COWI and Sinotech Team Up on Taiwan’s Super-Hot Geothermal Potential Taiwan’s geothermal story is moving from possibility to execution, and the COWI-Sinotech collaboration is a sign that the sector is entering a more serious phase of development. The partnership is focused on unlocking super-hot geothermal resources, which could improve project economics and expand the country’s clean-energy options. Introduction Geothermal has long been one of Taiwan’s most intriguing renewable resources because the island sits on active tectonic terrain with strong heat potential. What has held the sector back is not a lack of heat, but the difficulty of converting that heat into bankable projects at scale. The new collaboration between COWI and Sinotech points to a more technical, internationally connected approach to solving that problem. Why Taiwan Matters Taiwan has ambitious decarbonization goals, and geothermal fits neatly into the need for firm, low-carbon power. Unlike solar and wind, geother...
Enhanced geothermal systems are one of the few realistic paths to firm zero carbon power at scale, but they work by deliberately changing stresses in the crust, so induced seismicity is not a bug; it is a built‑in consequence that we have to manage, not eliminate.
Image: geothermal wells of power
The real question is whether we can design and regulate EGS so that most earthquakes stay tiny and useful as a reservoir diagnostic, and rare felt events stay within a risk envelope society will accept, with clear rules on who pays when something still goes wrong.
EGS and induced seismicity
Enhanced geothermal systems increase permeability in hot but relatively tight rock by injecting fluid under pressure, which raises pore pressure and shifts effective stresses on pre‑existing fractures and faults. When those faults are close to failure, even modest pressure changes can trigger slip, generating induced seismic events that range from microquakes only instruments detect to felt earthquakes like Basel in Switzerland and Pohang in South Korea.
A widely cited review in 2007 already noted that EGS‑related seismicity had rarely caused significant structural damage but had caused annoyance and public concern, especially when projects operated near urban areas.[1] Later case studies emphasise that the main risk is not thousands of tiny events but the small chance of a moderate event on a larger fault, which can undermine social licence even if the technical risk was judged low ex ante.
The Pohang lesson
The Pohang EGS project in South Korea was designed as a deep stimulation of granitic rock and became infamous after a magnitude 5.4–5.5 earthquake in November 2017 that injured residents and damaged buildings. Subsequent scientific investigations concluded that the mainshock was likely triggered by the project’s injections, which had reactivated a previously unmapped fault near the injection well.
The fallout was severe. Authorities permanently shut the project, commissions recommended against further EGS activity at the site, and the government set up compensation mechanisms for affected residents while debates about legal liability and negligence played out.For the global industry, Pohang became a cautionary tale: even one poorly characterised site with aggressive injection can reset public attitudes and push regulators toward very conservative approaches.
Why we use traffic light systems
To keep operations within acceptable bounds, most EGS projects now run under some form of “traffic light system” that ties injection decisions to real‑time seismic monitoring. In its simplest form, a TLS defines thresholds in terms of local magnitude or ground motion; events below a low threshold are green and allow normal operations, events above an intermediate threshold trigger amber responses such as reduced flow, and events beyond a higher threshold require red actions such as immediate shut‑in and pressure bleed‑off.
TLS frameworks appear explicitly in the US Department of Energy’s induced seismicity protocol and in European “good practice” guidance, which recommend project‑specific thresholds based on local hazard and vulnerability. However, they are inherently reactive: operators only change course after seismicity has already occurred, and fixed magnitude cut‑offs may be poorly tuned to local geology and risk tolerance.
Real‑time microseismic monitoring
Modern EGS projects therefore supplement TLS with dense microseismic networks that record events far below human perception Downhole and surface arrays routinely detect magnitudes down to about −1 or −2, allowing engineers to map fracture growth in three dimensions, see which faults are being activated and track how seismicity evolves as injection proceeds.
Reviews of recent projects argue that microseismic data should be treated as both a safety signal and an operational diagnostic.[8][1] High event rates with small magnitudes can indicate efficient creation of fracture surface area, while changes in magnitude‑frequency distributions or migration of events toward known larger faults can flag rising hazard.[8][4] In this view, microseismicity is not just noise to be minimised but a real‑time window into the subsurface that, if interpreted correctly, helps keep operations out of dangerous parts of the stress landscape.
Adaptive and AI‑informed TLS
Building on that, several research groups and regulators are exploring “adaptive” TLS that adjust thresholds and responses as more data accumulate.Instead of static trigger levels, these systems use evolving hazard estimates that incorporate microseismic patterns, fault models and operational history, tightening or loosening constraints depending on whether indicators point toward increased or decreased risk.
AI and machine learning add another layer by learning relationships between injection parameters, microseismic signatures and subsequent larger events across many datasets. For example, models trained on past EGS and wastewater‑injection projects can flag combinations of event rate, b‑value changes and spatial migration that often precede larger shocks, allowing operators to pre‑emptively reduce or pause injection instead of waiting for a threshold exceedance.[8][9] Early studies emphasise that such AI systems must be physics‑informed and used as decision support rather than standalone autopilots, but they show promise in turning TLS from a simple rulebook into a predictive control tool.
Engineering toward “safer” earthquakes
On the engineering side, best‑practice reviews converge on a few levers that can steer induced seismicity toward smaller, more frequent events instead of occasional larger ones. These include conservative site selection that avoids large, critically stressed faults near population centres; gradual ramp‑up of injection rates; limiting overpressures; and using cyclic or pulse injection strategies that promote distributed micro‑slip rather than allowing stress to build to larger failures.
Numerical studies and field data suggest that controlling total injected volume, injection depth relative to key faults and pressure‑time histories can reduce the likelihood of moderate events, even if microseismicity is abundant. However, authors stress that there is no free lunch: any project that significantly changes pore pressure in a seismically active region will create some probability of felt events, and risk cannot be strictly zero. What can be engineered is the distribution of event sizes and the probability of damaging ground motions, which brings us to the idea of “safe limits.”
Is there a safe limit?
Regulators have moved away from the idea of a universal safe magnitude and toward project‑specific risk criteria based on hazard and exposure. For example, Dutch and European guidelines for geothermal operations emphasise probabilistic seismic hazard analysis and acceptability thresholds framed as the annual probability that ground motions exceed levels likely to cause damage or significant nuisance.
Within this framework, small induced events are expected and tolerated, provided that their ground motions stay below vibration criteria derived from mining and construction standards and that the probability of more damaging motions remains below agreed‑upon thresholds such as one in ten thousand per year. The US DOE protocol similarly emphasises ground motion and risk rather than magnitude alone and positions TLS as one element within a broader seven‑step process that includes screening, outreach, hazard quantification, risk characterisation and mitigation planning.
So there is no single magic number, but there is an emerging consensus that “safe enough” means transparently quantified hazard, explicit risk targets, and operational envelopes designed to keep projects within those boundaries. Whether a given community accepts those boundaries is as much a political question as a technical one.
Liability and who pays
Pohang also crystalised the liability debate. Investigations there led to findings that the EGS project likely triggered the damaging earthquake, and South Korean authorities ultimately set up compensation schemes for victims, while discussions about operator responsibility and government oversight intensified. That experience highlighted gaps in pre‑existing legal frameworks for induced seismicity, which often treated it under generic nuisance or negligence law rather than sector‑specific regimes.
In response, emerging practice in Europe and North America is to embed seismic risk management and liability allocation directly in project approvals.Developers are typically required to: carry insurance that covers induced seismic damage; implement approved monitoring and TLS protocols; report seismic data to regulators; and accept that exceeding specified thresholds can trigger mandatory shutdowns or even permit revocation.[6][3][4] Some frameworks also contemplate shared responsibility, recognising that governments promoting EGS as public policy may bear part of the risk, especially for legacy damage or if state‑owned entities are involved.
Image: advanced geothermal systems networks
There is ongoing debate about whether induced seismicity should be treated more like nuclear risk or CO₂ storage, with explicit state backstops and dedicated funds, or left largely to private insurance and tort systems. For now, most regimes sit somewhere in between, with project‑specific licences that delineate responsibility but no universal compensation fund.
Can we engineer earthquakes safely?
Putting this together, the honest picture is nuanced. Technical literature is clear that EGS induced seismicity can be managed to levels that many experts consider acceptable, provided operators follow robust protocols, regulators enforce them and sites are chosen carefully.[1][4][2] At the same time, case histories like Basel and Pohang show that missteps at a few projects can generate outsized societal and political backlash.
The cutting edge consists of multi‑step protocols like the DOE’s seven‑step framework, dense microseismic networks, adaptive TLS, physics‑informed AI for fracture and seismicity forecasting and operational strategies aimed at fostering frequent tiny events instead of rare larger ones. These tools do not remove risk, but they can lower it and make it more transparent, which is essential for social licence and financing.
Ultimately, the question “can we engineer earthquakes safely” has two answers. Technically, we can substantially reduce and manage risk, though not to zero, and induced seismicity can even be turned into a monitoring asset if handled correctly. Societally, “safe” will be defined community by community; in some quiet regions, even minor felt events may be intolerable, while in others with high natural seismicity, carefully managed induced microseismicity may be seen as a reasonable trade for firm decarbonised power. The EGS industry will live or die on how honestly it engages with that trade‑off and how rigorously it uses the new toolkit of TLS, real‑time monitoring and AI‑guided fracture management to keep the ground shaking mostly where it belongs.
Comments
Post a Comment