Check Valve Leakage For PSV Sizing.

[skoutso]

Hi you all.

Currently there is a check valve in our plant that separates a low (600#) and a high pressure (900#) system. The LP side must be protected for check valve leakage, failure to close etc. The API leakage rates (for valve rating purposes) are too low and to my consideration do not apply for PSV sizing purposes. Any rule of thub or industry practice, standard recommendation etc for the calculation of the PSV for this service is welcome.

Thanks in advance.

 

[Latexman]

Fail wide open.

Good luck,
Latexman

 

[SeanB]

Just in case you need a second opinion, I agree with Latexman, you have to assume that the check has failed completely.

 

[CJKruger]

skoutso,

I seem to recall that I used a leakage rate of about 10% in the past. But, if reverse flow was a real concern, I used 2 check valves in series. Especially if the lower pressure system has a lower flange rating.

But I think the industry is changing and becoming more conservative (in line with what Latexman said). Recently I saw automated reverse flow protection on high pressure process units.

 

[EGT01]

Skoutso,

I'm not sure to what "API leakage rate" you are referring but have you looked at the latest edition of API RP-521 (5th edition, January 2007)? Unlike the previous editions, it now has a more complete discussion about the subject of check valve malfunction and offers some guidance for relief system sizing.

 

[SeanB]

CJKruger,
The guidelines I follow is with one check valve you assume it has failed completely. If you have two check valves in series you assume the first one has failed and the second one leaks at a rate of 10% of the forward flow.

 

[skoutso]

Thank you all.

EGT01.
I refer to API-598 that defines the leakage rates for metal-to-metal seated "healthy" check valves. I'll try to get a copy of the 7th RP-521 Ed. Thanks for the info.

SeanB-Latexman.
Check valves by construction "fail" in the closed position. "Fail" condition refers in this case to the loss of the actuating medium i.e. the forward flow. Since there isn't any control system that would drive the valve to the "wide open" position (control system failure), the assumption of having a "wide open valve" does not sound credible to me.

CJKruger.
I used to install two check valves in series and assume leakage though a hole with a diameter of 15% of the nominal line diameter (~2.5% open area). That results in a leakage rate of ~1 CFM per inch and 100 psi differential, a guideline that can still be found in some engineering companies procedures.
Yet the point is that the valve could theoretically fail anywhere between 0% and 100% open.

I keep searching and I'll let you know of any valuable info.

 

[SeanB]

Skoutso,

You are entitled to your opinion. But as a Process Design Engineer, who has worked on many large projects for different oil companies, I can tell you that we never take credit for singe check valve to prevent back flow from a high pressure source to a lower pressure. You can do want you want - but it would never pass a good Hazop.

 

[skoutso]

SeanB,

I agree. I always use two of them.

 

[Latexman]

Check valves by construction "fail" in the closed position. "Fail" condition refers in this case to the loss of the actuating medium i.e. the forward flow. Since there isn't any control system that would drive the valve to the "wide open" position (control system failure), the assumption of having a "wide open valve" does not sound credible to me.

I've seen corrosion, deposits, and polymerization cause check valves (yes, more than one) to seize in the open position. I pulled one old check valve that had no internals. I guess it's a good thing I'm not your reviewer!

Ditto on what SeanB said.

How are you going to check that your check valves have not failed open? Also, how are you going to check that one of your two check valves has not failed open, thus reducing your back flow protection to one check valve and, possibly, to an unacceptable level?

Good luck,
Latexman

 

[pleckner]

I'm going to agree with Latexman and SeanB.

But more interestingly, the latest version of API 521 is no longer listed as a Recommended Practice.

The official title of the document is:

Pressure-relieving and Depressuring Systems
ANSI/API STANDARD 521
FIFTH EDITION, JANUARY 2007

ISO 23251 (Identical), Petroleum and natural gas industries—Pressure-relieving and depressuring systems

Note that not only is it an ANSI/API Standard but It is also an ISO Standard. Depending on how you want to interpret this, my feeling is that the document now carries more weight than it ever did in the past. Meaning, it is more of a "this is how you should be doing it and you better have a good reason why not" rather than just a guideline.

 

[sshep]

The guidance on using two check valves is sometimes expanded to include the language "of different design" so as to reduce the chance of common failure mode.

For example if salt or gums can cause the flapper of one lift check to stick open, then including two may not significantly reduce the chance of backflow. This reasoning applies to corrosion related failure, etc.

This is my experience anyway, the experience of others may be different.

best wishes,
sshep

 

[CJKruger]

SeanB and Skoutso,
Yes, I think you may be correct in that I used 10% for two check valves in series. I couldn't quickly lay my hand on an actual design, so I'm not sure.

sshep,
Yes, we also try to (but not always) use "different designs" to reduce common mode failure.

pleckner,
Thanks, I did not know there was a 2007 editions.