12 replies to this topic

[ShaunHill]

I am currently involved in an evaluation of a flare system at a fairly old gas plant (built in late 1960's).
The question has come up as to whether we should be using rated PSV capacities for our evaluation or the "required" relieving capacity. Since the plant has been modified several times since it was first built, the rated and required capacities are very different.

API 521 Section 5.4.1.3 recommends that the laterals be sized based on rated flow of the PSV while the headers be sized based on the worst case cumulative required capacity.

When I am designing a new flare system I normally size the header for 0.5 Mach at required flow and then the laterals at 0.7 Mach for rated flow. I then calculate the back pressures using rated flow and confirm that design pressure of the flare system and back pressure on the PSV's is not a problem. Normally, if all the velocity criteria are met, there is no problem with back pressure for this case.

However, in the existing plant I am looking at, there are cases where the rated capacity of some PSV's is resulting in sonic flow both in the header and in the lateral. This results in a back pressure that exceeds the design pressure of the existing flare piping (the flare piping in this plant is only registered to 100 psig design). The existing PSV is a bellows design, so we can exceed the 10% of set pressure. Our current calculated back pressure (using rated flow) is about 20% of set pressure.

If this same system is evaluated using only the required flow, we don't reach sonic velocity in the header and the back pressure is significantly reduced. I don't really see how I can justify using different flows in different sections of the system because the back pressure calculated won't have a lot of meaning. The justification I've heard is that the rated flow only occurs for a short time. If it is greater than the requried flow, the system will depressure until the flow balances or the PSV reaches its reset pressure. Since it is a dynamic condition the flare header never really "sees" the rated flow. This is also a justification I've heard for sizing the flare knock-out drum and stack based on required flow. I'm not sure I really buy into this explanation since at high velocities we may only have a few seconds before the "real" flow is transmitted from the PSV to the end of the flare header.

My feeling on this one is that the difference between the rated and required flow is large enough here that if the PSV ever opened, it would reset shortly and start to cycle. As such, I am planning on changing the PSV to a smaller one that more closely matches the required flow for this case.

However, I would still like to ask whether using the rated or required capacity to calculate back pressure more realistic?

[rxnarang]

Shaun,

1) API 521 Part II Para 5.3:

"The rated capacity of a conventional spring loaded, balanced
spring loaded or pop action pilot-operated pressurerelief
valve should typically be used to size the atmospheric
vent piping or the discharge line from the pressure-relief
valve to the relief header. Common relief header piping in
closed discharge systems should be sized using the protected
system's required relieving capacity.
For a modulating pilot-operated pressure-relief valve, the
discharge piping can be sized using the required relieving
capacity of the system that the valve is protecting."

2) If the PSV is oversized then , as noted by you, the valve will chatter, with potentially diastrous consequences. You are correct in concluding that a smaller valve, with adequate relieving capacity, should be installed.

3) Limitation of velocity in the laterals. You mention 0.7 mach . I had raised a question separately in this forum on the logic behind this value. The issue has not been closed out. Nevertheless, I have seen laterals designed to sonic flow or close to 0.9 mach, without any problems at site later on. In this case one needs to use the correct analysis for computing pressure drops.

4) I have not seen main headers designed to sonic flow.

I hope this helps.

[ShaunHill]

Mr. Narang,

In point 1 of your reply you have just quoted the section of API back to me that I'd already refered too. (Although I realize now that I had been refering to the 1990 edition).

I recently started using Flarenet program for calculating flare hydraulics. I did notice that it includes an option for using rated flows in tail pipes. If this option is not chosen (and tailpipes not defined) it will calculate everything based on required flow. I noticed when I used this option that it still calculates the back pressure in the header based on required flow, and then the back pressure from the header to the PSV based on rated flow. This gives a slightly lower back pressure than if I set the required flow to equal the rated flow.

Anyways, I'm still wondering if this method is a realistic representation of the peak pressure in what is a dynamic situation.

On point 3, I have also seen flare laterals with sonic (Mach 1) flow at the rated capacity. Although I don't do this for any of the flare systems that I design, I accept that it will function as long as it can be shown that back pressure is not an issue. My feeling on the use of a number like 0.7 or 0.8 is that it allows a safety factor since we are normally calculating velocity in straight pipe and do not account for the formation of a "vena contracta" caused by turbulence at pipe elbows and sudden expansion into tees.

On point 4, I certainly would never consider designing a flare header for sonic flow. However, what I have been looking at is an existing plant that has a 10" flare header that leads to a 30" stack. They have one PSV that discharges into the 10" header via a 6" lateral and has a rated capacity in excess of 135 MMscfd. However, I believe that the requried flow for this PSV is only half that amount. This leads me back to point 1. Do I have a real problem with the header size because of the rated flow of this valve?

[rxnarang]

Mr Hill,

The lateral sized for rated flow and the header sized for required flow is the way I would go.

This leads me to an interesting point. What happens if one PSV is controlling the flare load? The above API statement is valid for the situation when a number of PSVs are relieving in a particular case ( e.g. fire zone has four PSVs). In that case not all the PSVs are expected to pass the rated flow simultaneously. That is the logic behind sizing the header for summation of all the required flows of PSVs expected to relieve together.

When the PSV relieves, it will pass the rated flow. No question about that.

Now in the case where one PSV is the controlling flare load , then the rated flow is seen by the header. In case the header is designed for the required flow, then it will be undersized.

Something to mull about.

Secondly about Flarenet. I hope you realize that although Flarenet reports pressure drop and velocities based on required flow for the main header ( which is per API guidelines), there is a bug in it when it does the lateral.

Flarenet calculates the pressure drop in the lateral based on rated flow, but does NOT calculate the velocities based on rated flow. It reports the velocity based on required flow only! This is a known bug and the vendor is addressing it.

API 520 requires the lateral sizes to be based on rated flows. I read this as considering the velocity criteria, in addition to the pressure drop criteria, when sizing the lateral.

The criteria given by you on limitation of 0.7 mach for sizing will not hold water when we have Flarenet. This programme is able to simulate all fittings pretty accurately. Yes, I do agree that there is safety factor involved.

Above are my general thoughts, in no particular order.

[benoyjohn]

I in support of Mr Narag that if single PSV governs the Header size then, it should be designed for the rated flow of that PSV.

Also sometimes if the scenarios itself is not having capacity to produce that kind of load then,
the sizing load need not be rated flow of PSV. Again this has to be studied case to case basis.

For laterals I found Mach number sonic at point where Lateral joins the main header,
this can be solved by providing proper supports , reinforecement at the junction point,
or the size of the lateral can be increased close to the joining point at the header.


Dvk

I do not agree with Narang. Relief valves need not always relieve the rated capacity.

In a normal relief scenario, the valve will begin to open when the inlet pressure is within 5% of its set pressure. The valve will be fully open and achieve rated flow at its overpressure (normally 10% of set pressure). When the conditions that created the increased pressure in the protected component are brought under control, the inlet pressure to the valve will decrease and the disk will close.

Thus if the conditions that created the increased pressure is brought in control then the relief valve may not relieve the rated capacity.

Regards
Benoy

[Nirav]

Hello,

To Benoy :

In my opinion, you are correct in your explanation about the function of PSV from the time it opens, gradually going up to its overpressure value and then coming down.

BUT the point from RxNarang is NOT at all INCORRECT. That is, we must size our header at "rated" capacity, IF it's single PSV scenario. Not only for PSV, but for any other system, we take into account "MAX Possible" flow that can pass through the system under the circumstances, while evaluating hydraulic aspects.
For PSV, it is necessary to take "rated" flow into account (based on various aspects already explained), because many times, "rated" flow is in excess of 10% of required flow.

To RxNarang :

Thank you very much for pointing out that "bug" in Flarenet. It might be known to many people, but i was not aware. Thank you again.

As fare as "mach no." is concerned, i would say, it's a matter of experience and how confident you are about your calculations and system. We can go upto 0.9, but only if we have "actual" isometrics available to us. But if we are designing 'new' facilities or plant, it is better not to exceed 0.7. Because, most of times, we only estimate the fittings and length of pipes based on lay out during initial stages.

Overall, i find it a very common, but a good discussion on the topic.
Thank you,

[gvdlans]

I noted that Rajiv Narangs post about the bug in flarenet dated from Nov 4 2004, i.e. more than 2 years ago. It may be interesting to know whether this bug has been solved by now? It should be, noting that Rajiv states that "This is a known bug and the vendor is addressing it."

[rxnarang]

I believe they have. A collegue reported that he had checked this very anamoly, and it was resolved.

Regards

[benoyjohn]

I believe they have. A collegue reported that he had checked this very anamoly, and it was resolved.

Regards

Hello Narang,

Need some clarification about your earlier posts.

" The above API statement is valid for the situation when a number of PSVs are relieving in a particular case ( e.g. fire zone has four PSVs). In that case not all the PSVs are expected to pass the rated flow simultaneously. That is the logic behind sizing the header for summation of all the required flows of PSVs expected to relieve together."

Can you clarify why the above four PSVs in the same fire zone will not pass the rated flow simultaneously if there is a fire in the zone?

Also

API recommends rated flow to be used only for sizing relief laterals but not the main flare header.
The code does not mention anything specifically about cases in which a single PSV controls the flare load.
Do you think they have missed this point?

Best Regards
Benoy

[pleckner]

Let's remember that API is not code, at least as far as RP520 and RP521 go. They are only recommended practices but are considered to be good engineering practice so we try to follow their teachings. And even API sometimes can't answer their own questions. In this case, they specifically state in section 5.4.1.3 Design of Relief Device Discharging Piping, "Simple rules cannot be expected to cover all installations. Good engineering judgment should be applied to select the flow basis most appropriate to each case. (See the ASME
Code, Section VIII, Division I, Appendix M-8.)"

So, yes, it seems sometimes we can't just ignore the fact that valves during a simultaneous relief may very well pass their rated flow, and a fire case may be just the scenario where this would still apply.

[wojtar]

What if we have only few relief valves sized for quite big capacity and numerous thermal reliefs. Is it reasonable to size the header for rated capacity of all those valves or maybe only for the main valves rated flows? I can't imagine the situation when all thermal relief valves are relieving simultanously.

[fallah]

What if we have only few relief valves sized for quite big capacity and numerous thermal reliefs. Is it reasonable to size the header for rated capacity of all those valves or maybe only for the main valves rated flows? I can't imagine the situation when all thermal relief valves are relieving simultanously.

wojtar,

Below statement from API 521 clarify the way you should go:

"Common header systems and manifolds in multiple-device installations are generally sized based on the

worst-case cumulative required capacities (instead of rated capacities) of all devices that can reasonably be

expected to discharge simultaneously in a single overpressure event (in other words, for certain scenarios, it

can be appropriate to assume some level of favourable instrument and/or operation response)."

Especial attention is needed to the point that for header you should consider required capacities instead rated ones.

Fallah

Edited by fallah, 07 December 2011 - 02:56 AM.