Network-Secure Consumer Bidding in Energy and Reserve Markets

Abstract

Electricity systems are undergoing a fundamental transformation from centralised generation to a distributed paradigm in which electricity is produced at a smaller scale by numerous distributed energy resources (DER). The replacement of centralised facilities by DER brings economic and environmental benefits. However, it also makes it challenging for the market operator to secure the system with sufficient frequency response in the absence of centralised facilities -- dominant providers of such services -- in the electricity markets.

Fortunately, the aggregate response of DER can fulfil systems' need for frequency reserve services. However, DER are operated within distribution networks whose technical limits are not accounted for within the wholesale market. This raises the question of how DER can participate in the energy and reserve markets while respecting the distribution network's constraints. To ensure network constraints, consumer and grid constraints / preferences should be modelled simultaneously within a large-scale optimisation problem. Yet, the need for scale, involvement of multiple stakeholders (grid operator and consumers) who possibly have conflicting interests, privacy concerns, and the uncertainty around consumer data and market prices make this extra challenging.

This thesis contributes to addressing these challenges by developing network-secure consumer bids that account for the distributed nature of the problem, consumer data and market price uncertainties. Note that when bidding in the market, consumers, and thus, the network operating point is not clear, as it depends on the dispatch in the energy market and whether a contingency occurs. Therefore, we ensure grid feasibility for operating envelopes that include any possible operating points of consumers.

We first use the alternating direction method of multipliers (ADMM) to enable network-secure consumer biding. Using ADMM, consumers optimise for their energy and reserve bids and communicate with the grid their required operating envelopes. The network then solves OPFs to see whether any constraint is violated and updates the ADMM parameters. Such communications continue until converging on a consensus solution. We learnt that our ADMM-based solution approach is able to maintain grid's constraints as long as consumers commit to their envelopes -- a requirement that might not hold due to uncertainty. Thus, we further improve our bidding approach by modelling uncertainties around solar PV and demand, using a piecewise affinely adjustable robust constrained optimisation (PWA-ARCO). We observed that not only is PWA-ARCO able to compensate for live uncertainty variabilities, but also it can improve the reliability of consumer bids, especially in reserve markets. We also extend our initial envelopes by enabling consumers to provide reactive power support for the grid.

We next enable consumers to bid (possibly) their entire flexibility by developing price-sensitive offers. Such offers include a bid curve chunked into several capacity bands, each being submitted at a different price. We identified that when the prices cannot be forecast accurately, the price-sensitive bidding approach can improve consumer benefit. To ensure network feasibility, instead of an iterative ADMM approach, we propose a more scalable one-shot policy in which the network curtails the part of the consumer bid that violates the network. Compared to ADMM, the one-shot policy significantly reduced the computation complexity at the cost of a slightly less optimum outcome.

Overall, this thesis investigates different techniques to provide network-secure energy and reserve market services out of residential DER. It expands the knowledge in the area of consumer bidding solutions, adjustable robust optimisation, and distributed optimisation. It also discovers a range of interesting future research topics, including distribution network modelling and uncertainty characterisation. Show more