Generation

How the generation fleet is built and dispatched

The generation fleet is modeled at the plant level. Each unit carries technology type, capacity, heat rate, fuel, and commitment parameters used in unit commitment and dispatch. SPP carries a large installed wind base, so renewable shapes are a primary driver of price formation.

Fleet inventory and zone mapping

The base generator inventory is sourced from EIA-860M, using the May 2026 monthly release as the latest available fleet snapshot to build on, with generation and fuel data from EIA-923. Each plant is mapped to a settlement zone by the following precedence order:

1. Balancing-authority code: where it maps directly to an SPP zone.
2. Settlement-node name match: the plant name matched to SPP’s real-time settlement nodes, giving the node’s physical zone.
3. Pricing node: the zone derived from the unit’s own LMP node.
4. Location fallback: the generator’s county, which is mapped to a zone.

60-mile radius rule: a settlement-node or pricing-node match only holds if the resulting zone has a county within 60 miles of the plant. This places generators in their physical locations rather than by ownership structure, which gives more realistic congestion behavior.

CSWS western footprint: the CSWS zone has a highly dispersed footprint, with the majority of its territory in eastern Oklahoma but a significant presence in western Oklahoma. This shape does not make sense as a single zone in a zonal model, so the western part is redistributed to neighbouring zones to give a more realistic pattern of congestion.

See the Capacity Expansion Model.

Thermal generation

Thermal units are represented with minimum and maximum output, start costs, and heat-rate-derived marginal costs. Heat rates and minimum-load values are sourced from EIA-860 and EIA-923, with per-technology fallbacks where plant-level data is missing.

• Minimum up and down times: applied to the units where they bind: combined-cycle gas and coal. Peaking and gas-steam units are freely dispatchable; nuclear runs as always-on (must-run).
• Start costs: technology-specific, combining start fuel and other per-start costs, and added to the unit commitment decision.
• Seasonal derates: summer and winter net-capacity derates are applied from EIA-860 ratings, with spring and fall months held at nameplate.

Each thermal unit’s marginal bid is its heat rate applied to delivered fuel price, plus any carbon cost and less applicable tax credits. Stepped bids above marginal cost capture mark-ups under tight system conditions.

Heat rate is the amount of fuel a unit burns to produce a unit of electricity, so a lower heat rate means a more efficient plant and a lower fuel cost per MWh. The model combines the heat rates reported in EIA-923 with commodity prices and a bid curve to form each thermal generator’s bid prices. Below we can see how heat rates vary significantly between generators and technologies.

• Combined-cycle gas is the most efficient thermal technology, clustering near 7 MMBtu/MWh, so it typically sets price below coal when gas is cheap.
• Coal and gas-steam units sit around 10 to 12 MMBtu/MWh.
• Combustion-turbine peakers are the least efficient and run only when prices are high enough to cover their fuel burn.

Retirements

Thermal units leave the fleet on their announced retirement dates, taken from the latest EIA-860M filings and from utility Integrated Resource Plans (IRPs). Coal units without an announced date are additionally retired at a fixed age, 62 years after their commercial operation date, so the aging coal fleet winds down over the horizon rather than running indefinitely. Two plants publicly cited as SPP carbon capture candidates, Gerald Gentleman and Antelope Valley, are exempt from the age rule. Gas units retire only on their announced dates, with no age rule. Around 15 GW of SPP coal retires by 2050 on this basis.

Outages

Generator availability is captured from SPP’s published “capacity on outage” feed, which reports hourly capacity out of service by fuel category across the footprint. These outages are applied at the system level by technology rather than to individual units, scaling available capacity to reflect realistic planned and unplanned outage patterns.

Renewable generation

Wind and solar are modeled with weather-year hourly capacity factor profiles. County-level profiles from NREL’s reV dataset are used where available and anchored to metered EIA-930 output for the weather year, so the modeled fleet reproduces observed SPP wind and solar generation. For weather years outside the reV coverage, EIA-930 generation provides the shape directly.

• Wind aging: a plant-level performance derate captures declining output as turbines age, using the capacity-weighted performance-by-age curve from Berkeley Lab’s Land-Based Wind Market Report.
• Curtailed wind is added back: EIA-930 records metered output, which is net of curtailment. Anchoring to it alone would strip curtailed energy out before dispatch and leave the model nothing to curtail. The model therefore adds SPP’s published wind curtailment back to the weather-year profile in every hour in which at least one zone’s day-ahead price was negative, spread over those zones in proportion to their wind output. Curtailment in hours when every zone priced positive is left out: it reflects congestion inside a zone, which a zonal network cannot reproduce. New wind builds take the footprint-average uplift.
• No explicit curtailment cap: with that energy restored, curtailment emerges economically from the supply-demand balance and transmission limits, rather than from a hard constraint.

A detailed treatment of capacity factor construction and validation is maintained separately.

Hydro

Hydro is dispatched to the water actually available, rather than freely against a price.

• Daily energy budget: each unit must produce its weather-year energy every day, taken from EIA-923 plant capacity factors. It cannot generate more than its water allows, or hold water back indefinitely.
• Mostly must-run: 90% of each hour’s expected output runs regardless of price, as most SPP hydro self-schedules rather than offering economically.
• Flexible remainder: the rest of the day’s energy is offered as a single block at a water value, up to the unit’s nameplate capacity. The water value is the capacity-weighted average short-run marginal cost of the dispatchable thermal fleet, recalculated each month so it tracks seasonal fuel prices. Because each day’s energy is fixed, the offer price decides when that share runs, not whether, so it moves into the highest-priced hours.

In the 2023 backtest the model produces 8.35 TWh of hydro against 8.34 TWh metered by EIA-930, follows the monthly pattern closely, and reaches an hourly correlation of 0.92. Without the must-run share, an energy budget alone empties hydro overnight and piles it into the evening peak, which flattens price spreads.

Hydro outages are not modeled, as SPP does not publish outage data for this generation type.

Nuclear

SPP’s nuclear units, Wolf Creek and Cooper, run as baseload. They are not cycled for economic dispatch, so the model bids them well below every other resource, letting them self-schedule and clear in every hour they are available. This deep bid also stands in for the federal nuclear production credit, which pays existing reactors on top of the energy price. Availability follows SPP’s nuclear capacity-on-outage feed, so refueling outages are reflected.

Storage

Battery energy storage uses each unit’s reported duration where available; units without a reported energy capacity inherit the capacity-weighted average duration of the existing SPP battery fleet, at 85% round-trip efficiency. Pumped hydro defaults to ten-hour duration. Storage is dispatched in the production cost model and, for individual assets, in the Dispatch Model.

Policy and incentives

• Wind Production Tax Credit (PTC): modeled as a bid credit on a declining phase-out schedule keyed to each project’s construction-start vintage (stepping down through 100%, 80%, 60%, and 40% of the full credit), allowing wind to bid below zero where it still earns the credit. The credit runs for ten years from a plant’s in-service date. A repowered plant restarts that clock from its repower year, identified from the US Wind Turbine Database retrofit records and EIA-860 planned repowers. That keeps about 3 GW of repowered SPP wind bidding below zero where it would otherwise have been treated as having exhausted the credit.
• Investment Tax Credit (ITC): applied on the investment side, reducing the effective capital cost of new solar and new storage in the Capacity Expansion Model rather than as a production bid credit.
• Power-purchase agreements: the contracted share of wind and utility solar, estimated from Berkeley Lab’s published PPA data for SPP projects, by technology and vintage, is re-bid at levelized PPA prices to reflect contracted output.

State renewable portfolio standards are enforced in the Capacity Expansion Model through tradable renewable energy credits, rather than in dispatch. No carbon price applies in the footprint: there is no regional cap-and-trade program such as RGGI and no state carbon price, so the carbon cost in each unit’s bid is zero.

Data sources