|M.Sc Student||Zargari Noa|
|Subject||A Modified Dynamic Programming Algorithm for|
Optimal Control of Energy Storage Devices with
Limited Ramp Rate
|Department||Department of Electrical and Computer Engineering||Supervisor||ASSOCIATE PROF. Yoash Levron|
|Full Thesis text|
Renewable energy sources are a possible solution to climate change, the depletion of
fossil fuels, and high carbon emissions. Such sources are also likely to provide easier
access to electricity in countries and regions that lack the natural resources.
While the advantages of renewable sources are well established, there are also considerable challenges in integrating them, since current power stations, infrastructure and operation principles are compatible with centralized fossil fuel based generation. Renewable sources are naturally intermittent, which makes maintaining the balance between generation and consumption more challenging. Another challenge is that renewable sources are not sufficient to provide the entire energy demand, so conventional generation is still an essential part of energy production. Renewable production might provide excess energy which cannot be used or stored. In the case of solar energy, since production peaks around noon, conventional generators are required to ramp-up their production, in order to meet the demand when solar energy is not available. This causes conventional generation patterns to resemble a duck-like shape, hence its name, the ”duck curve”. The required increase in generated power is often unfeasible due to technical or economic reasons. This ramp-rate or ramp-up constraint is considered a major limiting factor in integrating renewable sources. A simple solution to this issue is curtailing renewable production when demand is low. However, since this approach may lead to increasing costs of renewable production, it is considered a hindering factor in integrating renewable sources.
One recently suggested application of energy storage is to smooth fluctuations in renewable production. The main idea is to store surplus energy when demand is low, and then to use it when generation is difficult or expensive. To acquire the full potential of energy storage, effective optimization tools are required to consider factors such as costs and duration and quality of operation. Dynamic programming is an effective and well-known optimization tool, which has a major drawback - the significant computational burden entailed with applying it in systems with several state variables, as required in operating energy storage while considering the ramp constraint.
To address this challenge, in this study we propose an optimal control method, based on dynamic programming, that is specifically designed to achieve low numeric complexity when dealing with ramp-rate constraints. The main idea is limiting the number of computations by applying an on-going “trimming process”, which limits the number of energy values being scanned. These values are further reduced by applying standard dynamic programming, ensuring that the selected solution is the globally optimal one, if it exists. The Israeli grid is the chosen case study for this study. The generated power pattern is evaluated for the year 2030, based on annual increases in electricity demand, and on government goals for integrating renewable sources. Simulations demonstrate that using energy storage may facilitate in integrating renewable sources by effectively increasing the grid ramp capability. The proposed algorithm is applied to evaluate its benefit for large scale storage systems, and based on standard algorithm criteria such as computation time and memory requirements.