11 replies to this topic

[chejosh]

Hello all,

I recently started a new job which requires me to do a lot of relief valve updates, and I haven’t done much with relief valves in about 8 years, so I’m a little rusty.

Right now I have several small relief valves in ammonia service on a pipeline protecting against thermal expansion.

We have had several difficulties with these valves leaking, so I was planning on adding a rupture disc to the valves to prevent leakage. As such, I was going back and redoing calculations to throw in a derating factor for the rupture discs. Plus the calculations are in some cases 20+ years old and not well documented. I also wanted to harmonize some of the set pressures because they seem to vary in a few cases for no visible reason. Less varying set pressure = less rupture discs to buy, after all.

One of the valves in question is in ammonia service. Set at 330psig. The person who originally calculated it did it two different ways:

1) Assumed liquid relief and used solar heating to calculate a liquid relief rate.

2) Assumed it would boil and used latent heat to calculate a relief rate from the solar heating. Then multiplied the vapor volume by a liquid density and still used the liquid relief area calculation. None of this makes sense to me.

Would it be all right in this case to assume liquid relief? Or is this two phase flow since the ammonia will vaporize once in the atmospheric pressure? Does it even matter since the required relief rate will be so small for thermal expansion anyway?

Thank you

[latexman]

What size lines, and how long are they?  We need to get a feel for the trapped volumes.

[chejosh]

Well, I went out and gathered info on 6 valves this afternoon. They were all 2” pipe (aside from one that had a 1.5” section due to a flow meter). I did not add up all the lengths yet, but five of the six were fairly short (around 20 ft or less). The one from the pumps to the reactor was 250-300 ft.

I’ve already left work for the day so I may be misremembering some details, but for the one I was working on, I think the length was around 17 ft and I got a surface area around 12 ft2. I think the required relief rate I got was about 0.2 gpm. And the required area was about 0.002 in2

[latexman]

I have assumed the anhydrous ammonia temperature in the pipes is about ambient temperature and not refrigerated.  Please correct me, if wrong.

 

I have never seen a rupture disk/safety valve combination used for pipe ambient hydraulic relief.  Anyone else?  Depending on the pipe operation, the frequency of activation can vary greatly.  Continuous operations are vulnerable before start-up and after shutdown.  Batch operations are vulnerable then too, but also between batches.

 

How do you know the PSVs suspected of leaks are not bona fide activations of the PSV for high pressure?  Pressures in pipes are seldom recorded, except at critical locations.

 

For this one leaking ammonia thermal relief valve (TRV), is the line pressure recorded, what’s the normal operating pressure, what’s the line MAWP, is the line blocked in somewhat frequently, is the TRV of high quality or cheap, does the TRV have a soft seat o-ring?

 

The lines mentioned are small, short lines like inside battery limit process lines.  These are similar to those that API 520 and 521 refers to when sizing calculations are not required:

 

“4.4.12.2 Sizing and Set Pressure (API-521)

The required relieving rate is not easy to determine. Since every application is for a relieving liquid, the required relieving rate is small; specifying an oversized device is, therefore, reasonable. A nominal diameter (DN) 20 × DN 25 (NPS 3/4 × NPS 1) relief valve is commonly used. If there is reason to believe that this size is not adequate, the procedure in 4.4.12.3 can be applied. If the liquid being relieved is expected to flash or form solids while it passes through the relieving device, the procedure in 4.9.2 is recommended.”

 

“6.3.2 Thermal Relief Valves (API-520 Part II)

The discharge piping from thermal relief valves designed solely to protect against liquid hydraulic expansion due to ambient heating (including solar radiation) typically does not need to be sized to meet the built-up backpressure limits provided in API 520 Part I and as discussed in 6.3.1. The reason for this is that the rated capacity of these PRVs is typically larger by an order of magnitude (>10 times) than the required relief rate and the flow in the discharge line never reaches a steady state flow at the rated capacity. See 7.3.8 for additional discussion and cautions with thermal relief valves.”

 

Please check this out thoroughly with ammonia in mind!  It’s been 15 years since I worked with anhydrous ammonia.

 

One other fact I've learned over the years.  Control valves usually leak enough such that TRV's are not required.  Just be careful if you specify Leakage Class IV or V CV's!

Edited by latexman, 23 March 2021 - 10:02 AM.

[chejosh]

Thank you for your replies.

The ammonia is ambient to slightly above ambient (~100F)

The reason I suspect leaking versus activations is that the process is a continuous process. The valve only leak when we’re operating and no valves are shut. For some of the cases, there are nearby gauges that typically read around 210 psig, the valve with the lowest set point is at 275 psig.

The valves are soft seat. They’re threaded valves either made by Anderson-Greenwood or consolidated.

There are other companies with adjoining property lines that have just recently became occupied, and management is worried about the ammonia odor from leaking valves.

I have only started here in the past few months, but from what I have seen the standard operating procedure when a relief valve leaks is to remove it and cap it, send the valve off, and replace it when convenient. I want to get this practice to stop, and my thought was the valves wouldn’t leak if there was a rupture disc keeping the fluid off the valve. You are right though now that I think about it. Having an rd/psv combo in a thermal relief scenario seems a bit odd.

Edited by chejosh, 23 March 2021 - 10:06 AM.

[latexman]

Is the flow pulsating? Some PD pumps pulsate pretty bad. Any valves in line or in branches opening and shutting and propagating pressure waves?

Edited by latexman, 23 March 2021 - 10:56 AM.

[chejosh]

It varies.

Some of the valves are on the suction side of the pump, some on the discharge. Some are after the discharge of PD pumps, some are after centrifugal pumps.

Like I said, I’m pretty new here still, so I am not completely sure about valves. My understanding is once the process is up and running, there isn’t a whole lot of opening/closing of valves, perhaps just some modulating.

[latexman]

The reason I suspect leaking versus activations is that the process is a continuous process. The valve only leak when we’re operating and no valves are shut. For some of the cases, there are nearby gauges that typically read around 210 psig, the valve with the lowest set point is at 275 psig.

The valves are soft seat. They’re threaded valves either made by Anderson-Greenwood or consolidated.

There are other companies with adjoining property lines that have just recently became occupied, and management is worried about the ammonia odor from leaking valves.

 

A PSV will start to leak at about 90% of set pressure; an o-ring improves this to about 95%.  If you normally run at 210 psig, but it's a pulsating 210 psig, you may have transients up to the point it leaks.  Does the pressure gauge pulsate around 210 psig a lot, a little, or is it  rock  steady?    In liquid service the approximate reseating pressure of a PSV is 70% of the set pressure.  275 x 0.7 = 192 psig.  You may have a reseating issue (210 psig > 192 psig) if the PSV ever gets a dynamic pressure wave > 275 psig and opens momentarily, like pulsations from a PD pump, a valve closing too fast, a check valve shuts during a momentary flow stoppage, etc.

 

Good for the o-rings!  What's the MOC?

 

Yes, ammonia odor is strong.  That's bad, and good.  It warns you early and fast!

[chejosh]

That’s interesting that the reseating pressure is 70%. I thought it was 90%. From what I remember of the gauges I have looked at, they pulse a little, not wildly.

The soft seat MOC is epdm, I believe.

The pipe is 2” sch40 with 150# flanges. I think the MAWP of which is something like 720psig. If the issue with leaking is pulsing to near the set point and then being unable to completely reseat, I don’t see any reason I could increase the set point. I have no knowledge of why the current set points were chosen.

[latexman]

For a gas/vapor the reseat is about 93% of set pressure.

720 psig sounds high for 150 lb class flanges.

Edited by latexman, 23 March 2021 - 06:04 PM.

[Pilesar]

The class 150 flanges are limiting according to B16.5 flange specs. I doubt you can justify a PSV set pressure much higher than 275 psi with those flanges, especially if the temperature ever gets above 100 F..

[chejosh]

You’re right. I was thinking class 300 flanges. Sorry. The pipe spec has a pressure limit of 720psig up to 100F.

I have a few valves that are on 150#, 300#, and 600# flanges. I think all the relief valves are threaded though.

Edited by chejosh, 24 March 2021 - 09:07 AM.