Investigating Price Formation Enhancements in Non-Convex Electricity Markets as Renewable Generation Grows.

• Author: Daraeepour, Ali

1. INTRODUCTION

   Falling costs of variable renewable electricity (VRE, i.e., wind and solar) along with federal and state subsidies are driving rapid growth in VRE. Variability and intermittency in VRE supply results in more frequent and faster fluctuations in net demand (net demand = electricity demand--VRE generation), which increases the demand for operational flexibility on the grid. Operational flexibility is the ability to adjust to rapidly changing load and supply conditions. Figure 1 illustrates challenges faced by conventional generators when there are greater fluctuations in net demand: steeper ramps, shorter running times, and deeper turn-downs (Milligan et al. 2015). Ensuring reliable and economically efficient supply of net demand at higher VRE penetration requires conventional generators to startup, shutdown, and endure more frequent and significant ramp up and down events (Daraeepour, Patino-Echeverri, and Conejo 2019; Papavasiliou and Smeers 2017). It is not well-understood if the price formation process in wholesale energy markets today will appropriately remunerate and incentivize the greater levels of flexibility that will be required as VRE penetration levels grow. Using a custom-built scale-model of the PJM electricity grid operations, this paper explores how greater wind penetration affects the efficiency of conventional marginal pricing and its ability to remunerate operational flexibility. It also explores the degree to which alternative pricing schemes remunerate operational flexibility. To investigate these questions, this study simulates and analyzes wholesale electricity market operation outcomes for a continuous 365-day period with existing and alternative pricing schemes at different wind penetration levels. The investigation identifies and quantifies shortcomings of the existing pricing methodology. Producers' surplus in the energy market and the percentage share of revenues earned from out-of-market uplift payments are used to assess the market's ability to remunerate and incentivize flexible load-following performance. This study also offers comparative analysis of alternative pricing schemes that better incentivize flexibility.

   Section 2 provides a summary of market design and price formation principles in U.S. electricity markets and their shortcomings, elaborates on minimum uplift pricing schemes, and reviews relevant literature. Section 3 begins by presenting the two specific minimum uplift pricing schemes studied in this paper, proceeds to present the modeling framework used for simulating market operation outcomes, and ends with a description of case study simulations and related input data. Section 4 presents results and discussion. Conclusions and recommendations follow in Section 5.

2. REVIEW OF STUDIES ON PRICE FORMATION ISSUES IN U.S. ELECTRICITY MARKETS

   Designing more effective electricity pricing schemes requires a nuanced understanding of conventional market designs and the role that the price formation process plays in achieving the design objectives. This section reviews conventional market designs, highlights some shortcomings, illustrates market inefficiencies that such shortcomings pose, and discusses alternative pricing schemes that can overcome the inefficiencies.

   2.1. Conventional marginal pricing in U.S. markets

   Wholesale electricity markets in the U.S. are operated in a centralized manner by Independent System Operators (ISOs). An ISO serves as a social planner, managing the operation of energy, ancillary services, and (in most territories) capacity markets to ensure reliable and least-cost supply of electricity (Baldick et al. 2005). In the energy market, ISOs solicit demand bids and supply offers to clear the market. The supply offers are multipart, representing fixed commitment costs, i.e. no-load and startup costs, and marginal fuel cost of supply. The market clearing process relies on running multiple optimization models to determine socially optimal commitment (on/off) and generation-level schedules that will minimize short-run supply costs and equivalently maximize welfare. Prices that are used to settle energy transactions are also determined. Appendix A explains how a typical ISO uses Unit Commitment and Economic Dispatch models to calculate optimal generator schedules and prices.

   Pricing in the energy market follows principles of marginal pricing theory (FERC 2014; Helman, Hobbs, and O'Neill 2008), yielding prices that reflect the incremental cost of supplying the next MWh of electricity demand at all times and locations. Ideally, electricity supply offers would represent convexity, and the price paid to all generators would be the price of the highest offer that clears the market. Here, this is referred to as the true marginal cost of electricity. Conventional generators rely on their inframarginal rent--the difference between energy market revenues and short-run operating costs--to cover capital investments and fixed, going forward costs like insurance, property tax, and maintenance. If inframarginal rents are insufficient, a generator bids its revenue shortfall, aka missing money, to the capacity market where the price is determined by the last (highest) bid needed to meet projected future capacity requirements (Bowring 2013).

   2.2. Uncertainties and shortcomings with conventional marginal pricing

   A generator's costs partly consist of costs incurred to provide flexibility (Bowring 2013; PJM 2016). It is not well understood if the current market design will reward flexibility and motivate investment in flexible resources at higher VRE penetration levels. To encourage flexibility, a generator's profitability should not be compromised when providing flexible performance; otherwise, the generator will adjust its flexibility to strike a balance between its costs and the benefit it gains from flexible performance. In current market designs, capacity markets have no constructs to remunerate flexible performance because, by design, they remunerate aggregate capacity attributes characterized by installed capacity and forced outage rates (Bothwell and Hobbs 2017; Bowring 2013; PJM 2017). Moreover, the costs of providing flexible performance may lead to greater revenue shortfalls for a generator, thus undermining its chances of clearing the capacity market, which further discourages providing flexibility. Absent incentives for flexible performance in the capacity market, the energy market must be relied upon to send efficient price signals that reward flexible performance.

   Conventional marginal pricing schemes suffer from shortcomings in this respect due to the non-convex nature of US electricity markets. Non-convexities arise from fixed commitment costs, e.g. startup costs, ramp rate limits and minimum production constraints, that shape ramping and cycling behavior of these generators (Kuang, Lamadrid, and Zuluaga 2019; O'Neill 2009). In non-convex markets, conventional marginal prices may not reflect fixed commitment costs and may be insufficient to recover such costs for all committed generators (Sioshansi, O'Neill, and Oren 2008). See illustrative example in Appendix B.1. Also, counter-intuitively, prices can fall with increasing demand. Here, such situations are referred to as unrepresentative price events, i.e., when prices fall below the true marginal cost of electricity. Such events mainly occur when technical supply constraints, which characterize the flexibility of conventional generators, become binding. See illustrative example in Appendix B.2.

   The shortcomings in conventional marginal pricing reduce market efficiency and incentives for flexible performance in two ways. The unrepresentative price events shrink inframarginal revenues for all generators, making it harder to recover ramping costs and other fixed costs. In addition, the price suppression, along with fixed commitment costs not being reflected in prices, result in negative inframarginal rents for some generators and increase out-of-market uplift payments. The uplift payments erode producers' incentives to bid their actual operating costs or to improve their performance, e.g. heat rates and/or flexibility (Harvey 2014; CAISO 2013), and they distort investment decisions (Herrero, Rodilla, and Batlle 2015).

   Supplementing conventional marginal prices with uplift payments is essential in non-convex markets to ensure incentive compatibility. Absent uplift payments, generators would wish to deviate from an ISO's dispatch schedule to maximize their individual profits, resulting in higher costs to consumers and a loss of welfare (O'Neill 2009; Ring and Hogan 2003). Coutu and White (2014a, 2014b) provide examples that illustrate welfare loss from such pricing inefficiencies. See discussion of the link between incentive compatibility and uplift payments in Appendix B.3.

   Two types of uplift payments are relevant: make-whole payments (MWP) refer to uplift payments that ensure generators that follow an ISO's schedule do not operate at a loss; lost-opportunity cost payments (LOCP) make generators indifferent to following the ISO's schedules and deviating from those schedules to maximize their profits at the given prices (Eldridge, O'Neill, and Hobbs 2019; Gribik, Hogan, and Pope 2007). Henceforth in this paper we use LOCP, whose name reflects forgone profits, to refer to uplift payments that offset incentives for exceeding an ISO's schedules. We use MWP to refer to uplift payments that offset all incentives for under-generation relative to the ISO's schedules. Total uplift payment refers to the sum of MWP and LOCP.

   2.3. Performance of conventional marginal pricing at high VRE penetration levels

   Growth in VRE has the potential to exacerbate the shortcomings of conventional pricing by further eroding incentives for providing flexibility. Expected increases in load-following events at higher VRE penetration increases the fixed commitment costs that shrink inframarginal rents and...