An Experimental Strategy for Designing Tool Life Experiments

Abstract

An investigation of the tool life experimental environment and its effect on designing tool life tests is presented. The D -optimal design criterion is introduced as a tool life experimental strategy particularly suited to the tool life experimental situation typified by (1) the use of nonlinear tool life model forms, (2) constrained operability regions which are irregularly shaped, and (3) the inhomogeneous nature of tool life variation. Linear and nonlinear tool life models are examined with the D -optimal design criterion. Comparisons with the 3-level factorial design and the Central Composite Design are presented. A significant reduction in the required number of tool life tests is obtained when D-optimal designs are employed. In particular, improvement in the predictive capabilities of the model is obtained along the constaint boundaries where the economic optimal operating condition will most likely be located.