In the past decade, there has been sustained efforts around the globe in developing renewable energy-based generation in power systems. However, many renewables such as wind and solar are variable resources. They pose significant challenges to near real-time power system operations. This dissertation focuses on introducing and testing advanced scheduling algorithms for electric power systems with high penetration of variable resources. A novel predictive and optimal corrective look-ahead dispatch framework for real-time economic operation is proposed. This dissertation has four key parts. First, the basic framework of look-ahead dispatch is introduced. Different from conventional static economic dispatch, look-ahead dispatch is the fundamental function for future power system scheduling. Taking the whole dispatch horizon into account, look-ahead dispatch has a better economic performance in scheduling the resources in power systems. The decision-making of look-ahead dispatch is cost-effective, especially when handling with high penetration of variable resources. Second, we study the benefits of look-ahead dispatch in system security enhancement. An early detection algorithm is proposed to predict and identify potential security risks in the system. The proposed optimal corrective measures can be computed to prevent system insecurity at a minimized cost. Early awareness of such information is of vital importance to the system operators for taking timely actions with more flexible and cost-effective measures. Third, novel statistical wind power forecast models are presented, as an effort to reduce the uncertainty of renewable forecast to support the look-ahead economic dispatch and security management. The forecast models can produce more accurate forecast results by leveraging the spatio-temporal correlation in wind speed and direction data among geographically dispersed wind resources. Fourth, we propose a stochastic look-ahead dispatch (LAED-S) module to handle the high uncertainty in renewable resources. Even with state-of-the-art forecast technology, the near-real-time operational uncertainty from renewable resources cannot be eliminated. Given the uncertainty level, a conventional deterministic approach is not always the best option. The proposed LAED-S is able to judge whether a stochastic approach is preferred. The innovative computation algorithm of LAED-S leverages the progressive hedging and L-shaped method to produce good stochastic decision-making in a more efficient manner. Numerical experiments of a modified IEEE RTS system and a practical system are conducted to justify the proposed approaches in this dissertation. This framework can directly benefit the power system operation in moving from a static, passive real-time operation into a predictive and corrective paradigm.
In the past decade, there has been sustained efforts around the globe in developing renewable energy-based generation in power systems. However, many renewables such as wind and solar are variable resources. They pose significant challenges to near real-time power system operations. This dissertation focuses on introducing and testing advanced scheduling algorithms for electric power systems with high penetration of variable resources. A novel predictive and optimal corrective look-ahead dispatch framework for real-time economic operation is proposed.
This dissertation has four key parts. First, the basic framework of look-ahead dispatch is introduced. Different from conventional static economic dispatch, look-ahead dispatch is the fundamental function for future power system scheduling. Taking the whole dispatch horizon into account, look-ahead dispatch has a better economic performance in scheduling the resources in power systems. The decision-making of look-ahead dispatch is cost-effective, especially when handling with high penetration of variable resources.
Second, we study the benefits of look-ahead dispatch in system security enhancement. An early detection algorithm is proposed to predict and identify potential security risks in the system. The proposed optimal corrective measures can be computed to prevent system insecurity at a minimized cost. Early awareness of such information is of vital importance to the system operators for taking timely actions with more flexible and cost-effective measures.
Third, novel statistical wind power forecast models are presented, as an effort to reduce the uncertainty of renewable forecast to support the look-ahead economic dispatch and security management. The forecast models can produce more accurate forecast results by leveraging the spatio-temporal correlation in wind speed and direction data among geographically dispersed wind resources.
Fourth, we propose a stochastic look-ahead dispatch (LAED-S) module to handle the high uncertainty in renewable resources. Even with state-of-the-art forecast technology, the near-real-time operational uncertainty from renewable resources cannot be eliminated. Given the uncertainty level, a conventional deterministic approach is not always the best option. The proposed LAED-S is able to judge whether a stochastic approach is preferred. The innovative computation algorithm of LAED-S leverages the progressive hedging and L-shaped method to produce good stochastic decision-making in a more efficient manner.
Numerical experiments of a modified IEEE RTS system and a practical system are conducted to justify the proposed approaches in this dissertation. This framework can directly benefit the power system operation in moving from a static, passive real-time operation into a predictive and corrective paradigm.
