Modeling and analysis of the role of energy storage for renewable integration
- The renewable energy generated from wind and solar radiation is increasing rapidly. The power imbalance due to the mismatch between the available renewable generation and load is a significant challenge to the reliable operation of the electric power grid. When the renewable generation is too high, the excess power has to be curtailed. When there is not enough renewable generation, conventional generation from gas turbines can be used to balance the excess load. However, this offsets the environmental benefits of renewable generation, and there is operation cost incurred. Energy storage can be charged by the excess renewable generation and be discharged to meet the excess load and hence reduces the power imbalance. What are the optimal control policies of the electric power system when energy storage is present? How much power imbalance can energy storage reduce? How much cost of integrating renewable generation can energy storage reduce? How much energy storage is needed? This dissertation provides some quantitative answers to these questions by establishing and analyzing simple analytic models. The models are based on the multi-timescale operation of the electric power grid. We define the net load at an operating interval as the difference between the load and the renewable generation. The predictions of the net load are made at multiple timescales, such as, day ahead, hour ahead, and minutes ahead of the operating interval. The power imbalance at each timescale is the difference between the predicted net loads of adjacent timescales. The conventional generation and energy storage charging and discharging operations are scheduled to reduce the power imbalance at each timescale. We first consider the power imbalance problem at each timescale separately and formulate it as an infinite horizon stochastic control problem. The input is the power imbalance process, the controls are the charging and discharging operations of the energy storage, and the output is the residual power imbalance. We show that the greedy control policy of energy storage minimizes the average magnitude of the residual power imbalance. Modeling the power imbalance process as a sequence of independently and identically distributed Laplace random variables at short timescales and as a weakly dependent stationary process at longer timescales, we establish closed form expressions of the minimum cost functions for some cases. We corroborate the analytic results with numerical results using wind power dataset from National Renewable Energy Laboratory (NREL) and Bonneville Power Administration (BPA). To optimize the scheduled generation across multiple timescales, we model the electricity markets as follows. As the operating interval approaches, we have more observations enabling us to predict the net load at the operating interval more accurately. However, the price for scheduling generation for the operating interval at the electricity market increases. We wish to minimize the cost for scheduling generation for the operating interval at multiple electricity markets and the risk from the mismatch between the random net load and the scheduled generation at real time. This risk limiting dispatch problem can be formulated as a finite horizon stochastic control problem, and the optimal policy is a simple threshold policy. We extend the risk limiting dispatch framework to include energy storage. To simplify the analysis, we consider fast-response energy storage and slow energy storage separately. We first consider the risk limiting dispatch problem for a single operating interval in which the fast-response energy storage can be operated at real time during the operating interval to reduce the risk. We show that the optimal policy is still a threshold policy and develop efficient algorithms to compute the thresholds. Next, we consider the risk limiting dispatch problem for multiple operating intervals in which the slow storage can be charged by excess generation at one interval and be discharged to meet the excess load at a later interval. We develop two approximate algorithms and compare them using numerical examples.
|Type of resource
|electronic; electronic resource; remote
|1 online resource.
|Stanford University, Department of Electrical Engineering.
|El Gamal, Abbas A
|El Gamal, Abbas A
|Van Roy, Benjamin
|Van Roy, Benjamin
|Statement of responsibility
|Submitted to the Department of Electrical Engineering.
|Thesis (Ph.D.)--Stanford University, 2013.
- © 2013 by Han-I Su
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...