12 replies to this topic

[shan]

Hi Everyone,

We are sizing a HP Flare Knockout Drum for a 150,000 BBL/D production facility. Per API 521 6.4.3.6.7, the knockout drum should provide sufficient volume for 30 minutes of liquid holdup capacity. If we assume 50% liquid volume in the drum, the KO drum volume will be 150,000/6.3/24/60*30/50% = 992 m3 and dimension will be about 7.5 m ID x 22.5 m S/S (24’6” ID x 74’0” s/s), which is huge vessel. Is there any way to reduce the knockout drum sizes while satisfy API 521? HIPPS (High Integrity Pressure Protection System) is not in the consideration.

Shan

[Technical Bard]

The vessel doesn't need to be sized for 30 minutes of your facility oil throughput, unless you have a scenario where a relief valve or pressure control valve would divert the whole oil flow to flare (which would be a poor design).



You need to evaluate the contingencies that can occur in the facility that would result in a flow to flare. You then need to assess for each one what the LIQUID rate from that relief will be. In many cases, most reliefs are primarily vapour. You then take the largest liquid relief rate (adding simultaneous reliefs if necessary) and use 30 minutes of that rate. It should be much smaller. In fact, I have rarely seen flare KO drums sized on the basis of liquid holdup. It is usually the liquid droplet disengagement from the vapour that governs.

[Root]

Hi,

Size your KOD for 6 min hold up time fron BTL to HLL, you will get the reduced size of your KOD.

Hope this will help you.
Toor

[paulhorth]

Shan,

I agree with Technical Bard's reply. You do need to provide some protection from overpresssure from the source of inflow to the plant. If the total oil is coming in from a number of wells, then it is a reasonable basis to assume that the instrumented shutdowns on your first vessel (such as high pressure, or high level) will succeed in closing the wing valves on all except say one well. Then, the relief capacity required would be the flow from one well, which will be a lot less than the full flow. If the source is a pipeline, with a higher design pressure than the inlet vessel, with the wells a long way away, then you can't use this reasoning, and you have no choice but to protect the plant with a HIPPS type instrumented system (either two separate high level trips on the inlet vessel, or a 2-of 3 high pressure trip arrangement).
The 30 minute holdup time is intended to give sufficient time for operators to take manual action to shut off the flow, in the event of an emergency where the instrumented shutdowns have failed, before the flare drum fills up and oil comes out of the flare.
You could argue that this is excessive and that 15 minutes would be sufficient, but be careful, you would have to prove that an operator could respond, and safely access and operate a manual isolation valve, during a major emergency (such as a fire).
This is why Toor's reply proposing 6 minutes holdup time is totally unrealistic.

Paul

[shan]

Hi Paul/Technical Bard,
The relief scenario in our consideration is block outlet of production manifold, which is downstream of subsea well risers and upstream of HP separator. It is difficult to assess the exact operator response time. For example, 5 minutes may be OK for an experienced operator to identify and close the emergency hand valves while 35 minutes may still not be enough for a fresh guy to find the right page on the operation menu to deal the situation. This is why we applied the commonly accepted industrial code API-521 30 minutes liquid holdup as design criteria to protect us from possible legal issues.

My question now is if it is OK to take Flare KO Drum Pump capacity as credit. That is:
Flare KO Drum Liquid Holdup Volume = 30 Minute*[Liquid Inlet - Liquid Outlet (pump capacity) ], which made our KO Drum much smaller than based on the equation Flare KO Drum Liquid Holdup Volume = 30 Minute*Liquid Inlet.

Regards,

Shan

[paulhorth]

Shan,

In my view there is one and possibly two reasons why you cannot take credit for the flare drum pump running.
(1) the pump has to discharge to somewhere on the facility, either the export pipeline or possibly a downstream separator, and in the ESD situation this discharge is likely to be shut and not available.
(2) the power supply to the pump itself may be lost on ESD. You could put this pump on the emergency generator to get round this, but you would still have problem (1).

I am not clear about the relief case that you describe. In my experience it is usual to design the inlet manifold for the full shut-in pressure of the wells, so that the overpressure case will then apply to the first separator (not the manifold) and this is where the relief case arises. As I said in my earlier post, it is accepted philosophy by some oil companies to consider just one well, not the total production, as the relief case arising from blocked outlet of the first separator. This is on the basis that there are two means of shutdown - high pressure and high level, or two high levels -and one of these shutdown signals has worked to shut the well wing valves, with only one wing valve having failed to close on demand.

The 30 minute time in API RP521, while having a sound basis, is a recommended practice rather than being mandatory. On the job I am doing right now, the flare drum has been sized for 15 minutes flow from one of two wells. The oil in this case is being pumped by downhole pumps so that if the power is shut off (remotely) the flow will definitely stop.

Paul

[Root]

Shan,

Flare KOD is designed on 150 to 300 micron droplet size, therefore you can calculate hold up time between 5-15 minutes, keep in mind this not FSD this flare KOD.

Again this depend on your client and sole his discretion for hold up time.
Toor

[shan]

Hi Paul,

Please see the following for the clarifications for our discussed case.

1. The production facility is a FPSO. There is a big enough bad oil tank to store the Flare Drum Pump discharge volume.
2. The scenario is block outlet of production manifold without power outrage. Therefore, in my opinion, it is double jeopardy to assume loss of power supply of Flare Drum Pump. Also, the Flare Drum Pump is 100% standby (2x100% Pumps).
3. We have to install a PSV on production manifold because there is a shutdown valve between production manifold and the separator.

Regards,

Shan

[Technical Bard]

On the one FPSO I worked on, the production manifold was designed for the shutin wellhead pressure, to avoid this relief valve.

[paulhorth]

Shan,

Thank you - I now understand a liittle more about your facility. However you are not correct to say that the loss of power to the flare drum pump, on ESD, is double jeopardy. It is not. I will explain why.
You have told us that there is a shutdown valve between the manifold and the first separator. This valve will be closed on an ESD, I assume. The ESD can have several causes, such as Fire, or Gas Detection, or Failure of Main Power. Any of these causes will result in an ESD which will close that valve and give you a blocked outlet on the manifold. Fire, and Gas Detection, will in turn (or should) also cause a shutdown of main power through the ESD system.Thus, any ESD will cause the blocked outlet, and will also make the flare drum pumps unavailable, unless they are on the emergency generator supply. No double jeopardy but a common, single, cause.
The same ESD will probably also close an ESD valve from the flare drum pumps discharge to the tank on the FPSO, thus giving another reason why this outlet is not available.

So you will have a platform shutdown with a blocked in manifold. If any of the wells has failed to close, you will need that relief valve.

Now, if you do as I and Technical Bard have suggested, and design the manifold for the well shutin pressure, the relief case can be reduced from full flow, by fitting additional shutdown devices to the separator, as I have already explained.

Paul

Edited by paulhorth, 31 July 2011 - 06:19 AM.

[shan]

Paul

Thank you Paul for your suggestion. However, our production manifold is downstream of choke. Therefore, the design pressure is significant lower the well tubing shut in pressure. We have two 100% flare KO drums pumps, of which one is on power and the other one is on diesel.

Shan

[paulhorth]

Shan,
you said:

Thank you Paul for your suggestion. However, our production manifold is downstream of choke. Therefore, the design pressure is significant lower the well tubing shut in pressure. We have two 100% flare KO drums pumps, of which one is on power and the other one is on diesel.

Of course I understand that the manifold is downstream of the chokes. Every production manifold is. Of course I understand that this means that the operating pressure is a lot lower than the welll shut-in pressure.
Now please try to understand this:
What Technical Bard, and I, have been trying to explain to you, is that there is an advantage for you if you decide to make the design pressure of the manifold a lot higher than the operating pressure - make it equal to the well shut-in pressure.There are many manifoids in the world which are designed for API 2500 or API 5000 even though the operating pressure is much lower. It's only piping and valves, and the extra cost is traded off against the saving on the relief valve size, the flare drum size and the flare boom, if your relief case reduces to one well only.
Re read the earlier posts.

Running the flare drum pump on diesel is of no benefit since I think there will be an ESDV on its discharge which means you can't run it on a shutdown.

You asked for a solution for the large flare drum, which we have offered you at least twice. If you choose not to follow our suggestion then we cannot help you further.

Paul

[sheiko]

FYI, exact quote of API STD 521 section 6.4.3.6.7 is:

" The liquid holdup capacity of a flare knockout drum is based on consideration of the amount of liquid that can be released during an emergency situation without exceeding the maximum level for the intended degree of liquid disengagement.
This hold-up should also consider any liquid that can have previously accumulated within the drum that was not pumped out.
The hold-up times vary between users, but the basic requirement is to provide sufficient volume for a 20 min to 30 min emergency release.
Longer hold-up times might be required if it takes longer to stop the flow.

It is important to realize as part of the sizing considerations that the maximum vapour release case might not necessarily coincide with the maximum liquid.
Therefore, the knockout drum size should be determined through consideration of both the maximum vapour release case as well as the release case with the maximum amount of liquid."

Edited by sheiko, 02 August 2011 - 11:19 PM.