What Are the Challenges in Developing Enhanced Geothermal Systems? Observations from 64 EGS Sites (Pollack, Horne, Mukerji, WGC 2020+1)

Provenance: Peer-reviewed WGC proceedings, Stanford Energy Resources Engineering. The most systematic quantified failure catalogue of the EGS record to date. Counts are described by the authors as minimums. Retrieved via research agent; full text extracted with pdftotext.

Headline tallies (of 64 sites)

Failure class	Count	Notes
Shut down (temporary or permanent) by induced seismicity	6	Basel, Pohang, St Gallen, Landau, Insheim (partial), Soultz curtailment
Drilling / completion / well-integrity problems that raised cost, delayed or terminated	24 (37%)	“at least”
Reservoir creation or circulation failure (no connectivity, water loss, thermal short-circuit)	≥18
Still operating with improved output	29

Quote: “Drilling and plant operation issues, such as holes and cracks in wellbore casing, stuck drill strings, and well collapses, have increased costs, delayed or terminated at least 24 EGS projects.”

Induced seismicity

Pohang, Korea, Nov 2017, Mw 5.4 (Ellsworth et al. give Mw 5.5): US$52M structural damage, 90 injured — the only EGS quake to harm people. National EGS program cancelled.
Basel, Dec 2006, ML 3.4: non-structural damage, US$7M insurance claims, criminal prosecution (acquittal) of the operator’s director. Project abandoned.
St Gallen 2013, ML 3.5 after a gas kick; project ended mainly for low flow and low temperature.
Landau 2009, ML 2.7 and 2.4 during circulation → suspension; restart only after buying EUR 50M/yr liability insurance and cutting production to 55 L/s.
Insheim 2010, ML 2.2/2.4 plus a further M2.0 after mitigation. Soultz Mw 2.4 (2002), 2.9 (2003) → stimulation curtailed. NCPA Geysers (AltaRock) delayed by press criticism referencing Basel before stimulation even began.

Water loss and thermal short-circuiting

Water recovery in first circulation: Ogachi 3% (97% lost), best 32%; Fjallbacka 50% at 100 m spacing; Hijiori 45–70%; Rosemanowes ~70% loss; Falkenberg 60–100% loss; Fenton Hill 7–11.7% loss “set the standard for reasonable.”
Thermal short-circuits: Fenton Hill Phase I 75-day test fell 175 → 85 °C (reservoir likely a single ~100 m joint); Rosemanowes 80.5 → 70.5 °C; Hijiori 163 → ~100 °C; Bacman (Philippines) cut injection 187 → 70 kg/s after too-direct a connection cooled producers. Soultz GPK-3: one zone carries 70% of flow. Habanero: the whole reservoir is one pre-existing fault, which “resulted in significantly lower estimated values of potentially recoverable thermal energy.”
Pressure-dependent permeability (Rosemanowes, Fenton Hill): fractures close when pressure drops, so injection pumps must run continuously — “significantly lowering the energy efficiency potential of the site.” Parasitic pumping is structural, not incidental. Bad Urach: 2.9× transmissivity gain (2002) reverted to baseline by 2003.
Injectivity: Jolokia 7 L/s at 69 MPa wellhead (0.011 gpm/psi) after 14 days; Bradys 15-12 ST1 improved 30× (0.047 → 1.4 gpm/psi) yet still 7× short of the ~10 gpm/psi commercial threshold.

Drilling, well integrity, cost

Habanero-3 failed by stress-corrosion cracking and blew out uncontrolled for 28 days (2009); replacement well H04 cost AUD 50.5M (USD 34.4M); Geodynamics claimed future wells at AUD 16.5M.
Drilling + stimulation can exceed 50% of EGS project cost. Fenton Hill packers succeeded in 6 of 11 operations. Newberry 2012 stimulation fluid escaped through two casing leaks; regulators capped pressure at 210 bar. Coso 46A-19RD (327 °C) abandoned after a month failing to pull a liner. Savina-1 plugged at 2,640 m after differential sticking at 3,700 m.
Soultz economics: drilled 3,300 m more (1,700 → 5,000 m) for only 140 → 200 °C because the mid-section gradient is 5 °C/km; a shallower, cooler, naturally permeable target might have been more economic.

Scaling and corrosion

Genesys Hannover: halite precipitation clogged tubing, project suspended. Hijiori: CaCO₃ scale after thermal breakthrough. Northwest Geysers: chloride casing corrosion / steam leak halted circulation. Soultz: radiation-protection procedures for radionuclide-bearing scale on the reinjection line.

Finance

Paralana failed to raise AUD 5M equity and thereby lost AUD 13M ARENA + AUD 24.5M REDP grants. Australian EGS was also hurt by rig costs bid up by US shale, falling demand and policy uncertainty; remote sites (Paralana, Cooper Basin) needed long new transmission. Quote: “In the end, most abandoned projects also failed due to financial constraints.”

Companion: Lessons from Pohang (Stanford summary of Ellsworth et al. 2019; Lee et al., Science 2019)

Source: https://sustainability.stanford.edu/news/lessons-pohang-solving-geothermal-energys-earthquake-problem

Mw 5.5, 15 Nov 2017: largest earthquake ever caused by an EGS; >1,700 residents displaced; site permanently shut.
Total injected volume across five stimulations (PX-1, PX-2) was only ~12,800 m³ — small by oil-and-gas standards — yet it activated a previously unmapped fault crossing PX-2 at >4 km. Refutes the volume–magnitude (“McGarr”) limit as a safety guarantee.
Warning signs: small quakes continued and grew for weeks after injection stopped; the main event came ~2 months after the last stimulation.
Critique of traffic-light systems: tuned to prevent felt quakes, not low-probability high-consequence ones; Ellsworth argues for continuous evolving-risk assessment with authority to say “we’re done.” Pohang released ~1,000× the energy of Basel.
Quote: “water injected in the other well began to activate a previously unknown fault that crossed right through the well.”

Relevance to the cooling wiki

The risk ledger that any “geothermal as firm drive heat” plan must carry: seismicity governance, parasitic pumping, thermal drawdown that shortens the useful drive-temperature life, and the finding that projects die on finance after surviving the engineering. Contrast with Fervo’s 2024–2026 results, which address the flow and short-circuit problems (propped multi-stage fractures) but not the seismic or financial ones by construction.