Flare header purging: Oxygen ingress through rapid cooling 1

[ColourfulFigsnDiags]

Hi all, in this (now closed) thread thread124-21178 , TD2k mentions "Note, one other thing you may want to look at is if you flow hot gas down through the flare system. Once that flow stops, the hot gas in the system cools and shrinks which can pull air back in through the tip if the shrinkage rate is greater than your purge rate. We had to install a purge system for this case at one plant."

We are designing a very large flare network, and I am concerned that in the case of a rainstorm, the rapid shrinkage could result in an unacceptable amount of air drawn back into the system.

Does anyone have any experience with detirmining whether this will be a problem or not? We are using Methane as the purge gas, with a molecular seal in a 36-40" flare header.

Read the Eng-Tips Site Policies at FAQ731-376

 

[ColourfulFigsnDiags]

And while I have your attention, various posters here have mentioned purging rates of 0.02-0.03ft/s, but API 521 clearly states that "Most molecular seals are purged at rates of 0.5 ft/s to keep the flame out of the flare tip ..." The John Zink manual also says 0.25 to 50ft/s (the later I believe to be a typo for 0.5ft/s considering the amount of other typos in that document) ... but only if NO molec seal is used.

What is with the 10 time difference?

Read the Eng-Tips Site Policies at FAQ731-376

 

[Flareman]

ColourfulFigsandDiags

Not a small set of issues you raise.

Things called "seals" on stacks are usually given the minimum practical purge rate by the vendor.
For labyrinth type seals (incl what you call "molecular" but that's a trade name), the rate used to be in the range of 0.07 fps, but commercial (sales) pressures have driven that down and these days it tends to be given in the region of 0.03 fps. The labyrinth device causes a gas flow inversion and, theoretically, gives you an air/gas interface somewhere inside the seal. You replace the diffusion zone by constantly purging. The velocity is relatively unimportant regarding whether it can do its job of inverting the gas flow. As long as the rate is greater than the interface diffusion rate (~ 0.01 fps ?) it will work. What you sacrifice with the low velocity is the restoring time. When the wind blows or the atmospheric conditions change, the interface relocates itself in the seal. If it moves too far in, the inversion effect is lost, but will eventually be restored as more purge gas flows into the device. With a low velocity like 0.03 fps, the restoring time could be 5 to 10 minutes.

That all tends to be farly moot anyway, because rates like this are useless unless the flare is dormant and just waiting for start-up. Once the flare is live, you need to hold a velocity which doesn't generate a burn-back condition. That tends to be greater than 0.1 fps for small tips and as great as 0.5 fps for big tips (hence the API comment)
Go to my web site

http://www.geocities.com/flareman_xs

links|services for a paper which covers this (amongst other things)
I have some stuff on seals on the web site too.

When you set the basic system purge for the headers and laterals, you need to be aware of the minimum rate needed for the flare and not go below that. Then you can set about "guessing" the rate needed to sweeten the headers which, in total, may or may not be more than the flare minimum. The flare minimum can be distributed amongst the laterals and subheaders as you see fit.

In operation the flame needs flammable gas at a rate sufficient to support its burning outside the tip. You can set this using one specific and controllable flammable purge which the operators can tweak according to whether there is a concurrent relief condition or not. You don't need to waste this purge gas if there's is a big relief all the time anyway.

That's just the starting point.
Now we come to your real question.
The shrinkage issues are real and should be addressed. In the main, the problems which could be caused by a changing ambient condition tend to be dwarfed by the normal day-to-day flow conditions, which is why they are not a "big thing" in most cases, although you should certainly think about the possible ramifications.

The bigger issues are what happens if
- you have a long term hot relief which raises the system temperature and then stops
- you have a short term "puke" of really hot stuff which all cools because the pipline is much colder
- you have a condition which allows a lot of material to suddenly run to dew point.


The bottom line in all these calculations is to work out the rate of heat transfer from the material inside the pipe, to the pipeline itself as well as the rate of heat transfer from the pipeline to the atmosphere (is it windy??)

You will need to know the entire system volume, and the mass of the steel you are heating or cooling, as well as the internal and external surface areas (which are almost the same).

What you discover (estimate) is a rate of shrinkage for a specific set of (worstcase?) conditions. You then have to be prepared to match that rate of shrinkage with a gas input. But you only need it when you have a problem so you have to instrument your system to monitor for the problem and then blast away with supplementary purge only when needed.

If you have a water seal in the header it may be useful, to close up the header and prevent a backflow IF a suction condition occurs. The dipleg has to be long enough to support whatever suction you can pull.
See my water seal page at the same website.

The downstream section still has to be protected but you are dealing with a smaller overall volume. Remember that purges upstream of the water seal don't get to the flare when the seal is closed so you need a purge point downstream of the seal.

Also, if your instrumentation scheme includes a pressure monitor, which it may well do, bear in mind that most cells and instruments use an atmospheric reference pressure on one side of the cell. If you are purging a flare with a gas which is lighter than air, the header pressure relative to the atmosphere at the elevation of the header is affected on the inside by the physical height of the column of light gas inside the elevated stack, and on the outside by the equivalent column of heavy air, so the pressure "looks" like a negative even though it isn't really. You have to solve that one by careful use of DP cells.

Getting back to the Mol seal, it has NO BENEFIT relative to this shrinkage issue and in some ways complicates things, partly because of its added volume and partly because, if you DO have a flash back (it's not unknown) the flame running back through the seal reaches a detonation much faster than it would in a plain pipe. I know of at least one case where the seal was physically "blown" off the top of a flare by this effect (not to scare you!!)

Good luck, you have plenty to do.

David

 

[Bala0610]

Flareman

You mention : the lower purge rate is irrelevant in situations where the flare is on - the lower rate being useful only if the flare is idle and waiting to come on. This is readily understood.

Can we hence conclude that in cases where flare is continuously on (or on for a significant amount of time), the seal is not of any practical use…? To prevent back burning, the flow has to be at a rate equivalent to 0.1 to 0.5 fps or higher velocities. This is significantly higher than the purge rates. Hence, the seals are irrelevant...?

Secondly - can the labyrinth type seal be of any use in a horizontal position - especially in large dia lines (24" or higher) ? What will be the effect of having any entrained liquids (when these are not separated at a flare KOD upstream)? What exactly happens when back burning occurs ? Can this damage tip / upstream pipework including seal - how is the oxygen supply maintained a few meters u/s ?

Many thanks in advance

 

[Flareman]

Bala0610

I can't say "irrelevant", because there are situations where the labyrinth seal has benefits. I simply point out that the benefits are limited and are sometimes overstated or misunderstood. The judgement of whether YOU need a seal has to be your own (hopefully educated) decision.

You can't use a labyrinth seal in a horizontal position, but you could use one in a vertical position in a horizontal line as long as you understand what it's doing which is basically to act as a U trap where the "seal" is created by differing gas density (remembering to purge and counteract diffusion of course).
I don't know whether anyone has done research on orifice type "seals" in horizontal lines but the dynamics of increased velocity will be retained regardless of direction.

One of the recurring drawbacks of the labyrinth is that it collects liquid and becomes a liquid seal very quickly if not constantly drained (codicil: you need to seal the drain with its own liquid U trap)
Most problems occur due to water collection from the tip steam system (or maybe from very heavy rain). If liquids are being carried in the gas stream, the same velocity which gets them into the seal will probably get them out too, to be discharged at the tip (a whole other subject)

Burn back is just a condition where some air creeps into the tip along the edges and supports partial burning of some of the gas before it gets out of the tip. It occurs mainly at the interface and is more pronounced when the tip is large or has some irregularity which promotes an uneven flow pattern. With hydrocarbons, the partial burning effect causes the gas to drop free carbon which can make smoke and build a carbon lump inside the tip. The lump causes an uneven flow and further aggravates the burnback, until a large flow comes along and then it becomes detached and flies out of the tip as a red hot chunk (and perhaps starts a grass fire ??)

Usually, burn back stays within 1 - 2 diameters of the outlet, unless you have a really low flow rate. It can over heat the tip and reduce the service life. Refractory lined flare tips are generally somewhat resistant over the short term but you really don't want burn back in large unlined tips, and you need to take steps to prevent that by keeping the velocity high enough.

Look at the link in my previous response for a copy of a paper to help with this understanding.

David

 

[H20logged]

We have a small continuous purge from each flare lateral and an intermittent (I'll call this the supplementary purge which is activated on detection of flaring event) purge that is considerably higher and only lasts for a short period following a flaring event which is introduced into the header. The intent is to push the slug of hot HC out the stack. Our guideline recommends calculating the required supplementary flow based on the volume of the flare header, KO drum and stack, the maximum expected flare temp and the lowest expected ambient temp. When doing this calc I determine that we would only turnover the above volume less than once. Intuitively it would make more sense to changeover twice to ensure hot HC has been removed while maintaining a minimum velocity in the header and stack. Our existing control scheme also will re-initiate the supplementary purge until the flare header temp is >150°F.

Have you found anything that helps you better understand this?

 

[CJKruger]

H20logged,

Active the supplementary/temporary purge based on the pressure in the flare header.

 

[Flareman]

H2Ologged,
Your supplementary purge guideline clearly implies an expectation that there is a hot slug of gas which can contract or condense in the header and it's better to push it out than wait. The key to finding a suitable flow rate depends on the rate of heat transfer from the gas to the pipe and the pipe to the atmosphere. As you mention temperature differences, I have to suppose that this is part of your calculation procedure. Once you have a rate of shrinkage, you have to match or exceed that with the gas inflow rate. The volume calculation then tells you how long you have to keep purging to be sure that you have achieved your goal. Certainly one replacement volume seems a little sparse and more (2?) would be better, because you won't get the thermal calculations right and you need a safety factor.
I'm not clear about the trigger. Starting the supplement when the flare load starts seems backwards. You need it after the flare load stops.
Triggering by using a pressure sensor (per CJKruger) implicitly takes into account anything else which is in the system at the same time and doesn't waste gas when you don't have to. However, you DO have to be very careful about the STATIC pressures due to the chimney effect of the flare height, because most flares without a water seal pull a negative in the flare header (referenced against local ambient outside the pipe .. see my earlier missive).
The 150 degF reset is probably because whatever you are doing on a daily basis is enough to cover any shrinkage conditions from 150 degF down. When the temperature has been proven to go above that, the supplementary purge will be "armed".

I hope this is useful
David