Integrated topology-response surface-metaheuristic optimization framework for failure prevention and lightweight design of a wheel loader middle lever

Abstract

The present paper proposes a combined approach to structural optimization that is applied to the middle lever of a Dresser 515C wheel loader which experiences 85 kN of load from a hydraulic cylinder and is known for recurrent cracking at its vulnerable section. This approach uses topology optimization, response surface methodology (RSM) and metaheuristic algorithms within two approaches: a topology-based one where feasible light weight designs were created using the optimal material distribution obtained through optimization and a preventive expert-based one in which strategies aimed at eliminating stress concentrations from the intact arms are created. In each of the two cases, the RSM models of the mass, Von Mises Stress and maximum displacement were built in a Box–Behnken design in Minitab 17 and then optimized by four algorithms: ALO, DIO, HHO, and Puma. From topology-based strategies, Strategy 3 resulted in the minimum mass of 107.60 kg, a 5.9% reduction from the 114.323 kg reference geometry, using Puma in 0.62 s, while the best structural performance was provided by Expert Strategy 1, in which the stress and displacement were reduced from the reference values of 194.9 MPa and 1.766 mm to 185 MPa and 1.62 mm respectively using HHO in 0.29 s. The use of Box–Behnken designs reduced the required experimental runs to 27 (four-factor topology strategies) and 15 (three-factor expert strategies), a 67% and 44% reduction respectively compared to a full three-level factorial design. The FE model has shown that the critical zone of stress concentration in the reference arm has been effectively removed by the proposed methodology. The new method takes care of both weight reduction as well as avoidance of fracture in one go.