Design for Manufacturing of a Digital Hollow Section Beam Using Interlocking Connection

EasyChair Preprint 6542

Abstract

This paper presents a computational design strategy and manufacturing technique for a curved steel beam using interlocking connections. Although there have been developments in industrial robotic setups to automate steel beam bending, these processes rely on an iterative approach and are time-consuming and costly when there is a need for high accuracy. A shift from bending a whole section to 2D steel laser cutting and bending during assembly simplifies the fabrication processes, as well as reduces the weight of the structure. This work presents a novel methodology to create a three-dimensionally curved beam that is elastically pulled into position from discrete laser-cut sheets. The surfaces of the double curved beam are generated by defining the stiffener’s geometry and an axial curve through geometric modeling. The interlocking connection details are then applied to the edges of unrolled plates based on graph connectivity. Finite Element Method (FEM) form-finding simulates the deformation process of plates using contracting cable elements. Structural analysis and optimization are also performed to evaluate the residual stress under external loads to reduce the thickness of individual plates. The physical prototyping demonstrates geometric accuracy and laser-cut tolerance. The uniaxial tensile tests highlight the impact of welding on structural performance at the interlocking connection.

Keyphrases: - steel laser cutting, FEM form-finding, Optimization, active bending, elastic deformation, interlocking connections, manufacturing tolerances, steel construction, structural analysis, welding

@booklet{EasyChair:6542,
  author    = {Seyed Mobin Moussavi and Felix Kallwies and Arnold Walz and Philipp Rumpf and Stefan Böhm and Niklas Sommer and Stephan Völkers and Julian Lienhard},
  title     = {Design for Manufacturing of a Digital Hollow Section Beam Using Interlocking Connection},
  howpublished = {EasyChair Preprint 6542},
  year      = {EasyChair, 2021}}