Abstract
Wind energy in Denmark covers 42% of the total power consumption in 2015, and will share up to 50% by 2020. Consequently, the conventional power plants are decommissioning. Under the progress of the green transition, the national decision leads to underground many overhead lines in the future transmission system. These issues initiate the infrastructure constructions of the transmission system, i.e. transmission lines over 100 kV in the specific areas will be undergrounded. Many switchable and variable shunts will be placed in the grid for fully compensating the capacitive cables. In order to ensure a highly reliable transmission, e.g. balancing the generation and the consumption in large geographic regions, the exchange capacities will be enlarged by upgrading the interconnections. The Danish power system, the electricity transportation hub between the Nordic and continental European systems, driving by the market, is highly flexible. As more and more renewable energy integrating to the system, the voltage control in the future power system is becoming a challenging objective.
This project considering the future challenges is to design a robust Automatic Voltage Control (AVC) system that maintains the voltage inside the acceptable band, optimally reserves the reactive power in the continuous reactive power components, and be operated in an economic way.
The review of the existing AVC systems is presented in the thesis. A centralized voltage control system with decentralized fallback controllers is proposed, to address the challenges in the Danish transmission system. A national control center is expected to periodically dispatch the setpoints to the reactive power components in a closed loop. The optimal setpoints of the reactive power components are obtained from a decision making process, which formulates an optimization problem. The objective of the Danish AVC system is to minimize the operational cost including the grid loss and the regulation cost of discrete components while reserving the reactive power in the continuous reactive power components as much as possible. The cost of the grid loss and regulations are converted to the expense, coupling to the price signal from the spot market, which provides more precise and meaningful decisions than the existing AVC systems. Moreover, as a market-driven system, the transits in the main corridors vary according to the market price, which leads to frequent adjustments on the voltage profile. In order to taking into account the system variability, the proposed AVC system is coupled to the forecasting system, making decisions based on several snapshots in a time horizon. Such multi-stage optimization is verified superior than the existing AVC systems in term of the cost minimization. In the multi-stage optimization based AVC system, the forecasted snapshots representing the load flow in the upcoming stages are involved. Due to the high share of the wind energy being in the system, the forecasting errors exist in the predicted stages. The voltage uncertainty caused by the wind power forecasting errors is estimated, which is applied as a voltage security margin to further constrain the voltage magnitude in the optimization problem. The problem under the uncertainty is therefore converted to a deterministic problem, which is strongly tractable for improving the robustness.
In case the centralized AVC system is malfunctioning, e.g. telecommunications fails, the distributed fallback controllers will take the control responsibility to maintain the voltage in the acceptable band. The fallback scheme installed at substations monitoring local voltage, switch in/out shunts following inverse time characteristics, maintaining the voltage without telecommunications. The entire AVC system robustness is therefore improved.
In addition to the AVC system study, there is a chapter introducing the implementation of the advanced load flow calculation method, i.e. Holomorphic embedding load flow method (HELM). It can solve the load flow problem analytically without an initial guess, and thereby avoid the numerical non-convergence problem. In this way, the HELM algorithm can be an alternative method for the Newton-Raphson method for the load flow calculation to provide more insight on the system conditions.
This thesis focuses on the problem formulation and implementation. The AVC decision making and HELM algorithms are implemented in Matlab®, and the fallback scheme is modelled in DIgSILENT® Powerfactory. The functionalities of the proposed AVC system are simulated based on the measurement data from the Energy Management System in the Danish electricity control center, where the network model represents the western Danish transmission system. The studies provide the understandings of the proposed Danish AVC solution that is superior to the existing AVC systems in term of the cost minimization and the robustness, which shed light to the industrial applications.
This project considering the future challenges is to design a robust Automatic Voltage Control (AVC) system that maintains the voltage inside the acceptable band, optimally reserves the reactive power in the continuous reactive power components, and be operated in an economic way.
The review of the existing AVC systems is presented in the thesis. A centralized voltage control system with decentralized fallback controllers is proposed, to address the challenges in the Danish transmission system. A national control center is expected to periodically dispatch the setpoints to the reactive power components in a closed loop. The optimal setpoints of the reactive power components are obtained from a decision making process, which formulates an optimization problem. The objective of the Danish AVC system is to minimize the operational cost including the grid loss and the regulation cost of discrete components while reserving the reactive power in the continuous reactive power components as much as possible. The cost of the grid loss and regulations are converted to the expense, coupling to the price signal from the spot market, which provides more precise and meaningful decisions than the existing AVC systems. Moreover, as a market-driven system, the transits in the main corridors vary according to the market price, which leads to frequent adjustments on the voltage profile. In order to taking into account the system variability, the proposed AVC system is coupled to the forecasting system, making decisions based on several snapshots in a time horizon. Such multi-stage optimization is verified superior than the existing AVC systems in term of the cost minimization. In the multi-stage optimization based AVC system, the forecasted snapshots representing the load flow in the upcoming stages are involved. Due to the high share of the wind energy being in the system, the forecasting errors exist in the predicted stages. The voltage uncertainty caused by the wind power forecasting errors is estimated, which is applied as a voltage security margin to further constrain the voltage magnitude in the optimization problem. The problem under the uncertainty is therefore converted to a deterministic problem, which is strongly tractable for improving the robustness.
In case the centralized AVC system is malfunctioning, e.g. telecommunications fails, the distributed fallback controllers will take the control responsibility to maintain the voltage in the acceptable band. The fallback scheme installed at substations monitoring local voltage, switch in/out shunts following inverse time characteristics, maintaining the voltage without telecommunications. The entire AVC system robustness is therefore improved.
In addition to the AVC system study, there is a chapter introducing the implementation of the advanced load flow calculation method, i.e. Holomorphic embedding load flow method (HELM). It can solve the load flow problem analytically without an initial guess, and thereby avoid the numerical non-convergence problem. In this way, the HELM algorithm can be an alternative method for the Newton-Raphson method for the load flow calculation to provide more insight on the system conditions.
This thesis focuses on the problem formulation and implementation. The AVC decision making and HELM algorithms are implemented in Matlab®, and the fallback scheme is modelled in DIgSILENT® Powerfactory. The functionalities of the proposed AVC system are simulated based on the measurement data from the Energy Management System in the Danish electricity control center, where the network model represents the western Danish transmission system. The studies provide the understandings of the proposed Danish AVC solution that is superior to the existing AVC systems in term of the cost minimization and the robustness, which shed light to the industrial applications.
| Original language | English |
|---|---|
| Supervisors |
|
| External collaborators | |
| Publisher | |
| Electronic ISBNs | 978-87-7112-756-0 |
| DOIs | |
| Publication status | Published - 2016 |