Tolerance optimization by modification of Taguchi’s robust design approach and considering performance levels: Application to the design of a cold-expanded bushing

This paper defines a method for the optimization of design parameter tolerances. The general architecture of the proposed method is identical to that of the robust design reference method proposed by Taguchi but its content is different as the tolerances are considered as functions to be maximized here, while Taguchi’s method rather considers these tolerances as fixed data. Instead of looking for design parameters that minimize the sensitivity of some performance criteria, the design parameters are calculated so as to obtain maximal tolerance intervals, thus minimizing manufacturing costs. Performance criteria are then considered in terms of optimization constraints: each criterion gives rise to an inequality constraint that specifies the minimum level of performance that the designer wants to achieve. The possibilities offered by this method are illustrated through its use in the preliminary design of a cold-expanded bushing. In this case, tolerance optimization enables the allowable tolerances on the design diameters to be increased and performance levels are defined on the residual radial stress at the bushing/part contact radius and on the residual orthoradial (hoop) stress at the part inner radius.