BRENNEN

Industries / Energy, water and waste / Power generation / Geothermal

Engineered reservoirs

This work makes a reservoir where nature did not: wells are drilled into hot rock that holds little water, water is pumped in to open cracks, and it is then circulated between wells to bring the heat up; in a closed loop, the water instead stays inside sealed wells and never enters the rock.

7

Milestones achieved, from 1977 to 2026.

2

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1977

    Heat from dry granite

    In the 1970s scientists at Los Alamos National Laboratory drilled into hot granite at Fenton Hill, New Mexico, to test an idea they called hot dry rock. The IEA dates the first such pilot well to 1974; that well, GT-2, reached 2,932 m and 197 °C. The team pumped water in to open cracks and linked a second well to them, and from 1977 to 1980 it ran a series of circulation tests through this first reservoir. A 2015 review by Los Alamos authors calls the first, 75-day test "a first ever successful operation of an EGS system in the world," producing up to 4.3 MW of heat. A deeper second reservoir ran for 170 days in 1992 to 1995 at 3.9 to 4.7 MW of heat with negligible cooling, but at only 5 to 7 liters per second against a commercial target of 80 kg/s, and the wells were plugged in 1996.

    Lessonthe water did not split the rock along one new crack, as first thought; it opened a network of joints already in the rock, and later projects were designed around that.

    osti.goviea.blob.core.windows.netosti.gov

    Credit

    Los Alamos National Laboratory, led by geologist Donald W. Brown

  2. 1988

    A short circuit in Cornwall

    The Camborne School of Mines tested the same idea in granite at Rosemanowes Quarry, Cornwall, England, from 1977, at depths of 2 to 3 km where the rock was about 80 °C. The first reservoir lost too much water and could not get within a factor of ten of the 75 liters per second target. A second reservoir lost less water, but the water found a short path between the wells, and between 1985 and 1988 the water reaching the surface cooled from 80 °C to 55 °C. A 1991 review for the UK Department of Energy said £40 million had been spent, that no way to seal such short circuits had been shown, and that power from a commercial plant could cost ten times as much as conventional power.

    Lessonif circulating water finds one fast path, it cools the rock along that path and the rest of the reservoir goes unused.

    osti.gov

    Credit

    Camborne School of Mines; UK Department of Energy (Energy Technology Support Unit)

  3. 2008

    Soultz taps natural faults

    At Soultz-sous-Forêts in Alsace, France, a European project drilled three wells about 5 km deep into granite and stimulated them with water and acid between 2000 and 2007, raising the productivity of two wells about 20-fold and of the third 1.5-fold. A five-month circulation test in 2005 recovered about 25% of a tracer at one production well and 2% at the other, and the project concluded that its heat exchanger was connected to a deep natural reservoir. A 1.5 MW binary plant made its first kilowatt-hour in mid-June 2008, after down-hole pumps were installed because the wells would not flow enough on their own; tests on site found pitting corrosion from the salty brine, about 100 g of salts per liter. A later book chapter dates the first operation to 2009.

    Lessonthe reservoir worked by joining faults already in the rock, and it needed pumps and corrosion-proof equipment to deliver power.

    cordis.europa.eu

    Credit

    the European Economic Interest Group "Exploitation Minière de la Chaleur", with European Union funding

  4. 2013

    A pilot with no buyer

    Geodynamics built a 1 MW pilot plant on its Habanero wells in South Australia's Cooper Basin, where granite is hotter than 280 °C, and ran it for 160 days in 2013. At closure it was producing 19 kg/s of brine at 215 °C at the wellhead. The Australian Renewable Energy Agency (ARENA) calls it the first demonstration of geothermal electricity from an engineered system in Australia. The next stage, heat for future gas processing nearby, ended when Beach Energy declined to join in November 2015, and Geodynamics planned to plug and abandon all its geothermal wells in 2016.

    Lessonhot rock in a remote place needs a customer next to it, and when the nearby gas project fell through there was none.

    arena.gov.au

    Credit

    Geodynamics

  5. 2023

    Horizontal wells in hot rock

    Fervo Energy drilled two wells that turn sideways for about 3,250 ft next to the Blue Mountain geothermal plant in Nevada and fractured them in stages, sealing off one section of the well at a time, using methods from shale oil and gas, with fiber-optic cables in the wells to watch flow and temperature. In a 37-day test in April and May 2023, Fervo reported flows of up to 63 liters per second, water at up to 336 °F (169 °C) and a peak of 3.5 MW of electric power equivalent. On 28 November 2023 Google, which buys the power, said the project was operating and supplying the grid that serves its Nevada data centers, and a 2026 national laboratory report calls it the first commercial-scale EGS drilling and reservoir development pilot in the United States.

    blog.googleosti.govdoi.org

    Credit

    Fervo Energy (Jack Norbeck, Tim Latimer); Google; NV Energy

  6. 2024

    A public test reservoir

    Utah FORGE is the US Department of Energy's field laboratory for engineered geothermal systems near Milford, Utah, begun in 2015, by the IEA's account. In April 2024 its team fractured rock between two wells by perforating their steel casing and pumping in water. In August 2024 it circulated water between them for 30 days at 420 gallons per minute (about 26.5 liters per second); FORGE says nearly 90% of the fluid was recovered and the produced water stayed at about 370 °F (188 °C). FORGE posts its data on the Department of Energy's public Geothermal Data Repository, and a 2026 national laboratory report says FORGE has been largely successful in showing a process for building such reservoirs that others can repeat. A circulation test of up to four months began on 12 August 2026.

    utahforge.comosti.govgdr.openei.org

    Credit

    US Department of Energy; the Utah FORGE team (principal investigator Kristie McLin)

  7. 2026

    Closed loop falls short

    Eavor Technologies' project at Geretsried, Germany, was designed as four closed loops, each made of 24 side branches 3,500 m long drilled from the bottom of two 4,500 m wells, about 320 km of drilling in all, to give 64 MW of heat and 8.2 MW of electricity. Eavor said it began sending power to the grid in December 2025. A few months later the company disclosed that the site made only 0.5 to 2.0 MW gross, with practically no power reaching the grid, as reported by ThinkGeoEnergy. In October 2026 Eavor withdrew its two drilling rigs while it works on the project's commercial terms, including a firm contract for its heat, and Isar Loisach Naturwärme, the town's heat company, said it would reassess how to secure its heat supply.

    Lessonfirst power from a new design says little until the measured output matches the design.

    thinkgeoenergy.comeavor.comiea.blob.core.windows.net

    Credit

    Eavor Technologies

Watch for

  1. Commercial power at Cape Station

    Fervo Energy says the first 33 MW block of its Cape Station in Beaver County, Utah, synchronized to the grid on 24 September 2026 and entered commercial operation on 30 September, making 33 MW net, which it calls the first greenfield engineered geothermal development in the world to do so; it expects two more blocks by 1 January 2027 and another 400 MW in 2028. EIA's September 2026 inventory of generators lists one Cape Generating Station unit as operating (43 MW nameplate, 30 MW net summer capacity) and two more as planned for January 2027.

    Achieved whenEIA's monthly generation data or a buyer's report show the plant delivering close to the claimed output over at least six months.

    renewableenergymagazine.comeia.gov

    Claimed by

    Fervo Energy, September and October 2026

  2. Superhot rock at Newberry

    Mazama Energy says it created the hottest engineered geothermal system yet, with 331 °C (629 °F) at the bottom of its wells on the flank of Newberry Volcano, Oregon, by stimulating an old well and drilling a new 10,200 ft well to within 6 ft of its planned path, and that circulation tests showed the wells were connected. It plans a 15 MW pilot with horizontal wells in 2026 and to drill into rock above 400 °C.

    Achieved whenan outside party, such as a national laboratory or a peer-reviewed paper with released data, reports the production temperature and flow from a sustained circulation test.

    ktvz.com

    Claimed by

    Mazama Energy, October 2025

Open

  1. 01

    Score the FORGE forecasts

    A public scorecard for Utah FORGE's 2026 extended circulation test: collect every forecast of production temperature, flow and water loss published before the test began on 12 August 2026, add blind forecasts made from public data before FORGE releases its results, and score all of them against the measured data once posted.

    The brief
    Credit · unclaimed
  2. 02

    Check next-generation output claims

    A public tracker that sets what engineered geothermal projects say they produce against what grid records show: monthly net generation from EIA for Cape Station (EIA plant 68068) against Fervo's stated 33 MW per block, the change in output at the Blue Mountain plant in Nevada that Project Red was built to supply, before and after November 2023, and any further engineered or closed-loop plant as it appears in EIA's records.

    The brief
    Credit · unclaimed

Open 01 · the brief

Score the FORGE forecasts

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public scorecard for Utah FORGE's 2026 extended circulation test: collect every forecast of production temperature, flow and water loss published before the test began on 12 August 2026, add blind forecasts made from public data before FORGE releases its results, and score all of them against the measured data once posted.

Why it is still open

FORGE's principal investigator says that "there is still a great deal we do not know about thermal breakthrough and water loss over time," and that the test is meant to answer it. Forecasts exist, such as Lee and Ghassemi's 26-day and 365-day simulations of extended circulation, posted in September 2025, and Mina Khalaf's January 2026 preprint, which forecasts FORGE temperatures up to 60 months ahead, but a November 2026 search and a search of the Geothermal Data Repository found no challenge or comparison that scores them against the 2026 test.

utahforge.comgdr.openei.orgarxiv.org

Built with what exists

FORGE's public data on the Geothermal Data Repository (the August 2024 circulation test, tracer results and fiber-optic temperature and strain records); open reservoir simulators such as GEOS and OpenGeoSys; a low-cost AI model to pull forecast numbers out of papers and reports (see the Note "The jobs agents repeat"); Python.

  1. First step

    Download the summary plots in Geothermal Data Repository submission 1766 and the August 2024 circulation data, and write down each forecast's production temperature at day 90.

  2. Done looks like

    A table of at least five forecasts, each either published before 12 August 2026 or time-stamped before results are released, scored on production temperature (error in °C) and share of water recovered (error in percentage points) at days 30, 60 and 90 against FORGE's data.

Open 02 · the brief

Check next-generation output claims

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public tracker that sets what engineered geothermal projects say they produce against what grid records show: monthly net generation from EIA for Cape Station (EIA plant 68068) against Fervo's stated 33 MW per block, the change in output at the Blue Mountain plant in Nevada that Project Red was built to supply, before and after November 2023, and any further engineered or closed-loop plant as it appears in EIA's records.

Why it is still open

The Clean Air Task Force's 2025 database of 103 engineered geothermal projects was built from more than 400 papers, reports and press releases, and the report says publicly available data on operating performance is particularly limited. EIA lists Cape Station's first unit as operating from September 2026, and EIA's monthly generation data for September are due at the end of November 2026, though only plants in its monthly sample appear there before the annual release; no outside tally of delivered output has been published.

catf.useia.gov

Built with what exists

EIA-923 monthly generation files and EIA-860M monthly generator inventories; Python with pandas.

  1. First step

    Download the EIA-923 files for 2022 to 2026, pull the monthly net generation for the Blue Mountain plant (EIA plant 56982), and check whether Project Red's output is reported there or under a separate plant.

  2. Done looks like

    A monthly table for each plant from January 2022, refreshed each month, with capacity factor on EIA's net summer capacity and the gap between reported output and each company's stated output; for Cape Station, at least six months of data.