6 years old problem of PSV leaking at a petrochemical plant 1

[Mous1747]

this has been a long standing problem of PSV leaking, the PSV is installed directly on a pump housing at discharge, the PSV has been leaking and get replaced every month or two on an average. the PSV never lifted, it just leaks. there was never a detailed RCA done on this but the recommendation was always to relocate the PSV on the piping downstream from the pump.

the set pressure is 3900 psi with backpressure of 300 psi (CDTP 3600 psi) the pump will trip at 3650
the PSV is metal seated with SS trim

the PSV manufacturer say it's overpressure or pressure pulsation because he is seeing severe damage to the disc and nozzle seats, but that can't be true if the valve is not even lifting?
also it's not chemical compatibility issue because SS should be okay with ammonia and will not corrode in a month.

it's a reciprocating plunger pump so maybe the pressure pulsation generated can cause this? especially that the PSV installed directly on the pump housing can cause misalignment inside the valve?

thank you all in advance, your inputs will be greatly appreciated.

 

[mk3223]

Agreed to conduct a RCA for the incident before next PSV replacement. Additional to the possible issue of the pump pressure pulsation, IMO, it could be the type of PSV valve issue.

 

[georgeverghese]

Pump trip pressure is 93.5% of set pressure, so what type of PSV is this? - would suggest only a pilot operated PSV would be acceptable for this service. However, actual peak transient pressure in the pump casing may be somewhat higher than 3650psig and too fast for the trip pressure transmitter to pick up. Presume the trip PT is somewhere further down the line.
Even with a pilot op PSV, due to the peak pulsations, would suggest, that together with a well operated pulsation dampener and a triplex pump, the PSV set pressure be at least 10% higher than trip setting. Pulsations can be quite bad on simplex and duplex pumps, and would be more manageable on triplex double acting pumps.

 

[don1980]

With an incompressible fluid and a PD pump, leakage through a metal seated PSV doesn't surprise me. The seat damage that you're seeing is a clear indication that the force applied to the PSV is causing "simmer" which is the quivering of the stem that occurs just before PSV's lift (at about 92-95% set pressure). The PSV stem is in fact moving (lifting). It's not enough to describe as "PSV opening", but it's movement nevertheless. And it's likely that that simmering action occurs with each pump stroke. That's a clear and obvious explanation for the mechanical damage that you're seeing (seat damage). Assuming that you can't raise the set pressure, the best solution I can think of is to install a pulsation dampener between the pump and the PSV. I think that will solve this problem. There's a a lot of suppliers. Here's a link to one that I've used.

https://blacoh.com/sentry1.aspx?_vs...xHwivAGjwC_rQCAUyLxbIAxm5XmHqcJcaAn10EALw_wcB

 

[Mous1747]

actually the valve is designed for constant backpressure of 300 psi at outlet, so the actual set pressure CDTP is 3900 - 300 = 3600 psi

now, the pump trip at 3650 psi assuming there's no delay and happens instantly, but that will lead to the seat in the PSV to keep bounding( maybe that explains the damage we are seeing inside the PSV)

the pump manufacturer recommends to keep the valve on the pump body and not relocate it to the discharge pipe directly after the pump, they also said this 3-piston pump has +/-15 % pressure pulsation

i believe I need to lower the trip on the pump below the 3600 set pressure of the PSV? 10% below that?

but that still not taking in consideration the 15% pressure fluctuation, will the dampeners or resonators completely eliminate that 15% ?

 

[Mous1747]

I believe the pulsation dampeners need to be installed in both suction and discharge, correct? how much of the 15% pressure pulsation will be smoothed out with these dampeners

@georgeverghese: yes it's the 3-piston pump however the PSV is the conventional type, so i agree with you i should consider different types like the balanced-bellows or pilot type.

i am asking the pump manufacturer on their recommendation on what should i set the trip pressure at, then i increase the PSV set pressure 10% above it, assuming the dampeners will completely eliminate the 15% pulsation.


@don1980: thank you, your explanation make so much sense

 

[Mous1747]

max discharge pressure = 3556
trip = 3560
operating pressure = 3200
pump pressure pulsation +/-15 % of the 3200 = 3680
current PSV set pressure = 3600
with each pump stroke a simmering effect will happen at 92-95% of set pressure = 3312 - 3420
the pump manufacture is recommending we set the PSV at 3200*1.15*1.1 = 4048 psi, but that's much higher than the max pressure of 3556
i don't want to rely on the trip to protect the pump
E&I guys claim there's no delay in the trip pressure transmitter but i doubt it, there might be couple seconds delay

 

[georgeverghese]

With a pilot opererated PSV, given the recommendation from the pump manufacturer, you can set the PSV at
(3200/0.93)*1.15/0.98 = 4035psig, since the PSV can hold pressure without simmer at up to 98-99% of set pressure. Set pressure is also not affected by backpressure on pilot op PSV.
You have trip PT and motor overcurrent relay to trip pump before PSV blows, which is the standard sequence for pump protection. Good that you have a triplex pump, presuming it is double acting. Check that N2 pressure on discharge pulsation dampener is approx 0.7*3200 = 2200-2300 psig when not in operation.

Actual relief pressure for conventional PSV is affected by variable backpressure, so if backpressure is 200psig, PSV will blow at 200+(3900-300) = 3800psig, and will simmer at 0.93*3800 = 3530psig.

 

[Mous1747]

I am not sure a im able to interpret this formula correctly (3200/0.93)*1.15/0.98 = 4035 psig
why are we dividing by .93 and .98 ?

since this is not a pressure vessel that means i am not governed by the 10% accumulation pressure rule?

 

[Mous1747]

i understand you divided the 3200 by .93 due to the pressure accumulation, basically multiplying by 1.07 instead of 110% to reach full left.

but what is the .98 ?

if i would use conventional PSV for this application and taking in account the 300 psi back pressure, then the set pressure 4035 and the CDTP will be 3735 psi ?

 

[georgeverghese]

Agreed, division by 0.93 is not required for pilot op PSV. Division by 0.98 is to account for max op pressure to be no more than 98% of set press for a pilot op PSV. So with pilot op PSV, set pres = 3200*1.15/0.98 = 3775psig.
With a conventional PSV, set pressure would be 300+(3200*1.15/0.93)=3957psig, and CDTP=3657psig.
Am assuming 3200psig is max normal op pressure. What is the significance of 3556psig - is this the max pressure as permitted by motor current or some other mechanical constraint ? Can you raise this higher?

 

[LittleInch]

Looking at all of this my only comments to add are these:

The +/- 15% pressure pulse from the pump vendor is not an exact number. If you added a high speed pressure transmitter and oscilloscope it would be very interesting to see exactly what the peak, millisecond pressure actually is. I very much doubt it never goes above 15%. However your PSV will react to even very short pulses of pressure, passing very small but high pressure drops of liquid.

Your trip setting of 3650. Where is this instrument located and does it have some sort of smoothing function to avoid tripping on pulsation?

Basically you're just running too close to the wind here in terms of pressure and set points with a standard PSV. Pilot would be much better, but I think even then you really need to sort out the pressure pulsations.

Is the 300 psi back pressure constant? If so then remember the actual set pressure as installed is actually 3900 psi, not 3600 as you seem to chop and change within this post.

If not then you will get more very short term minute jets of liquid which will just destroy your metal seals - it's like taking a water jet cutter to them time after time.

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

 

[Mous1747]

i just checked with operation , the actual trip is 3560 ( not 3650 as i stated before)
alarm at 3400
operating pressure 3200

so pulsation pressure generated is 3200 *1.15 = 3680

now the question becomes, why the pump is not tripping on pulsation?

the answers i am getting from your replies above are:
1. "the peak transient pressure in the pump casing happen too fast for the trip pressure transmitter to pick up. Presume the trip PT is somewhere further down the line".
2. "the instrument have some sort of smoothing function to avoid tripping on pulsation"
3. "the instrument location"
4. maybe the transmitter is located after the discharge resonator ?

are there any explanation on why the pump is not tripping or alarming on pulsation?

 

[Mous1747]

the pressure transmitters we are using are actually mechanical pressure switches - diaphragm type
and they are installed right on the pump

 

[LittleInch]

Yes, but how are they mounted and do they have a "snubber" or a length of small bore pipe involved? Both will smooth out any peaks.

What sort of accuracy is a mechanical type pressure switch?

They will have an inbuilt inertia which won't see very short duration pressure spikes.

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

 

[georgeverghese]

Maybe the pressure switch was not triggered because the conventional PSV was relieving at a lower pressure (PSV simmer pressure = 0.9*3900 = 3510psig < 3560psig), assuming simmer pressure is as low as 90% of requested set press.
When you change to pilot operated PSV, the PSHH may trigger if its setpoint is < 0.98*PSV set pressure.
So the question remains, can you raise this PSHH setting to be say 3700psig when you replace with pilot operated PSV?
When you spec out the pilot op PSV, also ask for set pressure tolerance to be +/- 1% of requested set pressure. Else you will have to raise the requested PSV set press higher.

 

[Mous1747]

in fact because the back pressure dropped to 200 psi instead of the 300 the valve was originally designed for , so this will reduce the actual set pressure to 200+(3900-300) = 3800 psig
and since this is a metal seated and has 90% seat tightness, so yes it's more likely to begin simmering at 90% of the 3800 = 3420
so that make sense why the pump is not tripping at 3560 psi

the other issue now, is this pump is set to alarm at 3400 , which is higher than the 3420 , is that possible?

another thing, if i change this conventional PSV to soft seated instead of metal seats, because the soft seats has seat tightness of 95% instead of 90% in metal seats
will the soft seats handle this much pressure or will it get destroyed? (the temp and chemical compatibility will not be an issue when using soft seats)

 

[LittleInch]

I think the issue here remains that you don't really know what the very short duration peak pressure being seen by the PSV actually is.

It may be that where the PSV is mounted means it sees a higher peak pressure than the Pressure switch or transmitter do for the millseconds it occurs. Difficult to say with any degree of accuracy.

I don't think soft seats are your answer though - you're just running too close to the margins with a reciprocating pump.

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

 

[georgeverghese]

The conventional PSV option is not a good option is this case, given the variable backpressure. The pilot op PSV set point is not affected by backpressure and has a much higher simmer point, so dont see any reason why you want to continue with conventional PSV; loss of revenue with ammonia loss, exposing your maintenance crew busy to increased health and safety risk with PSV removal/ re installation.

 

[Mous1747]

LittleInch:

can you clarify little more on this?

"If you added a high speed pressure transmitter and oscilloscope it would be very interesting to see exactly what the peak, millisecond pressure actually is"

how and where these instruments will be installed ? any specific brand or vendor you may know? can it be installed on the inlet piping of the PSV ?