How geologic stress reshapes fracture intersections in fractured rock

Models used to calculate how pollution migrates through fractured rock typically assume that the points where cracks meet are passive, well-connected junctions.

That premise sits beneath almost every simulation designed to anticipate where contaminants will travel underground.

Yet those junctions are compressed by the Earth in the same way as the surrounding rock.

Under sufficient stress, an intersection can be squeezed to the point of near-closure, and the orientation of the stress is what determines whether it remains open.

Rock cracks that cross

Subsurface fractures act like fast lanes for fluids. Compared with the intact rock around them, they are far more permeable, drawing in groundwater along with dissolved pollutants.

When two fractures converge, flows with different chemistries meet, mix, and then move onwards into the wider network.

These meeting points-known as fracture intersections-play an outsized role in controlling how fluids disperse through an interconnected web of cracks.

Peter K. Kang, a geoscientist at the University of Minnesota, and collaborators investigated how these junctions respond when loaded by stress.

Cracks and cross sections

Rock at depth is always subject to geologic stress, and it is well established that pressure can change how fluids move through single fractures.

What occurs at the points where fractures intersect, though, has largely been inferred rather than observed.

Because researchers had not directly seen a real intersection deform under load, models have generally represented junctions as unchanging hubs.

Printing rock cracks

To observe deformation directly, the researchers created fracture networks they could precisely control. Using 3D printing, they produced small, intersecting networks within resin blocks, each with deliberately roughened surfaces.

The textured walls were intended to replicate how natural fracture faces touch, slide, and meet underground, while avoiding the variability of real rock that makes identical repeat tests difficult.

Each printed sample was placed in a compact apparatus that applied a steadily increasing downward force. Between loading steps, the team collected 3D X-ray scans to look inside the block, then charted the void space through the fractures gap by gap.

This approach builds on a growing body of work that images fractures as they deform; previous studies have tracked fracture behaviour inside stressed rock.

In the new experiments, intersections were oriented in two ways relative to the applied stress: a plus (+) alignment and a tilted cross (×).

A squeeze with direction

The direction of loading proved to be decisive. In the plus-shaped configuration, increasing pressure forced the horizontal fracture towards closure, while the vertical fracture largely remained open.

In the scans, bright regions of contact spread along the compressed fracture, marking where opposing walls were pressed together.

The tilted cross produced a less intuitive response. As expected, one fracture narrowed, but the other widened; its rough surfaces appeared to slide and climb over one another, effectively levering the aperture open.

With further loading, the intersection itself began to close. As stress increased, the clean X form tightened into two V-like channels linked by a single narrow opening.

The researchers refer to this constricted connection as the intersection throat. It became smaller with every additional loading step.

No one had previously watched an intersection change shape in this manner, indicating that stress can actively rework junction geometry.

Following the flow

Geometry alone was not enough, so the team also ran fluid-flow simulations through the scanned, stress-altered intersections.

They introduced plain water into one branch and water containing a dissolved tracer into another, then quantified how much tracer reached the far outlet by crossing through the junction.

In the plus configuration, the outcome remained straightforward. As the horizontal fracture tightened, flow into the still-open vertical fracture increased by roughly 80 percent under the heaviest load.

Mixing rose in a smooth, orderly way, broadly consistent with what standard models would predict.

The tilted cross told a different story. Initially, the widening of one branch allowed more tracer to transfer across.

But as the throat then constricted, mixing fell away towards zero.

Even at very slow flow rates-conditions where models often assume streams will still blend-the narrowed throat kept the two flows largely separated.

When the models break

For the plus-shaped intersection, conventional modelling assumptions worked well. For the tilted cross, they did not.

At high pressure, standard equations continued to predict substantial mixing, whereas the simulations indicated almost none.

The discrepancy stemmed from the narrow throat. Traditional junction models allocate flow based mainly on the width of each intersecting fracture, implicitly assuming the intersection remains open.

Once the throat tightens, however, it becomes the feature that governs the split and routing of flow, undermining a core assumption that underpins much fractured-rock research.

To address this, the team introduced a correction to the standard approach: an added parameter describing how much the throat has narrowed under stress.

With that additional input, the revised equations reproduced the simulated mixing behaviour, including in the conditions where the original models performed worst.

What this changes

The key result is straightforward. Geologic stress can reshape the crossings where fractures meet, and the stress direction determines whether a junction continues to mix fluids or instead isolates them by sealing.

Models used to predict contaminant transport in fractured rock, or to estimate how injected fluids move through geothermal systems, typically omit a force that can redraw the subsurface pathways.

The implications are particularly significant for carbon storage, which aims to immobilise carbon dioxide in rock by converting it into solid minerals.

Because stress influences where-and how strongly-fluids mix, an intersection that currently promotes reaction and blending could instead constrict and cut off exchange.

Being able to anticipate where such reactions will occur, and how the Earth’s slow compression shifts them, offers engineers a form of control that has largely been missing.

Underground intersections were never as fixed as earlier models assumed.