It's not that difficult, but I really don't have time to create and check something as I have a ton of design work to be completed. I would like to have something that will show Worst Case, RSS, Six Sigma, CP, Cpk, DPPM, and graph the results.
I am on board with tdalke. I have a excel spreadsheet that allows you to input the nom, plus and minus of dims and gives you the max/min clearance or interference.
I am concerned because tolerance stack-up is not absolutely simple, and I do not see how far you can automate it. You have to type in each tolerance, and you have to account for the direction each tolerance works in. This requires an operator skilled enough to recognize all this, as well as recognizing tolerances that are not reasonable. A tolerance of +/-0.1mm may work on your spreadsheet, but it will not work in your welding shop. Thermal expansion and contraction can be a problem too.
Automation works when it saves work and eliminates error. I do not see much work being saved here, and the errors are eliminated buy someone paying attention to what they are doing.
I agree that there are a number of things to take into account as I worked in manufacturing for over 17 years and fully understand things such as weld shrinkage, process capability, tool wear, operator skill level, etc.
What I am looking for is to check a number of tolerances inside of an assembly to ensure that mathematically it is possible. If it is not mathematically possible then there is no way for it to be manufactured. It also requires working with the vendors (machine shops, sheet metal shops, etc.) to understand their process cabilities and get as much input from them while in the design stage.
Your statement:
"I would like to have something that will show Worst Case, RSS, Six Sigma, CP, Cpk, DPPM, and graph the results."
I'm not sure you're going to get this from stackup analysis. Most of these quantities are process quality data that are developed after a process is implemented. These are tracking data and are used to determine how much processes change & improve by process changes.
From my own experience, one can develop an arithmetic stackup which is simply additive plus/minus values of all tolerances in a direction. Usually it isn't reasonable because the total stackup is huge. Besides, statistically one would probably not experience such wide variation in stackup, which sort of fits common sense. Juran's quality manual states that if you have capable in-control processes with Cpk's of 1.33 or better, then the arithmetic tolerance stack can be reduced through some sort (can't exactly remember the details) of a squareroot-of-sum-of-squares analysis.
Your analysis is the worst case senario. I want the RSS which is Root Sum Square. This involves adding all of the nominal values. Adding the root sum sqaure of the associated tolerance values, and then calculating the assembly tolerance using standard deviations based on the final assembly tolerance. From that you can calculate the CP, Cpk, and DPPM. From the CP or Cpk you can determine if your design is valid or not.
ex.
Part1 Overall = 1.000 ±.010
Part2 Overall = 1.000 ±.005
Assembly Overall = 2 ±.010
Normal distribution used
Assembly Worst Case = .015
Assembly One Standard Deviation = .003727
Assembly Statistical Tolerance = ±.011180
Cp = .894
Cpk = .894
DPPM = 7290.358
If 7290 defective parts per million is acceptable then the part tolerances are OK, but if not then the part tolerance need to be tightened, or the assembly tolerance needs to be opened up. In order to achieve a Cp of 1.414 both part tolerance would have to be ±.005.
I could but then I will have to double check it, create the graphs, etc. It is simple enough but I really don't have a couple of hours to do it and I figured that someone else may have already created one. Just trying to save some time...
A quick check in Google using [spreadsheet tolerance stack] came up with many hits. This web site seems to have a spreadsheet willing to give you. I did not go any further, as I do not care to end up on some mailing list.
I already check the formula's and they are all correct. It only took a couple a few minutes to do the manual calculations and veriy the results against each other.
