WorksEngineering
Impact of Wind-Solar-Battery Growth on Electricity Prices: Reserve Scarcity and Transmission Congestion
Abstract
This paper asks whether sequential wind, solar, and battery storage additions that eliminate unserved energy also remove modeled reserve scarcity and transmission congestion. Thirty-one cumulative portfolio states are evaluated on a modified IEEE 24-bus system for a constructed stressed day using a 24-hour day-ahead unit commitment model with network constraints, co-optimized energy and upward reserve, chronological battery storage, and an approximation of the Electric Reliability Council of Texas operating reserve demand curve. The thermal baseline has 106 MWh of unserved energy, whereas 450 MW of wind is sufficient to serve all demand. However, with 1,500 MW of wind, the hourly modeled reserve scarcity price still reaches $5,000/MWh because wind output declines after the system peak. Adding 1,800 MW of solar reduces the load weighted daily average modeled reserve scarcity price from $1,136.2/MWh to $101.2/MWh and confines modeled reserve scarcity to three evening hours. Adding 250 MW/1,000 MWh of battery storage further lowers the average to $4.3/MWh and reduces wind curtailment from 631 MWh to 306 MWh. Transmission congestion nevertheless remains: all residual wind curtailment occurs at bus 21, wind capture prices at buses 21 and 23 differ by $31.8/MWh, and identical battery storage plants yield normalized net energy market revenues of $201/(MWday) and $121/(MWday) at buses 19 and 21, respectively. These results show that unserved energy, modeled reserve scarcity, and locational congestion effects measure distinct operating conditions and should be evaluated separately but interpreted together.
Introduction
Adding capacity can remove an energy shortfall without resolving the next operating limitation. A power system may serve all demand during the peak while upward reserve remains scarce later in the day. Even when energy and reserve are sufficient in aggregate, transmission limits can curtail renewable output at one bus and separate resource values across the network. Evaluating a resource addition therefore requires attention to how much capacity is added, when it is available, and where it connects.
This distinction is increasingly important as electricity demand grows and new resources face long interconnection delays. The 2025 Long-Term Reliability Assessment of the North American Electric Reliability Corporation identifies resource adequacy challenges in 13 of 23 assessment areas and notes that additions expected in the near term are concentrated in solar photovoltaics and batteries [1]. Lawrence Berkeley National Laboratory reports more than 2,060 GW of generation and storage in U.S. interconnection queues and a median interval exceeding five years from interconnection request to commercial operation for projects completed in 2025 [2]. These conditions make it necessary to distinguish energy adequacy at the peak, reserve availability in critical hours, and deliverability through the transmission network.
Previous research provides established concepts for these effects. Large solar additions can shift the hours of highest net load toward the evening [3], while transmission constraints and spatial flexibility can materially affect renewable curtailment and operating outcomes [4, 5, 6, 7, 8]. Operational detail can alter the resource mix selected by planning models [9]. Market value, capture price, and value factor metrics characterize the economic value received by variable generation [10, 11, 12]; capacity credit methods quantify the contribution of variable generation and storage to resource adequacy [13, 14, 15]; operating reserve demand curves connect reserve availability with scarcity pricing [16, 17]; and storage studies show that temporal price variation and network location both affect storage value [18, 19, 20]. These outcomes are often examined separately, making their joint evolution under a common operating data set less clear.
This study follows a prescribed wind–solar–battery storage sequence under fixed load, renewable availability, outage, transmission, and operating assumptions. It asks when unserved energy disappears, how solar and battery storage change the timing and magnitude of modeled reserve scarcity, and how transmission constraints affect curtailment, capture prices, and battery storage energy market revenue after all demand is served. The sequence is not an optimized expansion plan; it is a controlled comparison in which one resource build-out level changes at a time.
We apply this design to a modified IEEE 24-bus reliability test system based on [21], [22]. Each of the 31 portfolio states is solved with the same 24-hour day-ahead unit commitment model. The model co-optimizes energy and upward reserve, represents chronological battery storage and a lossless direct current (DC) power flow network, and includes a piecewise constant approximation of the Electric Reliability Council of Texas (ERCOT) operating reserve demand curve (ORDC). After each mixed-integer solution, the binary decisions are fixed and the resulting convex quadratic program is solved again to obtain locational marginal prices and constraint dual variables.
The analysis makes three contributions. First, it isolates the operating response to sequential wind, solar, and battery storage additions on one constructed stressed day. Second, it reports systemwide outcomes, namely unserved energy, modeled reserve scarcity prices, and congestion rent, together with locational outcomes, namely wind curtailment, capture prices, and normalized net energy market revenue, within one internally consistent model. Third, it distinguishes the role of each addition in this case: wind removes the energy shortfall at the peak, solar and battery storage reduce the remaining evening modeled reserve scarcity, and transmission congestion continues to differentiate outcomes across buses.
Study Design
Evaluation Framework
Each portfolio state is evaluated in the same order. We first determine whether all demand is served, then examine the timing and magnitude of modeled reserve scarcity, and finally assess transmission congestion and outcomes at individual resource locations.
Let denote the set of buses, the set of transmission branches, and the set of hourly intervals. Each interval has duration h. The intervals are reported using hour ending notation, denoted HE1 through HE24. Total unserved energy over the study day is
where is unserved load at bus in interval , measured in MW. Thus, is measured in MWh, and indicates that all demand is served during the study day.
Modeled reserve scarcity is evaluated using the hourly modeled reserve scarcity price , defined in Section III-B. We report its hourly magnitude and the number of intervals with . Let denote demand at bus in interval . For any hourly or nodal price component , its load weighted daily average, denoted by angle brackets, is
Because does not vary by bus, its load weighted daily average follows from (2) by setting . It equals the modeled reserve scarcity component of gross load payment divided by total demand energy.
Transmission effects are evaluated using daily congestion rent and wind curtailment. Capture prices and normalized net energy market revenue are reported by location to show how network constraints affect individual resources.
Network and Candidate Resources
The study uses a modified IEEE 24-bus reliability test system based on [21], [22] and represents it as one balancing area. The study day represents a hot summer day with tight supply. Demand peaks at 3,100 MW and stays high into the evening. Wind and solar availability both fall in the evening, so renewable output is low while demand is still high. A 197 MW unit and a 100 MW unit are out of service for the whole day, and another 100 MW unit is derated to 50% of its rating from HE15 to HE21. These outages reduce available thermal capacity at the peak to 3,058 MW. Thermal capacity alone therefore cannot serve peak demand, and the evening hours test whether added resources provide enough upward flexibility after the peak.
The candidate portfolio contains 1,500 MW of wind at buses 21, 22, and 23, allocated in nameplate shares of 0.4, 0.4, and 0.2, respectively; 1,800 MW of solar at buses 3, 5, 9, and 19, allocated in shares of 0.3, 0.2, 0.3, and 0.2, respectively; and 250 MW/1,000 MWh of battery storage divided equally between buses 19 and 21. Each battery storage plant therefore has a 125 MW/500 MWh nameplate rating and a 4 h nameplate duration. Production incentives of $30/MWh for wind and $5/MWh for solar are represented as offer costs of −$30/MWh and −$5/MWh, respectively.
The superscripts w, s, and e denote wind, solar, and battery storage. For each resource type , the build-out level is the installed fraction of the corresponding candidate capacity. The same build-out level is applied at every candidate site of a resource type, preserving the spatial allocation above. For example, installs 450 MW of wind in total, while installs 125 MW/500 MWh of battery storage in total, or 62.5 MW/250 MWh at each battery storage bus. Figure 1 shows the candidate resource locations and the highlighted delivery interfaces.

Figure 1. Modified IEEE 24-bus system, candidate resource locations, and highlighted delivery interfaces.
Sequence of Resource Additions
Resources are added in the order in which they grew in ERCOT: wind first, then solar, then battery storage [23]. Adding battery storage last also allows its charging and discharging to respond to the net load profile of the complete wind and solar portfolio. Four case groups define this cumulative sequence. B1 is the thermal baseline, . B2 increases from 0 to 1 with no solar or battery storage. B3 fixes and increases from 0 to 1. B4 fixes and increases from 0 to 1. In B2–B4, the build-out level takes the values . Because adjacent case groups share an endpoint, the sequence contains 31 distinct portfolio states.
The load, outage schedule, renewable availability, transmission network, and operating assumptions are identical in every state. One build-out level changes at a time, while unit commitment, dispatch, reserve provision, battery storage operation, and line flows are reoptimized. Differences between adjacent states therefore describe operating responses along the prescribed sequence. The model is implemented in Pyomo (version 6.10.0) [24] with Python 3.12.10 and solved using Gurobi 12.0.3 [25]. Every instance reaches optimal status at a relative mixed-integer programming gap tolerance of .
Operating Model and Metrics
Day-Ahead Unit Commitment With Network Constraints
The operating model is a 24-hour day-ahead unit commitment formulation with lossless DC power flow, branch thermal limits, joint scheduling of energy and upward operating reserve, and chronological battery storage [26], [27]. Related day-ahead unit commitment formulations have also been applied to operating studies of renewable power grids [28], [29]. The model is formulated as a mixed-integer quadratic program with quadratic thermal production costs, startup and shutdown costs, minimum up and down time constraints, and ramp limits.
Let and denote the candidate wind and solar nameplate capacities at bus , and let denote their hourly availability factors. If and are the dispatched wind and solar outputs, then and . The difference between available and dispatched renewable output is curtailment.
Let denote the set of battery storage buses. At battery storage bus , and denote the candidate power and energy ratings. The installed ratings are and . Let , , and denote charging power, discharging power, and stored energy. Charging and discharging are each limited by the installed power rating, and a binary operating mode prevents simultaneous charging and discharging. Stored energy evolves as
where , giving a round trip efficiency of 86%. Stored energy is restricted to . The initial and terminal state of charge (SOC) are both fixed at 50% of the installed energy rating:
Thus, each battery ends the day with the same stored energy as it starts, and all energy it discharges is charged within the same day. The objective includes a throughput based cycling cost of $3/MWh, applied to both charging and discharging energy.
For branch , denotes active power flow in interval , and denotes the branch thermal limit. The flow satisfies . The voltage angle at bus 13 is fixed to zero in every interval. The modeled daily operating cost reported below is the optimized 24-hour objective value under the stated cost and penalty terms. After each mixed-integer solution, the binary variables are fixed at their optimal values and the resulting convex quadratic program is solved again; all dual variables reported in this paper are taken from this fixed binary solution.
Modeled Reserve Scarcity Pricing
The model co-optimizes energy and upward reserve [30]. Reserve scarcity is priced with the 2024 ERCOT ORDC [17]. To fit the test system, the reserve quantity axis of this curve is scaled by , the ratio of test system peak demand to the ERCOT reference peak. This scaling keeps reserve needs proportional to system size: a reserve level that is the same fraction of peak demand receives the same scarcity price in both systems. The price axis is unchanged, since ORDC prices measure scarcity value per MWh, which does not depend on system size. The modeled ORDC maximum is therefore $5,000/MWh.
The approximation covers reserve quantities up to MW and uses 16 segments indexed by . Each segment has width MW. Segment corresponds to the interval of available reserve and has midpoint . Its marginal shortfall penalty is the ERCOT ORDC value evaluated at the corresponding reserve quantity on the ERCOT scale, . Because scarcity value increases as available reserve declines, .
Let denote the set of thermal units and the set of quick start units. For thermal unit , denotes online upward reserve, denotes commitment status, and denotes maximum available output, which varies with the outage and derating schedule. For battery storage bus , denotes upward reserve from battery storage, limited by the discharging power headroom and by the energy stored above the minimum SOC. The variable denotes reserve shortfall assigned to segment , with . The hourly reserve balance is
The first three terms in (5) represent upward reserve from online thermal units, offline quick start units, and battery storage, respectively. The final term represents reserve shortfall relative to . The objective includes , so lower cost segments are used before higher cost segments as available reserve declines.
The dual variable associated with (5) in the fixed binary solution is the hourly modeled reserve scarcity price. It measures the marginal increase in operating cost from a 1 MW increase in in interval . Because h, the dual is numerically reported in $/MWh; when segment is marginal, .
The value of lost load, /MWh, exceeds the maximum reserve shortfall penalty. When available capacity cannot simultaneously serve demand and provide the full reserve quantity, the model therefore accepts reserve shortfall before leaving firm load unserved.
Prices and Reported Metrics
The dual of the nodal power balance in the fixed binary solution is the locational marginal price (LMP) . These prices are conditional on the optimized commitment and battery storage modes. The lossless network has no marginal loss component.
Bus 13 is used as the reporting reference bus. Define the energy component as and the congestion component as . The nodal price then satisfies
Equation (6) is an accounting identity under these definitions. The energy and congestion components depend on the selected reference bus, whereas their sum with the modeled reserve scarcity component recovers the nodal LMP. The components are reported as load weighted daily averages using (2).
Gross load payment is . This quantity is reported only for portfolio states with , for which all demand is served.
Let and be the duals of the upper and lower branch flow limits, respectively, and define . Daily congestion rent is
For renewable type with positive dispatched energy over the study day, the daily capture price at bus is
For and , normalized daily net energy market revenue is
Capture prices exclude production incentive revenue. Equation (9) includes energy market settlements but excludes reserve payments, cycling costs, and investment costs. Because the initial and terminal SOC are equal by (4), the 24-hour settlement is evaluated without net depletion of stored energy.
Results
Table I reports selected systemwide outcomes, Figure 2 shows the load weighted LMP component across all 31 states, and Table II reports selected locational outcomes for the portfolio with all candidate resources installed. Along the prescribed sequence, unserved energy is eliminated and modeled reserve scarcity falls substantially, while transmission congestion, curtailment, and locational differences remain.
| Case group | Portfolio state | Unserved energy (MWh) | Wind curtailment (MWh) | Load weighted daily average modeled reserve scarcity price ($/MWh) | Main observation | |
| B1 | Thermal baseline | 106 | 0 | 2,116.3 | Unserved energy is present. |
| B2 () | 450 MW wind | 0 | 0 | 1,817.3 | All demand is served; modeled reserve scarcity remains. |
| B2 () | 1,500 MW wind | 0 | 631 | 1,136.2 | Evening modeled reserve scarcity and wind curtailment remain. |
| B3 () | + 1,800 MW solar | 0 | 631 | 101.2 | Positive modeled reserve scarcity prices are confined to three evening hours. |
| B4 () | + 250 MW/1,000 MWh battery storage | 0 | 306 | 4.3 | Wind curtailment falls further; transmission congestion remains. |
Table I. Selected systemwide operating outcomes along the prescribed resource sequence. Resource additions are cumulative. All battery storage cases enforce equal 50% initial and terminal SOC.

Figure 2. Load weighted daily averages of the three LMP components in (6) across the 31 portfolio states, with B1 at the left end of B2. The symmetric logarithmic axis, linear between −$1/MWh and $1/MWh, displays values that are both positive and negative and span several orders of magnitude.
| Wind outcomes at selected buses | ||
| Bus | Capture price ($/MWh) | Wind curtailment (MWh) | |
| 21 | −7.3 | 306 |
| 23 | 24.5 | 0 |
| Battery storage energy market outcomes | ||
| Bus | Normalized net energy market revenue ($/(MW · day)) | | |
| 19 | 201 | |
| 21 | 121 |
Table II. Selected locational operating and energy market outcomes with all candidate resources installed. Wind at bus 22 has zero curtailment and is omitted.
Wind Eliminates Unserved Energy During the Peak Hours
The thermal baseline has 106 MW of unserved energy from HE15 through HE18, including 42.0 MW of unserved load in each of HE16 and HE17. With 450 MW of the candidate wind capacity installed, wind output is 43.2 MW in both peak hours and ranges from 31.7 MW to 43.2 MW over the four affected intervals. This is the first portfolio state with .
Modeled reserve scarcity remains after all demand is served. With all 1,500 MW of candidate wind installed, wind output in HE20 is 66.6 MW, or 4.4% of installed wind capacity. The hourly modeled reserve scarcity price reaches the modeled ORDC maximum of $5,000/MWh in HE18 and HE19 and is $3,315/MWh in HE20. Wind additions reduce the load weighted daily average modeled reserve scarcity price, but the evening decline in wind output leaves limited upward flexibility.
Eliminating the energy shortfall at the peak therefore does not eliminate modeled reserve scarcity later in the day.
Solar and Battery Storage Reduce Evening Modeled Reserve Scarcity
With all candidate wind installed, the modeled reserve scarcity price is positive in ten hours. Adding all candidate solar reduces its load weighted daily average by 91%, from $1,136.2/MWh to $101.2/MWh, and confines positive values to HE19–HE21. The highest remaining value is $1,904/MWh in HE20, when solar output has fallen to 36.0 MW. For this case, renewable availability during HE19–HE21 is more relevant to the remaining modeled reserve scarcity than total renewable energy over the full day.
Installing all 250 MW/1,000 MWh of candidate battery storage reduces the load weighted daily average modeled reserve scarcity price from $101.2/MWh to $4.3/MWh, a decrease of about 96% relative to the state with all candidate solar and no battery storage. Wind curtailment falls from 631 MWh to 306 MWh. The modeled daily operating cost of the state with full battery storage is $641,328. The battery
storage addition therefore reduces both evening modeled reserve scarcity and wind curtailment.
Figure 3 shows the hourly operation behind these outcomes. Solar output falls from 622.8 MW in HE18 to 190.8 MW in HE19 and 36.0 MW in HE20 while demand is still above 3,000 MW, thermal generation reaches its daily maximum of 2,732.6 MW in HE20, and battery storage discharges 162.3 MW, 171.8 MW, and 193.8 MW in HE19–HE21. Charging occurs in HE4–HE5, HE11–HE14, and HE23–HE24; the last interval coincides with recovering wind output and with the terminal SOC condition. Positive modeled reserve scarcity prices in this state occur only in HE20 and HE21, and these two hours account for the $4.3/MWh load weighted daily average in Table I.

Figure 3. Hourly operation of the portfolio with all candidate resources installed ().
However, Figure 2 also shows that lower operating cost does not always mean lower modeled reserve scarcity prices: the price rebounds at high battery storage build-out levels. This rebound is a unit commitment discontinuity. Because commitment decisions are binary, a small increase in battery storage can switch a unit off and change the price in a single step. Supplementary runs at finer build-out increments place this step between and 0.86. At this step, the added battery storage makes it economical to keep a 155 MW coal unit offline, which removes its reserve contribution. Two quick start units are committed at the same time and no longer count as offline reserve, and greater battery discharge reduces battery reserve headroom. With less reserve available, the load weighted daily average modeled reserve scarcity price rises sharply from near zero, even though operating cost falls and all demand remains served. Further battery storage additions lower the price again.
Transmission Congestion and Locational Outcomes
Transmission congestion persists after modeled reserve scarcity is substantially reduced. Across the solar and battery storage additions, gross load payment falls from $74.1 million with all candidate wind installed to $8.0 million after the full solar addition and $2.0 million after the full battery storage addition. Over the same stages, daily congestion rent remains between $478,000 and $687,000.
The load weighted congestion component in Figure 2 stays within along the entire sequence, while daily congestion rent remains in the hundreds of thousands of dollars. The two quantities describe different aspects of congestion. Positive and negative nodal deviations from the reference bus offset each other in the load weighted average, so measures the effect of congestion on the systemwide average price, which is small here.
With all candidate resources installed, all 306 MWh of wind curtailment occurs at bus 21, while wind at buses 22 and 23 is not curtailed. The daily wind capture price defined in (8) is −$7.3/MWh at bus 21 and $24.5/MWh at bus 23, a difference of $31.8/MWh. In hours with high wind availability, exports from bus 21 are limited by the interface to bus 15 (Figure 1), and the LMP at bus 21 is set by the negative wind offer cost. Much of the wind output at bus 21 is produced in these hours, so its capture price, which weights LMPs by output, falls below zero, while LMPs at bus 23 remain positive in the same hours. The concentration of curtailment and the separation in capture prices show that wind projects with the same technology and offer assumptions can face substantially different outcomes at different network locations.
The identical 125 MW/500 MWh battery storage plants also have different energy market outcomes. Their normalized daily net energy market revenues, defined in (9), are $201/(MW·day) at bus 19 and $121/(MW·day) at bus 21. The revenue difference reflects their nodal price profiles and the resulting charging and discharging schedules.
Conclusion
On the constructed stressed day, the thermal baseline has 106 MWh of unserved energy, and 450 MW of candidate wind is the first portfolio state that serves all demand. Modeled reserve scarcity nevertheless remains with all candidate wind installed. Adding all candidate solar reduces the load weighted daily average modeled reserve scarcity price from $1,136.2/MWh to $101.2/MWh. With all candidate battery storage installed, the
average is $4.3/MWh, wind curtailment is 306 MWh, and the modeled daily operating cost is $641,328.
Transmission congestion continues to shape locational outcomes after systemwide modeled reserve scarcity is substantially reduced. All residual wind curtailment occurs at bus 21; wind capture prices at buses 21 and 23 differ by $31.8/MWh; and the identical battery storage plants yield normalized net energy market revenues of $201/(MW·day) at bus 19 and $121/(MW·day) at bus 21, respectively. These results show that unserved energy, modeled reserve scarcity, and the locational effects of transmission congestion measure different operating conditions and should be reported separately but interpreted together when evaluating sequential wind, solar, and battery storage additions.
Acknowledgment
This work was supported by the National Academy of Engineering Grainger Foundation Frontiers of Engineering Grant. This work was completed in part with resources provided by the Research Computing Data Core at the University of Houston.
References
- [1]North American Electric Reliability Corporation, “2025 Long-Term Reliability Assessment,” NERC, Atlanta, GA, USA, Tech. Rep., Jan. 2026.
- [2]J. Rand et al., “Queued Up: 2026 Edition, Characteristics of Power Plants Seeking Transmission Interconnection As of the End of 2025,” Lawrence Berkeley National Laboratory, Berkeley, CA, USA, Tech. Rep., May 2026.
- [3]P. Denholm, M. O’Connell, G. Brinkman, and J. Jorgenson, “Overgeneration from solar energy in California: A field guide to the duck chart,” National Renewable Energy Laboratory, Golden, CO, USA, Tech. Rep. NREL/TP-6A20-65023, Nov. 2015.
- [4]L. Bird et al., “Wind and solar energy curtailment: A review of international experience,” Renew. Sustain. Energy Rev., vol. 65, pp. 577–586, Nov. 2016.DOI
- [5]A. V. Ramesh and X. Li, “Reducing congestion-induced renewable curtailment with corrective network reconfiguration in day-ahead scheduling,” in Proc. IEEE Power & Energy Soc. Gen. Meeting (PESGM), Montreal, QC, Canada, Aug. 2020, pp. 1–5.arxiv.org/abs/2002.01090
- [6]H. Wan, L. Fang, and X. Li, “Grid operational benefit analysis of data center spatial flexibility: Congestion relief, renewable energy curtailment reduction, and cost saving,” presented at the IEEE Power & Energy Soc. Gen. Meeting (PESGM), Montreal, QC, Canada, Jul. 2026.arxiv.org/abs/2511.08759
- [7]X. Li and Q. Xia, “Stochastic optimal power flow with network reconfiguration: Congestion management and facilitating grid integration of renewables,” in Proc. IEEE/PES Transmiss. Distrib. Conf. Expo. (T&D), Chicago, IL, USA, Oct. 2020, pp. 1–5.arxiv.org/abs/1911.12961
- [8]H. Wan and X. Li, “Data center spatio-temporal load flexibility in security-constrained unit commitment for enhanced grid efficiency and reliability,” in Proc. IEEE Ind. Appl. Soc. Annu. Meeting (IAS), Vancouver, BC, Canada, Oct. 2026.arxiv.org/abs/2605.18517
- [9]B. S. Palmi ntier and M. D. Webster, “Impact of operational flexibility on electricity generation planning with renewable and carbon targets,” IEEE Trans. Sustain. Energy, vol. 7, no. 2, pp. 672–684, Apr. 2016.
- [10]A. D. Mills and R. H. Wiser, “Changes in the economic value of variable generation at high penetration levels: A pilot case study of California,” Lawrence Berkeley National Laboratory, Berkeley, CA, USA, Tech. Rep. LBNL-5455E, Jun. 2012.
- [11]L. Hirth, “The market value of variable renewables: The effect of solar wind power variability on their relative price,” Energy Econ., vol. 38, pp. 218–236, Jul. 2013.
- [12]A. D. Mills and R. H. Wiser, “Strategies to mitigate declines in the economic value of wind and solar at high penetration in California,” Appl. Energy, vol. 147, pp. 269–278, Jun. 2015.DOI
- [13]A. Keane et al., “Capacity value of wind power,” IEEE Trans. Power Syst., vol. 26, no. 2, pp. 564–572, May 2011.DOI
- [14]S. Wang, N. Zheng, C. D. Bothwell, Q. Xu, S. Kasina, and B. F. Hobbs, “Crediting variable renewable energy and energy storage in capacity markets: Effects of unit commitment and storage operation,” IEEE Trans. Power Syst., vol. 37, no. 1, pp. 617–628, Jan. 2022.DOI
- [15]P. Denholm, J. Nunemaker, P. J. Gagnon, and W. J. Cole, “The potential for battery energy storage to provide peaking capacity in the United States,” Renew. Energy, vol. 151, pp. 1269–1277, May 2020.
- [16]W. W. Hogan, “Electricity scarcity pricing through operating reserves,” Econ. Energy Environ. Policy, vol. 2, no. 2, pp. 65–86, Sep. 2013.DOI
- [17]Electric Reliability Council of Texas, “2024 Biennial ERCOT Report on the Operating Reserve Demand Curve,” ERCOT, Austin, TX, USA, Tech. Rep., Nov. 2024.
- [18]R. Sioshansi, P. Denholm, T. Jenkin, and J. Weiss, “Estimating the value of electricity storage in PJM: Arbitrage and some welfare effects,” Energy Econ., vol. 31, no. 2, pp. 269–277, Mar. 2009.
- [19]H. Pandžić, Y. Wang, T. Qiu, Y. Dvorkin, and D. S. Kirschen, “Near-optimal method for siting and sizing of distributed storage in a transmission network,” IEEE Trans. Power Syst., vol. 30, no. 5, pp. 2288–2300, Sep. 2015.DOI
- [20]A. V. Ramesh and X. Li, “Network reconfiguration impact on renewable energy system and energy storage system in day-ahead scheduling,” in Proc. IEEE Power & Energy Soc. Gen. Meeting (PESGM), Washington, DC, USA, Jul. 2021, pp. 1–5.arxiv.org/abs/2103.13321
- [21]C. Grigg et al., “The IEEE reliability test system-1996. A report prepared by the Reliability Test System Task Force of the Application of Probability Methods Subcommittee,” IEEE Trans. Power Syst., vol. 14, no. 3, pp. 1010–1020, Aug. 1999.
- [22]A. V. Ramesh and X. Li, “Enhancing system flexibility through corrective demand response in security-constrained unit commitment,” in Proc. North American Power Symp. (NAPS), 2021, pp. 1–6.arxiv.org/abs/2006.11519
- [23]Electric Reliability Council of Texas, “2018 State of the Grid Report,” ERCOT, Austin, TX, USA, Tech. Rep., Apr. 2019. [Online]. Available: https://www.ercot.com/files/docs/2019/04/16/ERCOT_2018_State_of_the_Grid_Report.pdfDOI
- [24]W. E. Hart, J.-P. Watson, and D. L. Woodruff, “Pyomo: Modeling and solving mathematical programs in Python,” Math. Program. Comput., vol. 3, no. 3, pp. 219–260, Sep. 2011.
- [25]Gurobi Optimization, LLC, Gurobi Optimizer Reference Manual, ver. 12.0.3, Beaverton, OR, USA, 2025.
- [26]M. Shahidehpour, H. Yamin, and Z. Li, Market Operations in Electric Power Systems: Forecasting, Scheduling, and Risk Management. New York, NY, USA: Wiley-IEEE Press, 2002.DOI
- [27]M. Carrión and J. M. Arroyo, “A computationally efficient mixed-integer linear formulation for the thermal unit commitment problem,” IEEE Trans. Power Syst., vol. 21, no. 3, pp. 1371–1378, Aug. 2006.DOI
- [28]J. Lu and X. Li, “The benefits of hydrogen energy transmission and conversion systems to the renewable power grids: Day-ahead unit commitment,” in Proc. North American Power Symp. (NAPS), Salt Lake City, UT, USA, Oct. 2022, pp. 1–6.
- [29]C. Zhao and X. Li, “Optimal power scheduling for high renewables-integrated energy systems with battery storage,” IEEE Trans. Sustain. Energy, pp. 1–11, 2025.arxiv.org/abs/2309.08853
- [30]J. Zhao, “Energy and reserve co-optimization,” Intermediate Wholesale Electricity Markets (IWEM) training, ISO New England, Holyoke, MA, USA, Sep. 2024. [Online]. Available: https://www.iso-ne.com/static-assets/documents/100016/20240924-iwem-03-energy-and-reserve-cooptimization.pdf