🔬 physics

Maximum fluid-induced earthquake magnitude shifts from injected volume to stress control

This study identifies two distinct rupture regimes for fluid-induced earthquakes—self-arrested and runaway—demonstrating that while maximum magnitude is controlled by injected volume in the former, it is governed by effective background stress in the latter, thereby reconciling volume-based hazard models with the occurrence of unexpectedly large induced earthquakes.

Original authors: Roxane Ferry, Simon Guérin-Marthe, Corentin Noël, Jérémie Engler, Jean-François Molinari, Marie Violay

Published 2026-08-05

📖 7 min read🧠 Deep dive

CC BY 4.0

Original authors: Roxane Ferry, Simon Guérin-Marthe, Corentin Noël, Jérémie Engler, Jean-François Molinari, Marie Violay

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Earth's Ticking Time Bomb: Why Water Isn't the Only Story

Imagine the Earth's crust as a giant, slightly sticky puzzle. The pieces (tectonic plates) are constantly trying to slide past each other, but friction holds them in place, like a heavy book resting on a rough table. This friction creates a "grip" that keeps the book from sliding, even though someone is slowly pushing it. In the world of geology, this grip is called friction, and the force pushing the plates is called stress. When the push gets too strong, or the grip gets too weak, the book suddenly slides, releasing a massive burst of energy. We call this an earthquake.

Now, imagine you have a way to make that grip weaker without changing the push. If you squirt a slippery liquid (like water) between the book and the table, the friction drops, and the book might slide. This is what happens when humans pump fluids deep underground for things like geothermal energy or oil extraction. The water increases the pressure inside the tiny cracks in the rock, effectively "lubricating" the fault lines. For decades, scientists and engineers have worried: If we pump in a certain amount of water, how big of an earthquake could we accidentally trigger?

The old rule of thumb was simple: More water equals bigger earthquakes. It was thought that the size of the earthquake was strictly limited by how much fluid you injected. If you pumped in a bucket, you'd get a tiny tremor; if you pumped in a swimming pool, you'd get a bigger one. But this rule has been broken in the real world. Sometimes, people pump in a relatively small amount of water, and boom—a massive, destructive earthquake happens that shouldn't have been possible based on the volume alone. This paper dives into a laboratory to figure out why the old rule failed and what the real "trigger" is.

The Great Slip-Up: When Water Meets Stress

In this study, a team of researchers at the École Polytechnique Fédérale de Lausanne (EPFL) built a giant, high-tech "fault line" in a lab to watch earthquakes happen in slow motion. They used two massive sheets of clear plastic (PMMA), each 2.5 meters long, pressed tightly together. They squeezed them with a force of about 7.6 MPa (which is roughly the pressure you'd feel if you were 760 meters underwater) and then started pumping water into the crack between them at a steady rate of 0.1 cm³/s.

They didn't just watch; they set up a "stress trap." Before injecting water, they loaded the plastic sheets with different levels of "pre-stress"—basically, how close the sheets were to slipping on their own before the water even touched them. They tested everything from a gentle 50% of the breaking point all the way up to a tense 99%.

Here is where the story gets exciting. The team discovered that the size of the earthquake isn't just about how much water you pour in. It's about the tension already in the rock. They found two very different ways the fault can break, and which one happens depends on that pre-existing stress, though not in a simple "on/off" way.

1. The "Self-Arrested" Slip (The Safe Zone)

When the rock was under low stress (around 50% to 75% of its breaking point), the water acted like a gentle nudge. The rupture (the sliding part) started where the water was, but it couldn't go far. It was like a fire spreading in a damp forest; it sputtered along the wet patch and then died out because the surrounding rock was too "grippy" to let it go further.

The Result: The earthquake stayed small. Its size was strictly controlled by the volume of water injected. If you injected more water, the wet patch got bigger, and the slip got bigger, but it never escaped the water's influence. This is the "old rule" working perfectly.

2. The "Runaway" Rupture (The Danger Zone)

But when the rock was under high stress (above 85% of its breaking point), the story changed completely. Here, the rock was already teetering on the edge of disaster. The water didn't just nudge it; it was the final straw. Once the slip started, it didn't care about the water anymore. It was like a snowball rolling down a steep, dry hill. Even if the water only covered a tiny spot, the earthquake "ran away," ripping through a substantial fraction of the 2.5-meter plastic sheets, and in some cases, extending across the entire length.

The Result: These earthquakes were huge. Their size was not controlled by how much water was injected. Instead, it was controlled by how much stress was already stored in the rock. The water just started the engine, but the stored energy of the rock drove the car.

The Magic Switch: Finding the Tipping Point

The researchers used a mix of real experiments and computer simulations to find the exact "tipping point" where the behavior switches from safe to dangerous. While computer simulations suggested a sharp threshold around 53% of the reactivation stress, the real-world experiments revealed a more complex picture. Instead of a strict switch where one behavior stops and the other begins, the team observed a probabilistic coexistence. Under most testing conditions, both self-arrested and runaway events happened together. The initial stress didn't act as a rigid gatekeeper; rather, it modulated the likelihood of each outcome. High stress made runaway ruptures much more common, while lower stress favored self-arrested slips, but the transition was a shift in probability, not a single deterministic line.

This finding is a big deal because it explains why some small water injections cause massive earthquakes (like in Pohang or Basel) while huge injections sometimes only cause tiny tremors. It's not just about the volume of the water; it's about the stress state of the ground and the likelihood of a rupture escaping the fluid zone.

The team also calculated the maximum size of these earthquakes. For the "runaway" ones, the maximum energy (seismic moment) could reach up to 9 × 10⁴ N·m in their lab setup, which is far larger than what the water volume alone would predict. They showed that once a rupture goes "runaway," it breaks free from the fluid's control and is only stopped by natural barriers in the rock or the edges of the fault itself.

Why This Matters

This paper doesn't just say "be careful." It gives us a new way to think about safety. The old way of looking at earthquake risk was like checking how much fuel is in a car's tank to guess how far it will drive. This new research says, "Wait, we also need to know if the car is parked on a steep hill or a flat road."

If the ground is already under high stress (the steep hill), even a tiny bit of fluid injection can trigger a massive, runaway earthquake because the conditions are right for the rupture to escape. If the ground is relaxed (the flat road), you can pump in a lot of water, and the earthquake will stay small and contained.

The authors emphasize that we can't just look at the injection volume to predict the worst-case scenario anymore. We have to measure the stress in the ground. By understanding this "stress-controlled switch," engineers can better design safety protocols (like traffic-light systems that shut down operations if things get too risky) to prevent small slips from turning into runaway disasters. The paper confirms that while water is the spark, the stress is the fuel that determines how big the fire gets.

Technical Summary: Maximum Fluid-Induced Earthquake Magnitude Shifts from Injected Volume to Stress Control

Problem Statement
Fluid injection during subsurface operations (e.g., geothermal energy, wastewater disposal) can trigger seismic events ranging from microseismicity to destructive earthquakes with magnitudes exceeding 5. Traditional seismic hazard frameworks estimate the maximum magnitude ( $M_{max}$ ) based on the total injected fluid volume ( $\Delta V$ ), assuming that rupture propagation is self-arrested within the fluid-pressurized zone. However, unexpected large induced earthquakes at sites such as Pohang, Basel, and Oklahoma have exceeded these volume-based predictions. This discrepancy suggests that rupture propagation is not always bounded by the pressurized zone. While recent numerical and laboratory studies hypothesize that elevated background stress may promote "runaway" ruptures (where slip propagates beyond the fluid front), the physical transition between self-arrested and runaway regimes, and its specific implications for maximum magnitude scaling, have remained unresolved due to a lack of experimental apparatus capable of generating statistically meaningful datasets for both regimes under controlled conditions.

Methodology
The authors address this gap using a combined experimental and numerical approach:

Experimental Apparatus: A large-scale biaxial shear apparatus was utilized featuring a 2.5-meter-long fault interface created by two PMMA plates. The system allows for constant normal stress loading and constant-rate fluid injection at the interface.
Instrumentation: The setup includes 16 fluid pressure sensors and 9 strain-gauge rosettes to monitor fluid pressure diffusion, shear stress evolution, and rupture propagation in real-time.
Experimental Protocol: Experiments were conducted across a range of far-field prestress levels (50% to 99% of the reactivation stress). Fluid injection triggered a spectrum of slip behaviors, generating a dataset of over 35 rupture events.
Numerical Simulations: Fully dynamic Finite Element Method (FEM) simulations were performed using the open-source code Akantu. These simulations modeled the exact experimental geometry and material properties to isolate the mechanical threshold governing the transition between rupture regimes, free from the intrinsic heterogeneities present in the physical experiments.

Key Results
The study identifies two distinct rupture regimes governed by the effective average prestress ratio ( $S_{eff}$ ), which compares available driving stress to fault breakdown strength:

Self-Arrested Ruptures (Low to Intermediate Prestress):
- Occur when $S_{eff}$ is low.
- Rupture propagation is strictly bounded by the fluid-pressurized zone ( $L_{rup} \approx W_{pf}$ ).
- Scaling Law: The seismic moment ( $M_0$ ) scales deterministically with the injected fluid volume ( $\Delta V$ ). The authors derive equivalent "operator scaling" (based on surface-monitored injection parameters) and "experimentalist scaling" (based on measured pressurized zone width) for this regime.
- These events align with traditional volume-based hazard models.
Runaway Ruptures (High Prestress):
- Occur when $S_{eff}$ is high. In physical experiments, runaway ruptures were observed at prestress levels $\ge$ 85%, where they coexisted with self-arrested events, though they became the dominant mode at these high stress levels.
- Ruptures break out of the fluid-pressurized zone, propagating across the entire fault interface ( $L_{rup} = L_{max}$ ) driven by ambient tectonic strain energy.
- Scaling Law: The maximum magnitude is controlled by the effective background stress and the finite dimensions of the fault, decoupling from the injected volume. Interestingly, the study notes a "nucleation trade-off": higher injected volumes can lower the effective normal stress at nucleation, potentially reducing the dynamic stress drop for these events, but the propagation itself is no longer limited by fluid diffusion.
Transition Threshold:
- The experiments and simulations resolve the transition between these regimes, revealing a distinction between experimental observations and theoretical predictions.
- In the physical experiments, the initial fault prestress did not act as a strict deterministic threshold; instead, self-arrested and runaway events coexisted under most testing conditions, indicating a probabilistic modulation of rupture modes where runaway ruptures become more prevalent at high prestress.
- Conversely, the numerical simulations, which isolate the underlying mechanics by removing intrinsic heterogeneities, show prestress acting as a sharp deterministic switch. Above a critical value (~53% of reactivation stress), all ruptures are runaway; below it, they remain self-arrested.
- This numerically determined threshold closely matches the theoretical prediction by Garagash and Germanovich (2012), which posits that runaway rupture occurs when initial prestress exceeds the ratio of residual to peak friction coefficients ( $f_r/f_p$ ).
- The study provides the first direct experimental evidence for this theoretical critical prestress threshold, while clarifying that in real-world (heterogeneous) settings, the transition manifests as a shift in the relative likelihood of competing rupture modes rather than a strict binary switch.

Significance and Claims
The paper claims to reconcile volume-based hazard models with the occurrence of anomalously large induced earthquakes by establishing a framework linking rupture regime to maximum magnitude.

Bifurcation of Magnitude Control: The maximum magnitude of fluid-induced earthquakes is governed by a stress-controlled bifurcation. For self-arrested events, magnitude is a function of injected volume; for runaway events, it is a function of the ambient geomechanical stress state.
Hazard Assessment Implications: The authors assert that reliable estimation of $M_{max}$ cannot rely solely on injection volume. Instead, it must integrate fluid injection parameters with the in-situ stress state to determine whether a rupture is likely to remain self-arrested or transition to a runaway mode.
Validation of Theory: The work validates the theoretical bifurcation predicted by Garagash and Germanovich (2012) and demonstrates that the transition is fundamentally controlled by the background stress state, though the manifestation of this control ranges from a probabilistic coexistence of modes in heterogeneous experiments to a sharp mechanical switch in idealized homogeneous systems.

The study concludes that while volume-based models successfully predict the numerous small events confined to the pressurized zone, they fail to bound the magnitude of runaway ruptures, which are instead limited by geological barriers and the stored elastic strain energy of the system.