Single and multi-objective optimization of a gearbox considering dynamic performance and assemblability

Abstract

Today, most companies do not consider the assembly aspects of a product until late in the design process, and often do so in a manual way, mainly relying on the experience of the assembly engineers. This results in designs that comply with the performance requirements, but that have a suboptimal or unfeasible assembly process. This leads to design changes late in the product development process, causing significant extra costs. To solve this issue, a framework is presented, which is able to perform design optimizations considering both the assembly process and the performance of the future product, allowing trade-off analysis of the obtained solutions. The assembly process evaluation is automated with an algorithm that evaluates Design For Assembly (DFA) rules, enabling the assessment and quantitative evaluation of the ease-of-assembly of the whole product and/or individual assembly operations.The presented framework is used in the context of the design of a gearbox, in which total mass, assemblability and performance need to be optimized. The performance of the gearbox is determined by the axle‘s stiffness and first resonance frequency. The assemblability of the gearbox is evaluated based on 2 DFA rules. The first rule considers the number of unique types of standard parts used in the assembly and how they influence the complexity of the assembly process. The second rule evaluates the convenience of fastening standard parts (e.g., bolts or screws) using the corresponding tool during the assembly operation.First, the design problem is formulated as a single objective optimization in which mass and assemblability are optimized given constraints on the axle‘s stiffness and first resonance frequency. In each iteration of the optimization, a parametrized finite element model is used for the performance evaluation of the current design, while the assemblability cost estimation is based on the CAD model of the assembly. Second, a multi-objective co-optimization of the gearbox design is carried out were both performance and assemblability are optimized. By constructing the Pareto front an accurate and effective trade-off can be made between objectives of the gearbox design, which allows to reasonably choose the best design of the gearbox.