Choose Language Hide Translation Bar
julian
Community Manager Community Manager

Designing a Central Composite Design

Learn more in our free online course:
Statistical Thinking for Industrial Problem Solving

 

In this video, we show you two ways to create central composite designs in JMP. First, we use the classical Response Surface Design platform, and then we create the same design using the Custom Designer.

 

To start, we select DOE, then Classical, and then Response Surface Design.

 

In the responses outline, you can change the response name, change the response goal, or add additional responses. We’ll use the default settings for this video.

 

In the factors outline, two continuous two-level factors have already been added. We click Add to add one additional factor.

 

We click Continue to see the available designs for three factors.

 

You can create a 15-run Box-Behnken design with three center points, or one of four central composite designs. We select the 16-run central composite design with two center points, and click Continue.

 

You can choose where you’d like to place the axial values. This updates the axial value to use, but you can change this. We’ll use the default, On Face, with the axial value set at 1.

 

We select Make Table to create a randomized design table.

 

We’ll sort on the Pattern column so that we can better see this design. To do this, we right-click the column, and select Sort and then Ascending.

 

This design has eight factorial points. These points are a 23 full factorial.

 

There are two center points, and there are six axial points on the face of the design space.

 

Let’s look at the model we can fit with this design. To do this, we run the Model script in the data table.

 

JMP estimates a model with three main effects, three two-way interactions, and three quadratic effects.

 

The main effects are assigned the RS attribute. This tells JMP to run an extra Response Surface report when you conduct the analysis.

 

Now let’s create this same design using the Custom Designer.

 

To do this, we select DOE and then Custom Design.

 

We’ll add three continuous factors. So we enter 3 under Add N Factors, and then click Add Factor, Continuous.

 

We click Continue to see the Model outline. We want to estimate a model with main effects, two-way interactions, and quadratic effects.

 

So we select the three factors and click the RSM button.

 

The interactions and quadratic effects have been added to the model.

 

By default, JMP generates a design with 16 runs. This design will be identical to the central composite design we generated earlier.

 

To see this, we select Make Design and then Make Table.

 

Custom designs don’t have a Pattern column, but you can see that this design has eight factorial points, six axial points, and two center points.

 

This is a 16-run central composite design.

 

Note that you can create some classical response surface designs using the Custom Designer, but not all. 

 

However, the Custom Designer gives you much more flexibility in creating response surface designs, and these custom designs are often much more efficient than the classical designs.

Article Tags
Contributors