Control issues in oxy-fuel combustion
MetadataVis full innførsel
Combustion of fossil fuels is the major energy source in todays society. While the use of fossil fuels is a necessity for our society to function, there has been an increasing concern on the emissions of CO2 resulting from human activities. Emissions of CO2 are considered to be the main cause for the global warming and climate changes we have experienced in recent years. To fight the climate changes, the emissions of CO2 must be reduced in a timely fashion. Strategies to achieve this include switching to less carbon intensive fuels, renewable energy sources, nuclear energy and combustion with CO2 capture. The use of oxy-fuel combustion is among the alternative post- and precombustion capture concepts, a strategy to achieve power production from fossil fuels with CO2 capture. In an oxy-fuel process, the fuel is burned in a mixture of oxygen and CO2 (or steam), leaving the exhaust consisting mainly of CO2 and steam. The steam can be removed by use of a condenser, leaving (almost) pure CO2 ready to be captured. The downside to CO2 capture is that it is expensive, both in capital cost of extra equipment, and in operation as it costs energy to capture the CO2. Thus it is important to maximize the effciency in such plants. One attractive concept to achieve CO2 capture by use of oxy-fuel, is a semi-closed oxy-fuel gas turbine cycle. The dynamics of such a plant are highly integrated, involving energy and mass recycle, and optimizing effciency might lead to operational (control) challenges. In these thesis we investigate how such a power cycle should be controlled. By looking at control at such an early stage in the design phase, it is possible to find control solutions otherwise not feasible, that leads to better overall performance. Optimization is used on a nonlinear model based on first principles, to compare different control structures. Then, closed loop simulations using MPC, are used to validate that the control structures are feasible. It is found that compared to a conventional gas turbine cycle, it is possible to change loads much faster. And if the right control structure is applied, it is possible to operate at part load with just a small loss in overall effciency. A central part in all gas turbines is the combustion chamber. It is well known that thermoacoustic instabilities can be a problem in combustion chambers, leading to large high frequency (up to several hundred Hz) pressure oscillations. Such pressure oscillations are unwanted, but their amplitude can be reduced by use of active control. The control problem is however challenging, due to the high frequencies involved. Experimental results have shown that the same problem is inherent to oxy-fuel combustion, possibly the instabilities are even worse. In the thesis we present results on stabilization of a conventional combustion chamber by use of feed-forward microjet air injection. A look-up table is necessary to stabilize the process in this way, and the solution is not very robust. The use of feedback could probably improve the results. To study thermoacoustic instabilities and control for oxy-fuel combustion, we developed a linear low order model of oxy-fuel combustion. The model is based on a modal discretization of the wave equation, combined with a flame model based on heat release from a thin wrinkled flame. Analysis on the model shows that injection of CO2 is a promising actuator to stabilize the instabilities. The CO2 should however be injected somewhere upstream of the flame, to allow the injected CO2 to mix with the reactants before burning. Although the resulting time delay is small, it is a problem because of the high frequencies involved. We use a frequency domain method to include the time delay in the control analysis of a simple linear combustion model. An extension to the method is proposed and used to find a set of stabilizing controller parameters, and to analyze robustness of the controlled process. Extensive closed loop simulations on an advanced combustion simulator shows that the applied analysis is relevant, and provides useful information about how an oxy-fuel process should be controlled.