Design of Experiments
Analyze input and output variables to identify the critical few.
Design of experiments (DOE) is a complex but powerful method of validating your innovative solution during the design process or prior to entering full production. For example, we could run a DOE to identify the optimal settings for a new ultralight, high-mileage vehicle. The DOE would tell us how varying factors (such as tire pressure, fuel octane rating, speed, and road conditions) affect gas mileage.
DOE is an alternative to best-guess or one-factor-at-a-time experiments, which are time- and resource-intensive and may not produce the optimal solution in the end. By using DOE to test more than one factor at a time, you'll end up with better, more reproducible solutions in less time, and you'll expend fewer resources. However, the approach does require rigorous statistical analysis and should only be used with support from statisticians or others who have been trained in DOE.
Scenario: Suppose you're designing a small robot that picks up metallic objects (screws, staples, metal shavings, etc.)—and one of the key components for the robot is an electromagnet. DOE can help you determine the electromagnet configuration that will best meet your design criteria.
Responses are the outputs that you will study during the experiment. Identify the key responses you want to measure, but keep in mind that as the number of responses goes up, so does the ...