Projects · Energy Systems and Grid Analytics

Energy Systems

How Texas's changing power grid ripples through the rest of the energy system, from factories and cars to electricity prices.

Two charts. Left: stacked bars for 2020 to 2050 showing how Texas electricity generation differs between the solar-and-wind targeted scenario and the reference. Wind (blue) and solar (orange) add up to about 0.9 EJ above zero by 2035; natural gas (green), coal (yellow) and nuclear and other (pink) fall below zero by up to about 0.45 EJ; black dots show a net gain of about 0.4 EJ. Right: horizontal bars for 2050 energy use by sector: buildings +0.03 EJ electricity and -0.03 EJ gas; industry +0.18 EJ electricity, -0.14 EJ gas and -0.05 EJ refined liquids; transportation +0.07 EJ electricity and -0.14 EJ refined liquids.
What linking the two models shows for Texas: a grid scenario with more wind and solar (from ReEDS) changes the power mix a lot in GCAM-USA, but shifts fuel use in buildings, industry and transportation only a little. Differences are the solar-and-wind targeted scenario minus the reference, in exajoules.Rendered by the SEAR Lab from the committed GCAM-USA outputs in the code repository of Aghapour, Alavi, Atitebi & Jones (2025), IISE Annual Conference

In plain English

Texas runs most of its own power grid, ERCOT, and leads the country in wind and solar power. Planners need to know what that changing grid means beyond the power plants: for the price of electricity, for the fuels that factories and vehicles burn, and for communities that build their own generation. Answering that takes models that work together, because a detailed grid model sees little of the wider economy, and an economy-wide model sees the grid only coarsely.

The work began with a 2020 study of how community-owned wind and solar change a utility's costs and prices. A 2025 paper then linked NREL's ReEDS grid model to the economy-wide GCAM-USA model for Texas. Raziye Aghapour's work uses NREL's ReEDS model at county-level detail to test transmission plans for Texas: for example, whether the lines ERCOT has proposed for carrying West Texas wind and solar to the cities reduce wasted (curtailed) renewable power, compared with letting the model choose its own upgrades. She also screens which uncertain inputs, from technology costs to fuel prices, move ERCOT electricity prices most, and builds the tools that turn GCAM-USA demand projections into the hourly loads a grid model needs.

Main points

  • Links a detailed grid model (ReEDS) with an economy-wide model (GCAM-USA), so a Texas grid scenario can be traced into buildings, industry and transportation.
  • Looks at the grid from both sides: the utility that must meet demand every hour, and the communities and large customers that may generate their own power.
  • Student analyses have mapped Texas power plants by size, age and fuel and traced the transmission corridors in ERCOT's long-term plans, giving the models a checked picture of today's system.
  • Current work runs hundreds of grid-model scenarios to find which inputs, from fuel prices to electrification, matter most for ERCOT prices; results will be shared once published.
  • The tools behind it, including a converter from GCAM-USA demand to hourly grid-model loads, are being prepared as reusable lab code.

Interactive demoThe Texas grid →

Papers

A greener Texas grid barely changes what factories and cars burn

Aghapour, R., Alavi, S., Atitebi, O.S., & Jones, E.C., Jr. (2025). Texas power system evolution and sectoral energy demand: Contrasting sector-specific and multi-sector model outcomes. Proceedings of the IISE Annual Conference & Expo 2025. https://www.proquest.com/openview/fb47c6b94c33e0e25a48f3549c46ac0e/1

Detailed Texas wind and solar plans cut coal and gas power in an economy-wide model, but industry and transport still run on fossil fuels.

  • The solar-and-wind targets taken from ReEDS grow from 5.2 GW of solar and 23.1 GW of wind in 2020 to 75.2 GW and 88.6 GW in 2050, or 0.59 and 1.30 EJ of generation a year.
  • With those targets, coal generation drops sooner than in the reference scenario and natural gas generation is lower from 2025 on, while nuclear, hydro and rooftop solar barely change.
  • Electricity use rises slightly in buildings, industry and transportation, the only fuel that grows in every sector.
  • Industry and transportation stay dominated by natural gas and refined liquids; in 2050 the model's industry uses about 0.18 EJ more electricity and 0.14 EJ less gas, out of roughly 7 EJ in total.

Correction note. The abstract says buildings show the strongest shift toward electrification; the model outputs show industry gains the most electricity (+0.18 vs +0.03 EJ in 2050). Section 5.2 says transportation's gas use falls; in the data it rises slightly from a tiny base, and it is refined liquids that fall.

Testing Texas energy policy on the whole economy, not just the grid

Aghapour, R., Alavi, S., Atitebi, O., & Jones, E.C., Jr. (2025). Achieving sustainable energy transition: A system analysis approach to energy policy implementation. Proceedings of the IISE Annual Conference & Expo 2025. https://doi.org/10.21872/annual2025_6968

An EPA scenario tool traces how state energy policies would shift emissions, air quality, health and energy prices across Texas.

  • Uses GLIMPSE, the EPA's front end to the GCAM-USA model, to simulate Texas energy policies across the energy sector and the sectors that depend on it.
  • Reports how the policies change greenhouse gas emissions, air pollution, health outcomes and energy prices.
  • Adds a sensitivity analysis to find target levels for alternative energy sources.

What happens to a utility when communities build their own power

Jones, E.C., Jr. (2020). Decomposing systems: Illustrating the utility of distributed energy resources with decomposition techniques. Proceedings of the 2020 IISE Annual Conference. https://www.proceedings.com/content/076/076356webtoc.pdf

When communities build their own wind and solar, they pay less per kilowatt-hour, while the utility must raise prices for everyone else.

  • In every scenario the utility met demand with natural gas alone, because gas is available whenever it is needed and solar and wind cannot cover every hour.
  • Communities that built their own wind or solar paid less per kilowatt-hour than the utility's price, and still bought power from the utility when their own output fell short.
  • As more communities generated their own power, the utility sold less and had to raise its price to its remaining customers, a result that holds after the 2026 correction.
  • Splitting the system into one utility model and one community model let all 40 identical communities be solved with a single community problem.
Line chart of hourly availability over a year, with hours sorted from most to least available. A flat blue line for natural gas sits at 87%. A green wind line falls gradually from about 85% to near zero. An orange solar line starts at 100%, falls steeply and drops to near zero for the last 44% of hours.
Availability duration curves: each technology's hourly capacity factors for one year, sorted from highest to lowest. Gas is fixed at 87% in the model, wind varies but is rarely near zero, and solar is near zero in about 44% of hours. That is why the utility builds gas and communities lean on the utility as backup. Rendered from the public repository's synthetic input workbook (synthetic demand data; the paper used licensed Pecan Street data).Rendered by the SEAR Lab from the paper's public code and synthetic inputs (sear-labs/der-decomp-iise-2020)

Crop leftovers could make North Dakota's ethanol supply chain leaner

Alavi, S., Evans, T., Atitebi, O., Javaheri, A., Fand, Y., & Jones, E.C., Jr. (2024). Multi-source biofuel supply chain network optimization. Proceedings of the IISE Annual Conference & Expo 2024, 1725–1730. https://doi.org/10.21872/2024iise_7959

Adding corn stover and wheat straw to switchgrass lets a North Dakota ethanol network use 10 preprocessing plants instead of 28, and less land.

  • The optimal network opens six biorefineries of 380 million liters a year each, as in the earlier switchgrass-only model, in Cass, Grand Forks, Mountrail, Richland, Rolette and Walsh counties.
  • It opens 10 preprocessing plants, the minimum the model allows, instead of 28; crop residues ship straight from 10 farm counties to the biorefineries without being densified first.
  • Switchgrass still supplies 61% of the biomass, corn stover 38% and wheat straw 1%.
  • The switchgrass is grown on 29% of the available marginal land, against 61% in the switchgrass-only model, freeing land for other uses.

Correction note. The paper prints the network's total cost as $1.21 billion. Re-solving the published model gives $1,121,384,981, so the printed figure appears to be a transposition of $1.12 billion.

Three bar-chart panels comparing an earlier switchgrass-only model with this paper. Biomass mix: 100% switchgrass versus 61% switchgrass, 38% corn stover and 1% wheat straw. Preprocessing plants: 28 versus 10. Share of available marginal land used for switchgrass: 61% versus 29%.
The earlier switchgrass-only model (Zhang et al., 2013, as compared in the paper) versus this paper's multi-source network: where the biomass comes from, how many preprocessing plants are built, and how much of the available marginal land is planted with switchgrass. Both use six biorefineries.Rendered by the SEAR Lab from the committed solution of the published model in the paper's code repository (re-solved and confirmed); earlier-model values as quoted in the paper

Which assumptions drive wasted renewable power in Texas grid models?

Aghapour, R., Alavi, S., Chen, V.C.P., & Jones, E.C., Jr. (2026). What really drives our power system models? Identifying the assumptions that shape ERCOT power futures. Proceedings of the IISE Annual Conference & Expo 2026.

A structured screening of uncertain inputs to NREL's ReEDS model ranks what most changes renewable curtailment in ERCOT.

  • Applied a Morris (elementary effects) screening design to NREL's ReEDS capacity-expansion model to rank which uncertain assumptions most affect renewable curtailment in ERCOT through 2050.
  • The uncertainty ranges for technology costs, fuel prices and electricity demand were drawn from published projections.
  • Preliminary results, from the first of five planned trajectories (40 of 200 runs), point to natural-gas prices, wind capacity-factor assumptions, and the operating cost and efficiency of gas combined-cycle plants as the strongest drivers.
  • Several coal-related inputs had almost no effect in this first trajectory; the authors note that more runs are needed to settle the rankings and reveal interactions.

Grid modernization for a changing climate

Jeffers, R., Kushner, D., Ntakou, E., Paaso, A., Chalamala, B., Jourdier, B., Rahmatian, F., Fotuhi-Firuzabad, M., Jones, E.C., Jr., Araneda Tapia, J.C., Abbas, A., & Price, T. (2024). Enabling climate adaptation and mitigation through grid modernization (TR-124). IEEE Power & Energy Society technical report. https://doi.org/10.17023/02vn-4d23

An IEEE Power & Energy Society technical report on how the grid must change both to cut emissions and to keep working through extreme weather.

  • The grid must decarbonize, bringing in renewables and retiring fossil plants, while serving new loads from electric vehicles and electrified heating and cooling.
  • As more of daily life depends on electricity, the grid will need higher reliability and resilience.
  • The report argues that, for the electricity sector, cutting emissions and adapting to climate change are each impossible without the other, and is written to support practitioners.

Code and materials

People

  • Erick C. Jones Jr., PhD, PEPrincipal Investigator · SEAR Lab directorin
  • Raziye AghapourAuthor, Proceedings of the IISE Annual Conference & Expo 2025 2025 · Author, Proceedings of the IISE Annual Conference & Expo 2026 2026in
  • Sarasadat AlaviAuthor, Proceedings of the IISE Annual Conference & Expo 2025 2025 · Author, Proceedings of the IISE Annual Conference & Expo 2024 2024 · Author, Proceedings of the IISE Annual Conference & Expo 2026 2026
  • Oluwatosin Stephen AtitebiAuthor, Proceedings of the IISE Annual Conference & Expo 2025 2025 · Author, Proceedings of the IISE Annual Conference & Expo 2024 2024
  • Toryon EvansAuthor, Proceedings of the IISE Annual Conference & Expo 2024 2024in
  • Atusa JavaheriAuthor, Proceedings of the IISE Annual Conference & Expo 2024 2024
  • Yash FandAuthor, Proceedings of the IISE Annual Conference & Expo 2024 2024
  • Victoria C. P. ChenAuthor, Proceedings of the IISE Annual Conference & Expo 2026 2026

Current team: names to be added from the lab personnel sheet. Profiles marked in link to LinkedIn. More past and present lab members are on the SEAR Lab team page.

← All projects