Dinghuan Zhu — 2012-13 Fellow
With the increased connection of variable and intermittent renewable energy sources (RESs) such as wind and solar, the energy management mechanism to avoid an unacceptable increase in frequency deviations becomes much more challenging and needs to be revisited due to the limited prediction accuracy for these resources. Most of the renewable energy sources cannot be dispatched like conventional generators because their power output is highly dependent on environmental factors such as wind speed and solar irradiation. On the other hand, the fast ramping required for conventional generators to follow the RES power variations is increased significantly due to their variable and intermittent nature, which greatly increases wear and tear losses of the generating units.
The emerging large scale energy storage technologies are being recognized as a natural fit to overcome the challenges related to these variable and intermittent energy sources and have received increased attention. However, the question is how these new devices fit into the operational structure of power systems which has evolved over decades without consideration of large scale storage. Not only the power limits but also the energy limits on energy storage devices have to be taken into account.
This research work is dedicated to developing an optimal control framework for the energy management problem on multiple timescales to coordinate energy storage and intermittent resources and conventional generation. There are three major objectives. The first is to balance the power supply and consumption in real time, i.e. to maintain the frequency within an acceptable and safe region. The second objective is to optimally schedule conventional generation and energy storage to reduce the ramp rates of conventional generators and meantime to minimize the total operation costs of providing energy for the system. The third one is to maintain safe operation of storage devices.
We re-categorize the multi-timescale energy management into two levels: real-time frequency control and advanced economic dispatch. The level of real-time frequency control incorporates the main functions of the traditional primary control and secondary control and additionally takes into account the storage model and limits, while advanced economic dispatch on the timescale of several minutes is designed to optimally schedule generation and storage to serve the forecasted system demand in the most economical manner subject to various physical constraints and uncertainties of renewable generation.
We are planning to adopt two main methods to address the multi-timescale energy management problem. One is H∞-based robust control approach for real-time frequency control and the other is stochastic model predictive control (SMPC) for advanced economic dispatch. The robustness and stability of a dynamic system subject to bounded disturbances can be achieved simultaneously by the robust control whereas SMPC is able to optimize the expected system performance over a certain look-ahead time horizon using a model of the system under uncertainties.
In the next year, we will study and solve three key issues in the multi-timescale energy management problem. The first issue is related to the robust control design. The order of the yielding controller based on the standard robust control design is very high and makes the controller implementation impossible. We need to use techniques such as static output feedback to reduce the order of the robust controllers. The second issue is that the size of the resulting SMPC optimization problem is typically very large. We are proposing to apply the optimality condition decomposition to decompose the original problem into several subproblems and then to solve the subproblems in parallel. The third issue is regarding the interface between the real-time frequency control level and the advanced economic dispatch level. As the same energy storage device may be employed in both control levels, we need to investigate how to optimally coordinate the contributions of the storage device to the two levels so that the safe operation is still maintained and the two levels of control do not conflict with each other. The reason behind this issue is that the maximum power and energy that a storage device can provide is limited.