Multi-Disciplinary Tolerance Optimization for Internal Combustion Engines Using Gaussian Process and Sequential MDO Method

The internal combustion engine (ICE) is a typical complex multidisciplinary system which requires the support of precision design and manufacturing. To achieve a better performance of ICEs, tolerance assignment, or tolerance design, plays an important role. A novel multi-disciplinary tolerance design optimization problem considering two important disciplines of ICEs, the compression ratio and friction loss, is proposed and solved in this work, which provides a systematic procedure for the optimal determination of tolerances and overcomes the disadvantages of the traditional experience-based tolerance design. A bi-disciplinary analysis model is developed in this work to assist the problem solving, within which a model between the friction loss and tolerance is built based on the Gaussian Process using the corresponding simulation and experimental data. In addition, the formulation of the compression ratio considering those non-critical dimensions which actually affect the friction loss is proposed. Finally the multi-disciplinary tolerance design optimization problem is formulated and solved using a recently developed sequential MDO (S-MDO) method.