## Lesson 6: Optimize Product/Process Performance

Consider a simple example.

We would like to find optimum Backing conditions, Baking Temperature and Baking Time to get a desired Bulk Density between 1.15 and 1.3 and desired Moisture % between 40% and 50%. How would we find the optimum conditions?

One possible approach is to use a contour plot. We generate a contour plot of Bulk Density and Moisture %, as a function of Baking Temperature and Baking Time. See figure below.

In the above contour plot the light brown diamond shape shows the desired optimum region. If we look closely a 345 degree F bake Temperature and 35 Minutes Bake Time is an optimum.

But wait.

What about the other responses, Particle Size, Color, Weight, and Cost?

Should we draw 4 more contour plots?

For these four contour plots which Control-variables-pairs should we select?

A little bit of math, which I would skip, will show that we need to look at (8 x 7/2) x (6 x 5/2) = 420 pair-wise contour plots are possible. Hmmm ... too much work, too many options. A visual approach will not work!

We need to use a different approach.

As Albert Einstein, would say:

We cannot solve our problems with the same thinking, that we used, when we created them. We need to think in a higher plain.

Here is a set of optimums. The second column, "moisture_max" shows optimum conditions for Maximizing Moisture. First column lists the Responses and Control Variables values corresponding to that optimum.

Notice you cannot obtain these type of results using contour plots. FReD, the bot, thinks in a higher plain!

That concludes this short lesson. Lesson 6: Optimize for Product/Process Performance.

About Author