Using DoE Methods in Six Sigma Design and Manufacturing Projects
One of the most important consequences of implementing six sigma has been the increased use of DoEs by the design engineering community. DoEs can be used effectively to augment the traditional design engineering methods of computer simulation and analysis of worst- case design and materials selection. The DoE techniques outlined in this chapter can be used effectively for new product quality improvement as well as manufacturing process variability reduction. Several opportunities for using DoE for design engineering are:
• Worst-case study is the method by which engineers analyze designs using a combination of the worst case of the individual parts or materials specification limits. Design engineers might overspecify parts to tighter tolerances to ensure that they meet worst-case conditions. DoE methods can be used to analyze design tolerances, resulting in the proper specification of parts. Expensive tight tolerance parts should be used only when actually needed for the design to meet the specifications.
• DoE methods can be used in computer simulation of the design to obtain optimal results. The orthogonal array experiment conditions can be inputted into the simulation. The results could then be analyzed the optimal design.
DoES can be used in new products to solve some of the “black magic" type problems specific to electronic products, including the successful completion of environmental and transportation tests. Examples are reductions in electrical noise and radio frequency interference (RFI), and product mounting, shipping, and packaging techniques
• DoEs can be used effectively by multidisciplinary teams that need to work together to achieve performance to specifications for new products through trade-offs in design disciplines. A thermal printer case study is used to illustrate this use of DoE for new products in the next chapter.
• DoEs can be used for robust designs to achieve a linear region of performance of the factors for the quality characteristic. By selecting this linear region, the design is less sensitive to small factor changes, and hence less rigorous specifications can be used for the factors.
Conclusions
It has been shown, through several examples, that DoE is an excellent tool for optimizing designs by shifting the average characteristic(s) of the design to target and reducing variability. Both of these actions are very important in achieving six sigma quality. The mathematical background for DoE is a mix of tools of orthogonal arrays, designed experiments, and analysis of variance. There are several techniques in DoEs that should be thought out well in advance: the definition of the characteristics to be optimized, the selection of factors and levels, the treatment of factor interactions, the selection of experiment arrays, and how to simulate and measure variability and error.
An initial DoE project should be selected carefully to optimize a design that is relevant but not too complex. Careful hand calculations should be made to complete the analysis. Only after initial successes should software-based methods of analysis be attempted.