Model Predictive Control for Smart Grids
Project Leader: Prof. Dr. Lars Grüne
Contact: Prof. Dr. Lars Grüne
Dr. Philipp Braun
Project start: 2012 Project end: 2016
Project team:
- Prof. Dr. Lars Grüne
- Dr. Philipp Braun
External partners:
- Prof. Dr. Christopher M. Kellett, The University of Newcastle, Australia
- Prof. Dr. Steven R. Weller, The University of Newcastle, Australia
- Jun.-Prof. Dr. Karl Worthmann, Technische Universität Ilmenau
Support:
Universität Bayreuth, The University of Newcastle
Project description:
BACKGROUND
With recent advances in battery technology, it is expected that battery storage for storing electrical energy will be widely implemented on the residential level in the next decade. This project investigated - from a mathematical point of view - various optimization based ways for coordinating the use of such storage devices, mainly for mitigating the fluctuations in the generation of electrical power from renewable energy sources.
CONTRIBUTION TO THE MISSION OF MODUS
Within this project various hierarchical distributed optimal control algorithms for solving scheduling problems for charging and discharging large-scale networks of electrical storage units have been developed and analyzed. Primal and dual decomposition approaches, ADMM type and price based algorithms were investigated. These algorithms were implemented in a receding horizon fashion in order to yield model predictive control schemes. The use of formal methods allowed for rigorous convergence and performance results of the distributed optimization algorithms, thus rigorously ensuring proper functionality of these algorithms.
RESULTS
For a large variety of hierarchical distributed optimization algorithms and for different dynamic models for smart-grids, the project succeeded in establishing rigorous mathematical convergence results. Moreover, excellent performance of these algorithms in a model predictive control context was illustrated via extensive numerical simulations. For detailed results, we refer to Philipp Braun's PhD thesis.