Elimination of overpressure scenario on ASME pressure vessel

[met11]

Hi all,

I was hoping someone had some guidance on something that frequently comes up. Say you had an ASME VIII pressure vessel with a relief valve set at 150 psig. The vessel has an ASME relief valve which meets the requirements of ASME VIII including inlet/outlet piping requirements etc.

Now say there is a pump feeding into the vessel, and the pump is capable of generating 300 psig. However the pump has its own relief valve which is set at 150 psig, relieving back to the pump suction. However it is not an ASME relief valve. It is just a simple proportional relief valve sized for the capacity of the pump. The pressure the pump normally puts out is just the backpressure from the vessel plus whatever frictional losses are in the line.

My question is, there is the potential to block in the outlet of the vessel and have it become liquid-full if the level controls fail. Do you have to consider the case for the ASME relief valve on the vessel where the pump could overpressure the vessel? Or can you just assume that the proportional relief valve on the pump will do its job, and eliminate the scenario? Does the code allow you to do this? Seems like technically you're using a non-ASME valve to protect an ASME vessel against a certain scenario?

 

[EmmanuelTop]

In this particular case, we are talking about two simultaneous failures that have to happen and possibly lead to overpressure of the vessel:
1) Failure of the vessel level controls;
2) Failure of the pump discharge relief valve.

These are two independent systems and considering simultaneous failure of both would be a double jeopardy. So unless we are talking about really MAJOR consequences in terms of HSE or Economics if the combined event occurs, there should be no additional layers of protection required for this system.

In addition to very low likelihood of such a combined event, process safety time (PST) required for the vessel to be completely filled with liquid and consequently overpressured should be calculated. There are various guidelines out there, but as a general note if the calculated PST is more than 10-15 minutes, there is sufficient margin for operator intervention and mitigation of the top event.

If you want a documented decision process, LOPA will tell you what is the expected frequency and risk of unmitigated event (likelihood would differ for ASME- and non-ASME certified relief valve on the pump, hence the risk calculated in both cases would be different).

Dejan IVANOVIC
Process Engineer, MSChE

 

[zdas04]

I don't quite see it the same as Dejan sees it (pretty rare occurrence).

The constant pressure valve is not part of the system safety equipment. It could fail and it is likely that no one would notice for an extended period (the pump discharge pressure is actually controlled by the backpressure on the system). If you block the outlet then you could overpressure the vessel with the pump.

Those valves are notorious for running for years without anyone noticing that they are broken, so I don't think that this is a double-jeopardy situation. I wouldn't ever consider one of those valves to be a contributing portion of the safety equipment (I use them for control, but ignore their existence when I'm doing a credible scenario evaluation).

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[georgeverghese]

My recollection of these internal RVs' in these PD pumps is that accumulation is about 25% of set press. So the RV reaches full capacity only at 150psig +25%, which is a lot more than the 10% permitted per Code. Usually, credit for such an internal RV is not taken unless it is in some very low HSE risk service.

 

[Latexman]

The answer for me depends on the fluid being pumped. If it is clean, non-fouling, no solid content, non-reactive, has some lubricating properties, solvent like, etc., I can go along with Dejan. If it is dirty, has solids, reactive, no lubricating properties, resin/polymer like, etc., I can go along with David.

It depends on the probability of the liquid to form a pre-existing condition that could go undetected.

Good luck,
Latexman

To a ChE, the glass is always full - 1/2 air and 1/2 water.

 

[EmmanuelTop]

These are all valid comments/amendments. Still, we are talking about two independent systems that have to fail simultaneously in order to create prerequisites for hazardous event to occur.

Requirements for additional safeguards will depend on what is the consequence of vessel ovepressure (likely results in a flange leak). Again, unless it is a very toxic compound or perhaps a flammable liquid in an offshore environment, I doubt this will call for further assessment but LOPA should be used to confirm this anyway.

Dejan IVANOVIC
Process Engineer, MSChE

 

[Latexman]

If a system is knowingly prone to failure, AND failure is not alarmed/announced/easily detected, can it be considered an independent system (i.e. layer of protection)? The way I've worded it, I say no. The specifics of each situation do need to be analyzed though.

Good luck,
Latexman

To a ChE, the glass is always full - 1/2 air and 1/2 water.

 

[zdas04]

The pump constant pressure valve is not part of the safety system (and its maintenance schedule), it is a control device not a safety device. Since it is not part of the safety system there is no way to rely on its proper functioning in your LOPA, HazOp or any other PSM add-on process. As far as PSM is concerned it doesn't exist. Removing the PSV from the vessel with any surviving credible scenarios would violate API 520/521.

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[Latexman]

Understood. Some PD pumps come with a built-in or optional "relief valve". One company I worked for did not accept these, and we had to put PSV's on the discharge. Two companies I have worked for did accept these as bona fide PSV's. Is this different from a "pump constant pressure valve"? We may be speaking to different type valves.

Good luck,
Latexman

To a ChE, the glass is always full - 1/2 air and 1/2 water.

 

[zdas04]

The difference between the pump manufacturer's PSV's (which relieve somewhere in the suction stream, and are full open or tight shut) and the constant-pressure valves (which are always open a little bit dumping discharge back to the suction) is really just a design philosophy. I've had both of them work well, and I've have both of them not work worth a darn. They are both always on the list of "control" equipment maintenance (basically never) and not on the list of "safety" equipment maintenance (mandatory).

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[EmmanuelTop]

For a device or a system to be considered as a layer of protection, it is not mandatory to be a part of the "safety system". DCS/PLC control loop is one layer, operator intervention (upon hearing an alarm) is another layer of protection (if the mentioned process safety time provides sufficient time span for operator intervention) etc. Let's go back to the OP question and to the title of this thread - how close or remote is the possibility for overpressure of this vessel to occur? Can we eliminate it or not?

What I was saying in previous posts is that arriving to vessel overpressure conditions requires quite a few things to go wrong, and simultaneously: pump controls to fail, vessel level control to fail, and operator intervention to fail. Everything in the universe is possible, but in this industry we are interested in actual risk (= consequence X likelihood) of a hazardous event, and see if that risk is acceptable or not. So, in reality, the question is not whether the overpressure can happen or not (because everything can happen, it is just that likelihood of various events is different), it is how big is the total risk and whether that risk requires further reduction or not.

Dejan IVANOVIC
Process Engineer, MSChE

 

[zdas04]

So you would sign off on REMOVING the PSV from a pressure vessel because of the control equipment on the pump? You have no assurance that the valve on the pump will work (it is a secondary control and they often don't ever have a reason to come off the seat and often won't come off the seat after a few years closed and they occasionally get piped into places that can be isolated without killing the pump), so a closed outlet can overpressure the vessel. Not a PSV I'd put my signature on the code case to remove.

We don't know what other credible scenarios have been considered. I would call this one credible with the flow rate equal to the pump capacity.

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[nickelkid]

Zdas04 is absolutely correct.

 

[EmmanuelTop]

I would never remove the PSV, why? But I wouldn't look for additional PSV's or other safeguards if the pump is pumping e.g. demineralized water, or if the facility is remote and/or unmanned, or if the consequence of a leaking flange on the overpressured vessel does not create fire or explosion hazard with possible fatalities and prolonged production downtime - all these being just examples, of course.

Dejan IVANOVIC
Process Engineer, MSChE

 

[zdas04]

The OP asked:

can you just assume that the proportional relief valve on the pump will do its job, and eliminate the scenario?

I take that to mean that (absent any other scenario) this was heading towards removing the PSV.

David Simpson, PE
MuleShoe Engineering

In questions of science, the authority of a thousand is not worth the humble reasoning of a single individual. Galileo Galilei, Italian Physicist

 

[moltenmetal]

Even if there is no credible relief scenario, the way I read ASME VIII is that an ASME VIII vessel requires a pressure relief device in communication with the vessel in every case. If there is no credible case, that device has no sizing basis and can be the smallest one you can buy.

The only time I would sign off on the removal of the only relief device connected to an ASME VIII pressure vessel is if it could be demonstrated that a relief event or leak caused by a faulty relief device would represent a substantial risk of harm that cannot practically be mitigated by design, and even then I would do so extremely reluctantly. Even in systems which are protected "by design", i.e. by setting the MAWP above the maximum pressure expected in a worst-case failure event, I would STILL insist on installing a relief device unless as noted above.

If the question is not one of removing the only relief valve on the vessel, but rather relying on a proportional relief valve on the only legitimate source of pressure to that vessel to take the place of the code-required overpressure protection device: I might consider that on a small UM stamped tank full of a low hazard material or the like where the hazard to people is very small, and then only if it can be demonstrated that the installed relief device (though non ASME) is sized such that it will keep the overpressure within 10% under all cases. In our business we do frequently find ourselves in the situation of having to install properly sized proportional relief devices to protect small UM stamped tanks, because the smallest commonly available ASME relief valve (a C or D orifice)is disproportionately large and an oversized PSV would have negative consequences for the downstream relief handling equipment. There are people who peddle A and B orifice ASME type relief valves and there are cases where we're forced to use them, but in many cases a properly sized proportional relief device is a safer and more practical solution.

If the question is rather whether or not the ASME relief device installed on the vessel needs to be sized for a case generated by the failure of another relief device such as one on a pump, that's a matter for a proper, systematic hazard evaluation process. I couldn't make a hard and fast rule on that which was suitable for all circumstances.

 

[EmmanuelTop]

It seems like we all understood the OP question in a different way. My understanding of his query was whether it is "possible to eliminate the overpressure scenario caused by failure of level controls on the vessel".

Now, when re-reading the last paragraph in his post, "there is the potential to block in the outlet of the vessel and have it become liquid-full if the level controls fail" I am getting a bit confused because if the outlet is blocked then it does not matter if level control works or not, because the outlet is blocked anyway.

As usual, we would need a sketch of the system and a bit of clarification what the OP meant...

Dejan IVANOVIC
Process Engineer, MSChE

 

[don1980]

There's some good discussion here, but I'm not sure the OP's core question has been clearly answered. The question, as I understand it, is whether or not he is required to consider the pump scenario as a credible sizing case for this PSV. The answer to that is no. There are no ASME mandated sizing scenarios, nor any guidance at all in ASME about how to assess potential scenarios. ASME, like all other pressure vessel codes, leaves it to the user to determine the sizing basis (determination of credible scenarios). Pressure vessel codes tell us to install a relief device, but they intentionally leave it to us to assess the potential causes of overpressure and to decide which is the sizing scenario for the relief device. They understand that each situation needs a case-by-case risk assessment, and there's no way that a code can prescriptively mandate how this should be done without making a mess of things.

Documents like API 521 fill this gap by providing safe and practical guidance for assessing scenarios, but these documents also leave the final decision to the user.

So, there's no right or wrong answer to the question of whether the PSV should be sized for overpressure due to the pump. For me, this is clearly in the grey zone - I could go either way. To make a decision I would look at the particular fluid involved. How hazardous is this material if it is spilled, and how much could be potentially spilled? I would also look closely at how many failures would have to occur in order for this scenario to play out. I sense that this scenario would require more than just a failure of the relief device on the pump. The downstream vessel is probably level controlled, and it may have a pressure transmitter that would alarm if the outlet was blocked. If so, there's a good argument that this is double-jeopardy, which is not typically regarded as a credible sizing scenario. Regardless, the final assessment is for the plant owner to make based on their analysis of the specific risks of this case, and their risk tolerance.

 

[met11]

Thanks don1980. I do like what has been discussed in the thread, but your answer addresses my question (I must have written it too ambiguously).

I guess to give more background, this is a vapor service valve that is only certified for vapor service. So when sizing for a liquid scenario, I'm forced to apply the correction factor (Kp) from API 520 5.9.1. which derates the valve to the point that I would have to step up an orifice size. So if I count the scenario (assuming the pump external proportional relief valve isn't functioning), then I have to go up an orifice size. But if I don't count the scenario (assuming the pump relief valve is functioning properly), then I don't have to change the relief valve.

The relief device on the pump only has to meet the B31.3 requirements, but the vessel relief device has to meet ASME VIII. API 520/521 are now standards, so I wasn't sure which path to take here. The company requires that all applicable scenarios are documented, and put into the database.

 

[georgeverghese]

For a PSV to be in ultimate overpressure protection service, the relief device must be a Code constructed and approved device; it must be maintained and tested regularly; it must reach design relief capacity at 10% overpressure at the protected vessel. This internal relief device fails on all 3 counts. Just assume this internal device doesnt exist from a process safety perspective.