Every documented geothermal well that reached superhot (≥374 °C) or near-superhot
conditions, and above all what failed at temperature and why. A $4–13M
well shouldn't repeat a failure someone already paid for. Every record links to its public source.
Krafla, Iceland · Landsvirkjun / Iceland Deep Drilling Project
Max temp
900 °C
Depth
2,096 m
Year
2009
Accidentally drilled into magma; produced the hottest geothermal steam on record (~452 C wellhead), enough for ~36 MWe from one well.
What failed
Superheated steam (~452 C at wellhead) caused silica scaling and valve/wellhead erosion; the well ultimately had to be quenched. Surface materials, not downhole drilling, were the limiting failure. A casing collapse in 2010 led to abandonment in 2012.
Root cause: Surface materials/valves and silica scaling under ~452 C corrosive steam
Lesson for the next well
A single magma-proximal well produced ~36 MWe-equivalent of superheated steam — the prize is enormous; handling 450 C scaling/corrosive steam at surface is the challenge.
Larderello, Tuscany, Italy · Enel Green Power / DESCRAMBLE (EU H2020)
Max temp
517 °C
Depth
2,900 m
Year
2017
Hottest geothermal well on record (~517 C); advanced managed-pressure drilling and high-temperature measurement toward the brittle-ductile zone.
What failed
Ultra-high temperature (>500 C) at the bottom exceeded the survival envelope of standard MWD/logging electronics and elastomers; specialized high-temperature tools were required to sense the deepest section.
Root cause: Standard MWD/logging electronics and elastomers exceed survival envelope above ~500 C
Lesson for the next well
Managed-pressure drilling plus purpose-built high-temperature sensing can reach and log the hottest section; the K-horizon may mark a supercritical/ductile transition.
Kakkonda, Iwate, Japan · NEDO / Geological Survey of Japan
Max temp
500 °C
Depth
3,729 m
Year
1995
Measured >500 C at 3,729 m and located the convective-to-conductive (brittle-ductile) boundary near 3.1 km — a foundational data point for superhot reservoir behavior.
What failed
Logging tools read >414 C at 3,600–3,690 m; the bottom (>500 C at 3,700 m) could only be measured with melting tablets. Fractured, permeable rock ended near 3.1 km, where the granite turns ductile.
Root cause: Ductile rock behavior + permeability collapse above the brittle-ductile transition
Lesson for the next well
The fractured reservoir sits above ~3 km and below ~350 C; beneath it the Kakkonda granite is ductile and nearly impermeable. Superhot reservoirs there must be engineered, not found.
Reykjanes, Iceland · HS Orka / Iceland Deep Drilling Project
Max temp
426 °C
Depth
4,500 m
Year
2017
Confirmed a supercritical resource (426 C, 34 MPa) at 4.66 km — the deepest, hottest IDDP well; flow test blocked by completion integrity.
What failed
Reached supercritical conditions (426 C, 34 MPa after six days of heating) but could not be flow-tested: total circulation losses below ~3.4 km, and casing later found damaged just below 2,300 m during stimulation, blocking logging deeper — the well-construction gap the industry must close.
Root cause: Completion integrity + circulation losses at supercritical depth
Lesson for the next well
Supercritical resources ARE drillable; well construction and completion integrity at 400 C+ is the binding constraint, not the resource.
The Geysers, Sonoma County, California, USA · Calpine / Lawrence Berkeley National Laboratory
Max temp
400 °C
Depth
3,352 m
Year
2011
Created an EGS reservoir in rock up to ~400 C (max 750 F recorded) by pairing an injector (Prati 32) with a producer (Prati State 31).
What failed
Corrosive, chloride-bearing steam damaged the production liner of Prati State 31; testing was delayed until it was replaced with a high-alloy steel or titanium liner.
Root cause: Chloride-bearing (HCl) steam corrosion of standard liner steel
Lesson for the next well
Injecting into a ~400 C zone can create a productive reservoir, but acidic superheated steam rules out carbon-steel completions: plan for titanium or high-alloy liners from the start.
Nesjavellir, Hengill, Iceland · Reykjavík Energy (Orkuveita Reykjavíkur)
Max temp
380 °C
Depth
2,265 m
Year
1985
One of the first wells on record to exceed the critical point of water (>380 C, >22 MPa at 2,265 m); the hostile section was plugged back.
What failed
Reached supercritical conditions (>380 C, ~220 bar) at ~2.2 km, but inter-aquifer cross-flow and blowout risk made the deep section uncontrollable; the well was filled back to ~1.6 km.
Root cause: Well control at supercritical pressure/temperature with no casing or wellhead rated for it
Lesson for the next well
Supercritical fluid sits at only ~2.2 km beside young rift fissures — shallow enough to reach with conventional rigs. Well control and pressure-rated completion, not depth, were the limit in 1985.
Status: Filled back to ~1.6 km (abandoned deep section) · Source
Drilled ~4.1 km into fractured basement and recorded ~380 C near the bottom, approaching the K-horizon that may host supercritical fluid.
What failed
Temperature (380 C at 3,970 m) could only be taken during a drilling stop; sources disagree on the peak (350 vs 380 C), a sign of how hard reliable logging is at these conditions.
Lesson for the next well
Larderello's metamorphic basement stays fractured — and so permeable — down to ~4 km at ~380 C, unlike the ductile granite at Kakkonda. Rock type decides whether superhot rock needs stimulation.
Salton Sea geothermal field, California, USA · U.S. Continental Scientific Drilling Program / USGS
Max temp
355 °C
Depth
3,220 m
Year
1986
Measured ~355 C at ~3.2 km — the first multimillion-dollar U.S. continental scientific well — before casing failure cut off logging.
What failed
Casing failure from late May 1986 blocked logging below ~1.8 km; the 355 C bottom temperature is extrapolated, not directly logged. Brines carried ~25 wt% dissolved solids.
Root cause: Casing failure in hot, metal-rich hypersaline brine
Lesson for the next well
Hypersaline brine at 300–355 C destroys casing fast and blinds downhole measurement. The same brines are now a lithium source.
Fenton Hill, Jemez Mountains, New Mexico, USA · Los Alamos National Laboratory / U.S. DOE
Max temp
320 °C
Depth
4,500 m
Year
1980
Reached ~320 C at ~4.5 km vertical depth (completed May 1980); injector paired with producer EE-3 in the first deep HDR system.
What failed
Long-term flow testing (1992–93) produced only ~5.7 L/s at 183 C — far below rock temperature — so the engineered reservoir could not deliver commercial flow.
Lesson for the next well
The world's first hot-dry-rock reservoir at ~320 C proved heat can be mined from impermeable granite; low flow rates, not heat, were the barrier, and remain the core EGS problem.
Fervo's deepest and hottest well: 15,765 ft TVD, 3,290 ft bit run, and ROP above 300 ft/h past 15,000 ft.
What failed
No failure reported. Bottomhole temperature is projected (500–520 F, ~260–271 C) after thermal recovery, not yet measured; sources differ.
Lesson for the next well
Oil-and-gas drilling practice reached ~4.8 km of hot granite in 16 days (79% faster than the DOE baseline). Drilling speed is no longer the main barrier at ~260 C; the next step is the same speed at 400 C.
Ran a 1 MWe pilot for 160 days in 2013 (19 kg/s at 215 C wellhead; Habanero-4 flowed 39 kg/s open). Abandoned as uneconomic: too remote from a market to sell the power.
What failed
Habanero-2 and -3 were lost to stuck drilling tools and casing problems; Habanero-4 switched to reverse cementing after the Habanero-3 failure review, and cost ~A$50M against a A$48.5M budget.
The reservoir worked; the location did not. Casing and cement are the recurring technical weak link, but distance to load killed the project — site near buyers (today: data centers).
Milford, Utah, USA · U.S. DOE / University of Utah (EGI)
Max temp
225 °C
Depth
2,632 m
Year
2020
Max induced quake
M2
DOE open field laboratory: demonstrated directional drilling and reservoir creation in ~225 C granite; extensive public dataset.
What failed
Hard, abrasive granite drove high bit wear and lower ROP than sedimentary section; motor/tool life in hot hard rock a key cost driver (not superhot but the closest well-characterized hard-rock analog).
Lesson for the next well
Directional drilling and hydraulic fracturing are feasible at commercial-ish rates in ~225 C granite; the open dataset de-risks hard-rock EGS.
Soultz-sous-Forêts, Alsace, France · GEIE Exploitation Minière de la Chaleur
Max temp
200 °C
Depth
5,000 m
Year
2004
Max induced quake
M2.9
Pioneer EGS: ~200 C at 5 km in granite; established stimulation and induced-seismicity management practice still referenced today.
What failed
Induced seismicity during stimulation (felt events up to ~M2.9) constrained injection rates — the defining operational limit for engineered reservoirs.
Second Krafla well to intersect magma (2008), predating IDDP-1; the recovered glass is used to study the size of the Krafla melt body.
What failed
Drilled into magma at 2,571 m; melt returned to surface as quenched glass. The well could not be deepened through the melt.
Root cause: Unmapped magma body at shallow depth
Lesson for the next well
Magma was hit at Krafla in 2008, a year before IDDP-1 hit it ~2.5 km away. Shallow melt here is a repeatable target, not a one-off accident — and a mapping gap that drilling programs must plan for.
Drilled to 5 km in 2006; the felt induced earthquake halted stimulation and the authorities cancelled the project in 2009.
What failed
Stimulation triggered an ML 3.4 earthquake felt across Basel on 8 Dec 2006 — just before wellhead pressure was fully released — ending the roughly $100M project.
Root cause: Induced seismicity from high-pressure injection under a city
Lesson for the next well
Site selection and traffic-light seismic protocols must come before stimulation. Felt events under a city end projects regardless of the resource quality.
Kīlauea Lower East Rift Zone, Hawaiʻi, USA · Puna Geothermal Venture (Ormat)
Max temp
—
Depth
2,488 m
Year
2005
Hit a dacite magma body at ~2.5 km — unusual for basaltic Kīlauea — and explained a 2005 ground-uplift anomaly under the plant.
What failed
Intersected dacitic magma at 2,488 m. Melt flowed up the borehole and had to be redrilled several times, returning glass cuttings to surface.
Root cause: Unmapped magma body at shallow depth
Lesson for the next well
Drilling into magma did not trigger an eruption. Shallow magma bodies under producing fields can be intersected and survived — a data point for magma-proximal superhot concepts.
Menengai Caldera, Kenya Rift · Geothermal Development Company (Kenya)
Max temp
—
Depth
2,106 m
Year
—
A magma-proximal system in the Kenya Rift: wells to ~2.1 km, some hotter than 390 C, with magma inferred just below.
What failed
Glassy, rapidly quenched cuttings just over 2,000 m in MW-04 and MW-06 point to magma under the caldera floor, capping the usable reservoir.
Root cause: Shallow magma limits reservoir thickness to ~1.5 km
Lesson for the next well
Some Menengai wells exceed 390 C at ~2 km. With magma this shallow, the reservoir above it is thin; superhot targets here are close but must stop short of melt.
Figures are best-available public values and can be approximate; where sources disagree we
record the conservative value or leave the field blank. Last updated 2026-10-24.
Read the evidence methodology.
PUT THE REGISTRY TO WORK
From the record to your next well
The public registry is free to read. These services apply it to your own site, program, or product.
Site screening report$499 per site
A sourced desktop screen of one coordinate, benchmarked against the closest wells in this registry and what failed there.
©
Add a well