⚡ electrical engineering

Full-scale experiments reveal the spatiotemporal controls on flood-induced erosion of road foundations

Full-scale flume experiments demonstrate that flow velocity, rather than water depth, is the primary driver of flood-induced erosion in road foundations, revealing a critical mechanism that necessitates the development of velocity-explicit predictive models to better protect road infrastructure.

Original authors: Ringo Rocha Reboucas, Josef Keller, Jürgen Stamm, Jürgen Hackl

Published 2026-09-23

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ringo Rocha Reboucas, Josef Keller, Jürgen Stamm, Jürgen Hackl

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

When heavy rains swell rivers and flash floods surge through towns, the water does not just soak the ground; it carries immense force that can tear apart the very ground beneath our feet. For decades, engineers and planners have assessed the danger of these floods primarily by measuring how deep the water gets. The logic seemed sound: deeper water means more weight and more pressure, so deeper water must mean more damage. This approach has guided how we build bridges, reinforce embankments, and plan evacuations. However, this focus on depth overlooks a critical, invisible force: the speed at which the water moves. While deep water might push against a house, fast-moving water acts like a sandblaster, scouring away the soil and gravel that hold roads together. As climate change brings more frequent and intense storms, understanding exactly how water speed destroys road foundations has become a matter of public safety and economic survival.

A team of researchers at the Technical University of Dresden and Princeton University decided to look beyond the depth of the water to see what the speed of the current actually does to a road. They built a massive, full-scale experiment to watch a road foundation being eaten away by floodwaters in real time. Instead of building a tiny model in a bathtub, which often fails to capture the true behavior of large rocks and soil, they constructed a five-meter-long section of a real road foundation inside a large laboratory channel. This foundation was made of the same crushed stone and gravel used in actual German road construction, compacted to the exact density and shape of a real highway shoulder. They then subjected this road section to controlled flows of water that moved parallel to the road, mimicking the conditions of a flood that runs alongside a highway rather than washing over it.

The experiment revealed that the speed of the water is the true driver of destruction, not just the depth. The researchers found that water depth acts like a ruler, determining exactly how high up the side of the road the water can reach and where the attack begins. But once the water touches the road, it is the velocity, or speed, of the flow that decides how much damage occurs. In their tests, they ran water at speeds ranging from 0.8 meters per second to 1.2 meters per second. These are speeds that many existing safety models consider safe, often assuming that such moderate flows would not cause significant harm. The researchers watched as the water, moving at these "safe" speeds, began to carve deep trenches into the side of the road foundation.

The damage happened quickly and in a very specific pattern. The erosion concentrated just below the surface of the water, where the flow moves fastest. As the water speed increased from 0.8 meters per second to 1.2 meters per second, the damage became more severe and more focused. In just six hours, the faster water carved a hole nine centimeters deep into the foundation, removing a significant amount of material. The erosion did not happen evenly; it started at the top of the exposed slope and worked its way down, undercutting the road surface. This is a dangerous process because it removes the support holding the road up, potentially leading to a collapse even if the water never rises high enough to cover the road itself.

Interestingly, the researchers also observed that the water did not simply wash the dirt away and leave a clean surface. In some areas, the eroded material settled back down at the bottom of the slope, creating a pile of sediment. However, this deposition was chaotic and did not follow a predictable pattern based on the water speed. The amount of material lost to erosion was directly linked to how fast the water was moving, but the way the dirt piled back up was inconsistent. This distinction is crucial for engineers: they can predict how fast a road will be destroyed by knowing the water speed, but they cannot rely on the water to naturally refill the gaps or stabilize the slope.

The study challenges the long-held belief that water depth is the only number that matters when assessing flood risk to roads. The researchers showed that a road can be severely damaged by water that is not very deep, provided the water is moving fast enough. In their experiments, the most dangerous zone was the area just below the water's surface, where the speed of the current is highest. This means that even if a flood does not overtop a road, the water running alongside it can strip away the foundation, leaving the pavement vulnerable to the weight of traffic. The damage was most severe at the edges of the road where it meets rigid structures, suggesting that these transition points are the weakest links in the system.

These findings suggest that the way we protect our roads needs to change. Current safety models often assume that moderate water speeds are harmless, but this experiment proves that they can cause rapid and extensive erosion. The researchers argue that future safety plans must explicitly account for the speed of the water, not just its depth. They also highlight that erosion happens very quickly, often within the first few hours of a flood event, leaving little time for emergency repairs. This means that protection must be proactive, built into the design of the road from the start, rather than reactive. By understanding that speed is the primary weapon of floodwaters, engineers can design foundations that are more resistant to the specific forces that tear them apart, ensuring that our critical transport networks remain standing when the next flood arrives.

Technical Summary: Full-scale experiments reveal the spatiotemporal controls on flood-induced erosion of road foundations

Problem Statement
Floods pose a critical threat to road infrastructure worldwide, often undermining the granular foundations that support pavement. While field evidence suggests that flow velocity is a primary driver of damage, current predictive models and risk assessments predominantly rely on water depth as the sole hazard metric. This approach overlooks the specific mechanism of external erosion—the rapid removal of granular support material—which leads to structural instability. Existing assessments often treat flow velocity coarsely (e.g., as binary variables) or assume that moderate velocities pose negligible risk. Furthermore, current understanding relies heavily on theoretical analyses or analogies with riverbank behavior, which differ fundamentally from engineered road foundations in terms of geometry, material compaction, and boundary conditions. There is a lack of field-scale experimental data documenting how road foundations erode under varying hydrodynamic forcing, leaving damage models without a process-based foundation.

Methodology
To address these gaps, the authors conducted full-scale experiments in a high-performance morphological flume at the TUD Dresden University of Technology. The study utilized a 5-meter-long granular road foundation model constructed to standard specifications, featuring a trapezoidal cross-section with a 1:1.5 slope and a 0.6-meter-wide crest. The material consisted of Paleozoic diabase aggregate (0–32 mm), compacted to a dynamic deformation modulus of $E_{vd} = 69.75 \text{ MN m}^{-2}$ , representing a typical unbound base layer.

The experiments subjected the foundation to sustained, quasi-steady, fully turbulent, subcritical longitudinal flows. Two primary hydrodynamic scenarios were tested, generating streamwise mean velocities ranging from approximately 0.82 to 1.17 m s $^{-1}$ and water depths between 0.28 and 0.33 m. These conditions were maintained for durations up to six hours.

Instrumentation included:

Hydrodynamics: A high-resolution acoustic Doppler velocimeter (ADV) measured three-component flow velocities to derive mean velocity, water depth, and near-foundation turbulent kinetic energy (TKE).
Morphology: A scanning robotic total station (Nova MultiStation MS60) surveyed the foundation profile at multiple intervals (30 min, 60 min, and 90 min) to quantify surface elevation changes.
Analysis: Morphological changes were quantified as erosion/deposition volumes and maximum depths using Multiscale Model-to-Model Cloud Comparison (M3C2). Statistical relationships between hydrodynamic drivers and morphological responses were analyzed using LOESS smoothing, robust linear regression, and logarithmic temporal modeling.

Key Results
The experiments revealed distinct spatiotemporal controls on erosion that challenge current hazard representations:

Velocity vs. Depth Roles: A fundamental distinction was identified between flow velocity and water depth. Water depth determines the elevation at which the foundation is exposed to erosion attack (specifically, the zone just below the water surface where velocity peaks), whereas flow velocity governs the intensity and rate of erosion once exposure occurs.
Velocity-Dependent Erosion: Erosion severity was strongly controlled by flow velocity. In the tested range (0.8–1.2 m s $^{-1$ ), higher velocities caused erosion to concentrate in narrower bands and intensified the rate of material loss. Maximum erosion depths reached 9.1 cm within six hours at the higher velocity class (1.02–1.17 m s $^{-1}$ ), compared to 5.5 cm at the lower class.
Temporal Evolution: Erosion volume and depth increased logarithmically over time, indicating rapid initial adjustment followed by progressive attenuation. However, higher velocities amplified the cumulative rate of material loss.
Deposition Patterns: In contrast to erosion, deposition showed no systematic dependence on hydraulic forcing (velocity or depth). Deposition remained confined to a narrow band at the toe of the foundation and exhibited high variability without a clear functional relationship to the drivers.
Spatial Heterogeneity: Erosion was most severe along the upper to middle exposed face near the waterline. The longitudinal edges (shoulders) showed heterogeneous erosion patterns, with localized scour zones developing near boundaries, particularly where water levels remained close to the shoulder.

Significance and Claims
The authors claim that these findings provide the first process-based, full-scale evidence linking specific hydrodynamic drivers to the erosion of granular road foundations, free from scaling effects. The study challenges the prevailing practice of assessing flood damage to roads based solely on inundation depth. The results demonstrate that significant structural damage can occur at velocities previously considered safe (around 1 m s $^{-1}$ ), suggesting that current fragility models systematically underestimate the risk to road infrastructure.

The paper argues for a shift in hazard representation:

Explicit Velocity Modeling: Damage models must explicitly resolve flow velocity in finer classes rather than relying on depth alone or binary velocity indicators.
Mechanistic Failure Analysis: Understanding that velocity controls intensity while depth controls location allows for more precise failure analysis, particularly for shoulder erosion which can compromise structural integrity even without overtopping.
Operational Implications: The rapid onset and intensification of erosion imply that reactive interventions during high-magnitude events are often too late; proactive measures and monitoring of erosion depth as a proxy for sediment loss are essential.

The study concludes that the threshold for progressive erosion in road foundations can be crossed at velocities of approximately 1 m s $^{-1}$ , a condition that existing damage models leave unresolved. This work provides a mechanistic basis for improving large-scale models of flood-induced road damage and developing more effective protection strategies against intensifying hydrological extremes.

More like this