I am using a mixture design (optimal custom design ) for designing my experiments. I have 3 factors and all of those factors having different ranges, but totalling 100 always. While using the design-expert, I found lack-of-fit points to be 5. Can you tell me what does this means? Also, what could be a possible difference might I get in my responses if I use point exchange and/or coordinate exchange searches in this optimal custom design.