Otimização paramétrica de um flange de motor elétrico

Abstract

This work has the objective of implementing a structural optimization procedure in order to adjust the natural frequency of electric motors. This is accomplished by modifying some flange dimensions, which is the part connecting the motor to the system being driven. The implemented optimization algorithm is the PSO (Particle Swarm Optimization), which proved to be appropriate for the type of problem considered (non convex and with few design variables). The optimization formulation aims to get a motor/flange whose first natural frequency is equal to a specified value while observing the criterion of Rankine for brittle materials. Both the modal analysis and evaluation of the principal stresses are performed by a commercial finite element code. Due to the large computational cost associated to the combined use of PSO with an external finite element program, it is studied an approximation by least squares of the natural frequency and of the principal stresses. This procedure is effective and extremely efficient. Examples of flange designs obtained using the proposed formulation are presented and discussed, showing the validity of the procedure adopted.