasyChair, 2021}}