The objectives of DoE are to adjust the quality characteristics (or design or process output) to the optimum performance by properly choosing the best combination of factors and levels, as shown in Figure 7.1. This is accomplished by collecting maximum information from the DoE experiment results using minimum resources. The factors can be categorized to determine which factors effect the average, variability, both average and variability, or have no effect on quality characteristics. Figure 7.2 shows these possible effects. The results of a DoE experiment can be one of the following:
1. Identify the most important factors that influence the quality characteristic
2. Determine factor levels for the important factors that optimize de- sired quality characteristics (output responses)
3 Determine the best or most economic setting for factors that are not important
4. Validate (confirm) responses and implement in production or design
The success of an experiment is not determined solely by just achieving the desired quality level. Important information about the design or the manufacturing process can be gleaned from any experiment. This information can be put to use in future experiments or through using more traditional quality improvement processes such as TQM. Information gained from DoE can be listed as follows:
• The factors that are significant for influencing the quality characteristic average, reducing variability, or both, and which factors are not significant. If none of the factors are found to be significant, then the design of the experiment has to be repeated to include factors or levels not previously considered.
• The proper balance between average shift from target versus variability reduction by choosing the proper factor levels. The choices of certain factor levels can shift the average, whereas others can reduce the variability, or both. Although good results can be obtained by moving the average to the maximum or minimum possible or achieve a target for the design, this action can be tempered by selecting alternate factors and levels to achieve the greatest robustness in reducing variability. The quality loss function discussed in Chapter 6 can be used to make decisions based on economic considerations.
• The predicted experiment outcome can be determined when the design or production factors are set to the specified levels. Confidence intervals and the expected error can also be shown for the predicted outcome.
• The goodness of the experiment design and the proper selection of factors and levels can be evaluated by statistical analysis.