Laboratoire d’Hydraulique Environnementale, École Polytechnique Fédérale, Lausanne (EPFL). Avalanches pose a serious threat throughout the mountainous regions of Europe. In the French Alps alone there were 57 avalanche fatalities in 2006, more than twice the annual average over the past decade. In the past, researchers have studied avalanches using mathematical models based on data gathered from sensors placed on mountain slopes. While the sensors provide some insight into flow path, snow density, and impact pressure, the data is often unreliable and incomplete. Researchers Steve Cochard and Christophe Ancey take a different approach: they analyze avalanche behavior in a controlled laboratory setting.
Cochard and Ancey created a unique experimental setup consisting of a metallic frame supporting a reservoir, an inclined aluminum plane, and a horizontal run-out zone (Figure 1). At 6 meters long, 1.8 meters wide, and 3.5 meters high, the structure is the largest laboratory setup of its kind in the world.
In a dam-break experiment, about 100 liters of viscous fluid are released from the reservoir down the 4.5-meter-long plane. The researchers precisely control initial conditions, such as the fluid volume, density, and rheological characteristics, and boundary conditions, such as plane angle and surface roughness.
Within this controlled environment, the researchers measure the free-surface (or interface) profile and the spreading rate of the fluid.
To measure the free-surface profile, Cochard and Ancey developed a novel imaging system consisting of a high-speed digital camera coupled to a synchronized micro-mirror projector. The camera records how regular patterns projected onto the surface are deformed when the free surface moves. Using MATLAB and Image Processing Toolbox, the team developed algorithms to post-process the image data, determine the spreading rate, and generate whole-field 3-D shape measurements of the free-surface profile. They calculate the phase of the projected pattern, unwrap the phase, and then apply a calibration matrix to extract the flow thickness from the unwrapped phase.
Post-processing data from a typical two-minute experiment required almost a week of computational time on a Mac Pro Quad Core. To accelerate computation time, the researchers used Distributed Computing Toolbox to run the post-processing on a cluster of two Mac Pro computers with eight processor cores. This parallel setup reduced processing time to one day, enabling the team to conduct more experiments and apply more complex processing algorithms.
Today, Cochard, Ancey, and their colleagues are able to measure the free-surface evolution and the spreading rate of a gravity-driven flow—a major achievement. They are confident that the results are scalable to actual geophysical flows, and are currently developing a 3-D model to simulate fluid avalanches.
Published 2007 - 91472v00