1.2 n-FACTORIAL

 

• These experiments vary parameters with the basic procedure below,

1. Identify process variables (inputs) and dependant variables (outputs). Outputs should be continuous values.

2. Select discrete values for the inputs. The most basic approach is to pick a high and low value for each.

3. Create a data collection table that has parameters listed (high/low) in a binary sequence. Some of these tests can be left off (fractional factorial experiment) if some relationships are known to be insignificant or irrelevant.

4. Run the process using the inputs in the tables. Take one or more readings of the output variable(s). If necessary, average the output values for each of the experiments.

5. Graph the responses varying only one of the process parameters. This will result in curves that agree or disagree. If the curves agree then the conclussion can be made that process variables are dependent. In this case the relationship between these variables requires further study.

6. Calculate the effects of the process variable change.

7. Use the results of the experiment to set process parameters, redesign the process, or to design further experiments.

 

• e.g. 3-factorial DOE for springs in last section