Stochastic Control Theory Optimization of Energy Systems

Research output: Book/ReportPh.D. thesisResearch

1 Downloads (Pure)

Abstract

Future low-carbon societies will be driven by renewable energy sources (e.g. wind and solar power). This will flip the characteristics of our power system from a production-tracking-consumption paradigm, to a consumption-tracking-production paradigm. This will increase the need of complex coordination of our power consumption as power grids require a strict balancing between power production and consumption. This dissertation investigates the potential of applying nonlinear model predictive control algorithms to solve complex power market coordination problems, where flexible consumers leverage the more volatile balancing power prices and thereby in-directly help neutralizing production and consumption imbalances. This dissertation only considers continuous-discrete stochastic models. Paper A provides a tutorial on how to formulate the entire algorithm-stack of nonlinear model predictive control algorithms where system dynamics are governed by stochastic differential equations with discrete-time observations. The techniques introduced in Paper A are applied in four case-studies relating to energy systems. Paper F introduces a new filtering technique that generalizes the observational model to contain general likelihood models. Paper B and Paper C propose an optimal control problem to operate the aeration equipment at wastewater treatment plants with the criteria to minimize the operational costs and the accumulated nutrient concentrations of the discharged effluent. Paper D considers the operation of a non-invasive ice-tank module added to a small retail refrigeration system located at Danfoss’ test-facility in Nordborg, Denmark. It is shown that the icetank is an efficient method for curtailing the power consumption of the refrigeration system, without re-arranging and modifying the entire piping and general hardware infrastructure. Paper E presents an optimization technique to optimally leverage
the volatile power prices observed in the Northern European regulating power market using a Vanadium redox-flow battery. This
paper shows that the payback time of investing in grid-scale flowbatteries are approximately seven years, when implementing this
optimization-based trading strategy.
Original languageEnglish
PublisherTechnical University of Denmark
Number of pages117
Publication statusPublished - 2021

Fingerprint

Dive into the research topics of 'Stochastic Control Theory Optimization of Energy Systems'. Together they form a unique fingerprint.

Cite this