Universität zu Köln

Kienscherf, Philipp Artur ORCID: 0000-0001-7459-986X (2026). Essays on Market Design and Algorithms for Distributed and Sustainable Energy Systems. PhD thesis, Universität zu Köln.

PDF Kienscherf_Philipp_Artur_Dissertation_PUBLISHED.pdf - Published Version Download (5MB)

Abstract

The dissertation "Essays on Market Design and Algorithms for Distributed and Sustainable Energy Systems" investigates how the large-scale electrification of transport can be coordinated through digital markets and intelligent algorithms. As electric vehicles become ubiquitous, millions of charging decisions intertwine with the real-time dynamics of power grids, creating a cyber-physical system. The dissertation finds that this challenge can be addressed by more means than engineering grid capacity: it proposes new market designs and machine learning methods that translate physical constraints into economic signals, allowing autonomous software agents to coordinate their behavior efficiently and sustainably. The work is built on the insight that large-scale EV charging is a distributed decision problem. Each driver cares about having enough charge when needed, yet the collective charging pattern determines local grid stress, costs, and emissions. Instead of relying on centralized schedulers or detailed travel forecasts, the work develops decentralized mechanisms in which intelligent agents learn to act on real-time prices and limited local information. Through multi-agent reinforcement learning, household charging agents observe only their state of charge, time of day, and previous prices but still learn to bid effectively for charging power in repeated auctions. Simulations on realistic low-voltage grids show that simple linear learning agents can reach near-optimal outcomes within a few percentage points of a full-information benchmark while avoiding the instability and training burden of deep neural networks. This establishes a foundation for self-organizing, bottom-up control of electric loads. Moving beyond individual grid situations, the thesis then explores how market architecture itself can be made trustless and transparent. It develops a blockchain-based bundle trading market that enables EV owners to buy and sell time-specific charging rights without a central auctioneer, using smart contracts that guarantee correctness and auditability. The system achieves the same efficiency as centralized clearing but adds resilience and privacy, illustrating how distributed ledger technology can support critical energy services. The concept of market-based coordination is further generalized to decentralized autonomous organizations (DAOs) that manage both physical and financial resources. Here the dissertation designs a mechanism in which agents not only trade physical bundles but also issue contingent financial claims, allowing them to share risk while preserving autonomy and data privacy. The mechanism converges to the same equilibrium a risk-neutral central planner would choose, demonstrating that even complex, stochastic resource allocation can be governed through decentralized markets. The final part of the dissertation widens the lens to the national scale. Using Germany’s planned Deutschlandnetz of fast-charging stations as a natural laboratory, the dissertation develops a spatial competition model to understand how the location of stations, regional demand, and regulatory price caps shape prices and investment incentives. The analysis shows that uniform national price caps can unintentionally reduce service in sparsely populated areas and that lax ownership constraints can foster market power, harming consumer welfare. Policy options include regionally differentiated price caps, dynamic tendering processes, and real-time data-sharing requirements to foster both competition and equity. What unites these strands is a sociotechnical vision of the future energy system. Rather than treating markets and technology separately, the dissertation demonstrates that market design, machine learning, and physical infrastructure must be co-designed. By coupling incentive-compatible mechanisms with adaptive algorithms, it shows how autonomous agents can collectively manage scarce resources and uncertainty, achieving reliable and cost-effective charging without heavy-handed central control. The work thus provides a conceptual and methodological blueprint for digital energy platforms that are at once economically efficient, technologically scalable, and sustainable.

Item Type:	Thesis (PhD thesis)
Creators:	Creators Email ORCID ORCID Put Code Kienscherf, Philipp Artur philipp.kienscherf@gmail.com https://orcid.org/0000-0001-7459-986X UNSPECIFIED
URN:	urn:nbn:de:hbz:38-801598
Date:	2026
Language:	English
Faculty:	Faculty of Management, Economy and Social Sciences
Divisions:	Faculty of Management, Economics and Social Sciences > Business Administration > Information Systems > Professorship for Informations Systems and Operations Research
Subjects:	Economics Technology (Applied sciences)
Uncontrolled Keywords:	Keywords Language Smart Chaging Energy Markets Blockchain Reinforcement Learning Market Design Mechanism Design Smart Markets English
Date of oral exam:	3 February 2026
Referee:	Name Academic Title Ketter, Wolfgang Univ.-Prof. Dr. Lu, Yixin Prof. Dr. (PhD) Valogianni, Konstantina Prof. Dr. (PhD)
Refereed:	Yes
URI:	http://kups.ub.uni-koeln.de/id/eprint/80159