Taguchi Methods: Benefits, Impacts, Mathematics, Statistics and Applications
Download citation file:
Engineers apply experimental optimization methods to find input factor settings that improve system output responses. There are four major experimental optimization methods used in engineering applications.
1. Trial-and-error experimentation: This problem-solving approach uses experience and the judgment of engineers as well as production process operators. Typically, this approach solves emergent problems in production processes. Very limited experiments (two or three) are conducted when solving these problems. When the output responses meet or exceed the targets, the trial-and-error experiments stop immediately due to timing and cost concerns.
2. One-factor-at-a-time experimentation: In this approach, engineers conduct experiments on several potentially significant factors. However, only one factor is varied in each experiment; the other factors are fixed at default levels. In addition, the experimental environment and conditions are tightly controlled to determine the effects of these factors. This type of experimentation is used in academic research or in scientific studies conducted at universities.
3. Full factorial experimentation: In this approach, engineers conduct experiments on all possible combinations of all factor levels. A factorial experimental design is done for a small number of factors, but it is not practical for a large number of factors.
4. Traditional fractional factorial experimentation: In order to reduce the total number of experiments, engineers conduct experiments using fractional factorial designs in industrial experimentation. Orthogonal arrays like the , , , or belong to this type of experimental design. In these designs, specific subsets of all possible factor level combinations are evaluated in the experiment.