Relief valve question 1

[safcon]

A question has come up in a process hazards analysis (PHA) that I'm conducting for a client. I'm confident that I know the answer to this, but I wanted to run it by the forum for confirmation.

A relief valve (standard spring loaded) is fitted to a vessel. The MAWP of the vessel is 6,500 psig. The relief valve is set at 6,500 psig and bench tested at that pressure (with no back pressure on the relief valve). When in service, the relief valve must discharge into a sealed system running a constant 500 psig. I contend that the spring pressure and the back pressure are additive, and that the relief valve, when in service, won't lift until the vessel reaches at least 7,000 psig (the 6,500 psig spring load plus the 500 psig back pressure).

I further believe that the 7,000 psig lift pressure may actually be a conservative estimate. Because the relief valve plug is cone shaped (with the small end of the cone facing the vessel), more surface area of the plug is exposed to the 500 psig of back pressure. Assuming a 6" diameter flow port with a 3" taper to the plug diameter, the larger, discharge side of the plug will have a 9" diameter, and a surface area of 63.6 square inches. Opposing that force will be the 6" diameter of the process side of the plug with a surface area of 28.3 square inches. The difference in area ratios will mean that the force on the larger surface will be multiplied by 2.25. Since the larger (9" diameter) surface sees 500 psig, there will actually be an additional 1,125 psig (2.25 times 500 psig) required on the smaller (6" diameter) surface of the cone to reach stasis.

If this is correct, the 6,500 psig of spring pressure will be added to the equivalent 1,125 psig of back pressure giving an actual relief valve lift pressure of 7,625 psig. Is this correct?

 

[TD2K]

Is this a conventional relief valve or some sort of speciality valve?

All the corrections I've seen for conventional relief valves is that any constant backpressure increases the effective set pressure on a 1:1 basis. So, if the vessel MAWP is 6500 psig and the backpressure is a constant 500 psig, the PSV should be set for 6000 psig. In service, its effective set pressure is then 6000 + 500 = 6500 psig.

The simple way to check on the area factor you have raised is to check the vendor's documentation and see what it says. If you still have questions, I would talk to their technical support people.

 

[safcon]

This is a conventional valve on the discharge of a reactor feed pump - liquid service. Since the reactor frequently has pluggage problems on its discharge (stopping all pump flow), redundant relief valves to prevent overpressure may be appropriate here.

Vendor documentation may not be available since this is an old process and many of the pumps and valves are made by companies no longer in business and some original documentation is missing. Valve recalculation will surely be in order..

 

[MortenA]

I agree with your findings: The PSV will lift at a pressure of 7000 psig when there is a constant backpressure downstream of 500 psig. I dont think the port size will matter. The back pressure "rests" on the valve seat. The seat may have some difference in size "upstream/downstream" but i believe that the manufactors "even this out" (the stem is not subjected to the higher pressure).

Best regards

Morten

 

[quark]

What you said is correct. The lifting force should always be morethan the spring force + the force due to back pressure.

What you should calculate is the surface of the cone(if the disc is conical) and then use the vertical component of the net force. If fully agree with MortenA that manufacturers may take care about it. How about a word with them?

Regards,

 

[safcon]

Thanks to all for the valuable feedback. For the time being, the PHA team will make a recommendation that the relief valves be reevaluated. The relief valve manufacturer will also be contacted.

 

[sailoday28]

safcon STATES Since the reactor frequently has pluggage problems on its discharge (stopping all pump flow), redundant relief valves to prevent overpressure may be appropriate here.

Are safety/relief valves the solution on an event that frequently happens?
Perhaps control valves, filter/strainers,etc should be looked at.

 

[safcon]

The process itself is pluggage prone. At the pressures that the process operates at, filters/strainers are probably not feasible anyway.

 

[dcasto]

A pilot operated PSV will handle this situtation.

 

[safcon]

The solution selected by the PHA team was to recommend controls on the reactants upstream of the reactor so that the supply pressures (individually & collectively) couldn't exceed the MAWP of the reactor. This solution avoids the pluggage problems that only occur once the reactants combine.

As to the answers posted, I've been told (in the mechanical eng. forum) that the relief valve's ACTUAL lift pressure as given in the original post would be exactly 7,000 psig. Why do the differences in pressures over different areas not apply? It seems to me that the spring pressure, the "backpressure," AND the differences in areas would all apply. To date, quark is the only one who has agreed with this.

 

[EGT01]

Safcon,

As mentioned in earlier posts, I believe your confusion is about the area over which the difference in pressure actually exists and it is only the area of the seat that matters.

[tt]
___________
\ /
Seat -----\ /-----
\_____/

[/tt]
You may want to review API RP-520 Part I in the section where they discuss the difference in balanced bellows valves versus conventional valves. Though their discussion covers the forces acting on a relief valve disk as opposed to a plug, I think the principles would be the same.

 

[safcon]

I don't have access to API RP-520. Might you summarize? Thanks.

 

[StoneCold]

Safcon
Do you have a way to bench test the relief valve. I think you are right about the area influence on the relief pressure. But a bench test under similar conditions or even at a much reduced pressure would prove out the situation.

Good luck and Kudo's to you for even asking what is going on. Most people just take the easy way out.

StoneCold

 

[EGT01]

Safcon,

Take a look at this link...

http://www.spiraxsarco.com/learn/default.asp?redirect=html/9_2_01.htm

It gives basically the same information you can find in API RP-520 about the difference between the forces acting on a conventional safety relief valve and balanced safety relief valve.

 

[JimCasey]

1. Yes, a conventional spring-loaded safety relief valve will lift as setpoint+backpressure. In this case, 7000 psi.

2. Nobody has mentioned that rated flow is at set pressure + 10%.

3. If the backpressure is CONSTANT 500 psi, then you can reset the valve to 6000 psi so that in actual operation it operates at 6500. A new spring may be required, but any ASME certified valve repair facility will be able to deal with this.

4. If the backpressure is variable, the valve must be referenced to atmospheric. A Bellows-balanced valve is used for this. The bellows is constructed to have the same effective area as the disc. and the back side of the bellows is welded to the disc and vanted to atmospheric pressure. Backpressure then has no effect upon setpoint but there may be an effect upon capacity.

5. A piloted valve also might be useful for variable backpressure service, but since this is on a liquid that has a tendency to solidify, the pilot would probably have plugging problems.

6. A NONFRAGMENTING rupture disc under the SRV is allowed by the code. A rupture disc also functions because of the differential pressure across the disc, so backpressure would superimpose upon the setpoint. A reverse-buckling disc such as the Continental Ultrx is nonfragmenting and can withstand backpressure..Also the Ultrx has been tested with many SRVs so the combination capacity factor is usually very close to 1.0 instead of the code mandated .62 for an untested combination. (I have no affiliation with CDC) The code requires a telltale connection between the rupture disc and the SRV to monitor if the rupture disc has blown. If you can vent this connection to atmospheric and use a small ball-check to seal it if the Rupture disc blows, then you will keep the liquid out of the valve until the valve has to perform its emergency function.

 

[safcon]

The plot thickens.

The relief valve in question is situated on the discharge line of a pump. The discharge relief valve vents into the pump's suction line. The suction line is normally at 250 psig.

The pump's discharge normally runs 5,500 psig. The pump's discharge relief valve is set at 6,500 psig (bench tested to atmospheric). The suction line has a relief valve that opens at 500 psig (also bench tested to atmospheric). That suction relief valve goes to a vent stack through a long vent line.

The lift pressure of the discharge relief valve would normally be 6,750 psig plus the differential area effect. The suction header pressure, however, immediately rises to 500 psig, and begins relieving through the suction relief valve to the vent. At that point, the discharge relief sees 6,500 psig of spring pressure plus 500 psig of suction pressure plus any flowing resistance (probably in excess of 10% per valve).

To make things worse, four pumps share common discharge and suction headers, and all relief valves eventually go to the same flare KO pot, which has seen hydraulic fill conditions in the past. The KO pot is rated at 150 psig, and has no local overpressure protection. The KO pot then discharges into a second (aluminum!) flare pot rated at 47 psig, again, with no local overpressure protection. Then the vent line goes through a single 16" CSO manual block valve. The line then splits into two parallel 10" headers, each with their own CSO manual block valve. Then the system finally reaches an atmospheric vent.

Since it is a liquified gas being pumped (NH3), the temperature can drop significantly when the liquid state flashes to vapor. The PHA team was particularly concerned with the aluminum vessel because of aluminum's tendency to stress crack and fail catastrophically.

Thoughts?

 

[MortenA]

EGT01: Thank you for your reference to sprirax's webpage about PSV's. Pretty good! Figure 9.2.2 (and text) illustrates exactly what i tried to describe in a previous post concerning liftigt pressure as a function of backpressure.

Another good source (often referred to in these pages) is:

http://www.tycovalves-na.com/ld/CROMC-0296-US.pdf

Best regards

Morten