A Six Sigma Approach to Improving Exhaust Valve Reliability in a Stoichiometric Natural Gas Engine

Six-sigma is a data-based problem solving methodology that is used to reduce production variation in order to improve product quality. It has typically been used by successful companies to optimize operational processes and assist with waste reduction. Over the past several years, Dresser Waukesha has incorporated Six-sigma tools into several different areas of product engineering. This paper will illustrate how Dresser Waukesha has used Six-sigma tools to assist with product design and validation processes for reliability improvements of exhaust valves. Several specific examples of Six-sigma tools will be provided as the improvement project follows the DMAIC process. The scope of this discussion will be limited to using Six-sigma tools in the measure, analyze and improve phases of the project, where product engineering is most involved. In the measure phase, design of effective ratings charts that transform subjective visual assessments into quantifiable data for purposes of evaluating the effects of specific design parameters on critical wear metrics will be shown. Also, gauge repeatability & reproducibility for precise wear measurements will be discussed. During the analyze phase, the use of box plots, design of experiments tools and ANOVA statistical analyses for material selection, valve guide length and stem clearance determination will be explained. During the improve phase, an example of using the Pugh Matrix to compare different analytical concepts will be provided. Finally, an evaluation methodology that uses statistical capability analyses instead of traditional analyses of individual valve wear to validate long-term reliability and to minimize required test time will be explained.