Misconceptions about Using Design of Experiments, Part 3

The last update you received discussed the second of the three misconceptions about using Design Of Experiments. The three are, the belief that DOE doesn't apply when all of the factors are discrete (categorical), thinking that responses are factors, and believing that factor constraints prevent the use of DOE. This update will discuss the third misconception. The exercises you performed in your DOE class allowed all factor levels between low and high. In the real world this isn't always possible. Factor constraints can make certain factor level combinations undesirable or impossible. Here's an example: You have a laser drilling process. You can pulse laser light at a target to drill a hole. You can adjust both the power and the number of pulses of light. If you have high power, you won't need many pulses. Too many might obliterate the target! If you have low power, you may need many pulses. Too few will barely dent the target. Since it doesn't make sense to run low pulses with low power or high pulses with high power, it looks like we can't use DOE here. The good news is, we can use DOE very successfully here. Custom designs can avoid the undesirable factor level combinations. To see how this works, let's assign some levels to our factors and their constraints. (These levels are hypothetical — you would need to understand your laser to choose realistically.) Power: 1 to 6kW Number of Pulses: 1 to 101 Never pulse fewer than 21 times at low power. Never pulse more than 11 times at high power. You can use either JMP 8 or Gosset to create a custom design for these conditions. A design for a full quadratic model looks like this: Power Pulses 1 21 6 7.54 1 60.19 3.693 10.23 1.462 55.46 3.662 10.35 3.581 54.53 1 101 3.435 52.22 Notice that you are never asked to run an undesirable combination. If you created your design with JMP 8, the constraints will automatically be applied in the analysis. If you used Gosset, you will need to account for these yourself. ( Objective DOE customers — give me a call and I can help you.) Here are some examples: You want to bake a cake with no more than 3 spices. You have 6 spices to choose from. Which spices should you use, and how much of each should you use? You have 2 materials, A and B. Depending on which material you choose, another factor will need to vary over a different range. How can you account for this in a design? You have two mixtures. The first mixture contains the second mixture as one of its components. How can you specify this in a design? You need to have a constant (or nearly constant) ionic strength while varying certain ion concentrations in your design. How can you do this? Constraints can look very complex. Don't let this stop you — odds are very high we can create a custom design for your situation. You can learn to create your own custom designs using either JMP 8 or Gosset. Objective Design of Experiments workshops will teach you to use DOE in your work. Design of Experiments is a fundamental technique for industrial experimentation. You will learn to apply DOE easily without excessive math and theory. We will help you be even more successful!