Torque Converter Stress Analysis by Transient FSI Technique

Abstract

The performance of a torque converter has been one of the most important areas of improvement for an automatic-transmission equipped automobile. Improving the torque converter’s performance and efficiency is key to saving fuel consumption, which is an important consideration with recent environmental awareness. Moreover, improving the overall automobile performance has led to more compact and lightweight transmissions. With the growing space constraints, the evolution of the torque converter has been towards smaller and more elliptical shapes. Since the smaller blades within the torque converter still have to endure the same engine torque, more strength is required of each blade of the pump, the turbine and the stator. There has been much research carried out to predict hydrodynamic performance and to understand the flow field inside a torque converter either experimentally or analytically using Computational Fluid Dynamics (CFD). However, none of the research has focused on the strength of the torque converter components — the blade, the shell and the core. The previous method for evaluating the blade strength had been to apply a simple, centrifugal pressure load on the blade using Finite Element Analysis (FEA). This method is no longer adequate for predicting blade stress since the pressure distribution on the blade is now known from CFD results. In this work, the fluid-structure interaction (FSI) technique is used to determine the deformation, which is indicative of the stress level of the blade, the shell and the core. In addition, this research compares the computational results from a model containing all blades to a conventional model of a single blade with axial symmetry. Analysis of the model containing all blades shows a completely different deformation mode than the single-blade model, especially for the pump blade. The differing results suggest that using a single-blade model analysis is less accurate for examining the torque converter structure.