Coal Mill Modeling for Monitoring and Control

The more fluctuating renewable energy sources like solar and wind are integrated into the electric grid, the more it is desirable that conventional power plants actively contribute to compensating the fluctuations. For coal fired power plants, the response time of the coal mills is critical for the overall reaction time to changing demand. Model predictive control is a possible way to improve the dynamic response of coal mills. Moreover, dynamic models may be used to estimate unobserved process variables such as the particle size of the ground coal. Such information is valuable for optimizing the performance and minimizing the operation and maintenance costs of the pulverizers. A third benefit comes when using a combined model of an entire array of coal mills, in order to distribute the load in an optimal manner. In order to obtain models, we show a suitable nonlinear gray-box approach and demonstrate how parameters can be identified via least squares fitting.