I was recently asked by a customer whether or not an explosion as the result of a fuel rich firebox in a fired heater was a credible hazard that they should be designing their heaters to address. The customer is spending a lot of money on stack gas analyzers and the associated maintenance and testing to implement shutdowns that would detect a fuel rich condition, using O2/Combustibles measurements of the stack gas, and then shutting off the fuel gas if the concentrations exceeded a safe limit. I informed them that a firebox explosion due to a fuel rich firebox is definitely a credible scenario, as I am personally aware of a handful of these explosions (of course, the parties that were subject to these explosions shall remain nameless) but the risk associated with these explosions is often grossly overstated in many simple LOPA assessments, and implementation of a shutdown to avoid the consequences of these events is often unnecessary.
The story begins in the early 2000’s when there was a large move to the use of NOx heaters. These heaters and the associated burners were much “tighter” than older designs and had a lot less tramp air entering the firebox, resulting in more efficient and environmentally friendly combustion. I also noted a rash of firebox explosions (using the term very loosely) in our refining customer base. The accident scenario was nearly identical in almost all of the cases. What typically happened is that a natural draft furnace had its “air flow” set by adjusting registers and dampers to optimize the amount of excess O2 as measured by a stack gas analyzer. At a subsequent time, the of firing called for in the process increased, increasing the fuel gas consumption, but since the damper and register positions were set manually, they did not change, resulting in decreasing levels of excess oxygen in the flue. While the excess O2 decreased and combustibles increased, the oxygen never gets so low that the flame is actually extinguished. Eventually the low O2 draws the attention of operations. At this point, the correct action from operations would be to cut the fuel gas until the excess oxygen returns to a safe level, then the “airflow”, via dampers and registers would be increased, and finally the fuel rate would be returned to the rate called for by the increased process demand.
Unfortunately, a lot of times a poorly trained or poorly motivated operations staffer blows it. Instead of cutting fuel the operator responds by just giving the heater more air. If operations rapidly gives the heater more air, the inrush of excess oxygen meets up with the very hot unburned fuel and CO, and boom! Luckily for most of the people that have wandered into this scenario, the “explosion” is usually quite minor. The reason is that this reaction is mass transfer limited. The reaction rate, and the resultant release of energy, can only occur as rapidly as the oxygen and fuel can mix. The reaction occurs in a wave up through the firebox as the fresh air makes its way upward, along with the existing flue gas, up the heater and out the stack. This results in an event that is more akin to a glorified flash fire than a true deflagration, let alone an explosion. In my experience, the consequence of these events is limited to damaging the stack dampers and/or the inlet registers. In no case was there any damage to the physical structure of the heater, nor was there any personnel injury from any of these events.
So back to the original question. Is a firebox explosion as the result of a fuel rich firebox credible? Yes. But, the magnitude of these events is actually quite small, and the probability of their occurrence when operations staff is properly trained to cut fuel gas when oxygen is low, instead of increasing oxygen, the probability of their occurrence should also be low. As a result, risk analysis often yields the conclusion that automatic heater shutdowns to address this scenario are not necessary.