InsightFab
Knowledge Base/DOE: Not Just Running Experiments, But Hitting the Most Targets with the Fewest Bullets
DOE9 min read

DOE: Not Just Running Experiments, But Hitting the Most Targets with the Fewest Bullets

A full factorial experiment with four factors and three levels each requires 81 runs. Using a Taguchi L9 orthogonal array, only 9 runs are needed, saving 89%, while simultaneously estimating the main effects of all four factors without mutual contamination.

Scenario

In an injection molding process, warping is a defect, and your supervisor asks you to find the optimal parameters. You have four adjustable factors: mold temperature, melt temperature, holding pressure time, and cooling time, each with three levels. Your colleague suggests, "Try them one by one, fix the other three and adjust only one."

This would require 4 × 3 = 12 experimental runs. However, the problem is that there might be interactions between these four factors – the effect of adjusting mold temperature might only appear when melt temperature is high. Trying them one by one will never find the truly optimal combination.

The DOE approach is to design 9 experimental runs, simultaneously estimating the main effects of the four factors.

In Plain English

The core logic of DOE (Design of Experiments) is: to arrange experiments in an "orthogonal" way, allowing the effect of each factor to be estimated independently, without interference.

"Orthogonal" means that each level of every factor appears an equal number of times throughout the experiment. This ensures that the calculated "factor effects" are not contaminated by changes in other factors.

A Taguchi L9 orthogonal array, with 4 factors and 3 levels, only requires 9 experiments to estimate the influence of each factor. A full factorial experiment would require 81 runs, saving 89%.

The S/N ratio (Signal-to-Noise ratio) is key to the Taguchi method: it's not just about getting the mean close to the target, but also minimizing variation. There are formulas for three scenarios: "larger is better," "smaller is better," and "nominal is best." A higher S/N ratio indicates a more robust set of parameters.

Practical Judgment

  1. First, create "factor effect plots": the factor with the steepest line slope has the largest impact.
  2. Find the combination of factor levels that yields the highest S/N ratio.
  3. Conduct a confirmation experiment: if the predicted value differs from the actual value by > 20%, it indicates an interaction, and a full factorial experiment should be used instead.

Situations where the Taguchi method is not suitable: If you already know that two factors have a strong interaction (e.g., temperature and time often do in chemical reactions), in this case, you should switch to the Response Surface Method.

How InsightFab Does It

In JMP, designing a DOE requires selecting the orthogonal array type, manually entering experimental results, and then interpreting the ANOVA output tables. In InsightFab, after selecting factors and levels, the experimental table is automatically generated, and after entering the results, it directly tells you the optimal combination and the predicted improvement.

Golden Quote

"Full factorial experiments are brute force; DOE is strategy. In production lines with limited resources, strategy is always more valuable than brute force."

Want to try it yourself?

Every tool mentioned in this article is available on InsightFab — just upload a CSV to analyze.

Go to Tools

More in this category

DOE Taguchi Method: Finding Optimal Parameters in 9 Experiments, Not by Luck

A full factorial design for 4 factors at 3 levels requires 81 runs, while a one-factor-at-a-time approach, though shorter at 36 runs, misses crucial factor interactions. The Taguchi L9 method offers a highly efficient alternative, completing the analysis in just 9 experiments and achieving an 89% reduction in experimental effort.

Response Surface Method: Finding the Optimal Process Sweet Spot

After DOE identifies the significant factors, RSM (Response Surface Methodology) helps pinpoint the precise parameter combination that optimizes the outcome. This marks a crucial transition from understanding 'which factors are important' to determining 'what settings are best'.

Full Factorial Design vs. Taguchi Method: How to Choose DOE, Choosing Wrong Wastes a Month

When factories need to conduct experiments, the choice between Full Factorial and Taguchi L9 is a critical trade-off between cost and precision, not a matter of preference. This article clarifies the applicable scenarios, advantages, disadvantages, and selection logic for both methods.