Power system adequacy: on two-area models and the capacity procurement decision process

Abstract

In this work, we explore methodological extensions to modelling practices in power system adequacy for single-area and two-area systems. Specifically, we build on top of some of the practices currently in use in Great Britain (GB) by National Grid, framing this in the context of the current technological transition in which renewable capacity is gradually replacing a considerable share of fossil-fuel-based capacity.

We explore two-area extensions of the methodology currently used in GB to quantify risk in single-area models. By doing this, we also explore the impact of shortfall-sharing policies and wind capacity on risk indices and on the value of interconnection. Furthermore, we propose a model based on the statistical theory of extreme values to characterise statistical dependence across systems in both net demand (defined as power demand minus renewable generation) and capacity surpluses/deficits (defined as power supply minus demand), looking at how statistical dependence strength influences post-interconnection risk and the capacity value of interconnection. Lastly, we analyse the risk profile of a single-area system as reliance on wind capacity grows, looking at risk beyond the standard set of risk indices, which are based on long-term averages. In doing this, we look at trends which are overlooked by the latter, yet are of considerable importance for decision-makers. Moreover, we incorporate a measure of the decision-maker's degree of risk aversion into the current capacity procurement methodology in GB, and look at the impact of this and other parameters on the amount of procured capacity.

We find that shortfall-sharing policies can have a sizeable impact on the interconnector's valuation in terms of security of supply, specially for systems that are significantly smaller than their neighbours. Moreover, this valuation also depends strongly on the risk indices chosen to measure it. We also find that the smoothing effect of parametric extreme value models on tail regions can have a material effect on practical adequacy calculations for post-interconnection risks, and that assumed independence between conventional generation fleets makes capacity shortfall co-occurrences only weakly dependent (in a precisely defined sense) across areas despite much stronger statistical dependence between system net demands. Lastly, as more wind capacity is installed, we find multiple relevant changes in the (single-area) system's risk profile that are not expressed by the standard risk indices: in particular, we find a substantial increase in the frequency of severe events, extreme year-to-year variability of outturn, and a progression to a system with fewer days of potentially much larger shortfalls. Moreover, we show that a high reliance on wind introduces a substantial amount of uncertainty into the calculations due to the limited number of available historic years, which cannot account for the wide range of possible weather conditions the system could experience in the future. Lastly, we also find that the a higher reliance on wind generation also impact the capacity procurement decision process, potentially making the amount of procured capacity considerably more sensitive to parameters such as the value of lost load.