Incomplete flare combustion

[magsar]

Has anybody ever met the problem of incomplete flare combustion, sometimes even flame out, due to high volumes of inert gases such as nitrogen and steam being flared? It's particularly challenging during the time the plant is shuting down with numerous pieces of equipment being steamed out. An obvious solution is to monitor the BTU value of the flare gas and to add fuel gas as necessary. We have been working for q quite while trying to find a suitable tool for flare gas BTU monitoring. I am just wondering if anybody else shares the same problem/concern.Would appreciate your thoughts/comments!

 

[nickelkid]

A well designed pilot(s) with thermocouples will let you know what is happening. Many types are available and all applications.

 

[magsar]

Good point Robinchandy, pilots are on the top list re flare reliability. But the function of pilots is to provide a reliable source of ignition, not the heat content for keeping the main flame on. So even if you have well designed pilots, it is still possible that the combustion is not complete due to low heating value of the flare gas. That's why EPA requires that the heating value of the flare gas shall be kept at 300 BTU/SCF or above to get a satifactory destruction efficiency.

 

[Flareman]

The problem of flame out is much more common than people like to admit, even with continuous pilots which, by the way, are mandatory in the US (per EPA 40CFR60.18)

The 200 Btu/scf limit for unassisted flares and 300 Btu/scf limit for assisted flares, set by the above CFR is, indeed, intended to mitigate against poor combustion and flame out but it assumes that someone is able to monitor what's happening.
Also, with all assisted flares, there is a tendency for operators to set up the steam (or air) flow on any given day and then leave it alone for months, regardless of the flare flow rate, on the basis that when the plant manger can't see the flame, he won't be calling on the phone. That's a dangerous state of mind because if you can't see the flame, it may not be there. With steam assisted flares particularly, I've had people who say that they "hide" the flame in the steam plume. Forget it! If you can see the steam there ain't no flame!

Continuous pilot flames are very small (only about 50,000 to 100,000 btu/h each) and they can sometimes be easily snuffed out by large quantities of inert Nitrogen or steam coming from the flare tip and falling over the pilot flame. At least one manufacturer makes a design specifically intended to resist external snuffing but only as a special. If you are monitoring pilots (with thermocouples or optical devices) you may think about an automatic rapid reignition system.

In any event, you need to identify where the inert is coming from. Is it in the process flow or is it the assist medium?

If it's the latter, you need to educate the operators and, probably automate the steam addition, or tweak an existing automatic control to back off the flow and give you a permanently visible flame.

If it's in the process flow, you need to be able to boost the CV of the total. You might be able to do this by identifying specific inert flows and pre-empting them with a fixed fuel gas input. Otherwise, if you want to do this on-the-fly, you need to monitor the relief gas in the header and react to whatever is happening. That means using a high speed analyser. A company called Thermo ONIX claim to have one which will do this.

http://www.thermo.com/eThermo/CDA/Products/Product_Detail/1,1075,10292-136,00.html

I have never used one so this is not an endosement but you could check it out.


David