Design of experiments to investigate blade geometric effects on the hydrodynamic performance of torque converters

Abstract

Torque converters are key components in automatic and hydrodynamic transmissions. Power is transmitted through the reaction force of fluid on cascades; thus, the geometry of the blade is essential to torque converter performance. The traditional one-dimensional blade design approach becomes inefficient for modern torque converter design because torque converters are highly coupled turbomachines and the flow is three-dimensional. In the present research, a novel six-parameter blade camberline design was developed to describe the overall shape of the blade. A full two-level factorial design was conducted with computational fluid dynamics (CFD) simulations on each component to determine the sensitivity of design variables and investigate the relationship between design parameters and hydrodynamic performance. The design variables were reduced from 18 to nine after the screening design. A quarter-fractional factorial design was performed on the selected primary design variables to explore the first-order interaction effects between different wheels. Then a response surface was generated for each component to provide a substitution model for further optimization. A series of torque converters with various design parameters were fabricated and tested to validate the important effects determined in the design of experiments (DOE) process. It is found that CFD in combination with DOE is able to precisely capture the correlation between design variables and hydrodynamic performance. A base torque converter was optimized based on the DOE studies and the result was tested. Pronounced improvement in powertrain performance and fuel economy were observed.