Date of Award

2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

College of Technology

Abstract

The demands on the automotive industry for increased reliability and reduced noise have a direct effect on automotive axle requirements. This demand translates to increased precision in automotive axle manufacturing, including axle assembly where accurate positioning of hypoid gears and setting of proper bearing axial load is the most challenging process. To achieve the accuracy required, the axle assembly often includes select fit shims to control gear position and bearing preload force. The shim selection process integrates a measurement system into the assembly process that includes; in-process measurements of components and subassemblies as inputs, and audit measurements of each assembly to confirm gear position by backlash and bearing preload by torque to rotate as outputs. Understanding the correlation of in-process measurements to audit measurements is an essential part of optimizing the shim selection process. The purpose of this research was to define and assess a method to correlate input measurements as independent variables to audit measurements as dependent variables in an axle assembly system. This correlational study developed and assessed an axle shim selection process model as a predictor of variance in the dependent variables of backlash and rotational torque. To account for errors affecting shim selection the measurement uncertainty framework was used. The study included three steps. The first step developed an uncertainty model of an existing axle assembly measurement system using the standard uncertainty propagation method. The second step evaluated the ability of the model to simulate the production process and predict process capability using a Monte Carlo Method (MCM) simulation. The third step applied the model to assess effects of a measurement error for the axle cover in-process measurement. The results of this study suggest that an uncertainty model can correlate input and output measurements in the shim selection process. Through regression analysis of reworked axles, a statistically significant linear correlation between shim thickness change and the dependent variables of backlash and bearing torque to rotate was identified. The coefficients from the regression analysis combined with the measurement uncertainty components were included in a predictive MCM simulation. Results from the simulation were compared to production data to evaluate the effectiveness of the model at predicting system performance. The model simulation did predict system first time acceptance through the shim selection process, MCM results were within 0.8% of backlash and 0.2% of bearing torque to rotate when compared to sample production data. Though there was a statistical difference in the prediction of backlash, the effect was not practically significant. The study identified that factors not directly associated with the assembly measurement process are a significant contributor, repeatability GR&R studies alone were insufficient in explaining the overall process error. Hypothesis testing suggests that the application of the measurement uncertainty framework that includes the effects of statistical and non-statistical contributors to process error can predict automotive axle shim selection process capability.

Download

Share

COinS