Compressor Reverse Flow Relief Sizing 1

[121202]

I am interested in hearing about the opinions of experienced process engineers regarding their experience on reverse flow cases in compressor installations. I am reviewing an oil and gas plant that has two parallel trains of compression. On the discharge side of each compressor there is a non slam check valve and a shutdown valve. When a compressor trips its shutdown valve closes isolating it from the parallel train. However, if the shutdown valve fails to close, then only the check valve is available to limit back flow from the operating parallel train. Flow leaking past the check valve could pressurise the suction side of compressor loop and could ultimately overpressure it. In my opinion it is therefore necessary to eliminate the relief case, by installing an additional dissimilar check valve on the discharge of each compressor or alternatively install a relief valve on the suction side of the compressor capable of accomodating the leakage rate past the check valve. I'd be interested in hearing about past experience with this type of situation.

If relief valves are installed to protect against reverse flow then the question then becomes what basis should be used for sizing? There is a great deal of uncertainty regarding leackage rates past check valves and determing them with any accuracy is difficult. Any experinece on sizing relief valves in these situations is very welcome.

 

[MortenA]

Are the two trains totally isolated and only connect in first and last stage - or are interstages also interconnected?

Best regards

Morten

 

[Samiran]

I hope a HAZOP was done to analyze this situation. What do the close-out reports say of this ? Think of the recycle valve - if that opens inadvertatently will it not overpressurize the suction ? I think a suction relief valve would be designed anyway whatever is the governing flow for the design. It could either be a leakage through check valve (typically 10% considered) or the sudden opening of recycle valve.

Samiran

 

[roker]

hi,

are the check valves "back pressure check valves"?

 

[121202]

Each train has three compression stages (LP,MP&HP). The trains are connected at the discharge of the LP and HP machines (there is a single Glycol Absorber between the LP and MP Compressors). There is no connection between the discharge of the MP Machines; they deliver directly to the suction coolers of the HP Compressors. At the discharge of each stage there is a check valve (non slam type) and a shutdown valve. The non slam check valves are mokveld type or equivalent and are relatively good quality. The system piping is fabricated in corrosion resistant alloy and therefore the system is likely to be relatively clean (i.e. corrosion products that could cause check valve malfunctions are not really an issue).

There is a scrubber on the suction side of each compressor. Each scrubber is fitted with a fire relief valve. The PSVs on the scrubbers can comfortably handle a flow equal to 10% of the design flowrate of the parallel train (each train can deliver 80000 kg/h and all scrubber PSVs can pass 35000 kg/h or more). Various upset and trip situations have been simulated using dynamic simulation to evaluate control responses. No overpressure situations were predicted on rapid opening of the compressor anti-surge valves, although as expected the suction pressure and electric motor load did increase.

Any additional comments most welcome

 

[brainstorming]

From experience with one compression train, I haven’t seen a relief valve considering backflow. Check valves are installed to avoid such scenario to happen otherwise u will end up with so many relief valves in pumps, compressors..etc
With respect to kickbacks they are designed for high pressure drop and no possible of overpressure in case of opening.
As I mentioned check valves would be enough for protecting from reverse flow and no risk of leakage if the system is not dirt or fouling service.
By assuming failure of check valve, relief valves installed in the train would cater such reverse flow.

Regards

 

[Montemayor]

121202 & all:

There is a process failure in this thread to identify the TYPE of compressor being discussed. If the subject compressor is a reciprocating type, this thread and its subject fall by the wayside. All reciprocating compressors inherently are dependent on "check valve" action in order to perform their work function. In this case, the subject compressor would have 6 levels of up stream "check valve" action that would have to be defeated before the upstream, 3 rd stage discharge gas could reach the 1st stage suction. This, of course, implies 3x double jeopardy and is not credible.

If we are talking about a centrifugal compressor, that is a totally different situation. I'm waiting for an answer to the practical question: "Was a HAZOP was done to analyze this situation. What do the close-out reports say of this?"

 

[121202]

The compressors are centrifugal. The HAZOP report did not recommend any additional protection from reverse flow other than the existing non slam check valve.

 

[EGT01]

121202,

Here's what API RP-521 has to say, they have a section that specifically addresses check valves. Under the section "Potentials for Overpressure"....

"The failure of a check valve to close must also be considered."

Though their example is given for a pump with potential to overpressure the suction vessel, I would think the same philosophy would apply for your compressor.

"A single check valve is usually considered acceptable unless a potential exists for backflow of high-pressure fluid to create pressures that exceed the test pressure of the
equipment. In these cases, one should consider providing a secondary device to minimize the potential for a reversal of flow. The device may be a nonreturn valve, a power-assisted check valve, a second conventional check valve, or similar equipment."

So it would seem there is industry practice that you could rely on to say your system is adequate but I would suggest you review the discussion yourself.

However, in general terms, my experience has been to not take credit for check valves in safety critical situations because they are usually not tight shutoff valves, some may require a significant differential to work properly, and they are not likely on a routine maintenance schedule to help ensure reliable operation.

In some cases (not a compressor) where a shutdown system was provided in addition to a check valve, I have seen the shutdown system installed as a high integrity system with redundancy in the controls and increased level of preventive maintenance. In those cases, the risk was considered too great to just rely on a check valve and routine shutdown system. Hopefully your HAZOP team has considered and documented the level of risk and its acceptability.

Can't say I have experience with the Mokveld brand of check valve but seems I've seen reference to them as being high quality as you indicated. In fact the Mokveld web site also has a good discussion about high intergrity control systems for overpressure protection.

http://www.mokveld.com/

In a way, I see a check valve in series with your shutdown system as a similar arrangement for protection. Maybe as an additional excercise you may want to apply the type of analysis used when determining the safety integrity level for a HIPPS to assess if your system meets the needed risk reduction.

 

[brainstorming]

I forget to mention SIL "high integrity system shut down" that protect the compressor from high pressure in case of block discharge case that much severe than backflow.
Then SIL and check valve are the only recommended solution for your case !

 

[121202]

EGT01,

Thanks for your concise and professional input, it is appreciated. I am familiar with the paragraph in API 521 that you are referring to.

Just to clarify, in my particular situation for backflow to occur there needs to be a failure of both the shutdown valve (a fail closed ball valve) and the non slam check valve. The question is, does the shutdown valve constitute the "secondary protection" referred to in API 521?

 

[Montemayor]

121202:

The paragraph that EGT01 correctly cites is very specific: "The device may be a nonreturn valve, a power-assisted check valve, a second conventional check valve, or similar equipment."

I would challenge the safe operation of an automated block valve as a mitigating instrument - even if it were a ball type. You must ensure that the detection, as well as the action time required to mitigate the over-pressure is sufficiently short and effective. I believe you will find that in the case of compressible fluids and centrifugal compressors the required reaction time is so small, that no available block valve + actuator device is applicable. This would explain the API 521 wording with regards to check valves. I would opt for multiple check valves in series, located in the discharge line. This type of installation simulates the characteristics of a reciprocating compressor - which I already mentioned.

 

[EGT01]

121202,

As always, Art Montemayor provides a valuable perspective and it may not be possible for even a high integrity instrumented system to meet the respone time that is needed for your system.

Just as additional information from API, the following is from their technical interpretations website and relates to the previously referenced sections of RP-521....

http://committees.api.org/standards/tech/reti.html

http://committees.api.org/standards/tech/docs/521ti.xls

"Experience has shown that a single check valve is not an effective means for preventing overpressure by reverse flow from a high-pressure source. Experience has also shown that when properly inspected and maintained, series check valves are sufficient to eliminate significant reverse flow. However, some check valve seat leakage may still occur. The user needs to determine if this leakage is significant to warrant additional protection measure such as isolation valves."

 

[brainstorming]

I agree with EGT01 of providing a check valve and a shut off valve integrated with shut down system with (2 out of 3 pressure transmitter input signal to shut down system located at the discharge line"
I have seen mega compressors with 5 stages and flow of 200 m3/h having such protection (i.e. a single check valve and trip system at the discharge line)

Regards

 

[121202]

ETG01,

The link to the API 521 interpretation that you provided appears to suggest that a shutdown valve together with a check valve in series is adequate protection for backflow. Is that your interpretation?

For clarity the shutdown valve on the discharge of the compressor is driven shut by the compressor UCP when it trips. The shutdown valve takes approximately 15 seconds to close.

Rgds

 

[EGT01]

121202,

I can't say I interpret API's response to the technical inquiry as referring to a check valve plus shutdown valve, but from their response, I think it is clear that 2 check valves in series would be "sufficient to eliminate significant reverse flow" with added protection by isolation valves to mitigate the leakage of the check valves.

My original thought was that your compressor shutdown valve might could be considered similar equipment to a power assisted check valve

http://www.weirvalve.com/weir/valves/home.nsf/Page/Products_atmorcomprescheck

in order to meet the secondary device requirement as suggested in RP-521. But I still had some reservation about taking credit for that arrangement and that was why I suggested further risk analysis similar to that applied to a HIPPS, to see if your existing arrangement could qualify as acceptable without modifications.