Combustion involves inherently unsteady processes. Swirler vortex breakdown, turbulence, and heat release are all unsteady or time dependent phenomena. Unsteady heat release leads to unsteady density and pressure. If the pressure fluctuations are amplified, enough energy can be present in the system to cause hardware damage in the combustion system and surrounding components.
This article is extracted from the paper, “Gas turbine emissions improvements by advances in design, analysis, materials, manufacturing and control technology” by David M. Stansel of Solar Turbines at the 2018 Turbomachinery Symposium.
In a diffusion flame system, there exists great heterogeneity of fuel air ratio through the primary combustion zone volume. This leads to temperature spikes and pressure spikes that are randomly distributed and normally only generate low levels of acoustic energy as they cancel each other out.
To minimize NOx, a DLE system strives for perfect premixing. In a well-mixed, homogenous system, all the heat release happens at very-nearly the same temperature. Any disturbance to the system can then be amplified by a uniform rise or fall in temperature and pressure throughout the flame zone.
Since a DLE combustion system uses less wall cooling, the acoustic boundaries are hard and provide little damping. The reflection of the pressure fluctuations back to the fuel injector can cause a periodic rise and fall of fuel/air ratio, which in turn causes the heat release to fluctuate periodically. If the feedback loop becomes in phase, the pressure fluctuations are amplified and damage may occur. If the feedback loop is out of phase, then the system is damped and no damage occurs. This requirement for feedback to be in-phase for amplitudes to rise is referred to as the Rayleigh criterion.
It is very easy to predict the frequency of the system response, knowing the system dimensions and temperatures. Predicting the amplitude of the dynamics is much harder to do a priori. Flame location, stiffness of the fuel system, and natural system damping all come into play.
Adjusting hardware dimensions can change the system response, getting the feedback out of phase and quieting the system. But since a gas turbine operates over a range of loads, ambient temperatures, and with a variety of fuels, it can be difficult to avoid worrisome acoustic amplitudes under all conditions.
Solving this problem has become the single biggest technical challenge for the designers of DLE combustion systems. Unfortunately, the easy solutions often lead to compromises in one of the other key performance parameters. Raising the amount of pilot fuel, reducing air to the primary zone, or intentionally causing the premixing to degrade have all been successful at reducing oscillations amplitudes – and increasing NOx.
Increasing the amount of cooling air to enter the combustion primary zone helps damp the pressure reflections from liner walls, but also leads to compromises in turndown stability and/or CO emissions. Several approaches have been proposed to solve this problem.
One technique, for instance, alters the location of fuel injection to break the feedback loop. A change in premixer geometry could similarly impact the transport time in the fuel injector and hence break the feedback loop. Others have demonstrated actively pulsing a small amount of fuel out of phase with the oscillations can provide active damping. Still others have demonstrated adaptive control systems that adjust parameters such as fuel splits or air splits when dynamic activity is measured. Passive dampers, such as the resonators attached to a combustion liner have also been employed with success.
OEMs have developed proprietary approaches to tune systems, or damp them from destructive dynamic pressure fluctuations.