Scientists 3D print rocks to study catastrophic avalanches

In a former military bunker near Davos, Switzerland, scientists at ETH Zurich and the WSL Institute for Snow and Avalanche Research SLF have rebuilt the Lötschental valley at a scale of 1:577 — complete with the village of Blatten, the Birch channel climbing toward the Kleines Nesthorn — and then deliberately sent a destructive torrent of water, sand and clay cascading down its slopes. The miniature catastrophe, captured on a gleaming blue 3D-printed landscape, is far more than an exercise in model making. It represents a new frontier in the study of alpine mass movements, one in which physical experiments at unprecedented fidelity are being used to calibrate and test the computer models that Swiss communities will rely upon when deciding whether their homes, roads and infrastructure lie in the path of the next rock avalanche.

The emotional and scientific context of the project is impossible to separate from the Blatten disaster of 2025, when a real rock avalanche nearly obliterated the village and buried vast areas of the valley. Johan Gaume, Professor of Alpine Mass Movements at ETH Zurich and head of the research group of the same name at SLF, and his colleagues are now recreating that event in miniature. A hatch at the top of the model opens, and a carefully composed mixture of water, sand and clay rushes down the valley from the direction of Kleines Nesthorn, spreading and settling across the printed terrain in an eerie echo of what happened at full scale. “We want to use models like this to find out how mixtures of different materials move in the natural environment,” Gaume explains. The simulation of the Blatten landslide — with different colours symbolising different speeds — allows researchers to compare a forecast calculation performed before the actual event with a recalculation based on the data known afterwards, a comparison that exposes exactly where predictive models succeed and where they falter.

Physical modelling of this ambition has rarely been attempted. The Blatten model is the first of its size, and its construction reveals the extraordinary labour behind seemingly simple experiments. The researchers used specially programmed software to select the landform, break it down into small sections, and print the terrain segments piece by piece — each measuring approximately 50 by 50 centimetres. Fifty-six printed parts, requiring roughly 100 days of printing time, were then assembled, coated and painted inside the former military bunker that now belongs to the SLF. “This is a vital step to achieve the desired roughness and optical properties,” Gaume clarifies. Surface roughness is not a cosmetic detail: the friction between a flowing mass and the terrain over which it travels is one of the dominant controls on runout distance, and any discrepancy between the real mountain and the laboratory replica would propagate directly into the models being tested.

The resulting installation is imposing by laboratory standards. The flow channel includes an inclined section of 4.5 by 2 metres and a horizontal runout zone up to 6 metres long. The printed topographic model measures up to 5.4 by 4.5 metres, covering approximately 12 square metres of surface area. The system can accept a maximum input mass of one tonne, though current experiments use an input volume of 48 litres. Every experiment is monitored by an array of sophisticated instrumentation — numerous cameras, lasers and sensors, as well as a 3D scanner — which together measure parameters such as flow depth, speed, runout distance, deposition patterns and impact dynamics. A force plate and pore pressure sensors record the mechanical interaction between the moving mass and its bed, providing quantitative ground truth that no field observation of a natural disaster could ever hope to capture.

The material itself is as carefully engineered as the terrain. Before each run, the researchers precisely calculate the composition and the required proportions of the individual components, then blend the mixture in a bucket. The experimental slurries — water, sand and clay in the current campaign, with ice mixtures planned as the work develops — are tipped into a box at the top of the model before the hatch opens and the event is triggered. This control over composition is critical, because real rock avalanches are rarely simple. In mountainous terrain, they entrain water, melt glacial ice, pick up saturated sediment, and evolve dynamically from dry granular flows into dense, muddy slurries with completely different rheological behaviour. A model that can accurately simulate such mixtures on a laboratory landscape can be trusted far more when it is scaled up to forecast genuine hazards.

Alongside the physical model, Gaume’s group develops virtual models on computers that simulate scenarios of what can happen during mass movements in the Alps, from debris flows to rock avalanches. The aim is that in the future, officials responsible for natural hazards at cantonal and municipal levels will be able to use these tools to determine whether infrastructure or even entire settlements are at risk. But the reliability of any computational model ultimately rests on the quality of the data used to validate it. “For these models to be reliable, they must be thoroughly tested in carefully controlled experiments,” Gaume says. These are precisely the experiments now running on the 3D-printed Blatten landscape, where every variable — slope angle, material composition, initial volume, release height — can be held constant or systematically varied in a way that nature never allows.

The scientific payoff extends well beyond validation. Each experimental run yields key insights into how mixtures of water, ice and other solid materials move, what happens when they strike an obstacle such as a bridge pier or an avalanche barrier, and how far a mass travels after encountering rising ground. Impact dynamics in particular remain poorly constrained in existing hazard maps, because direct measurements at real disaster sites are almost always destroyed by the events they are meant to record. In the bunker laboratory, a force plate survives every impact, delivering data that will directly inform the design of protective structures in vulnerable valleys across the Alps.

Gaume emphasises that the facility is designed for endurance rather than a single campaign. “This is a long-term investment in research, not a short-term experiment,” he explains. In the future, researchers will be able to mount other areas and landforms onto the substructure, transforming the laboratory from a single-landscape facility into a modular platform for hazard research. Numerous projects are already planned to investigate various effects, from erosion along the flow path to impact against obstacles, and to the distance the masses travel up counter slopes — the run-up behaviour that often determines whether a debris flow overtops a retaining embankment and reaches inhabited ground. The modularity of the printed terrain means that topographies of other endangered Swiss valleys could, in principle, be reproduced and subjected to the same battery of controlled tests.

At present, one physical question occupies Gaume’s attention above all others: terrain curvature and its influence on flow dynamics. Curvature — whether the land is concave, forming a natural channel that focuses and accelerates the flow, or convex, spreading and decelerating it — is a first-order control on where avalanches and debris flows deposit their material. Because the 3D-printed landscape reproduces the real curvature of the Birch basin, experiments can isolate its effect with surgical precision, adjusting the path while holding material properties constant. The resulting datasets will feed directly into the numerical codes used for hazard assessment, sharpening the forecasts that determine zoning decisions and evacuation planning in mountain communities.

The broader significance of the work lies in its timing. Climate change is intensifying permafrost degradation, glacier retreat and slope destabilisation across the high Alps, increasing both the frequency and the magnitude of rock avalanches and related mass movements. The Blatten event demonstrated how quickly an entire village can be lost, and how urgently accurate, physically grounded forecasting tools are needed. By marrying 3D printing technology, precision instrumentation and computational simulation, the ETH Zurich and SLF team has created a facility in which disaster can be rehearsed, measured and understood — without cost to a single human life. Each opening of the hatch at the top of the miniature valley brings researchers one step closer to answering the question that mountain communities everywhere are asking: where, and how far, will the next one go?

News Publication Date: 8-Sep-2026

Web References: Not provided

References: 3D-printed rock avalanche. EurekAlert! Earth Science. https://www.eurekalert.org