Overview for Create Response Surface Design (Box-Behnken)

Use Create Response Surface Design (Box-Behnken) to create a designed experiment to model curvature in your data and identify factor settings that optimize the response. Box-Behnken designs are useful when you know you need to model curvature in your data, because these designs usually have fewer runs than central composite designs with the same number of factors. Box-Behnken designs also do not include design points at the extreme settings of all factors, which may be more appropriate for certain processes. For more information, go to What are response surface designs, central composite designs, and Box-Behnken designs?.

When you create a design, Minitab stores the design information in the worksheet, which shows the order in which data should be collected. Once you have collected your data, use Analyze Response Surface Design to analyze the data.

For example, a chemical manufacturer is trying to maximize the yield of a catalytic reaction. The researchers used a Box-Behnken design to study the effects of time (Time) and temperature (Temp) on the yield (Yield) of the reaction. Because the data was collected on two different days, the design contains two blocks.

This Minitab worksheet shows a portion of the design. The researchers perform the experiment by collecting data using the order shown in the RunOrder column.

C1 C2 C3 C4 C5 C6 C7
StdOrder RunOrder PtType Blocks Time Temp  
4 1 2 1 100 150  
1 2 0 1 80 140  
2 3 2 1 100 145  
7 4 0 1 90 150  
3 5 2 1 80 145  
5 6 2 1 90 145  
6 7 0 1 90 145  

After collecting the data, the researchers enter the response data in an empty column in the worksheet and analyze the design.

Many of the choices you make when you create a design depend on your overall experimental plan. For more information, go to Phases of a designed experiment.

Where to find this analysis

Stat > DOE > Response Surface > Create Response Surface Design

When to use an alternate analysis