Step 3. How to vary the right factors – by the right amounts
To find out what happens when you change something, it is necessary to change it. (Box, Hunter, & Hunter)
Have you ever looked at an experiment and wondered how you could possibly decide which factors to test? In this post you’ll learn how to clear up this uncertainty. You’ll learn how to decide what experimental factors to vary, and by how much.
In Step 2 we made the key decisions on what properties should be measured. We determined that we are able to measure them with sufficient accuracy and precision. We can now make the equally important decisions on what factors (independent variables, parameters) to vary – and by how much. Our perils here are two:
- We have too many factors, we’ll never be able to test all the possibilities!
- We’ll miss something important!
To avoid these perils, we need to start with:
Relevant Background Knowledge
What are the known theoretical relationships, practical knowledge, and results of previous experiments?
The purpose of this information is:
- To establish a context for the experiment and a clear understanding of what new knowledge can be gained.
- To motivate discussions about the relevant knowledge.
- To uncover experimental regions of particular interest and regions that should be avoided. This narrows down the number of factors that led to overload.
Every effort must be made to get maximum participation from all involved parties – the experimental team, management, and customers. The flowdown exercises in Steps 1 and 2 are admirable places to begin to assemble the team.
Tools such as brainstorming process mapping and Fishbone Diagrams (Figure) are invaluable for sparking the discussion among the team on the possible root causes that need to be studied. It is best to be radically inclusive, both of team members and of ideas at this point.
The cumulative expertise of the team can then be used to prioritize the (usually large) list of factors. One useful tool is multivoting, which is effective for narrowing a large list of possibilities to a smaller list of the top priority factors.
In multivoting, each team member is given several (often 5) votes to distribute among the list of possibilities. When the votes are tallied, the top priority items usually become clear. The results often prompt useful discussion as well
A Strategy for Using the Factors Right
The top priority factors must then be classified by Experimental Type. Is each one:
- quantitative (temperature, flow, concentration,…)
- qualitative (type of material, technique, machine,…)
- formulation (concentration, adding up to a constant such as 100%)
Each of these factor types requires a different experimental plan and data analysis!
Normal Level and Range – Use the known operating ranges as a starting point to determine the experimental settings. If there are no known ranges, trial runs are appropriate.
Measurement Precision and Setting Error – Set the control factors so the difference between them is much greater than their setting error. DOE calculations are often based on the assumption that there is no error in the values of the control factors. All the error is assumed to be in the responses. If the error in the control settings is large, more samples will have to be taken.
Proposed Settings – Set the levels of the factors far enough apart that it is likely they will have a visible effect – but not so far that the system will be out of reasonable or safe operating range. Hard core operating knowledge is at a premium here!
You now have a DOE strategy that
- captures the factors that are relevant to the organization’s goals and objectives, and
- contains the best understanding of all the team members as to which factors are important.
In the next section, we’ll use this information to get a first estimate of the resources required.
If you want to jump right to the whole strategy, contact me at +1 413 822 5006 or firstname.lastname@example.org!