RSS Tolerance Stack Analysis involving differing number of washers

[Colclst6]

I have a gap of interest in an RSS tolerance loop that involves two loop items that can each be filled with one washer minimum and three washers maximum. The number of washers needed will depend on system installation tolerances that are not part of the tolerance loop being analyzed. Is it correct to assume the loop items can treated as varying from the value of one thin washer to the value of three thick washers? A thin washer is a standard washer produced at minimum thickness and a thick washer is a washer produced at the maximum thickness. A colleague suggested to analyze the gap three times with one washer, then two washers, and finally three washers due to the discrete thicknesses of the washers. This approach seems to remove the statistical variation of using one, two, or three washers. Is this second approach valid?

 

[3DDave]

Since you appear to be using selective fit the typical tolerance loop/stack won't work. The questions that seem important are:

1) How often will the assembler have to try more than X washers before finding a fit?
2) How often will X washer selections be insufficient to get the correct fit?

The typical approach would be to use a Monte Carlo simulation where there is a randomized gap and random washers are chosen until that gap is filled within whatever remaining gap is achieved. You will need to know the statistical distribution for the assembly gap to be filled. You will also need to define a procedure for selecting thin or thick washers that the assembler will use for making that decision.

 

[Colclst6]

I am attaching a file with the two different tolerance analyses. These analyses are used to determine the range of a seal compression. This joint is two cowl doors latched together with a specific preload in the latches. The latch preload is set and then any resultant gap between the doors is filled by adding washers. The nominal number of washers is two per door. Depending on what it takes to achieve the specified latch preload and close the gap between the doors, the total number of washers can range from one to three per door. The washer number on both doors needs to be the same. A thick or thin washer is just the thickness variation allowed by specification for a .016in washer. I am trying to find the correct methodology for doing a tolerance analysis for this type of situation.

 

[3DDave]

The typical approach would be to use a Monte Carlo simulation where there is a randomized gap and random washers are chosen until that gap is filled within whatever remaining gap is achieved.

There is no "stack" that is suitable for this situation. As soon as there is a step "The assembler selects washers to achieve the specified preload," which is not a dimension, then there is no stack to be calculated.

With even three washers you cannot RSS the result, if that is the intention, unless the entire tolerance loop is included, but that already includes a spring force, which isn't a dimension, so that loop cannot be completed.

Write a program that simulates the assembly process, making selections based on the expected distributions, noting that an entire box of washers are likely to be all the same thickness having been punched from a single sheet of material, ruining the idea they can be averaged to some middle value.