Frequent PSV internal wear and leakage

[eic13]

Hi all,

We have two PSV in seal oil service that are subject to failure/wear. The system in question has been operating since mid-1980’s and is original kit.
The two PSV’s are mounted on the discharge of two seal oil pumps; one PSV on the duty pump, one on the standby pump.
Reviewing the inspection reports, the PSV on the standby pump has been ‘leaking’ at every inspection since 1998. Inspections vary in frequency but are in the vicinity of once every 5 years.
The ‘standby PSV’ reports state phrases such as “seats badly worn/hammered/dented; machined. Recommend new plug and cage. Seats and disc holder in poor condition.”

The ‘duty PSV’ is not as bad. 2004 and earlier reports seem satisfactory. 2007 onwards note leakage with only hand lap of nozzle and disc required. Only 2014 has mentioned the seats in poor condition.

My understanding is that the standby pump is used infrequently. If it is relevant, the duty pump is turbine driven; the standby is electric driven. The system operates 24/7 and is only stopped for maintenance/turnarounds or if the plant trips.

The feed into the seal oil pumps (and hence into the PSV’s) is already filtered a little further upstream, so that would suggest against damage from media/flow (?).

When I’ve visited on site (starting this year) the PSV’s don’t appear to be chattering or making any other unusual noises.
The PSV inlet piping is warm to touch as expected; the outlet significantly cooler (still slightly warm but presumably from conduction/ambient conditions). This suggests the PSV’s are currently seated, however a leak may be so small that it is not enough to heat up the discharge.

If it is relevant, the pump discharge pressure is normally around 103 bar; PSV setpoint is around 115 bar.

PSV’s are identical. The model number suggests ‘other materials’ but according to inspection reports the internals are SS with body being CS.
They are a spring operated PSV (no pilot).
PSV discharge is into a pipe which dumps into the top of a reservoir tank (open to atmosphere). Each PSV has a separate outlet pipe (no combined header).

The issue is obviously the unexplained wear and frequent leaking of the PSV’s and that they may not function as a pressure relieving/safety device in an overpressure event.
Next year when the PSV’s come out for inspection there is likely to be replacement parts needed. However I’m now told the valves are obsolete (no spares available).

I’m looking into current models however I would like to solve the current issues prior to changing the installation.

Has anyone experienced anything similar with PSVs?

Thanks,

 

[georgeverghese]

A conventional PSV will begin to simmer at as low as 90% of set pressure per the API 520. So MAX operating pressure should preferably be less than 103barg. You havent said what the max operating pressure is.

If max op pressure exceeds 90% of set pressure, then would suggest switching out to a pilot operated type, which typically can operate at up to 95% of set pressure at least.

Also check for possible contamination of the seal oil with corrosive elements that may be in the process stream in the shaft seals of the process gas compressor.

 

[eic13]

georgeverghese, thanks for the suggestion.
Max operating pressure can be as high as 104 barg. Typically it sits around the 102-104 range.
I have been looking at the possibility of reducing the operating pressure of the pumps, in order to move away from the PSV setpoint. However one of our sister plants has a similar setup and they operate much closer to the RV setpoint, with no reported issues. They have conventional PSV's also.

The gas compressor is physically quite far away from the seal oil pump PSV's, so I would have thought it unlikely for any contamination back towards the seal oil pumps? The seal oil being at higher pressure than the process stream would make this even more unlikely? We have filters downstream of the SO pumps; upstream of the compressors, so this would stop any back-contamination of solids. But i take it you are talking about corrosive elements entrained in the oil.

 

[georgeverghese]

The seal oil will come into contact with the process gas at the shaft seal in the old fashioned type oil seals - corrosive elements in the gas and other heavy hydrocarbons will dissolve into the seal oil, and these will be present in the seal oil tank. The rotating machinery engineer in your plant may be able to tell you how this happens.

In modern compressors and expansion turbines, we dont use these oil seals due to safety reasons - that has ben the case for some 20years now. Numerous cases of fires at seal oil tank exhausts in the past due to flammable gas in the seal oil tank vent lines. Would you have H2S in the process gas also? If so, were the PSVs' specified for operation with trace amounts of dissolved H2S and / or other mercaptans? What about mercury ?

Pls note there is a permissible 3% tolerance on calibrated set pressure, if I'm not mistaken. That would mean actual SP could be as low as 111.6barg, which could result in PSV simmer at as low a pressure as 100.4barg. Check the calibration records for these PSVs.

If I remember correctly, Anderson Greenwood, in their PSV catalogues, claim their conventional PSVs' can operate at up to 95% of SP - and other conventional PSV manufacturers' dont make this same claim in their
catalogues..

 

[LittleInch]

What type of pump is it?

Do the operators make any modifications that cause the supply to be turned off or restricted, even for a short time without turning pumps off?

also if the system is designed to work at 103/4 bar when warm, what happens on start-p when it is cold and hence higher viscosity = higher discharge pressure.

Might only be for a short time, but if on start up the valves go into a popping mode until it all warms up, then you have the issues of valves not re-seating.

Why a similar plant operates and works differently could be down to any number of factors - maybe the instrument tech adds a bit onto the set point to avoid the issue, maybe the system operates in a subtly different manner, different gas, different vendors PSVs, shorter line length
who knows.

for conventional PSVs though I agree with George, you're on the ragged edge here and inspections every 5 years is a long time to be able to see any cause.

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

 

[hacksaw]

Standby pumps are routinely operated to insure operation when needed, you need to examine how your system is operated and the controls to make sense of the valve damage causes.

 

[gerhardl]

Hello eic13,

You refer to two plants with 'identical' components, one working OK, the other one failing, giving wear and leakage. If the valves are equal, something else is different. The difference could be anything, from proper calibration to operation procedure, placement of valves and differences in layout, difference in temperature and exact composition of fluid etc. etc.

Damaged seats indicates one or several opening and sudden shutoffs and even hair-scratches on the seats can give worsening leakage over time.

Without proper local inspection it is impossible to give a cause, or possibly several in combination.

Why is relief pressure so near working pressure? Is it the pump itself that require this, or the selected relief piping and valve class?

My advice would be as given above: correct dimensioned new valves and best possible calibrations, recalculate or control necessary relief pressure. Possible new valve and piping class. As of present the relief pressure is too close to working pressure and will almost certain give problems. Also have a close look at layout, gas pockets could give problems and hammering.

Best: Use a pilot operated valve, possibly one with (at high cost!) pilot operating fluid sealed from process fluid. Soft closing pilot operated valves will also protect from wear if opened repeatedly.

Cheaper solutions to be selected if consequence of leaking PSVs is small.

Bursting discs with alarm when bursting in front of PSVs is possible but here hardly a good solution, both because of pressure differences, and the 24/7 requirement without stop for PSV maintenance.

 

[eic13]

Thanks for the responses everyone. Some good suggestions/leads to follow that I’ll have to follow up with the operators and other engineers.

Georgeverghese,

Thanks for the further info. Having had another look at the flow diagrams I see the contaminated seal oil pots drain back to the main oil reservoir. So as you said, any dissolved elements would be present in the main tank and recirculated through the system.
I’m not sure regarding presence of H2S, Hg, mercaptans; I will check this out further with process engineering.

LittleInch and Hacksaw,

Pumps are positive displacement (screw).
Good questions; I will discuss with Operations.

The PI immediately upstream of the standby PSV has data logging. Looking at the trends for the past couple of years 90% of the time the pressure is around 12 bar. There are spikes up to ~120 barg. And a few instances of operating at ~100 barg. The data is still ‘raw’; further clarification with Operations planned to see what is going on.

Gerhardl,

The plants are not identical; only similar. Apologies if I gave the impression they were exactly the same.

Good question regarding the working pressure; as an aside I am currently looking into the possibility to lower the operating pressure by 10 bar. However the compressor seals (iso-sleeve) may need ~100 barg in certain events e.g. settle-out/shutdown (I am clarifying with the vendor/rotating engineers what the min. requirements are).

Thanks for the ideas on new PSV/pressure protecting devices. Something I will look into once I the cause of the valve damage is narrowed down.

Gas pockets could be an issue. Thanks for pointing this out.

 

[georgeverghese]

Presume you're aware that seal oil pressure should be kept above max settleout pressure at all times.

 

[eic13]

georgeverghese,
I was unsure about this requirement, as the listed settle-out pressure is higher than the rated discharge pressure of the compressor. So theoretically how can this pressure be reached?
I asked the some of my rotating equipment engineers, but have yet to get any responses on how the settle-out pressure is calculated, and whether it is appropriate.

 

[georgeverghese]

Agreed, settleout pressure cannot be greater than discharge pressure in most cases. May have to query the source personnel for this value on how this was derived, and to get errors corrected. Obviously, one of the preconditions for max settle out pressure would be that suction pressure is at its highest normal operating prior to the settleout event.

In refrigeration compressors, initial settlout pressure would increase somewhat as the blocked in gas contents slowly warm up to ambient.

 

[LittleInch]

Ok,

So normal discharge pressure has changed a bit, was 103, now normally 12 bar?

I don't know loads about seal oil systems but can only imagine this is impacted by the pressure seen at the seal face. The 120 bar figure will be a full open relief event.

As these are PD pumps feeding this system the relief valves will be sized for full flow relief and located immeadiately D/S the pumps. It would be good to see where they are relative to the pumps and NRVs as it is not easy to see how the standby pump relief valve is seeing the pressure from the duty pump.

Why one and not the other - Just variance between valves and setting. might be worth thinking about an actuated isolation valve on this system to isolate the standby unit.

I'm curious as to how the design / relief pressure has been set - 115 barg seems a little odd.

Depending on your max operating pressure required, an actutated control valve would be feasible to return fluid to the tank at a pressure set below the relief valve setting. These valves are designed to throttle whereas relief valves aren't.

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

 

[eic13]

georgeverghese,

I am querying how the settle-out was derived, however since the units were designed in the early 1980's I am not too optimistic on getting a decent answer.

LittleInch,
Normal operating discharge pressure of the duty unit is 103 barg (maybe a little higher at the pump, as this reading is after the filters located downstream of the pumps).
Normal pressure of the standby unit is 12 barg at the pump discharge. This is just the suction header pressure going through the standby pump casing and to the NRV. The NRV prevents the standby pump seeing the pressure from duty pump (103 barg).
NRV is located less than a foot from the pump discharge flange (straight pipe). The PSV is about 3 feet from the pump discharge flange (2x 90 deg bends in between).
Re: 115 barg setpoint; the best theory i have is that it is 110% of system design pressure.

An actuated control valve is possible however it would add additional complexity to the plant, so probably be a last resort.

Also, 120 barg may be the limit of the pressure transmitter so actual pressure may have been higher (hopefully not, if the RV's worked!).

 

[georgeverghese]

Okay, if the settleout calcs are buried somewhere at the bottom of a dark and abandoned stairwell in your office, time to get a new calc done up.

 

[LittleInch]

Hmmm,, Apologies I missed the "upstream" part of your post, but equally you didn't tell us originally that there was an NRV between PSV and PI... Anyway, If your system design pressure is actually a Class 600 limit, which is what it sounds like, setting the relief valve above design pressure is not that common.

SO I don't think there are many places left here - your relief valves will be sized for full flow pressure relief so in "normal" operation could easily "pop" to relieve pressure or "simmer" when you get even 1 or 2 bar above your normal operating pressure.

Increase in viscosity or drop in temperature might make a difference also how they operate the system. Then your relief valves are working in a mode for which they are not really designed, i.e. semi continuous flow and will either not reseat properly or will wear due to the very high velocities present at the sealing face as the gap is small.

Sounds like time for a re-design to me or increase monitoring of the valves, once every 5 years is too long, especially for items over 30 years old.

Let us know how it goes or if more information becomes available.


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

 

[eic13]

Unfortunately settle-out calcs are not even in our office (I have checked); hopefully the OEM can provide details.

The PI on the standby pump is upstream of the both NRV and PSV. NRV is on a separate branch to the PSV; they are in parallel not series.
Sorry it's a bit hard to sufficiently explain the system, without the drawings. Thanks for the assistance in the first instance.
PSV inlet flange is 1500#.
Initially i was surprised to see a PSV setting above the design pressure. But from what i gather from others it is not uncommon. My guess regarding the original design philosophy is they covered this under the B31.3 'allowances for pressure variations' clause.

Thanks for the heads up; it sounds like it is not uncommon for PSV's to simmer and not reseat if there are variances in the pressure, temperature, viscosity etc.
I thought specifying a hardface seat/disc material may reduce wear on the surfaces.
Given the issues experienced with the PSV's i will be recommending more frequent inspections.

Yes i will post any developments, however this may take some time (think months away).

Again thanks for the insights.

 

[georgeverghese]

In the reply to LI, the PI is downstream of the filters, while the PSV is upstream. Do you know what dp is incurred at the filters? So the pressure at the PSV locarion would be max 104barg + filter dp.

The Anderson Greenwood catalogue claims their conventional soft seated PSVs' can operate at up to 95% of SP, while the metal seated ones can used at up to 90% of SP only. You may want to call up these people.

Good luck.

 

[eic13]

There are numerous PI's on the system, but only a couple of them can log data. One being downstream of the filters, another being upstream of the standby pump PSV.
The filters typically operate with a DP of 0.3-0.7 bar according to data logging.

Thanks for the advice. We have been dealing with AGW's supplier recently for this job, so will make the enquiry and see what they can offer.