Direct current distribution systems (DCDS) are a promising alternative to alternating current (AC) systems because they remove AC--DC conversion between sources and loads that cause energy losses. Compared to AC systems, a DCDS has higher power capacity, energy efficiency and reliability, and no need for synchronisation---suitable where a large amount of renewable power is generated and consumed locally in DC.
A DCDS has unique features that affect its implementation: low system inertia, strict power limits and power--voltage coupling. Hence, simply applying markets designed for AC cannot guarantee a DCDS's supply security and voltage stability. This dissertation aims to identify DC-tailored local market designs that facilitate a DCDS's operational efficiency and reliability under uncertainty.
To identify promising DCDS market designs from all feasible options, we developed and applied a comprehensive design framework for local electricity markets. It is based on an engineering design process of identifying goals, determining design space, testing and evaluation. Whereas previous studies focused on individual commodities, we widened the scope to include the role of market architecture. Its main element is the choice of sub-markets for energy delivery, the provision of DC-substation capacity, and voltage regulation. For each selected sub-market, we analysed the design options for the general organisation, bid format, allocation and payment, and settlement. Considering the design complexity, we performed three rounds of market design according to the agile development principle: a qualitative assessment, a quantitative analysis without uncertainty, and a quantitative analysis under uncertainty.
In Step 1, we analysed the design options and identified three types of DCDS market designs according to the above framework, each featuring a unique architecture. First, the integrated market (IM) design explicitly links three sub-markets (for energy, substation capacity and voltage regulation) to incorporate all system costs into energy prices. It aims to create price signals that encourage prosumers to resolve congestion and voltage issues, but the challenges are privacy concerns and sophisticated market clearing. Second, the locational energy market (LEM) design relieves congestion with nodal prices--by linking the energy and substation capacity markets--whereas a system operator regulates the voltage. Third, the wholesale energy price (WEP) market design passes such prices directly to local prosumers, whereas the system operator resolves all network issues.
In Step 2, we quantitatively analysed how the market design addresses DC technical characteristics, such as volatile energy prosumption that challenges DC-substations. We built a deterministic optimisation model to evaluate three market designs, with a one-minute resolution to reflect the local prosumption volatility. Recognising that both total demand and demand flexibility may increase significantly in the future, we included a high share of electric vehicles (EVs) to test the market robustness. Simulations of a realistic urban DCDS demonstrated that the IM and LEM designs manage network congestion and voltage deviation even with a large share of EVs. It is found out that the main challenge to distribution-level market design is network congestion, mainly due to flexible prosumption at low-price hours. Voltage deviation and cable power capacity are not limiting factors of an urban DCDS market design. However, simply passing wholesale prices to local prosumers (like in the WEP design) is discouraged, as it may cause severe congestion and substantial flexibility investments.
In Step 3, we demonstrated the economic efficiency and reliability of the LEM design also under uncertainty. The performance of a local energy market is dominated by the uncertainty from stochastic local power prosumption, fluctuating wholesale energy prices, and unforeseen EV availability. We presented a novel agent-based model to evaluate the LEM design's performance in realistic scenarios. This model describes typical electric-vehicle user preferences and their bidding strategies with different levels of range anxiety. To stress-test LEM, we created challenging scenarios with a high share of solar generation and EVs. It performed efficiently and reliably in simulations, based on the high-resolution 2018 Pecan Street database and the IEEE European Low Voltage Distribution Test Feeder, even with a high share of EVs. We demonstrated that regardless of the bidding strategy, the LEM achieves efficient DCDS operation, as long as the network constraints are not too tight. Hence, we conclude that the simple LEM design---with only price--quantity bids and DC-substation capacity constraints---is the best feasible option among the three designs.
Although both DCDS technologies and the concept of local energy markets are still under development, we presented viable market solutions based on the best practices in the emerging DC technology, thereby clearing its market-side implementation barrier. The most economically-efficient yet technically feasible market design, at least in urban DCDS applications, is the LEM design. It supports fast market clearing and real-time control over flexible devices to resolve DC substation congestion. Other market designs, namely the IM and WEP, were proven to have practical limitations.
In the future, we recommend testing, improving and verifying the LEM design in field tests with real prosumers and various flexibility sources. This dissertation made assumptions and simplifications on both the technical system and the market operation, thereby leaving room for further development. First, the optimisation model and the agent-based model could be improved to enable more realistic market simulations. Second, a simple, user-friendly yet efficient agent module should be developed to enable high-frequency energy transactions in a DCDS. Third, follow-up research should estimate upon prosumers' bidding and investment incentives: the impact of additional price components---transmission and distribution system costs, national taxes and levies. Fourth, we should also evaluate the influence of prosumer values---including privacy, energy equality and energy self-sufficiency---on the local energy market design.