Injection-rate effects on failure in a fluid-saturated granular fault gouge

This paper combines analytical theory and numerical simulations to demonstrate that fluid injection rate governs fault-gouge failure by creating pressure heterogeneity, where slow injection causes uniform weakening while rapid injection preserves strength in distal regions, thereby offering a refined framework for predicting seismicity in geotechnical operations.

The Gist

Imagine the Earth's crust is like a giant, cracked sidewalk. Inside those cracks, there isn't just empty space; it's packed with crushed rock, sand, and dirt. Geologists call this "fault gouge." Now, imagine someone starts pumping water into this crack. This is a common practice for things like geothermal energy or waste disposal, but it carries a risk: the water pressure can push the crack open, causing the ground to slip and triggering an earthquake.

This paper asks a simple but tricky question: Does it matter how fast you pump the water in?

The researchers found that yes, it matters a lot. If you pump slowly, the water spreads out evenly, and the crack slips easily. But if you pump very fast, you actually need more pressure to make the crack slip.

Here is how they figured this out, using a mix of math and computer simulations.

The "Crowded Room" Analogy

Think of the fault gouge (the crushed rock) as a crowded room full of people (the grains).

• The Goal: You want to get everyone to move to the side (slip).
• The Water: The water pressure is like a "push" trying to make the people move apart.
• The Expansion: As the people try to move, they naturally spread out and take up more space (this is called "dilation").

The Two Scenarios

1. The Slow Pour (Slow Injection)
Imagine pouring water into the room very slowly. The water has plenty of time to seep through the crowd and reach every single person. The pressure becomes uniform. Everyone feels the push at the same time, the whole crowd loosens up together, and the room slips easily. This is what old theories predicted: more water pressure = easier slip.

2. The Firehose (Fast Injection)
Now, imagine blasting water into the room from one side at high speed.

• The Gradient: The people right next to the hose get soaked and pushed hard immediately. But the people at the far end of the room? They are still dry and standing firm.
• The Bottleneck: Even though the people near the hose are ready to move, the people at the far end are still holding the line. The whole room can't slip until the farthest people are pushed hard enough.
• The Result: To get the whole room to slip, you have to crank up the pressure at the hose to a much higher level than you would with a slow pour. The fast injection creates a "pressure gradient" where the push is strong in one spot and weak in another.

The "Loose Sand" Effect

There is a second twist. As the rock grains try to slide, they don't just slide; they jumble and rearrange, making the layer slightly thicker (dilation).

• In the computer simulations, the researchers stepped the pressure up slowly, letting the grains settle after each step.
• They found that as the grains rearrange and the layer gets "looser," the material actually gets weaker.
• However, because the fast injection creates those uneven pressure zones (strong near the hose, weak far away), the "weakness" of the loose sand doesn't help the whole fault slip until the pressure is high enough to overcome the strong, dry spots at the far end.

The Mathematical "Recipe"

The authors created a new math formula to predict exactly when the fault will slip. Their formula says the pressure needed to cause a slip depends on three things:

1. How fast you pump: Faster pumping = higher pressure needed.
2. How long the fault is: Longer faults = much higher pressure needed (because the pressure has to travel further to reach the weak spots).
3. How fast the water moves through the rock: If the rock is very porous (water moves fast), the pressure equalizes quickly, and you need less pressure.

Why This Matters (According to the Paper)

The paper concludes that we can't just use old, simple rules that assume water pressure is the same everywhere.

• The Trade-off: If you inject fluid very fast, you might need higher pressure to trigger a slip, which sounds safer. However, because you are pumping fast, you might reach that dangerous pressure threshold sooner in time.
• The Size Factor: The length of the fault is a huge factor. A short fault behaves like the "slow pour" (uniform pressure), but a long fault behaves like the "firehose" (uneven pressure), making it much harder to predict when it will slip.

In short, the paper shows that speed changes the rules. Pumping fast creates uneven pressure zones that act like a "holding pattern," requiring significantly more force to break the fault than pumping slowly does. This helps engineers understand that the size of the fault and the speed of injection are critical factors in preventing or predicting induced earthquakes.

Technical Summary

Technical Summary: Injection-rate effects on failure in a fluid-saturated granular fault gouge

Problem Statement
Fluid injection into the subsurface for energy extraction, waste disposal, and resource development is a known driver of induced seismicity, often reactivating gouge-filled faults. While the reduction of effective normal stress via pore-pressure diffusion is the standard mechanism for fault weakening (Terzaghi's effective stress law), the role of the injection rate in controlling failure remains poorly understood. Classical models often assume uniform pore pressure, yet recent studies indicate that heterogeneous pressure distributions can significantly alter failure thresholds. Specifically, rapid injection may create strong pressure gradients, leaving distal fault regions strong and delaying failure, whereas slow injection allows for uniform weakening. Furthermore, unlike bare-surface faults where hydraulic connectivity dominates, gouge-filled faults involve complex granular responses, including dilation (volume expansion) during the pre-failure phase. This dilation has competing effects: it can cause transient pore-pressure drops (dilatant hardening) or, in the context of progressive grain rearrangement, lead to a reduction in packing density and shear strength (weakening). The interplay between injection rate, pressure diffusion, and granular dilation in fluid-saturated gouge requires a quantitative framework.

Methodology
The authors employ a dual approach combining analytical theory and coupled numerical simulations:

1. Numerical Simulations: A coupled solid–fluid code is used, integrating a Discrete Element Method (DEM) for solid grains (modeled as 2D disks with linear elastic frictional contacts) and a continuum pore-fluid solver (finite differences on an Eulerian grid). The setup simulates a pre-stressed, fluid-saturated granular layer representing a fault gouge, confined between impermeable walls. Fluid is injected at lateral boundaries at controlled rates (20–200 kPa/min) in discrete pressure steps to allow equilibration between pressurization and dilation, preventing transient pressure drops. Simulations are run across varying pre-stress levels ($\tau/\sigma_n$) and normalized layer lengths ($L/d$).
2. Analytical Theory: An analytical model is derived based on a pore-pressure diffusion equation that includes a "dilative sink" term. This term accounts for the volumetric expansion of the granular skeleton as it approaches failure. The model assumes a linear relationship between the dilation rate and the pressurization rate. The solution to this diffusion equation with spatially uniform boundary conditions and a sink term predicts the evolution of pore pressure profiles across the fault.

Key Results

• Rate Dependence of Failure Pressure: Numerical simulations demonstrate that the boundary pressure required to trigger fault failure ($P^b_{fail}$) systematically increases with the injection rate. This confirms that rapid injection delays failure compared to slow injection.
• Heterogeneous Pressure Distribution: The analytical solution reveals that rapid injection generates strong spatial pressure gradients. The pressure is highest near the injection boundaries and lowest at the center of the fault ($x=L/2$). Consequently, global failure is controlled not by the average pressure or the boundary pressure, but by the least-pressurized (strongest) region of the layer.
• Scaling Laws: The analytical model predicts that the failure pressure scales linearly with the injection rate ($\dot{P}_b$) and quadratically with the fault length ($L^2$), while varying inversely with hydraulic diffusivity ($\alpha$) and depending on the product of the effective bulk modulus and fluid storage capacity ($K\beta$).
• Role of Dilation: The simulations show that dilation increases with pressure, following a power-law scaling ($\Delta h/h_0 \sim (\Delta P^b/\sigma_n)^{3/2}$). The effective bulk modulus ($K$) decreases as the layer dilates, reflecting progressive weakening of the granular packing. The analytical model bounds the simulation results by treating $K$ as an effective constant, though the simulations exhibit sub-linear behavior attributed to the progressive reduction of the friction coefficient as dilation increases.
• Validation: The analytical predictions successfully bound the numerical results across three orders of magnitude of grain-packing rigidity ($K\beta$). The model reproduces experimental observations that classical uniform-pressure effective-stress theory fails to capture.

Significance and Claims
The paper claims to provide a quantitative framework linking grain-scale physics (granular dilation and rearrangement) to fault-scale failure under fluid injection. By deriving a rate-dependent failure criterion, the study explains why higher injection rates require higher pressures to trigger reactivation in gouge-filled faults.

The authors emphasize that this rate dependence arises from diffusion-controlled pore-pressure heterogeneity. In slow injection scenarios, pressure equilibrates, leading to uniform weakening. In rapid scenarios, strong gradients persist, and the fault remains stable until the central, least-pressurized region reaches the failure threshold.

Regarding practical implications, the paper suggests that operators might potentially manage reactivation risks by controlling injection protocols. While faster injection raises the failure pressure threshold, it also reduces the time to failure. The authors note that this rate–pressure trade-off could be advantageous in specific operational contexts (e.g., cyclic injection), but they explicitly caution that field applications must account for natural fault heterogeneity, complex geometries, and thermal effects, which are simplified or neglected in their 2D model. The study concludes that failure criteria derived from laboratory scales cannot be directly extrapolated to field scales without accounting for diffusion-controlled scaling effects, particularly regarding fault length.