Monte Carlo Simulation
Monte Carlo Simulation
(OP)
Hi,
I need some help. I'm trying to run a true position tolerance analysis using the Monte Carlo method. I have googled it and haven't found much help. can anyone recommend or have some useful information on how to do one of these simulations using excel?
I need some help. I'm trying to run a true position tolerance analysis using the Monte Carlo method. I have googled it and haven't found much help. can anyone recommend or have some useful information on how to do one of these simulations using excel?
RE: Monte Carlo Simulation
this has the advantage that if one setting for one factor causes a failed run, at least it is likely to be just one run, instead of several in structured experiment.
Cheers
Greg Locock
New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?
RE: Monte Carlo Simulation
- tends to excessively explore the near centre region, which is often not very interesting
-needs more runs than say D optimal
-statistically it is not as neat - analysis involves a PCA type approach instead of just adding and subtracting columns
advantages
- if you decide to curtail, or expand, the number of runs then it is still meaningful, whereas in a taguchi style designed experiment it is difficult (but not impossible) to make use of runs from a smaller experiment.
-less likely to wipe out many runs due to one bad setting for one factor.
Cheers
Greg Locock
New here? Try reading these, they might help FAQ731-376: Eng-Tips.com Forum Policies http://eng-tips.com/market.cfm?
RE: Monte Carlo Simulation
The Excel funtion to use is typically NORMINV(probability, mean, standard_dev). This returns the inverse of the normal cumulative distribution for a specified mean and standard deviation, along with a weight factor (the probability). If you use RAND() in place of probability, and use the expected process capability to predict the standard_dev (such as the total tolerance divided by 3, assuming a 3-sigma process), you can create a series of cells corresponding to each dimension or variable in your stack. Then you have the computer run thousands of trials and show the overall average and standard deviation.
A monkey wrench in your scenario will be if the position tolerance has any modifiers such as "M" -- the extra bonus or shift tolerance can be accounted for, but it gets a little tricky.
John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems
http://www.gdtseminars.com