Design of Experiments (DOE)

In all processes there is a great advantage in knowing how changing a system variable will change the system output. Sometimes we can be intuitive. If we press the gas pedal on our car we will go faster. If we change the color of the car, we may feel comfortable saying it won't effect our speed. But what if we are asked to comment about our speed in relation to where we buy our fuel?  At this point we will likely have to run a few practical tests. How well we plan and execute our tests will determine the accuracy and validity of our test results.

Design of Experiments is a formal technique for evaluating the effects variables have on a system response. The technique takes into account the interactive nature of some variables and provides quantitative information regarding the significance a variable has in a given process.

A DOE does not attempt to explain the physics behind a process but simply the response of the process to changing inputs. Whereas engineers and scientists would feel most comfortable with a set of equations that would predict the behavior of a process, the DOE method measures the response of the process with different variable settings, and constructs a set of equations describing the process behavior. This allows the experimenter to select variable settings to provide the best output for the tested process. The output could be one or more responses such a feature size, cycle time, or surface finish.

Noble Plastics currently uses MiniTab software for DOE analysis. A more detailed description of the process may be found on their site.