Adding a 7th parameter to an existing DOE

[#1]

Hello,

I have a DOE with 32cases for 6 parameters. the simulations have been already done long ago but i would like to add a 7th parameter to the model. how can i do that? should i start a new DOE from scratch or can i keep the already existing cases and add a few new experiments?

Thank you!

[Answer in english or french is ok :-)]

[#2]

If you know the levels of the new parameter in the existing 32 runs, you can simply add a new column in the data table with the name and levels. You can add this parameter to your model and re-fit it.

[#3]

In addition to Mark's suggestion, it depends on the simulation. How long does it take to run the simulations? Do the simulations require a significant amount of computing power? If they are quick and cheap, have at it. Design a new experiment. You have a 32 treatment experiment with 6 factors, so obviously not a full factorial. How many levels tested for each factor? Why not run a full factorial since it is simulation?

Not sure how well the simulations relate to reality, but that discussion is for another time.

[#4]

Thank you for your answer! A simulation is 3 days average... I found how to add the new parameter but I don't really understand how to verify the robustness of the new runs plan. I'm a beginner with this software. the param is continuous between 1.02 and 1.15 (lambda for an excess of air in combustion)

[#5]

Where you able to add a data column for the 7th parameter? Were you able to enter the level of this parameter in each of the previous 32 runs? If so, now you are ready to inform JMP that X7 is a factor because it was not included in the original design. Right-click on the X7 column header and select Column Info. Click Column Properties and add the Coding property. It should look like this when you finish:

Set the Low and High values to be the factor range for your simulations. Click Column Properties again and select Design Role. I assume it is a continuous factor:

Lastly, click Column Properties again and select Factor Changes:

That's it for X7. Click OK. Now, augment the design with new runs. Select DOE > Augment Design. Assign X1-X7 columns to the Factor role. Assign Y (or whatever the name is for the response column) to the Response role. Click OK.

I assume that the original experiment used a space-filling design, as is normal for simulation experiments. You can use the dedicated Space Filling method on the next dialog:

Now enter the number of additional runs your require.

Note that there is no logic behind the default number of runs, so feel free to enter a number that suits your budget. If X7 did not vary during the initial 32 runs, then a large number of runs is recommended. If it did vary, then a smaller number might be sufficient to model it successfully.

I am not sure what you mean by "verify the robustness of the new runs plan."

See the JMP documentation for augmenting a design.