· Product evaluation should be focused on comparison with the engineering requirements and also on the evolution of the function of the product.
· Products should be refined to the degree that their performance can be represented as numerical values in order to be compared with the engineering requirements.
· P-diagrams are useful for identifying and representing the input signals, control parameters, noises, and output response.
· Physical and analytical models allow for comparison with the engineering requirements.
· Concern must be shown for both the accuracy and the variation of the model.
· Parameters are stochastic, not deterministic. They are subject to three types of noises: the effects of aging, of environment change, and of manufacturing variation.
· Robust design takes noise into account during the determination of the parameters that represent the product. Robust design implies minimizing the variation of the critical parameters.
· Tolerance stacking can be evaluated both by the additive method and by statistical means.
· Both analytical and experiment methods exist for finding the most robust design
