"the property RunsPerBlock is not set for a blocking factor"

[#1]

JMP 17.2.0

The issue with DOE is as follows:

• Screening design with 3 continuous, 1 categorical and 2 blocking factors
• The Custom design module rejects the blocking factors and reports: "the property RunsPerBlock is not set for a blocking factor. Skipping this factor"
• I can define the blocking factors as categorical but that seems to defeat the purpose of “blocking”

What is the recommended approach?

[#2]

Hi @MichaelR1,

Did you generate your design with a JMP platform, like Custom Design, or did you generate it elsewhere and then try to analyze/evaluate it ?
It seems you miss (at least) one column property for your blocking factors. See the different properties for Factors.

For blocking factor you should have the following column properties when designing your DoE :

• Value Order (1, 2, 3, 4... depending on the number of block you have): This property determines the order in which levels of the factor appear.
• RunsPerBlock (you can set this value in the design generation to 2, 3, 4, 8 or any other specific value for the number of runs per block): This property indicates the maximum allowable number of runs in each block.
• Design Role (it should be set to "Blocking" after design creation): This property reflects how the factor is intended to be used in modeling the experimental data.
• Factor Changes (can only be set as Easy in design creation): This property indicates how difficult it is to change factor settings in a designed experiment. 

These properties are automatically created if you create your design with any DoE JMP platform. If you have imported the design from elsewhere, you need to add and set the right column properties so that the role and specificities of each of your factors are well identified by JMP for the modeling.

Note that depending on your need and intended design, it is also possible to generate random blocks at the end of design creation. This possibility is disallowed if you have defined one (or more) blocking factors :

Hope this answer will help you. If not, could you share more info about your design, goal, possibly the JMP file for the design you're trying to create, ... ?

[#3]

Thanks Victor,

I used the JMP Custom Design module. First, I prepared a data file of factors and a separate data file of responses. I went over each factor in the file using "standardize attributes" and defined: data type, modeling type, design role, and factor changes. Following that, I loaded the factors into the Custom Design module by "Custom Design" -> Load Factors. When I tried to run the design, I got the above error message. Is there a better way to do this?

Enclosed is an example Factor data file.

[#4 · Accepted solution]

Hi @MichaelR1,

You can also create your factors directly in the Custom design platform.

In your factors table, some informations are missing :

• Blocking factors X_1 and X_6 : You need to add the column property RunsPerBlock and define an adequate maximal number of runs allowed per block. I choose as an example 4 and 8.
• Continuous factors X_3, X_4 and X_5 : You need to add the column property Coding and define the high and low values for each factor. By default JMP will choose the lowest level as the min level, and the highest as the max level.

It seems you have defined some factors with various complexity to change. Blocking factors can only be set as "Easy", as the introduction of a block will create a datatable with randomized runs inside each block, but the blocks will be kept "in order" in the datatable : all runs from block 1, then runs from block 2, ... If you have various complexity to change the factors, you may be interested into Split-Plot designs, instead of adding several blocking factors : Split-Plot Experiment

As an example, attached is your datatable with the column properties added and the resetting of Factors Change to "Easy" for all factors. You should be able to create a design without error messages with it. You can fine-tune and correct some properties values to adapt it to your use case.

Hope this answer will help you,