Surrogate-Based Dimensional Optimization of a Polymeric Roller for Ore Belt Conveyors Considering Viscoelastic Effects

The roller is one of the fundamental elements of ore belt conveyor systems since it supports, guides, and directs material on the belt. This component comprises a body (the external tube) that rotates around a fixed shaft supported by easels. The external tube and shaft of rollers used in ore conveyor belts are mostly made of steel, resulting in high mass, hindering maintenance and replacement. Aiming to achieve mass reduction, we conducted a structural optimization of a roller with a polymeric external tube (hereafter referred to as a polymeric roller), seeking the optimal values for two design parameters: the inner diameter of the external tube and the shaft diameter. The optimization was constrained by admissible values for maximum stress, maximum deflection and misalignment angle between the shaft and bearings. A finite element model was built in Ansys Workbench to obtain the structural response of the system. The roller considered is composed of an external tube made of high-density polyethylene (HDPE), bearing seats of polyamide 6 (PA6), and a steel shaft. To characterize the polymeric materials (HDPE and PA6), stress relaxation tests were conducted, and the data on shear modulus variation over time were inserted into the model to calculate Prony series terms to account for viscoelastic effects. The roller optimization was performed using surrogate modeling based on radial basis functions, with the Globalized Bounded Nelder-Mead (GBNM) algorithm as the optimizer. Two optimization cases were conducted. In the first case, concerning the roller’s initial material settings, the designs found violated the constraints and could not reduce mass. In the second case, by using PA6 in both bearing seats and the tube, a design configuration was found that respected all constraints and reduced the roller mass by 15.5%, equivalent to 5.15 kg. This study is among the first to integrate experimentally obtained viscoelastic data into the surrogate-based optimization of polymeric rollers, combining methodological innovation with industrial relevance.