Modeling and Optimal Control of Biomass Power Plant for Better Steady-State Error and Disturbance Rejection

Abstract

This thesis describes the development and testing of a dynamic system model and optimal controls for a solid biomass fuel power plant.The motivation for the model is to derive optimal Multi-Input, Multi-Output (MIMO) controllers intended to improve the steady-state output error of the power plant states and help reduce the effect of disturbances in the fuel stream. This model is developed primarily from first principles with some testing on a specific power plant used as a case study. The sensitivity of the model to the parameters is investigated. The model is used to derive Linear Quadratic Gaussian (LQG) servo controllers in a cascade feedback arrangement to control all of the steam states, combustion states, and outputs. The optimal controllers were implemented on a power plant and tested to compare their performance to the original hand-tuned Single-Input, Single-Output (SISO) controllers running the plant. The two control schemes were tested and compared for output error performance and disturbance rejection. The optimal controllers show a marked improvement over the SISO controllers in both categories. The standard deviations of the outputs are computed and compared. The optimal controllers reduce the output errors by 48.9\% and 20.6\% on average for steady-state and disturbance rejection for the four outputs, respectively.