Thermoacoustic modeling and control of multi burner combustion systems

Abstract

Thermoacoustic interactions in industrial combustion systems are difficult to model because they involve complex interactions between several physical mechanisms. In order to obtain dynamic models of such systems, a hybrid approach is used: numerical, experimental and analytical techniques are combined to describe the system. The system is modeled as a modular network, where the input-output relation of the modules can be based on analytic models, experimental data or numerical analysis. The modules are represented as state-space realizations. A modal expansion technique is used to obtain a state-space representation of the acoustic propagation through complex 3-dimensional geometries. The modal expansion can be based on an analytic model (for relatively simple volumes), or on a finite element analysis (for geometries of any complexity). Modules that are very complex, such as the acoustic behavior of the combustion process itself, are modeled using a combined experimental and analytic approach. The method is not restricted to symmetries of any kind: configurations with geometrically or operationally different burners are simulated. The state-space network approach allows for time domain simulations, including non-linearities. An active controller has been synthesized for an (hypothetical) annular multi burner combustion system. The controller uses spatial filtering to decompose the acoustic field to its individual modes. The modes are then controlled using an H∞ control algorithm. Time domain simulations of this control system demonstrate the effectiveness of this method, even in the presence of non-linear saturation and parametric errors.