🔬 physics

How effective normal stress oscillations advance failure in fault gouge: frequency dependence, non-failure window, and the role of dilation

This study uses coupled Discrete Element-fluid dynamics modeling to demonstrate that sub-critical effective normal stress oscillations can trigger fault gouge failure across most frequencies via dilation-induced strength deterioration, except within a distinct intermediate "non-failure window" (30–200 Hz) where the mechanisms of low-frequency ratcheting and high-frequency dynamic dilation do not overlap.

Original authors: Pritom Sarma, Einat Aharonov, Renaud Toussaint, Stanislav Parez

Published 2026-07-03

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Pritom Sarma, Einat Aharonov, Renaud Toussaint, Stanislav Parez

Original paper licensed under CC BY 4.0 (http://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

Imagine a fault line in the Earth's crust not as a smooth, solid crack, but as a thick layer of sand and gravel (called "fault gouge") sandwiched between two giant, rough walls. This layer is the "grit" that holds the two sides of the fault together.

The paper investigates what happens when you wiggle this sand layer back and forth by changing the pressure pushing down on it. Specifically, the researchers asked: If you shake the pressure up and down, can you make the sand slide and cause a "failure" (like a mini-earthquake), even if you never push hard enough to break it under normal, steady conditions?

Here is the breakdown of their findings using simple analogies:

1. The Setup: The "Sandwich" Experiment

The researchers built a computer model of this sand layer. They applied a constant "shear" force (trying to slide the top wall sideways) and then started oscillating the "normal" force (pushing down on the top wall).

The Rule: They never pushed down hard enough to crush the sand or pull it up so hard that it would slide on its own. They stayed strictly below the "breaking point."
The Surprise: Even though they stayed below the breaking point, the sand did start sliding and failing in most cases.

2. The "Goldilocks" Frequency (The Non-Failure Window)

The most interesting discovery is that the speed of the wiggling (the frequency) matters immensely. The researchers found a strange "Goldilocks zone" where the sand refuses to slide, no matter how much they wiggle it.

Too Slow (Low Frequency): The sand slides, stops, slides, and stops. It's like a person trying to walk through deep mud. They take a step, sink a little, then try to pull their foot out. If they wiggle slowly, they have time to sink (dilate) and get stuck, but eventually, they slip.
Too Fast (High Frequency): The sand starts sliding continuously, almost like it's turning into a liquid. This is like shaking a jar of sand so violently that the grains bounce off each other and the whole mass flows easily.
Just Right (The "Non-Failure Window"): Between roughly 30 and 200 wiggles per second, the sand refuses to slide. It just vibrates in place. It's like trying to push a heavy box across a floor: if you push too slowly, it sticks; if you shake it too fast, it slides; but if you vibrate it at a specific "annoying" middle speed, it just sits there, locked in place.

3. Why Does This Happen? (The Two Mechanisms)

The paper explains that the sand fails for two completely different reasons depending on how fast you wiggle it.

A. The "Ratchet" Effect (Slow Wiggles)

The Analogy: Imagine a ratchet wrench. You turn it forward, it moves a tiny bit, and then you turn it back, but it doesn't quite go all the way back. Over many turns, it slowly creeps forward.
The Science: When the pressure wiggles slowly, the sand grains have time to rearrange and "climb over" each other (dilate) during the low-pressure phase. When the pressure goes back up, they don't fully settle back down. This tiny, irreversible "creep" happens every cycle. Eventually, the sand gets so loose (porous) that it loses its grip and slides. This is driven by the shear (the sideways sliding force).

B. The "Acoustic Fluidization" (Fast Wiggles)

The Analogy: Think of a crowd of people in a hallway. If they just stand there, they are stuck. If you start shaking the floor violently, the people start bouncing and bumping into each other, creating chaos. Suddenly, the crowd can move through the hallway easily because everyone is agitated.
The Science: When the pressure wiggles very fast, the grains don't have time to settle. Instead, the rapid shaking creates powerful "inertial forces" (like being thrown around in a car crash) and pressure waves that agitate the grains. The sand grains start bouncing and vibrating so much that they lose contact with each other, turning the solid layer into a fluid-like state. This happens even without the sideways sliding force.

C. The "Gap" (The Middle Speed)

The Analogy: The "Non-Failure Window" is the gap between these two tricks. The wiggles are too fast for the "Ratchet" to work (the grains don't have time to climb and get stuck), but too slow for the "Acoustic Fluidization" to kick in (the shaking isn't violent enough to make the grains bounce). The sand is trapped in a stable state.

4. Water vs. Dry Sand

The researchers tested this with dry sand and with sand soaked in water.

The Result: They found that water and dry sand behave almost exactly the same way regarding when they fail.
The Difference: The wet sand is just a little slower to react. When the wet sand tries to expand (dilate), the water gets squeezed out, which temporarily holds the sand together a bit longer (like a suction cup). This creates a slight delay, but the overall pattern of "Slow = Fail," "Middle = Safe," and "Fast = Fail" remains identical.

Summary

The paper concludes that how fast you wiggle the pressure is a critical control knob for fault stability.

Slow wiggles cause failure by slowly loosening the sand grain-by-grain (Ratchet).
Fast wiggles cause failure by violently shaking the sand into a fluid (Acoustic Fluidization).
Medium wiggles create a "safe zone" where the fault stays locked.

This helps explain why some earthquakes might be triggered by slow, long-period waves while others are triggered by fast, high-frequency shaking, and why some frequencies might actually prevent a slip from happening.

Technical Summary: How Effective Normal Stress Oscillations Advance Failure in Fault Gouge

Problem Statement
Fault stability in the Earth's crust is governed by the effective normal stress ( $\sigma'_n = \sigma_n - P$ ), where $\sigma_n$ is total normal stress and $P$ is pore-fluid pressure. While the effects of monotonic pore-pressure changes on induced seismicity are well-studied, the response of granular fault gouge to cyclic variations in effective normal stress remains poorly constrained at the grain scale. Natural phenomena (tides, passing seismic waves) and engineered systems (pulsed injection) subject faults to oscillatory loads. Previous work established that cyclic normal stress can weaken faults below the static Mohr–Coulomb threshold, but the behavior across a wide frequency spectrum, the existence of stable frequency windows, and the specific grain-scale mechanisms driving sub-critical failure were not systematically addressed. Furthermore, it remained unclear if pore-pressure oscillations in saturated gouge produce the same response as direct normal-stress oscillations in dry gouge.

Methodology
The authors employed a coupled Discrete Element Method (DEM) and continuum fluid dynamics solver to simulate a two-dimensional, fluid-saturated granular fault gouge layer. The model tracks grain-level micromechanics and pore-fluid pressure evolution.

Setup: A granular layer confined between rigid walls, initially in a dense, over-consolidated state (porosity $\phi_0 \approx 0.185$ ), subjected to a constant shear stress ( $\tau = 0.2\sigma_0$ ) below the static failure threshold.
Forcing Protocols: Two complementary protocols were tested across a frequency range of 0.5 to $10^4$ $1 0^{4}$ Hz, with peak loading kept strictly below the quasi-static Mohr–Coulomb failure threshold:
1. Dry Control: Direct cyclic oscillation of normal stress ( $\sigma_n$ ) on a dry layer.
2. Wet Protocol: Cyclic oscillation of pore pressure ( $P$ ) on a fluid-saturated layer (with constant $\sigma_n$ ), testing Terzaghi's principle of effective stress equivalence.
Parameters: Simulations utilized realistic grain properties (Young's modulus, friction coefficient) and high permeability ( $k = 10^{-11}$ m $^2$ ) to ensure rapid pore-pressure diffusion relative to grain momentum diffusion.

Key Contributions and Results
The study identifies a non-monotonic frequency dependence in fault stability, revealing three distinct failure modes and a robust "non-failure window."

Three Failure Modes and a Non-Failure Window:
- Mode I (Low Frequency, $f \sim 0.5\text{--}1$ Hz): Cyclic failure and arrest. The system slips and arrests in synchrony with the loading cycle.
- Mode II (Intermediate-Low Frequency, $f \sim 5\text{--}10$ Hz): Cyclic sliding without arrest. The layer slides continuously but modulated by the cycle, never fully re-jamming.
- Non-Failure Window ( $f \sim 30\text{--}200$ Hz): A distinct regime where the system exhibits only small-amplitude cyclic creep. No macroscopic failure occurs despite the sub-critical loading.
- Mode III (High Frequency, $f \ge 500$ Hz): Failure leading to continuous sliding. The system transitions to sustained sliding after a period of apparent stability, independent of individual load peaks.
Equivalence of Dry and Wet Protocols:
Both protocols produced identical regime structures (the three modes and the non-failure window) at the same frequencies. This confirms that pore-pressure oscillations and direct normal-stress oscillations act as equivalent forcings via Terzaghi's principle. The primary difference was a time delay in the wet case due to dilatant hardening (transient pore-pressure drops during dilation), but the fundamental stability boundaries remained unchanged.
Mechanisms of Sub-Coulomb Failure:
The paper attributes failure to the gradual deterioration of the friction coefficient $\mu(\phi)$ driven by shear-induced dilation. The mechanism varies by frequency:
- Low Frequencies (Shear-Driven Ratcheting): When the oscillation period $T$ is much longer than the granular momentum-diffusion timescale ( $\zeta_v$ ), each cycle allows sufficient time for shear-driven micro-slip to cause irreversible dilation. The layer does not fully re-compact before the next cycle, leading to a "ratchet" effect where porosity accumulates cycle-by-cycle until a critical failure porosity is reached.
- High Frequencies (Vibration-Driven/Acoustic Fluidization): When $T \ll \zeta_v$ , the shear-driven ratchet is suppressed. Instead, rapid oscillations generate amplified seepage forces (in wet cases), stress gradients, and inertial forces. These dynamic forces agitate the grain assembly, inducing dilation and rearrangement (acoustic fluidization) independent of the shear stress, eventually leading to failure.
- The Non-Failure Gap: The intermediate window represents a regime where oscillations are too fast for shear-driven ratcheting (insufficient time for dilation accumulation) but too slow for vibration-driven forces to become large enough to agitate the grains.

Significance and Claims
The authors claim these results provide a grain-scale mechanistic basis for:

Sub-Coulomb Failure: Explaining how faults can fail under oscillatory loads that never exceed the static strength threshold.
Dynamic Earthquake Triggering: Suggesting that the efficiency of dynamic triggering by seismic waves depends on the interplay between the wave frequency and the fault gouge's intrinsic timescales ( $\zeta_v$ ). The existence of a non-failure window implies that not all frequencies are equally effective at triggering slip.
Cyclic Injection Protocols: Offering a theoretical framework for optimizing fluid injection in geothermal and CO2 sequestration projects. The results suggest that selecting oscillation frequencies within the "non-failure window" could potentially mitigate induced seismicity, whereas other frequencies might paradoxically promote failure at sub-critical pressures.

The study emphasizes that the boundaries of these regimes are not universal but scale with system size (specifically gouge thickness, as $\zeta_v \propto h^2$ ). Consequently, laboratory results must be scaled to field conditions where fault zones are significantly thicker, shifting the frequency windows to lower values. The findings are specific to the high-permeability regime studied; low-permeability gouges may exhibit different behaviors due to slower pore-pressure diffusion.