Spring constant for gaskets (gasket creep)

[506818]

I'm using Belleville washers with PTFE gaskets to account for gasket creep and handle vibrations on flanges. I want to work out torque values to use. The idea is that the Belleville washers will deflect enough to account for the gasket creep in the PTFE. I'm seeing this as springs in series and I have the information for the Belleville washers, but not the gaskets. How would I work out both the expected creep/compression set as well as the spring constant for the gasket? Gaskets to be used are Gore 800 series.

 

[506818]

Thinking about this further, if springs are in series, would the deflection of the gasket be the same regardless of the Belleville washer and vice versa if the same load is applied?

Therefore, if the torque>gasket torque required AND deflection of Belleville>gasket creep, it should work as intended?

 

[3DDave]

Be aware that Belleville washers have high variations in spring rate. We obtained some from a reputable maker and found that +100% to -50% of rated rate variation could occur. We eventually put then on a spring tester and recorded the deflected height for the desired force and string tagged the set so the assembler could match the height. This will be a better result than using torque.

I expect you have some metal cover to distribute the screw loads to the gasket, so the spring rate of the gasket will depend on deflection of the cover and the way it distributes load to the gasket. Also, use a hardened washer under the edge of the Belleville washer to keep the edge of the Belleville washer from digging in.

 

[IRstuff]

The spring equation would seem to imply otherwise, to me. F = -kx, so if the gasket crept, the Bellevilles would have less compression, so less restoring force?

The attached seems interesting, since it implies that certain forms of the washer are capable of retrograde spring constants, i.e., increasing force with decreasing compression over a limited range.

https://www.bellevillesprings.com/disc-spring-characteristics/

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[506818]

3DDave:
Did not know about the significant variation in Bellevilles. In this case, the quantities are too large to measure each one. They will be supported on flat metal washers, but the base flange is PVC. So that won't transfer the loads as much as metal flanges will, but should be sufficient gasket compression to seal (otherwise PVC flanges wouldn't be a thing).

IRstuff:
That is the intention; gasket creeps and the Bellevilles decompress to occupy the gap. The Bellevilles we use have a ratio of 1.2 so should be fine on that graph. I'd also heard about the deflection percentage so good to see an actual source for it!

 

[GBTorpenhow]

Therefore, if the torque>gasket torque required AND deflection of Belleville>gasket creep, it should work as intended?

I would say rather "AND deflection of Belleville>>>gasket creep" because if the Belleville deflection is on the same order of magnitude as the gasket creep the creep deflection could significantly unload the Belleville. I.e., make sure that the Belleville load in the the deflected (post creep) state is enough to keep the gasket compressed.

Also don't forget the bolt itself which acts as a very stiff spring in parallel with the Belleville spring (when its not flat) which may need to be considered depending on the magnitude of the creep deflection WRT to bolt length, if the initial bolt load is larger than the Belleville flat load, etc.

 

[IRstuff]

The Bellevilles we use have a ratio of 1.2 so should be fine on that graph.

I think you missed my point; you want to be closer to 2.0, since you want to maintain nearly- constant compression force, regardless of the deflection; so also closer to the right side of the curve, ideally, since that would also increase the compression force if the gasket creeps

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529 Entire Forum list

http://www.eng-tips.com/forumlist.cfm

 

[bcd]

You should not compress disc springs beyond 70 to 80 % of their height. If the springs are scragged, then you know they can be compressed to 100% without breaking. However, the load vs. deflection curve increases significantly once you go above 70 to 80% compression. Above this range, the load vs. deflection equations are not accurate.

 

[r6155]

I have never seen a belleville washer on pipe flanges.
And using torque........ridiculous.
What are you trying to invent?

Regards

 

[GBTorpenhow]

These are used regularly in cryogenic piping - Massachusetts state laws even mandate spring washers on LNG piping flanged connections.

The cone height to thickness ratios for Belleville spring washers designed for flanged connections are usually quite low, giving a very linear spring rate as shown in the above diagram. To get significant deflection they may need to be stacked in series, but this lets them be compressed to flat without damage.