FUNDAMENTAL PRINCIPLES BEHIND ROBUST DESIGN

1. The functioning of a product or a process is characterized by signal factors, SF, and response factors, RS, which are influenced by control factors CF and noise factors, NF.

2. In a robust product or process the RF's are accurately meeting their target values (FR s in axiomatic design) as functions of the SF's under the control of the CF's independent of the NF's.

3. The robustness of a product or a process can be increased through the choice of operating values for the SF's and the CF's (parameter design) and/or additional design parameters with functions improving in the accuracy of the RF values in relation to the target values (tolerance design).

4. A quality loss function is defined in order to be able to quantify the penalties associated with deviations in the RF's from their required target values.

5. The principles of parameter design and tolerance design are together the principles of robust design. Parameter design is the primary principle and is not associated with any additional cost. Tolerance design means the addition of extra design and is associated with extra cost. Tolerance design is only needed if parameter design is not enough in order to improve the accuracy in the target values of the RF's. The cost for tolerance design is balanced against the decrease in quality costs according to the quality loss function.

6. Parameter design is utilizing non linearities in the SF's and CF's in order to set their values so that the influence of the NF's on their values has enough low influence on the RF's. This leads to a more narrow distribution but also a dislocation in relation to the target values. In order to align this again a factor with a linear influence on the RF has to be found and used in order to calibrate the RF to be aligned with the target value.

7. In order to define useful values for the SF's and CF's tests with different values on the factors have to be carried out. The tests are done on the product or process in question or by simulation. For each factor two or three values are used. Three values have to be used in order to find useful non linearities. In order to limit the number of tests and also to avoid interdependencies between the factors that are tested a set of test arrays (orthogonal arrays) are designed and recommended to be used in the planning of the tests. The different arrays contain different number of factors and test-values.

8. Statistical analysis of the test result gives the base for deciding set values for the SF's and CF's leading to a more robust design. If this is not enough tolerance design has to be done.

9. The tests are carried out in the normal environment for the product or process in order to get exposed to the real noise levels.

This list of fundamental principles is set op by prof. Gunnar Sohlenius March 10.1998 after reading Quality Engineering Using Robust Design by M.S.Phadke 1996 and after conducting a course in robust design together with Roland Langhe’at KTH in Stockholm.