Relief valve maintenance philosophy

[robsalv]

Broadly describe your company's relief valve maintenance philosophy.

My work place was using an empirical approach that took performance and reported condition into account when setting overhaul intervals... but we're moving to a SIL based PRV management process that doesn't take into account any condition data gleaned from the overhaul.

The applied lambda is either based on actual failure data or taken from provided tables that relate to an assessment of service condition - clean, dirty, fouling etc., and/or type of valve. Apparently these are industry standard lambda's.

Every PRV in our plant is treated with the same priority as we consider them the last line of defence before an uncontrolled loss of containment.



How does your work place manage PRV's?

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

 

[MikeClay]

The goal is to prevent loss of containment due to overpressure. Any SIL based system is based on probability of failure on demand. One of the aspects is the testing frequency. In general, the type of service is used as a first pass guess at the testing frequency. This is true whether it is based on SIL tables or other tables based on generalized industry experience.

Once operating/testing data is available, testing frequency is adjusted (normally increasing the frequency of fouling/dirty service if needed). Any maintenace system employed to manage the reliability of these devices must be able to handle this situation. Whether it is just a frequency, class, or lambda ajustement.

I'm not sure what you mean by priority. But if it means that the PRV must be tested/maintained by its scheduled date, then there should not be a problem. However, if it means that all PRV must be routined on the same schedule, then you will either be overtesting the clean service vavles or under testing the fouling service valves.

--Mike--

 

[lizking]

Look into an API "risk based inspection" (RBI) methodology.

 

[robsalv]

My apologies for the apparent confusion, perhaps I should elaborate.

I'm understand how a SIL or "probability of failure on demand" philosophy works. What I'm curious about is what are the various methods/philosophies that organisations around the world use to manage their pressure relief valves. Is there a move towards SIL? Is there a move towards risk ranking the consequence of a PRV failure and setting intervals based on that? Is the emperical approach based on condition and performance still valid?

API576 is still focussed on a condition and performance basis for example. (I'm not aware of committee work to bring RBI principles into API576.)

What I mean by priority is that every PRV is considered as safety critical - i.e., the last line of defence before an uncontrolled loss of containment, as such their due date is very important and meeting due dates is a key KPI.

My concern with the SIL process is that PRV's are mechanical devices. They are subject to dynamic environments and multiple degradations. Categorising them in a clean or dirty service category and then ignoring condition data seems fundamentally opposed to good practice. In particular, if a PRV fails it's testing for an identifiable condition based caused, that failure is discounted for setting the following inspection interval.

At this point in time, the SIL philosophy for PRV's is not sitting well with me.


Cheers

Rob


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

 

[MikeClay]

My experience is that initial PRV maintenance schedules are based past experience with similar PRV. If there is no experience with similar services, then the initial schedule is based on a clean/dirty/fouling service basis.

However, once a PRV is pulled for maintenance and is found not performing, the cause for failure is determined. If the cause is determined to be due to the nature of the fluid, or unknown, the inspection frequency is reduced.

--Mike--

 

[robsalv]

What constitutes a failure Mike?

If a valve passess in service and cannot be pop tested on the bench due to leakage, can you use that data point to increase the overhaul interval or not?

What philosophy does your work place use? An empirical one, a risk based one, or something else?

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

 

[MikeClay]

-Valve fails to open
-Valve opens outside of tolerance
-Material fouling/restricting the relief device inlet
-Any noted mechanical failure

If for any reason, the interval is increased.

--Mike--

 

[robsalv]

Thanks Mike - that looks likes the the same list as ours, but we add if the PSV was known to have not lifted within tolerance in the plant.

I don't understand your last sentence.



To all the contributors in this thread, thanks for adding input. Apologies in advance, but I'm going to cross post the query to get a wider selection of views.

Cheers

Rob


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

 

[786392]

Dear Rob,
Just an explaination envisaged for Mike's last sentence above ;he has presumably indicated more than usually expected/anticipated time interval.
Between 'Popping up' and 'Re-seating back' of any Relief valve in question.

Best Regards
Qalander(Chem)

 

[robsalv]

786392, thanks for your explanation, but that still doesn't make sense to me.

The length of time between pop and reseat is essentially irrelevant. It's the quanta of the actual pressures which is important. When we "float" our steam boiler safety valves, we certainly determine the reseat pressures then... but then again, the valves are sitting on the ultimate full flow test rig, a industrial sized package boiler!

The other thing with that, is that on a limited volume test bench, how does one measure the reseat pressure? Once the SV pops, it reseats right away as there's no large capacity to relieve.

Thanks 786392

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