Flame oscillation after a release through flare system

[mflr]

We are experience flame oscillation after a release in our flares, any ideas of how to avoid the oscillations?, those oscillation could let air enter to the flare?

We have a normal elevated flare with a molecular seal and purge gas.

Thanks a lot.

Excuse my english

 

[sshep]

You say afterwards- what are you relieving? Is it possible to get some liquid carry-over out to the flare? I have seen several flare systems exhibit instabilities (often a puffing action with or without smoke) due to different ways to trap liquid in the system downstream of the knockout drum. It isn't supposed to be that way by design, but reality can be different. If your problem persists after flaring is over, I suspect that liquid somewhere is a likely cause.

Although most of these systems are supposed to be free draining, yours would not be the first operating unit which had to develop administrative procedures and fixes to deal with liquid related problems. These include ways to drain liquid from the base of the flare stack, or "burp" a flare by deliberately venting a high volume shot of vapor from a suitable part of the process. Note: the considerations of a burping approach include evaluating where the drops of liquid could fall (i.e. on parked cars, near water outfalls, etc).

Just a thought- sshep

 

[mflr]

Let me try to give an example, When a relief valve is actuated, the system start flaring normally, the problems begin when the valve close, when this happen the flame appear and disapper (ciclical fluctuation) from the tip of the flare. I think the phenomenon is cause by someting like hammer pressure or surge. What I am looking for is:

1.- Have someone expirienced the same thing?
2.- What cause the fluctuation?
3.- How can I avoid the problem?
4.- could Flare shutdown because the fluctuations?

Thanks for your time

 

[mflr]

I forgot to tell that the gas is totally dry.

 

[sshep]

Is it possible that these cycles originate at the relief valve. Conventional relief valves are not modulating, and will cycle on and off in a deadband around the setpoint if called on to handle a low load. Is this flame instability a problem with only one relief valve, or with all? When does the instability go away in terms of process conditions? If you are relieving some particular equipment frequently, you might consider using a control valve to the flare as a first and last out for low load relief- control valves do not chatter.

At the flare tip I am sure there are many different flame dynamics which can occure with unsteady flow. Does it look like cycles of low velocity gas being ignited off the pilot and then burning away? What is the cycle time (i.e. 1/2 second, 5 seconds, 30 seconds)?

As to your questions: Can unsteady flame dynamics put out the flare?- possibly, but I am sure you have procedure to relight in such a contingency. Can these effects cause air to back down the stack (equipped with mol seal and purged)?- I think this is unlikely.

 

[mflr]

The valve is not chattering, the fluctuations happends even if you open and close a manual valve, imagin the following: You have a 1200Psi system running, you open a litle bit a valve which connect that system to the flare, when you do that inmidiatly the flare start flaring as it is supposed to be, now whe you close the valve (remember is a manual valve) the flame down to a normal size and disapper (I think is inside the tip in that moment) after 1 or 2 seconds it appear again and disappear 1 or 2 seconds later, after 4 or 5 cicles the flame goes to the normal state. What I think is happening, when you suddenly close the valve (stop the source of gas) a negative presure is created in the flare header ( because the gas is lighter than the air and it is moving fast inside the pipe diretly to its way to be flared), so the gas returned to the pipe (and maybe sucking air from outside through the tip). I really dont know if what I thinking is true or not (I am electronic engineer) but seems to be a good explanation, but how I could call that effect? How to avoid it?

 

[jalvarez]

mflr:
In a plant I was working in the past we have an experience like yours, I would say exactly the same.
The reason for this problem was assigned to water waves in the seal drum, immediatly upstream the flare, and as you explain, immediatly after a release. The pulsing was in cycles of two seconds, approximately, and I always thought that the cycle time depends of the length of the seal drum;
as there is nothing inside the drum to eliminate the energy of the waves, the two seconds pulsing last for something like one hour.
It was almost eliminated adding some baffles in the drum, like the vertical baffles of a heat exchanger.
During ten years, we never experienced a flare lit-off or back flow, despite that in some moments the flame dissapeared completely. Pilot burners were working OK. As we changed the flare to a floor one, we forgot the problem.
Anyway, we had some problems with the refractory inside the flare, but it was thought that the caus for this was a reabsorption of the flame due to the wind, not to the pulses, because the amount of flames during the pulses was not important.
Hope this helps.
Have a safe and healthy day
J. Alvarez

 

[EGT01]

mflr,

If you do have a water seal drum, I believe there are certain aspects of the seal design that are important to avoid flare oscillations just as JAlvarez points out.

Here's some links you may want to look at

http://www.geocities.com/flareman_xs/index.html

http://www.gba-flares.com/ancilliaries1.htm

http://www.nao.com/arrestors_liquid_1.htm

 

[mflr]

We don't have water seal, just a KOD and molecular seal.

 

[jalvarez]

You don't have a water seal drum. Surprise.
I don't give up.
If you don't have a water seal, you maybe have a nitrogen injection through a pressure control valve (maybe another gas, or purging fuel gas that assures a positive pressure in the header). After a release, it could happen that the pressure change in the flare header, the valve injecting the nitrogen (or f. gas, etc.) begin to move and its positioner have some histeresis or malfunctioning. Suggest to check this (bypass the positioner, put the valve in manual, etc.).
Have a safe day
J. Alvarez

 

[Flareman]

mflr

Sorry to be so late with this reply. I just got back onto the site after some time away.

What I think I see you say is that
- the flare normally has no flame
- when you get a relief valve lifting you get a visible flame
- when the valve closes, the flame dimishes, then disappears, then returns, - disappears , - returns etc.

If this is the case then what you see is the flame burning out the residual gas in the flare. The flame will burn down to the top of the tip and then begin to burn inside. It can go perhaps 1 metre inside before there is not enough air to burn the gas and the flame goes out. Some air then goes into the tip and, in a few minutes a mixture of air and gas comes up to the top and ignites from the pilots. That then burns back down, perhaps, 3 or 4 metres before it goes out. In another 5 minutes, the mixture comes up to the top and starts again. Next time it might burn down 10 or 15 metres .. and so on.
Sometimes, some of these events become flash-backs and there is an audible (explosive) noise.
The way to prevent this is to always have a base flow rate of flammable purge gas at a rate which does not allow internal burning.
I have a representative formula published at

http://www.geocities.com/flareman_xs

navigate to purge | burnback.

You amy already have some purge on the system. if this is inert (Nitrogen ?) it will eventually prevent the mixture from flashing back because it suppresses the flammability of the mixture.

If you don't think that I have correctly understood your problem, submit another post and I will try to get back to you.


David