How to analyse a swing check dished disc strength during back flow. 2

[D4VE]

Hi All

Long time reader, first time poster! I'm hoping I can dip into this huge knowledge base for some answers to a problem we cant solve.

A bit of background first for context;
I'm a design engineer for a valve manufacturer. We are currently applying for the API monogram for our range of butterfly, ball, and swing check valves. All our current products were designed way back before I was even thought of and they have not changed much since, a lot of key staff have come and gone and various corporate groups have owned us along the way. This has left our design data in a bit of a state to put it mildly. My job at the minute is to re-write all the design reports and calculations for the aforementioned valves in preparation for the audit in a couple of weeks.

So to my problem. I am having trouble analysing the disc for swing check valves during back flow. For smaller sizes, the disc is (more or less) flat and treating them as a flat circular simply supported plate, as per Roark, is fine. However, from 6" to 10" the disc is dished/ellipsoidal in shape and analysing as a flat plate does not work. We know they do work, we've been making them for decades! I looked at treating it as a pressure vessel head, but this gave an unusually low stress I don't trust (it was 60% lower than the quick FEA I did). I think it's due to the fact that this assumes the outer circumference to be fixed (as in welded the the vessel body) and not simply supported on the seat as is the case in a valve. We cannot find any legacy data on our servers and I've searched the internet with no luck.

Please can anybody help with this analysis of what is essentially a dished/ellipsoidal circular plate, simply supported around its outer edge, with a pressure/UDL on the convex side, class 150?

Thanks in advance for your time, he says hopefully...

 

[jmec87]

Hi D4VE,

Roark's 7th Edition Table 13.1 Case 4a is one approximation that you could consider, but it is also likely imperfect. I doubt that there is a better Roark-style equation available for your flapper geometry.

Seeing as you already know from experience that these designs work and this is mainly documentation for the audit, I recommend checking what the applicable API design requirements are for the clapper. You might only have to document your engineering practices and acceptance criteria for the design (this would be the case for an API 6A swing check valve disc), in which case an approximate calculation and/or a quick FEA could be enough.

 

[D4VE]

Hi jmec87

Many thanks for your reply.

I have considered that case which you suggested and it gives a very low stress. Looks like your 2nd suggestion may be the way to go, which is unfortunate as I would really like to find out how analyse it properly.

 

[LittleInch]

Maybe go looking for how they analyse manhole covers (or whatever they are called - submarine hatch?) on submarines? They always seemed dished to me.





Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[D4VE]

LittleInch - good thinking! I'll see what I can find out.

 

[SPDL310]

I believe there is a section in the thin disk section of Roark's that talks about using an average thickness of a plate to analyze a plate of varying thickness, which may give you a better result. The API Auditor is there to see API design limits on allowable stress are followed. They are not an AI, will likely not be a trained engineer. API wants to se that calculations have been preformed, and input values for pressure, temperature, ect. are consistent with the design requirements. A complete FEA report for your disk would satisfy the API requirements.

 

[D4VE]

All,

Thanks for taking time out to help me with your suggestions.

I think I have found an acceptable solution. As I was looking at the bonnets for these valves, treating them as dished covers as per ASME VIII Mandatory Appendix 1-6, it dawned on me that maybe I could use this for the disc but with an external pressure. Treating it as a formed head per UG-32/UG-33 gives some results that tie in with what I was expecting from the FEA. I also found a research paper analysing a hatch door from LittleInch's suggestion that uses the same ASME equation. I feel a bit daft not doing this from the start as it seems like the pretty obvious thing to check with hindsight.

 

[jmec87]

D4VE, thanks for the update. Sounds like you found a good solution.