| Increased global competition among manufacturing organization has generated considerable interest in mechanical tolerancing research. Even though many significant achievements have been made over the last few years, several fundamental problems still exist. Current tolerancing methods cannot deal with the geometric tolerance design problems. They also fail to account for the non-normal nature of process random variations. In addition, the current methods neglect the impact of tolerances on the product performance.;In this research, a series of related investigations are described to solve the above problems. A new representation scheme for geometric tolerance, based on vectorial tolerancing, is proposed; an appropriate technique to analyze tolerance accumulation is devised. A new probability distribution, named as Finite Range Probability Distribution Function (FRPDF) is proposed to model the manufacturing data. This distribution is easy to use and is highly flexible to model wide variety of distribution shapes that are commonly encountered in manufacturing. The statistical analysis of tolerance accumulation and prediction of assembly yield are achieved by a technique based on numerical integration. This technique proved to be more accurate and faster than conventional methods such as method of moments and Monte-Carlo simulation. Finally, a new method for the synthesis of geometrical tolerances based on robust design philosophy is presented. This methodology can concurrently design product parameters and their optimum tolerances while considering factors such as performance variation, manufacturing cost and assembly yield. |
