Hierarchical MPC for Energy Management of Multi-Energy Systems

Hierarchical MPC for Energy Management of Multi-Energy Systems: Case Study Based on a Power-to-X Concept

Kaya, Oguzhan (TU Delft Mechanical, Maritime and Materials Engineering; TU Delft Delft Center for Systems and Control)

Keviczky, T. (mentor)
van der Roest, Els (graduation committee)
Vries, Dirk (graduation committee)

Delft University of Technology

Mechanical Engineering | Systems and Control

2020-02-18

All over Europe, the expansion of renewable energy sources is quickly proceeding, fueled by environmental and political motives. The power generated by renewables is heavily subject to the intermittency of the source, e.g. the availability of wind or solar irradiance. Consequently, electrical grids that rely on renewable sources alternate between periods of excess power availability and periods with lack of power production. Excess power is often curtailed or exported, whereas the shortage of power production must be imported or produced through the deployment of more expensive production units. Power-to-X strategies aim to utilise the excess power from renewables more effectively, by converting power to another energy carrier within the grid, e.g. heat or hydrogen. By allowing the transition of power to another energy carrier, the system is transformed into a so-called multi-energy system. Therefore, the system takes into account the multiple energy carrier's system characteristics and loads in an integrated way. This formulation allows for optimal scheduling of energy flows in the system while taking into account characteristics of each energy carrier, e.g., storage characteristics, time-varying costs or production emissions. The main focus of this thesis to deal with the tasks of an energy management system for the aforementioned system using Model Predictive Control. The Model Predictive Control framework allows real-time optimal scheduling while incorporating data-driven forecasts of future loads and generation in the grid. Due to the stochastic nature of these forecasts, this thesis also looks into extensions of Model Predictive Control that can cope with uncertainties. Furthermore, to handle the different timescales of the grid dynamics, Model Predictive Control for multi-timescale systems is investigated. In particular, the performance of a Heuristic Model Predictive Control scheme and a Hierarchical Model Predictive Control scheme on the control of a simulated Power-to-X based energy system are compared. The simulation is based on a conceptual Power-to-X system based on historical data of the Dutch energy sector. Based on this case study, a statement is made about the suitability of the Power-to-X principles to future Dutch sustainable neighbourhoods. Moreover, a statement regarding the economic viability of the presented concept is made based on the simulations.

Model Predictive Control
Power-to-X
Energy Management System
Aquifer Thermal Energy Storage (ATES)

http://resolver.tudelft.nl/uuid:28157eaa-7b4d-456c-8402-db00f651a12f

Student theses

master thesis

© 2020 Oguzhan Kaya