Abstract
The main goal of the project has been to evaluate the potential of a coordinated master control for a solid fuel power plant in terms of tracking capability, stability and robustness. The control strategy has been model-based predictive control (MPC) and the plant used in the case study has been the Vattenfall power plant Idbaecken in Nykoeping. A dynamic plant model based on nonlinear physical models was used to imitate the true plant in MATLAB/SIMULINK simulations. The basis for this model was already developed in previous Vattenfall internal projects, along with a simulation model of the existing control implementation with traditional PID controllers. The existing PID control is used as a reference performance, and it has been thoroughly studied and tuned in these previous Vattenfall internal projects. A turbine model was developed with characteristics based on the results of steady-state simulations of the plant using the software EBSILON. Using the derived model as a representative for the actual process, an MPC control strategy was developed using linearization and gain-scheduling. The control signal constraints (rate of change) and constraints on outputs were implemented to comply with plant constraints. After tuning the MPC control parameters, a number of simulation scenarios were performed
More>>
Citation Formats
Boman, Katarina, Thomas, Jean, and Funkquist, Jonas.
Model-Based Power Plant Master Control.
Sweden: N. p.,
2010.
Web.
Boman, Katarina, Thomas, Jean, & Funkquist, Jonas.
Model-Based Power Plant Master Control.
Sweden.
Boman, Katarina, Thomas, Jean, and Funkquist, Jonas.
2010.
"Model-Based Power Plant Master Control."
Sweden.
@misc{etde_1004335,
title = {Model-Based Power Plant Master Control}
author = {Boman, Katarina, Thomas, Jean, and Funkquist, Jonas}
abstractNote = {The main goal of the project has been to evaluate the potential of a coordinated master control for a solid fuel power plant in terms of tracking capability, stability and robustness. The control strategy has been model-based predictive control (MPC) and the plant used in the case study has been the Vattenfall power plant Idbaecken in Nykoeping. A dynamic plant model based on nonlinear physical models was used to imitate the true plant in MATLAB/SIMULINK simulations. The basis for this model was already developed in previous Vattenfall internal projects, along with a simulation model of the existing control implementation with traditional PID controllers. The existing PID control is used as a reference performance, and it has been thoroughly studied and tuned in these previous Vattenfall internal projects. A turbine model was developed with characteristics based on the results of steady-state simulations of the plant using the software EBSILON. Using the derived model as a representative for the actual process, an MPC control strategy was developed using linearization and gain-scheduling. The control signal constraints (rate of change) and constraints on outputs were implemented to comply with plant constraints. After tuning the MPC control parameters, a number of simulation scenarios were performed to compare the MPC strategy with the existing PID control structure. The simulation scenarios also included cases highlighting the robustness properties of the MPC strategy. From the study, the main conclusions are: - The proposed Master MPC controller shows excellent set-point tracking performance even though the plant has strong interactions and non-linearity, and the controls and their rate of change are bounded. - The proposed Master MPC controller is robust, stable in the presence of disturbances and parameter variations. Even though the current study only considered a very small number of the possible disturbances and modelling errors, the considered cases are good indications of robustness. - The Master MPC controller does not require excessive computational requirements and should be possible to implement in a control operation environment. On a standard desktop simulation computer, computation of a control input takes around 40 ms, which is 75 times faster than the sampling time of 3 seconds. - Even though the tuning of an MPC controller is straightforward in many respects, for example balancing the different controlled variables, the procedure also include some less obvious parameter settings and some experience on the power plants and on MPC is needed to tune the optimization problem parameters in the right way. The target groups for this study are both plant management, showing the potential of model-based solutions for the plant master control, and process engineers, giving valuable information on implementation issues for model-based predictive control solutions}
place = {Sweden}
year = {2010}
month = {Aug}
}
title = {Model-Based Power Plant Master Control}
author = {Boman, Katarina, Thomas, Jean, and Funkquist, Jonas}
abstractNote = {The main goal of the project has been to evaluate the potential of a coordinated master control for a solid fuel power plant in terms of tracking capability, stability and robustness. The control strategy has been model-based predictive control (MPC) and the plant used in the case study has been the Vattenfall power plant Idbaecken in Nykoeping. A dynamic plant model based on nonlinear physical models was used to imitate the true plant in MATLAB/SIMULINK simulations. The basis for this model was already developed in previous Vattenfall internal projects, along with a simulation model of the existing control implementation with traditional PID controllers. The existing PID control is used as a reference performance, and it has been thoroughly studied and tuned in these previous Vattenfall internal projects. A turbine model was developed with characteristics based on the results of steady-state simulations of the plant using the software EBSILON. Using the derived model as a representative for the actual process, an MPC control strategy was developed using linearization and gain-scheduling. The control signal constraints (rate of change) and constraints on outputs were implemented to comply with plant constraints. After tuning the MPC control parameters, a number of simulation scenarios were performed to compare the MPC strategy with the existing PID control structure. The simulation scenarios also included cases highlighting the robustness properties of the MPC strategy. From the study, the main conclusions are: - The proposed Master MPC controller shows excellent set-point tracking performance even though the plant has strong interactions and non-linearity, and the controls and their rate of change are bounded. - The proposed Master MPC controller is robust, stable in the presence of disturbances and parameter variations. Even though the current study only considered a very small number of the possible disturbances and modelling errors, the considered cases are good indications of robustness. - The Master MPC controller does not require excessive computational requirements and should be possible to implement in a control operation environment. On a standard desktop simulation computer, computation of a control input takes around 40 ms, which is 75 times faster than the sampling time of 3 seconds. - Even though the tuning of an MPC controller is straightforward in many respects, for example balancing the different controlled variables, the procedure also include some less obvious parameter settings and some experience on the power plants and on MPC is needed to tune the optimization problem parameters in the right way. The target groups for this study are both plant management, showing the potential of model-based solutions for the plant master control, and process engineers, giving valuable information on implementation issues for model-based predictive control solutions}
place = {Sweden}
year = {2010}
month = {Aug}
}