Re: How do I create a simulation with +/-10% bias and 30% SD with an existing data set?
Oct 15, 2015 6:07 AM(1037 views)
Have you tried to set spec Limits for ABRASION and MODULUS in the Simulator Settings. Then you can use the things described there and see how the CPK/CP as well the Simulation data will Change in respect to the spec Limits:
As soon you have the table you can run the Distribution script on the left top window of the table:
You see the Distribution for each step. In the example you have 100 simulations per mean step. If you do not see the capability Analysis you can get it from the red triangle (hotspot) next to Distributions SILICEA MEan = ..., Holding the control key while pressing the hotspot and selecting the cap. Analysis will distribute this Analysis to all of the other distirbutions as well, so you do not have to do it manually.
One side effect of the spec Limits Setting in the Simulator is that you then have Access to the Defect Profiler under the Simulator hotspot. Which will allow you also to determine if a mean shift or a std Deviation shift should be get after as it will improve things. There is a Quality Paper on this talking abut tolerance Analysis and the defect profiler using a slightly different tiretrad data set, but similar to follow. Here is the paper presented on the JMP Discovery Meeting 2010 (JMP | Discovery Summit 2010 - Abstracts).