Laboratory-scale dam-break study of gravity currents with basal entrainment: PIV measurements of a viscous newtonian fluid over a horizontal bed of the same ﬂuid.

Geophysical gravity flows such as avalanches and debris flows belong to a special class of hazardous environmental event, in which a mixture of solids and fluids (e.g. debris and mud, snow and air) flow as a liquid and may run out much further than expected over a slope less steep than the critical angle of repose. Exchange of material between the overriding flow and a loose bed layer underneath is known to affect the characteristics of such a flow however it is not well understood how this plays a part, nor the mechanism of such an exchange, due to the difficulties of seeing inside a flowing mass in the field and experiments. Research has been carried out on an entraining viscous gravity current in an laboratory flume using Particle Image Velocimetry. The aim was to investigate the flow properties in a vertical slice of fluid in the downstream direction, far from the side-walls, when a flowing layer overrides a basal layer of the same material. The parameter varied here is the volume of fluid released from the reservoir, which affects the height and velocity of the encroaching fluid current. An interface was identified between the overriding fluid and the initially stationary bed fluid. PIV was used to show the evolution of the velocity and shear fields throughout the system compared to the location of this interface and how this changes with the volume of fluid released. The overriding fluid displaces the bed fluid, either by plunging or spilling into the erodible layer depending on the initial volume, thus inciting the stationary material downstream and below to move. Initially, a large amount of bed material is suddenly mobilized by the front of overriding fluid and little deposition occurs. Shortly after this we see the clear development of a depositional region (with low velocities, predominantly in the downwards direction) and an eroding region (with high, almost uniformly horizontal velocities) within the front of material that enters the bed, separated by a region of high shear. For larger released volumes, multiple shear bands develop, thus showing distinct regions with different rates of material deposition.

Effect of available entrainable material on a viscous gravity current including run-out characteristics and internal flow properties.

Christophe Ancey, Belinda Margaret Bates

It has long been accepted that entrainment of loose material by geophysical gravity flows such as dense snow avalanches and debris flows may change their behaviour significantly. Run-out distances and bulk-flow velocities are notable examples of susceptible behaviours. It is still disputed how this has an effect but it has been noted that the availability of entrainable material is a principal parameter. Laboratory and numerical results are studied side-by-side to demonstrate the effects of a finite erodible bed of varying length and depth, which is placed in the path of a flowing gravity current. Both the current and the bed are composed of the same material. Natural geophysical flows are simulated as idealized viscous gravity currents at zero degrees inclination in order to study the link between the internal dynamics and the bulk features in the simplest case. In the laboratory, a PIV configuration using a laser sheet allows the visualization of a vertical stream-wise cross section of the flow in the transition region from rigid to erodible bed, far from the side-walls. This allows the study of the velocity field within the cross-section of the flow in the entrainable region. Run-out speeds and distances are measured after the current exits the erodible bed and flows over a rigid base once more. A relationship is sought between the released volume, the erodible bed dimensions (that is, length and depth) and the run-out characteristics of the flow. This bulk run-out behaviour is investigated with reference to the internal flow dynamics as measured by PIV. This work is supplemented by results obtained modelling the same system using the open source CFD software OpenFOAM. We were able to track the front of the current during the flow and found that even the presence of a shallow entrainable bed (3 mm deep) significantly advanced the run-out front compared to the no-bed case. A further increase in bed depth led to a slight increase in run-out. The length of the bed was also a strongly controlling parameter, highlighting the change in flow characteristics across this region, and underlining the necessity to investigate the internal dynamics when a gravity current flows over an entrainable bed.

2013

Dam Break of Newtonian Fluids and Granular Suspensions

Nicolas Andreini

The objective of this thesis was to increase our understanding of two-phase geophysical flows (e.g. debris flows) by providing velocity profiles in idealized laboratory avalanches. To that end, we developed a new experimental platform made up of an inclined flume coupled to an imaging system to measure velocity in granular suspension. The inclined flume was 3.5 m long and 10 cm wide and could be inclined from 0 to 35°. A reservoir with the capacity for 10 l of fluid was located in the upper part of the flume and closed with a pneumatic controlled gate. Velocity profiles were obtained using Particle Image Velocimetry (PIV) and index-of-refraction matching of the solid and liquid phases. We used transparent PMMA beads with mean diameters of 200 µm and the interstitial fluid was composed of a mixture of three fluids. The interstitial fluid was adapted in order to match the refraction index and the density of the solid phase. Using pulsed laser and a high speed camera we were able to measure velocity profiles at frequencies up to 1000 Hz with very good precision. Two additional cameras tracked the front position along the flume with a frequency of 30 Hz and a spatial resolution of 1 mm. Prior to acquiring data on the granular suspension, we tested our system on Newtonian fluids. Eight flow configurations were selected with different fluids (glycerol and triton X100), different slopes and different released masses. Velocity profiles were found to be parabolic far from the front as well as very close to the contact line. However, near the front, quantitative theoretical predictions given by lubrication theory diverged from experimental results. Velocities were significantly overestimated (∼ 400%) by the theory at low Reynolds numbers (Re < 2) and slightly underestimated (∼ 10%) at high Reynolds numbers (Re > 8). Very good agreement with theory far from the front indicated that the accuracy of the setup was good (reliable calibration procedure and image processing methods). Experiments on granular suspensions revealed a variety of behaviors depending on the particle concentration, the slope and the mass released. At solid fractions up to 45%, suspensions behaved as homogeneous viscous fluids. For the duration of the experiment, it was not possible to detect any inhomogeneity due to migration or sedimentation. In the range of shear rate tested and with the precision allowed by the setup no shear thickening or shear thinning was observed since velocity profiles remained perfectly Newtonian. For slightly more concentrated suspensions (up to 55%), we found that the flow dynamics at the bulk scale could still be described using a viscous theory. However, at the local scale, migration gave rise to concentration inhomogeneities producing a blunted velocity profile. The shape of the blunted profile was well described by the Mills and Snabre migration model coupled to a Krieger-Dougherty effective viscosity. However, magnitudes of the velocities were largely overestimated, most probably because we fitted the effective viscosity at higher shear rates. Above 55%, small released masses with high solid fractions stopped after a finite time and separation between fluid and solid phases occurred. The solid frame stayed at rest while the fluid seeped through the granular media eroding the front. For larger released masses, we observed successions of different regimes: After an inertial regime and a pseudo-viscous regime, the flow slowed down, corresponding to a new regime in which the shearing was localized in a thin layer at bottom and there was no shearing of the front. At the same time, we observed that the free surface deformed and became wavy. Fractures developed on the top of the flow and, if they grew sufficiently, modified the local velocity field substantially. Finally, at longer time (≥ 4 min) an intermittent motion (stick-slip) was observed with phases during which the suspension was flowing in a quasi-steady regime and phases during which the suspension was at a halt.

EPFL2012

Viscoplastic dambreak waves: review of simple computational approaches and comparison with experiments

Christophe Ancey, Nicolas Andreini, Gaël Epely Chauvin

We investigated the dam-break problem for Herschel-Bulkley fluids: a fixed volume of a viscoplastic material (a polymeric gel called Carbopol ultrez 10) was released and flowed down an inclined flume. Using Particle Image Velocimetry techniques, we measured the velocity profiles far from the sidewalls, the front position as a function of time, and the flow depth evolution at a given place. The experimental data were compared to three models of increasing complexity: the kinematic wave model, an advection diffusion model (lubrication theory), and the one-layer Saint-Venant equations. Surprisingly, the best agreement was obtained with the simplest model (kinematic wave model) even though it could not capture the details of the head profile (regarded as a shock wave, i.e., a discontinuity). Lubrication theory (the advection diffusion model) performed well from a qualitative viewpoint. Computed velocity profiles and depth evolution were in reasonably good agreement with data, but this model overestimated initial acceleration, which resulted in a systematic difference between theoretical and experimental curves of the front position over time. This shortcoming was not fixed when using a more elaborate model (Saint-Venant equations), rather it was exacerbated. The relatively modest performance of the more elaborate models was intriguing (for Newtonian liquids, the best agreement was obtained with the most sophisticated model).