Predictive Control of an Integrated Energy Storage Facility

The research activity of this PhD proposal is aimed at designing and testing advanced control methods for the real-time management of the above-mentioned IESTV facility, that is being set up as a part of the University of Leicester participation to ERA, with a total equipment budget of £395k. The facility will include state-of-the art energy storage devices, and give the opportunity to design and test advanced algorithms that can have an impact on real-life implementations. The combined use of multiple storage systems [1] (more precisely: electrochemical batteries, supercapacitors, compressed air energy storage, and pumped hydro storage) can simultaneously exploit all of their advantages, but requires the ability to manage their complex interaction in real time. Storage systems have been studied in recent years, mainly in the fields of automotive engineering (electric and hybrid-electric vehicles) and microgrids [2-3]. However, the number of experimental implementations of innovative management strategies is still limited, due to the lack of availability of advanced testing facilities. The proposed PhD activity will be aimed at designing a strategy that takes into account the practical characteristics of the IESTV.

The PhD activity will consist of three main steps, each corresponding to one year of activity:

Year 1: study of the theoretical aspects of MPC and of modelling of storage elements. Derivation of a mathematical model of the IESTV suitable for control purposes. The derivation of this model will be based on running experiments (e.g., measuring the variation of the battery state of charge in different charging/discharging scenarios), collecting data from them, and using data analysis techniques to obtain suitable dynamical models. This is a crucial step, which will benefit from the experience of Dr. Lefley in modelling and control of energy storage devices.
Year 2: definition of a computer simulation of the IESTV, and experimental validation of the derived model. Testing of a simple MPC control law in simulation. This phase will require the use of numerical optimisation algorithms for defining, in real time, the best way to manage the storage facility.
Year 3: Experimental implementation of the MPC control law on the IESTV, and refinements. This phase will require controller tuning, plus possibly a refinement of the models identified during Year 1, if necessary.

The energy flow to/from the IESTV will depend on the power demand, and the power produced by photovoltaic renewable sources. The designed management system will carry out the tasks of (a) deciding in real time how much power to be draw from the grid, or possibly “selling” electricity to the grid itself in case of excess storage, and (b) how to split the storage between the elements taking operative constraints into account (e.g., limiting the charge/discharge cycles on batteries, or the depth of charge/discharge of the different elements). The management problem will be solved by continuously predicting the consumer demand and the future power flow from renewable energy sources (via weather forecast), and re-planning the optimal strategy in order to minimise the total energy consumption, or the overall expense for buying electricity from the grid.

This approach is known as Model Predictive Control (MPC), a control technique that has been applied to many practical problems in industry, and is now, for instance, a standard in petrochemical plants. MPC is the main research interest of Dr Rubagotti, who has already experimentally applied this technique for managing a different storage facility within his previous assistant professor position, as PI for grant “Integration, Automation and Control of Renewable Power Sources” (274k USD, 3 years project). In this project he coordinated a group of 5 researchers and 3 academics: this experience (which led to the publication of [5]) will be a valuable asset for the considered PhD activity.