I'm looking to develop a DoE using an I Optimal design for 3D printing process parameters. However, I'm looking at two Run parameters (I'm using the Design Expert software): "Lack-of-fit points" and "Replicate points".
By default, they both have a value of 5. But from my understanding, the "Replicate points" is a parameter that affects the optimality the more the points are replicated while "Lack-of-fit points" affects the relationship between the input factors and the responses.
Now, by increasing either by a factor of 1, the total runs of course increases also by a factor of 1.
Question:
However, in terms of the significance and other points that are generated during the Analytical stage, has there been any works that discuss the importance of these factors on the accuracy behind an I-Optimal DoE in the end? Surely, you wouldn't want to increase the points by a large amount, otherwise you might as well do a Full Factorial, haha!
Of course, I do understand that this question is indeed quite open ended because the outcome of what I'm asking hugely depends on the type of research being conducted.
Any info and advice would be greatly appreciated!
Hi Ezekiel,
It's all a matter of how high you need the resolution of your DOE to be. For example, if you are having extremely high precision (low variance in replicates always) then there is barely a need to have elevated numbers of LOF/replicate points.
A more sophisticated approach is to check the Fraction of the Design Space (FDS) graph and see if the resolution for your purpose is sufficient after you have created the DOE. You can check this video from statease and make yourself familiar with FDS: https://www.youtube.com/watch?v=6CUa_n6ZEj4
- Badges
- Science topic