Radial stackups that accurately account for assembly clearance and shift? CETOL

[thezack123]

Hi there!

I’m having quite a bit of confusion over the correct way to do radial stackups that involve two or more concentric components. I noticed that solidworks tolanalyst and Sigmetrix CETOL provide different answers for a simple example assembly with two concentric components. This led me to try to investigate why these commercial software packages provide different results. I've attached a document which explains the issue with a simple example with two cylinders.

I started by trying to do the stackup by hand to check which was correct and try to isolate the issue. I’ve seen examples online where people calculate the distance between two concentric components using a stackup table, performing the analysis radially. Generally these stackup tables account for the gap distance that is the result of the size difference between the two components, or account for the amount that the center axis of both components can shift with respect to eachother. Or they do a worst case analysis which can give an accurate number for either the min or max gap, but not both (without producing two separate stackups and tables) I have not yet seen an example where somebody does a single stackup that accurately calculates the minimum and maximum gap.

I am trying to understand this because I am working on a prototype which involves multiple concentric components, for a turbocompressor assembly. I’d like to be able to perform an accurate radial stackup to determine whether interference is possible at any assembly condition. In order to consider the effect of multiple components on a given gap, one must be able to calculate it correctly for a simple example.

Two concentric components, an 8mm diameter shaft with +0, -.2 tolerance, fitting into a 10mm diameter hole with tolerance +.2, -0. I’m trying to find a way to accurately do a stackup that finds the minimum gap between the two components. My real application has 6 components which contribute to the stackup, so any innacuracy in my method will result in a wildly innacurate stackup for my application.

I'm looking for some feedback to help me figure out the best way to do these radial stackups for two or more concentric components.

Thank you!

 

[pmarc]

I’ve seen examples online where people calculate the distance between two concentric components using a stackup table, performing the analysis radially. Generally these stackup tables account for the gap distance that is the result of the size difference between the two components, or account for the amount that the center axis of both components can shift with respect to eachother. Or they do a worst case analysis which can give an accurate number for either the min or max gap, but not both (without producing two separate stackups and tables) I have not yet seen an example where somebody does a single stackup that accurately calculates the minimum and maximum gap.

Although I am not sure that your approach will be useful in more complex cases (where you would most likely have to include some additional location relationships between components), here is how it should be when a single worst-case stack is considered.

#1 | -3.95 | +/- | .05 |
#2 | +0.10 | +/- | 1.10 |
#3 |+5.05 | +/- | .05 |
-----------------------------
|+1.20 | +/- | 1.20 |

MIN = 0.00; MAX = 2.40

There are two numbers hidden behind the line #2:
1) +1.2 = +0.1 + 1.1, which represents the maximum possible radial shift of inner part inside the outer part in one direction from the nominal position (say up), which is when ID of outer part is 10.2 and OD of inner part is 7.8, (10.2-7.8)/2=1.2;
2) -1.0 = +0.1 - 1.1, which represents the minimum possible radial shift of inner part inside the outer part in the opposite direction from the nominal position (down), which is when ID of outer part is 10.0 and OD of inner part is 8.0, (10.0-8.0)/2=1.0.

************
Without knowing what exactly you did in CETOL (i.e. how you prepared the analysis and with what settings), it will be hard to tell why it is giving you -0.2 for minimum possible gap. I personally haven't worked with CETOL at all, but for what it is worth I did the analysis in VIS VSA (Variation Analysis) and I got expected MIN=0 and MAX=2.4 when everything (distributions for sizes and assembly float options) was set up to extreme.

 

[greenimi]

I am not familiar with CETOL either, but what I do know is that CETOL does not take in consideration the form error. That is the fact.....confirmed with some people that got experience with that piece of software.

And full disclaimer: I did not check pmarc's numbers, but I fully trust him on this. (I suspect he considered the applicable form error).

 

[greenimi]

To support my point imagine a simple washer dimensioned and toleranced for size (OD and ID) and the relationship between them (position at MMC, LMC or RFS).
The min/MAX between OD and ID is straight forward and has been discussed many times here on the forum.
I was told that these calculated values of min/MAX cannot be confirmed in CETol because the form error of the applicable features is not considered by the software. In other words you would have a discrepancy between CETol and manual calculation.

 

[3DDave]

There's the CETOL Details tab, the Sensitivities tab, and the Contributors tab. Looking at those should expose the error.

CETOL is (used to be) a vector chain analyzer where each element is forced to be in some location. Your analysis is of a situation where, choosing the wrong features, you've probably managed to create case where the axis is forced to a location where it might be for an LMC shaft and hole, but allowed for the MMC shaft and hole radius.

tolanalyst is likely to have purpose written equations to prevent that particular situation but will fail to be able to do complicated ones.

The minimum in your case is always zero if you are allowing the shaft to rattle in the hole.

 

[thezack123]

Thank you for your help and responses! I'll keep looking into this.