Generator Revenue Sufficiency in Electricity Markets with Variable Renewable Generation

摘要

In electricity markets, the ‘capacity adequacy problem’ arises when electricity supplier revenues are insufficient toincentivize investments in an adequate level of new capacity [1]. Such revenue shortfalls may stem fromadministratively determined price caps that are implemented to alleviate the potential for exertion of market powerduring periods of supply scarcity. While price caps limit consumer exposure to unbounded prices, they also limit therevenue available to generators during these periods. As a result, generators may not be properly incentivized tocontribute to system reliability during periods of extreme supply scarcity. In addition, new generators may bediscouraged from entering the market at all, possibly resulting in insufficient capacity to serve demand and supportsystem reliability over the long term.These issues are further augmented by the rapid expansion of variable renewable generation (VRG), such as windand solar, which introduce two additional challenges to system operation. 1) VRG resources have near-zero, zero oreven negative marginal costs (due to subsidies) and therefore reduce wholesale electricity prices when they providethe marginal unit of generation in a power system. 2) VRG resources have variable output profiles with limitedpredictability. As a result, flexible resources become more important in a power system with large amounts of VRG,which must maintain additional reserve capacity to ensure system reliability in periods when generation deviatesfrom the forecast. Under many existing market design mechanisms the provision of reserve capacity and otherancillary services are not always appropriately compensated. Therefore, higher reserve requirements may alsoincrease the revenue shortfall experienced by generators and intensify the capacity adequacy problem.Power markets in the United States and around the world have implemented a number of different policies toprovide generators with sufficient revenue streams so as to motivate generation investments and new entry to themarket with the aim of ensuring long-term capacity adequacy [2]. In this paper, we build on our previous work [3],[4] and investigate the impact of the following policies on revenue sufficiency and capacity adequacy:1. Fixed Reserve Scarcity Pricing (FRSP) – The system operator sets target levels for various reserveproducts. In the event that a given reserve target is not reached, the price of that particular reserve productis set to an administratively determined scarcity cost.2. Operating Reserve Demand Curves (ORDC) – The system operator values reserve capacity on the basis ofa continuous demand function, which is based on a probabilistic assessment of the contribution of reservestowards system reliability. As reserve levels increase, their marginal value and market price decreases. Thisapproach was recently implemented in the ERCOT system in the state of Texas [5].3. Capacity Payments (CP) – The system operator provides revenue to generators for having availablecapacity, independent of the amount of electricity they generate.4. Uplift Payments (UP) – The system operator implements a mechanism to compensate generators over aperiod of time (e.g. one day) during which they were instructed to generate, but received insufficientrevenue to cover their fixed and variable operating costs. This occurrence is a consequence of non-convexcost elements that are not reflected in energy prices.MethodsWe apply a mixed-integer linear programming model [3] to minimize the cost of generation unit expansion, hourlycommitment and dispatch, and reserve provision in the ERCOT power system. Thermal generation units are groupedinto four characteristic types and represented by integer variables. This dramatically reduces runtime as compared toa binary formulation that explicitly tracks individual units [6]. The expansion model is applied to analyze the fourpreviously described market policies in the ERCOT system for a future state with VRG penetration that varies from10% to 40% of total generation. We also conduct a sensitivity analysis around the fuel price of natural gas.ResultsWe find that with baseline parameter assumptions, all new generation capacity is developed in the form of naturalgas combustion turbines under each market policy. Optimal expansion plans are similar under the FRSP and ORDCformulations, and more additional capacity is developed when capacity payments are provided.The average wholesale electricity price decreases with increasing wind penetration under all market policies. This isprimary due to the fact that wind units increasingly supply the marginal unit of generation. In contrast, reserve pricesincrease with higher wind penetration, but this effect is generally not as significant as the impact on energy prices.While average electricity prices are comparable under both the FRSP and ORDC formulations, the ORDC approachresults in a smoother spectrum of prices with fewer extreme prices spikes. Hourly prices exceed $100/MWh during823 (out of 8760) periods under the FRSP approach and only 92 periods under the ORDC approach at the 40% windpenetration level. When capacity payments are implemented in the absence of any other revenue mechanismsaverage electricity prices are much lower, and the hourly price exceeds $100/MWh during only a single period whendemand is curtailed. With a fixed capacity payment of $40/kW-year, generator profits are consistently lower thanthey are under either the FRSP or ORDC mechanism.Profits for nuclear, coal, and wind units generally decrease with increasing VRG, while profits for natural gas unitsgenerally remain consistent under all market policies. The natural gas units are less exposed to lower electricityprices during off peak periods and they receive additional revenue due to increased reserve prices. Uplift paymentsare calculated ex-post and therefore do not impact the optimal expansion plan in our analysis. Natural gas unitsreceive the greatest benefit from uplift payments due to their higher relative operating costs. Coal units receive upliftpayments only when wind penetration is 20% or greater, and nuclear units only when wind penetration is 30% orgreater. Payments are also only triggered for nuclear and coal units during days when wind generation is high andnegative energy prices result. Natural gas units receive relatively greater uplift payments at all wind penetrationlevels, however these transfers still account for a fairly small fraction of their total unit revenue under the FRSP andORDC, typically on the order of 2-3%. Uplift payments for natural gas units are generally 3-5 times greater underthe capacity payment framework, however total revenues are still less than under FRSP or ORDC.A sensitivity analysis reveals that these results are highly sensitive to the cost of natural gas as a fuel source. Whennatural gas prices reach $10/MMbtu (compared to $5.15/MMbtu under baseline assumptions), new nuclear and coalunits are included in the optimal expansion plan at the lower wind penetration levels. Higher natural gas prices alsodrive up average electricity prices and increase the profitability of coal and natural gas units. A reduction in thenatural gas price to $3/MMbtu does not change the optimal expansion plan compared to baseline conditions, butdoes reduce the profitability of existing coal and nuclear units.ConclusionsOur results indicate that energy markets with ORDC or FRSP can both ensure revenue sufficiency and capacityadequacy. However this will depend on the design of each framework and the choice of key administrativeparameters, such as the reserves scarcity prices and the definition of the ORDC itself. We show that an ORDCimplementation can be structured to result in optimal unit expansion plans and generator revenue levels that arecomparable to those obtained under a FRSP implementation. The ORDC formulation results in a more continuousspectrum of wholesale electricity prices with fewer large price spikes when scarcity events occur. This could reduceoperational and investment risks for generators and may also help alleviate concerns over potential marketmanipulation. The capacity payment policy leads to more investment in generation capacity, but increasing revenuesufficiency problems, particularly for base load generators, as energy prices are reduced due to more capacity beingavailable. Uplift payments provide an additional source of generator revenue, but as they are allocated ex-post afterthe market clears, they do not directly impact energy prices or investment decisions in our formulation.