Hi Tim,
Yep, when you want to see various “build scenarios” for example. Using “Crystal Ball” you can model assemblies of parts from there underlying distributions. The method assumes “random assemblies”. I have done this quite a few times and am always surprised to see the “mathematical simulation” model the “truth”. (IE the actual data generated from the assembly over time, nearly always fits your original model) The nice thing about it is you can “experiment” with specifications and in turn lead on to process parameters that you need to have, such as a particular X bar or sigma. With Crystal ball you can move the pointer around, which is the confidence interval, and look at “regions” where you dont want to be in the design. Great tool for turning “traditional” worst case engineering tolerancing on its head. I am mostly applying it on manufacturing scenarios, but just use your imagination for others !!
Cheers
Jeff.
.
