Root Sum Square Fixed Fastener Tolerances Analysis 1

[KENAT]

(Oops, somehow this ended up in the wrong forum I'll try again)

Tried Googling this site & the web without apparant success.

I want to do an RSS type analysis on fixed fastener hole patterns.

Basically the equations in ASME Y14.5 Appendix B4 but slightly less conservative.

I could probably work it out from first principles but my initial efforts led to brain ache .

I happen to think that for my application worst case is usually most appropriate but in some cases might be interested in applying RSS for comparison.

Any help, or even (gently worded) contstructive criticism pointing out that I clearly don't know what I'm on about, is appreciated.
KENAT, probably the least qualified checker you'll ever meet...


KENAT, probably the least qualified checker you'll ever meet...

 

[drawoh]

KENAT,

I hacked around with a lot of this a few years ago, and I left an article on my website.

http://home.eol.ca/~hgibson/tolerances

Or...

http://home.eol.ca/~hgibson/tolerances.pdf

I worked this out before I noticed the appendices in ASME Y14.5M. I like my way of doing it. I need to clarify the Positional Tolerances with Zero Positional Error section.

Does any of it help?

JHG

 

[KENAT]

Thanks drawoh, I'm just leaving (POETS day) but I'll take a look Monday, or over the weekend if I get a chance.

I originally did if from notes by Gary Whitmire before I found it in ASME Y14.5!

KENAT, probably the least qualified checker you'll ever meet...

 

[KENAT]

Thanks drawoh,

I found working from first principles helps give a better feel for what's going and allows me to use the appendix B equation with more understanding.

I looked through and didn't see RSS applied to hole position/size, did I miss anything?



KENAT, probably the least qualified checker you'll ever meet...

 

[btrueblood]

Kenat...

I share your pain. I don't think I've read a good, comprehensive treatise on how/when to apply the RSS rule for tolerance stackup analysis. Generally, you need to know that the tolerance, or some multiple of it, represents the 3-sigma gaussian distribution of the real part condition, and then you can add all of the tolerances in question in one coordinate direction; it's not clear to me that you can add those errors in 2-dimensions in all cases (depending on the process used to make the parts). The principle, I believe, breaks down if the error is not a normal Gaussian distribution about the "zero" or perfect condition. Think of three independent bolt holes that must line up - the vector sum of the errors must be less than the difference of the minimum hole size and max possible fastener size...the problem is with the hole size since you have a min. size called out it affects the distribution, skewing it from a normal dist. Just one example, and a minor one.

I know of at least one case that I analysed where RSS would have been overly conservative (but where a true statistical analysis showed we had some wiggle room that was less than a worst-case stackup). I have, when necessary, resorted to a "Monte Carlo" analysis, using random numbers from the excel spreadsheet function or similar, to analyse a multi-variable, 2-D or 3-D tolerance stackup, basing my assumptions of the error distribution on how the parts were actually being produced/inspected/sorted.

 

[KENAT]

Yeah I've found some references for conventional stacks and that type of thing. I even have a spreadsheet from a colleague that puts in the factors for 6 sigma etc. All of this I more or less get though I use it so infrequently I have to re-learn it every time!

I was thinking of simplifying it to a 2 hole situation and going from there, using the min & max for spacing & hole size.

Any comments welcome.

KENAT, probably the least qualified checker you'll ever meet...

 

[drawoh]

KENAT,

You did not miss anything. I was not sure what RSS was. I did a Google search on your message title, and I did not see anything. I searched again on "root sum square tolerance", and found quite a lot of stuff, including some message threads here.

JHG

 

[KENAT]

Sorry lazyness on my part, I'd put it in the title and didn't type it out again.

KENAT, probably the least qualified checker you'll ever meet...

 

[WHITMIREGT]

When people talk about 6 sigma such as developed by Motorola, it is my understanding that ± 3 sigma is the same thing as the 6 sigma. Please clear this up for me. Thanks

 

[KENAT]

WHITMIREGT, I looked at this a few months ago and understood it for a few weeks but now I've lost track. I think what you said is correct but I can't be sure.

I'm not necessarily looking for explicit 6 Sigma but even just basic root sum square/root mean square as a less conservative analysis.

I'm always finding issues on hole patterns using the formula in your notes (which are basically the same as those in 14.5 unless I'm mistaken) and then people tell me "well we've been building it (or the design it's copied from) for X years/months/units with no problem".

I'd like to be able to go back and say either "well using RSS there's no problem so that combined with the history suggests low risk so we'll leave it" or "even using RSS which is arguably optimistic then there's a problem so we really do need to fix it".

Maybe this is too simplistic approach but I'm tired of being the bad guy pointing out theoretical problems they haven't yet encountered in reality and getting them all upset at me.

KENAT, probably the least qualified checker you'll ever meet...

 

[ShaggyPE]

Kenat,
I don't have specific training or references for this method, but this is a method that my company has used for determining the root mean square tolerance. I have heard this method called adding the tolerances in "quadrature" as well.

Rather than adding T1+T2+T3 you would SQRT(T1^2+T2^2+T^3). This definitely results in less conservative numbers than a worst case stackup.

Again, I don't know how valid this method is, but we use it here.

-Dustin
Professional Engineer
Certified SolidWorks Professional

 

[KENAT]

Thanks Shaggy, I understand that that is the principle behind root sum square or what ever you choose to call it. For a simple stack I have done that.

I'm talking specifically about hole patterns like in Appendix B of 14.5 as mentioned above.

KENAT, probably the least qualified checker you'll ever meet...

 

[wallyd999]

Wow - you made me open a book to see what your subject is. Being in the aerospace arena if we need to, we will grind the holes to the position. It gets expensive but the parts travel well when in use and they do meet the design intent.

Thanks for the information I’m getting to like this forum.