Climate Policy and Strategic Operations in a Hydro-Thermal Power System.

• Date: 01 September 2023
• Author: Moghimi, Farzad Hassanzadeh

1. INTRODUCTION

   Future climate packages in OECD countries envisage deep decarbonisation of the power sector by relying upon variable renewable energy (VRE) sources, viz., wind and solar power. For example, the European Green Deal aims for a 55% reduction in overall greenhouse gas emissions by 2030 relative to 1990 levels (European Commission, 2021). This 2030 target is in itself a pathway to a climate-neutral continent by the year 2050 with further electrification of the heating and transport sectors. In practice, such a transformation of the power sector requires unprecedented investment in flexible resources, e.g., storage, and transmission capacity to mitigate spatio-temporal imbalances between supply and demand.

   While such a challenge could have been surmounted in the era of vertically integrated utilities that coordinated all aspects of power sectors, the past forty years have further seen deregulation of electricity-industry services such as generation and retailing that were suitable for competition (Wilson, 2002). Meanwhile, the distribution and transmission services have been left largely state regulated due to their natural-monopoly characteristics. Thus, there is the potential for conflicting objectives among distinct agents as power companies aim to maximise their profits via generation investment and operations, which may not be aligned with society's objective of maximising social welfare in pursuit of climate goals. Consequently, policymakers require an understanding of how market participants' incentives, e.g., to behave strategically, affect economic and environmental outcomes.

   Instances of such conflicts stemming from the sustainable energy transition are prevalent in the real world. For example, the German Energiewende has rapidly decarbonised the power sector by replacing thermal generation with VRE. However, the initial lack of transmission capacity to connect the wind-rich sites in the north of the country with the load centres in the south (Kunz, 2013) has led to support payments to gas-fired plants (Rintamaki et al., 2016). Effectively, the price-decreasing impact of VRE generation has made precisely the flexible plants unprofitable without support payments. Likewise, in California, the so-called "duck curve" has bolstered the need for flexibility as solar generation increased from 912 GWh in 2010 to 27,265 GWh in 2018 (California Energy Commission, 2018; California Independent System Operator, 2016). As a result, the net-load profile exhibits a steep ramping requirement during three evening hours (17:00-20:00), which has increased even further in recent years (California Independent System Operator, 2021). Since solar overgeneration during the daytime depresses electricity prices, gas-fired plants, which could provide the ramping capability during the evening, are rendered unprofitable.

   In this context, energy storage, especially involving existing hydro reservoirs, could facilitate the integration of VRE capacity by moving energy from periods of abundance to scarcity (Electric Power Research Institute, 1976). Not only social welfare but also storage profit would increase from such temporal arbitrage, thereby obviating some peak-load generation capacity. However, this conventional perspective ignores the reality of a deregulated power sector in which firms with large market shares impact equilibrium prices through their storage operations. With a focus on hydro reservoirs, Crampes and Moreaux (2001 )'s stylised model demonstrates how hydro producers that behave a la Cournot use their allocated water more in the off-peak period than price-takers do. While both price-taking and Cournot hydro producers select generation in each period in order to maximise profit, the price-taker treats the price in each period as if it were exogenous. By contrast, the Cournot producer internalises the impact that its own production would have on the price in each period. The consequence of this distinction is that the price-taker's first-order necessary condition for profit maximisation is that the prices (equal to marginal utilities of consumption) across periods would tend to be equated, whereas the Cournot producer's first-order necessary condition for profit maximisation is that the marginal revenues across periods would tend to be equated. Thus, relative to a perfectly competitive use of water, a strategic one a la Cournot overproduces (underproduces) during off-peak (peak) periods, thereby lowering (raising) the price during off-peak (peak) periods. In effect, Cournot producers use less water in periods with inelastic residual demand by shifting that water to periods with more elastic residual demand. This propensity was illustrated in a case study of the California power system (Bushnell, 2003) even if a strategic hydro producer had to generate as much energy from its reservoirs over the study's time horizon as a perfectly competitive one. A Nash-Cournot model of pumped-hydro storage operations in Germany further indicated the potential for welfare losses vis-a-vis a no-storage case, especially if strategic storage operators also owned generation capacity (Schill and Kemfert, 2011). Storage operations' impacts on social welfare under various ownership structures are summarised via stylised equilibrium models under both perfect (Sioshansi, 2010) and imperfect competition (Sioshansi, 2014).

   While the aforementioned literature probes how strategic storage operations may distort strategic producers' incentives, it does not take C[O.sub.2] emissions and VRE into consideration. More recent work accounts for the interaction between climate policy and strategic storage operations. Virasjoki et al. (2016) model the uncertainty in VRE output in a case study of a transmissionconstrained Western European test network. They find that if firms behave a la Cournot, then they use their pumped-hydro storage to "dump" output from their ramp-constrained thermal plants, i.e., effectively withholding generation while circumventing ramping costs. In a Nash-Cournot analysis of the New York-Quebec interconnection, Debia et al. (2021) find that the main result of Bushnell(2003), i.e., temporal arbitrage by large hydro producers, follows through in spite of a regulatory constraint that prevents "spilling" of water. Moreover, they show that more expansive regulation that prohibits strategic hydro producers from additionally exporting away their "excess" generation, cf. Quebec's heritage pool (Hydro-Quebec, 2021), mitigates temporal arbitrage via reservoirs but exacerbates spatial arbitrage by pumped-hydro operators. Using a tighter regional C[O.sub.2] emission cap, Debia et al. (2021) also note that incentives for such strategic behaviour are likely to increase under future climate policies. Addressing a hypothetical 100% VRE power system with storage, Ekholm and Virasjoki (2020) uncover that strategic spilling of storage-enabled VRE output by Cournot producers could have more deleterious consequences for social welfare than manipulation of prices by storage alone. However, they remark that withholding of VRE generation in conjunction with storage would be more easily detected by regulators than temporal arbitrage of the type conducted via storage alone and that increased storage capacity in the hands of strategic firms could worsen welfare impacts. In a similar vein, Andre's-Cerezo and Fabra (2022) use an equilibrium approach to assess how market power by storage operators could distort both investment incentives and generation dispatch.

   By contrast, such distortions from the exercise of market power in hydro storage to integrate VRE capacities are thought to be less of a concern in the Nordic region. Indeed, the Nordic electricity market is generally held up as an exemplar due to its day-ahead market-clearing prices that are generally close to marginal costs. Amundsen and Bergman (2006) posit that the Nordic region's large hydro reservoirs and spatial integration dilute the potential for market power. As a result, the Nordic countries' ambitious targets for zero net emissions of greenhouse gases by the year 2045 (Regjeringen, 2019; Regeringen, 2021) seem well placed with sufficient flexibility from hydro reservoirs to cope with VRE intermittency. However, empirical analyses of the Nordic electricity market have yielded evidence of market power by hydro producers in Swedish price zones (Tangeras and Mauritzen, 2018), strategic reporting of unit failures by gas- and oil-fired power plants (Fogelberg and Lazarczyk, 2019), and system-wide withholding a la Cournot (Lundin and Tangeras, 2020). These findings are based on 2011-2013 data and do not build explicit computational NashCournot models of strategic operations. A separate strand of the literature that does use game-theoretic frameworks to identify the incentives for strategic behaviour in the Nordic region typically treats hydropower as a generic flexible resource and ignores reservoir operations. For example, Bj0rndal et al. (2014) examine the impact of nodal versus zonal pricing, Virasjoki et al. (2018) assess the strategic role of combined heat and power plants, and Rintamaki et al. (2020) identify how a flexible producer may create congestion in the balancing market to its advantage. Hence, an explicit representation of reservoir operations (F0rsund, 2015) is absent from game-theoretic policy analyses of strategic behaviour in the Nordic region under current and future climate packages.

   Given this research gap, our scientific contribution is to identify the mechanisms through which future climate policy may exacerbate strategic behaviour by hydro reservoirs. We generate insights about the impacts of climate policy and strategic operations in the Nordic power system by tackling the following research questions (RQs):

   RQ 1 How are social welfare, reservoir operations, and C[O.sub.2] emissions affected by market power in the current Nordic power system ?

   ...