Application of DOE Methods to RPM-Domain Data for Hydraulic Steering Pump NVH Improvement 2003-01-1431

The present work demonstrates the application of Design of Experiments (DOE) statistical methods to the design and optimization of a hydraulic steering pump for NVH performance. DOE methods were applied to RPM-domain data to examine the effect of several different factors, as well as the interactions between these factors, on pump NVH. Whereas most DOE analyses typically consider only a single response variable, the present work considered multiple response variables. Specifically, pump NVH performance curves for several pump rotational orders over a range of shaft speeds were analyzed. Thus, it was possible to determine the effect of the factors in question over the entire speed range of pump operation, rather than a single speed or setting.

Statistical methods were applied to determine which factors and interactions had a significant effect on pump NVH. These factors were used to construct an empirical mathematical prediction model for NVH performance. Lastly, this model was experimentally verified and used to select the pump characteristics and design features yielding improved NVH performance. The results of the DOE were objectively confirmed in vehicle-level interior noise testing and subjectively using jury evaluation methods.