This week, I attended the American Flame Research Committee’s Combustion Symposium in Houston where I presented a paper on radiation modeling.
Most of the papers presented were about industrial flares. If you live near a process plant, you must have seen these large stacks reaching into the clear blue sky. At the end of these stacks are large flames that can be seen from a distance. For most urban area residents, these flares create a concern about public health and safety especially if there is some black smoke as well in the fireball.
So, for a day and a half, I sat in there and listened to the experts in this area – John Zink, CANMET, Zeeco and Hamworthy Combustion, to name a few. There were a few papers from the US Environmental Protection Agency (EPA) as well that dealt with regulations. There were a lot of questions, for obvious reasons, from the flare operators regarding the EPA guidelines on flare operations. I would highlight a few here.
1. The first and important criterion that EPA looks at is the net heating value of the vent gas. This can vary if the vent gas composition changes during the operation of the flare. As a result, operation parameters need adjustment.
2. For flares that are assisted with steam injection (to draw ambient air into the combustion zone), the steam-to-vent-gas-mass ratio is important.
3. Other parameters also come into play such as the flare tip nozzle injection diameter, central steam to upper steam ratios, wind conditions, etc.
The bottom line is that there is no universal setting of the steam flow rate to vent gas flow rate that would achieve complete combustion – the target of this flare operation. For flare operators, this is not a good news. Ideally, they want to have a look-up chart to control the flare operation. This is possible but not always. Each flare is different and requires custom settings and intelligent controls. The monitoring equipment still lacks a few things. The IR cameras can capture an image of the flame but they can’t “see” the species in it. So, two identical flare pictures can have very different CO concentrations. Another issue is with the color of the flames. A yellow flame is not necessarily a bad flame as long as it achieves the desired reduction of HC species to CO2 and H2O.
The ideal flare will not have any “flame” i.e. you can not “see” it. We call it flameless flare. And this is possible to achieve with the recent advances in automatic feedback combustion control systems. I saw some example of this flameless operation of the flare.
Now, down to my last point. The industry is using new tools and the use of CFD is one main component of it. Almost all papers showed the use of a CFD tool in the design process of these flares. The overall problem is difficult and requires capturing of transients in the flare using advanced turbulent flow models. Chemistry predictions is another area of importance. For example, Philip J. Smith from the University of Utah has worked on modeling a lab flare in a horizontal domain and for this small scale experiment, he ended up using thousands of compute nodes and several weeks to produce a few seconds of the actual flare operation. So, this is a challenging problem to solve but has a lot of promise in terms of understanding the intricacies of turbulence and chemistry interaction in industrial flares.