The growing share of renewable energy sources drives the need for increased flexibility in the energy systems. The flexibility provision from thermal plants is limited by the boilerâ€™s thermal inertia as a bottleneck. Advanced controllers, such as model predictive control (MPC), have been identified as potential flexibility enablers. Fuel properties are crucial input for controllers. This work investigated the feasibility of using the properties obtained online by using near infrared spectroscopy based soft sensor to further improve the control performance. The performance of the existing proportional integral (PI) controller is compared with those of 2 feed forward (FF) MPC controllers. Both FF MPCs have significant improvement compared to PI controller and the FF MPC based on the full elemental composition shows the best performance due to more complete fuel information. There is a potential for revenues improvement with advanced control up to 1050 euros for one operation day.
Keywords biomass fuel, model predictive control, feed forward, near infrared spectroscopy, soft sensor, dynamic model