Parallel Relief Valves

[na5rc]

It recently came to my attention that several operating companies are opening parallel, 100% sized, relief valves to the process when the design was to leave one closed and one open to facillitate maintenance. This is apparently becoming a standard practice.

If more than one relief device is applied to a process and operated in parallel, the set points are staggered by 5% and designed to not allow an overpressure greater than 16%. These devices are sized such that the combined capacity does not exceed around 140% of the required capacity.

When parallel relief valves are installed and each is sized for 100+% of required capacity, the intent is to provide a spare that is closed to the process; this facillitates maintenance and testing. It is very bad practice, even dangerous practice, to open both of these valves to the process. In all fairness, this is done because it is perceived to be safer. In reality, it is not.

Without going into detail here, flare loading and pipe stress calculations are made based on P&ID's and the philosophy at the time of design. To open both valves to the process when this was not intended can result in a number of unwanted consequences to included:

- Transient Flare System overload - Capacity and Pressure
- Pipe Support Damage
- Relief pipe damage
- Damage to relief valves
- Less than required relief capacity
- Liability: Plant Operating not in accordance with approved P&ID's

I have looked at this in detail and have more information on the subject.

 

[skearse]

If the intent is to have one online and one offline then no credit should be taken for calling this a "multiple valve installation"-both valves should be sized for 10%/21% OP and set pressures should be equal and set at MAWP.

 

[JLSeagull]

Among the potential issues with two 100% relief valves both lined up is an actual reduction in capacity. This could occur if using conventional or bellows relief valves as they likely chatter thus don't relieve as much as expected. A good design for those intending to use redundant relief valves could be including a switchover valve to permit maintenance. These are a specialty three-way valve with one or the other PSV in service.

 

[jalvarez]

One of the potential problems is related with the analysis of contingencies and the possibility of chattering. As an example:
Suppose the designer end up with a maximum flow for the worst contingency of 100, and the valve (and its spare, that was thought here only for maintenance reasons) was built according to this. But one of the other contingencies that the designer calculated occurred at, let's say, 35% of that maximum. Everything is OK, no chattering is supposed to occur, as the limit is 25% (in general, one can say that if one contingency have an associated flow lower than 25%, multiple PSV's must be considered).
Now, and obviously, if both valves are in service, each one will take care of ~17.5% of the flow, and we'll have chattering problems with both. Imagine two PSV's chattering in the same structure... wow, yes, it's extremely dangerous.
So, I think that for these cases, the designer must consider interlocking the PSV's.
Anyway, the sequence for valves that has been interlocked, in general permits to stop at any point. So, at least a well explained procedure is also needed...
Have a safe day

 

[robsalv]

This is a typical install at my work place and we manage the OP's concerns by installing one valve at a time OR using an interlock so that at least one valve is in service at any time.

It's well understood here that having both on line isn't actually better.

- - - - - - - - - - - - - - -
"Life! No one get's out of it alive."
"The trick is to grow up without growing old..."

 

[na5rc]

I have placed a paper on this subject up on Scribd.

http://www.scribd.com/doc/18656992/Relief-Valves-in-Parallel

 

[robsalv]

Marking up the PID to show CSO and CSC against the inlet block valves of a spared installation is a really poor solution IMO. Does one need to raise a "control of change" when swapping the spare into active service?? The concept of having a CSC valve in a relief path just plain feels wrong.

Operators do read PID's, but they aren't taking them out in the field with them just to swing some valves or line up a process. Relying on the PIDs for SV line up configuration is therefore flawed.


As I wrote earlier, either have only ONE SV installed at a time (this works for us since we have inlet and outlet block valves) or both SV's installed but have the inlet block valves interlocked so that only one can be open.

Some other alternatives include directorial signage, improved education/training or standing orders that spell out how to line up spared SV installations.


On the matter of training, safety is a strong motivator, and I think if you explained to operators that they just made their plant LESS safe by lining up both SV's, they'd soon be dissuaded from doing it. SV chatter is generally understood to be a bad thing - by having both SV's lined up, Operators are guaranteeing chatter and therefore just made their plant less safe.




- - - - - - - - - - - - - - -
"Life! No one get's out of it alive."
"The trick is to grow up without growing old..."

 

[MortenA]

k5ajr - I think your statement is a gross overkill! I have been working in O&G process design for 15 years now and in all my time parallel PSV has been the standard (North sea and arabian gulf).

The operational advantages massively overshadow the disadvantages. Its normally handled by interlocking but special components that allow only one valve to be open.

Best regards

 

[na5rc]

MortenA

You are absolutely correct. There is no problem with interlocking the valves. This thread has to do with opening both valves to the process at the same time when only one was intended by design. Please see my paper at

http://www.scribd.com/doc/18656992/Relief-Valves-in-Parallel

Your experience in the North Sea and the Arabian Gulf parallels my experience in the Arabian Gulf and SE Asia and the Caspian. We are in agreement.

best regards,

 

[MortenA]

Ahh sorry didnt catch that. I see your point if both are left open.

Best regards

Morten

 

[MortenA]

I came up with one more argument (Couldnt find it in your list):

Maintenance. Damage to the seat or piston is likely to occur when the valve pops and re-seat. If both valves or "live" then the likelyhood is that both will lift and the risk of experiencing damage to the spare thus much higher.

Best regards

Morten

 

[Compositepro]

I have no position on relief valves but I've read your paper and offer these comments.

Most of your concerns seem related to relief valve design deficiencies in that if a valve operates it can become damaged or chatter. Relief valves must be sized for worst case scenarios so it is quite likely that in an actual event it will operate at much less than full flow. So valves should work properly over the full flow range.

Your logic about setpoint accuracy eludes me. You state that two valves set the same will actually operate at different pressures, albeit by an unknown amount. So then the valves behave just as you recommend (or rather ASME). When you have different setpoints you know which valve will open first but you still don't know what the actual differential will be with any greater accuracy.

You include as one purpose of a relief valve is to avoid civil lawsuits. This is an emotional point which should not be in a technical paper. A lawsuit would come one of your other points occurring, like an injury. Logically, a relief valve could cause a lawsuit by discharging.

My reaction to your arguments is that the problems you talk to should be addressed through better design and engineering of relief valves. This would, of course, increase their cost. Are we stuck with traditional technology that was developed in an era when low cost solution to problems was far more important than it is now? Or is there even really a problem? I don't know.

 

[na5rc]

Compositepro,

Thank you for your comments.

My concern is not with sizing deficiencies at all. My concern is with taking two, individually proper sized valves and opening them to the process when the design calls for only valve being open to the process. Opening to the process refers to the block valve under the relief valve.

With regard to setpoint, it is impossible to make two valves operate at exactly the same setpoint. You just can't get that setpoint accuracy and repeatability. They typically vary by a few percent. I do not make a recommendation here other than not opening both valves to the process. It is exactly for this reason of setpoint uncertainty (and other reasons) that the practice of opening both valves to the process should be avoided.

Sometimes the mention of civil liability is the only way to get attention. The practice I describe will eventually cause damage and/or personal injury or death. This is a certainty. I do want this fact clearly understood. I want this practice stopped.

I'm sorry, but better design of relief valves is not an issue. Relief valves have a very good design and no design improvement can remedy this misapplication of equipment.

In summary, this is a serious problem in practice by several large operating companies in specific regions. The solution to this problem does not cost a single Yankee dollar. All that is required is to close a block valve and revert back to the original, approved, and proper design.

 

[na5rc]

I am currently working on a simulation to demonstrate the issues associated with this practice.

 

[Drexl]

Interesting thread indeed. I'm not convinced there would be a problem because the valve is supposed to work within the full range but otherwise i agree that the design should be followed.

 

[robsalv]

Drexl, I'm surprised that you're not convinced.

A prime cause of chattering is an SV providing more capacity than required for the lift event. Do you agree?

SV design contingencies are usually special events (cooling water failure, fire, blocked in etc) so as a result, MOST lift events are plant upset overpressures which generate vapours/liquids at well below design rates.

Now put TWO 100% sized SV's into service... that's 200% of the design contingency capacity and the question for you is, how is chattering not the gauranteed outcome of ANY lift event in this case?

- - - - - - - - - - - - - - -
"Life! No one get's out of it alive."
"The trick is to grow up without growing old..."