The present work addresses the influence of internal drainage systems on failure by overtopping of earth dams. It is based on data collected in six homogeneous dam breach laboratory experiments. Different drainage systems were tested: (a) toe drain only and (b) complete internal drainage system—chimney filter, horizontal blanket and toe drain. The tests were performed with the same embankment body material, reservoir dimensions and inflow conditions but different relative compactions. We analyze the breach hydrograph and the stages of dam breaching. The failure of dams with complete and fully functional drainage systems exhibits marked differences, the most conspicuous of all being the longer duration of the initial erosion stage. This is accompanied by a lower increase rate of its breach discharge. We argue that this is the combined result of the upstream migration of the locus of hydraulic control and of a higher resistance to erosion of the downstream shell, kept unsaturated by the drainage system. Hydraulic control moves upstream, to a less morphologically active cross‐section, as a result of channel deepening once the chimney drain is uncovered and eroded. Once the erosion merges with the cavity where the chimney filter was, marking the end of the initial erosion stage, the rate of erosion increases and becomes comparable to that of dams with toe drain only. Our argument about the longer duration of the initial stage is not applicable to latter stages—the total time before full collapse is not necessarily increased.
The dataset provided in this paper refers to an experimental campaign conducted in Laboratory of Fluid Dynamics (LTDF) of the Free University of Bozen-Bolzano at NOI Techpark aiming to understand the movement of granular material in fluids of low viscosity and density exhibited in debris flows.
A granular system is a collection of macroscopic particles that interacts through dissipative collisions and enduring contacts. It can exhibit gas, liquid or solid behaviour. These systems present phase transitions and coexistence of different phases. As for solid-liquid transitions, there is vast literature in thermal and athermal systems but no universal models of first-order or second-order phase transitions.In particular, dry granular flows (the movement of granular material in fluids of low density and viscosity) can serve as models of debris flows. Mechanical clogging occurs when the mass of granular material is stopped in from of slits or orifices in check dams. There is currently not enough knowledge on the processes that lead to clogging. In this research we conducted a series of 31 laboratory experiments of dry granular flows constricted through a vertical gap, adjacent to the side wall, mimicking slit dam conditions. The granular material was composed of monosized polystyrene particles (Ø1.8 mm). The width of the slit was 2 particle diameters. The granular mass was released suddenly in a 1.5 m long chute, tilted at 20°. Instrumentation included two high-speed cameras (300 fps), located upstream, at the gate location, and downstream, at the slit location. Instantaneous velocities were obtained with PTV at the chute wall. In this work we discuss the behaviour or mean longitudinal velocities and of granular temperatures when the clogging occurs. The start of the clogging process was identified as the ts – solidification instant, this instant is defined by the moment the first particles stop moving.It is shown that the statistical distribution of ts is probably not heavy-tailed. It has a positive asymmetry [0.410] and low flatness [-1.369]. Analysing 0.133 s before and after the solidification instant, it is shown that the mean velocity and the granular temperature of the granular system is constant up to 0.033 s before ts while the solid volume increases. It is not clear which portions of the system are in a gas phase and which are in a liquid phase.The dissipative nature of the system becomes apparent from ts – 0.033 s. It is postulated that the rate of collisions has substantially increased with the increase of the solid fraction. It is expected that the rate of dissipation of fluctuating energy is a non-linear increasing function of the volume fraction. Hence, from ts – 0.033 onwards, a decrease in granular temperature (granular cooling) becomes evident. A reduction of the mean velocity becomes apparent at the same instant. The decrease of the fluctuating kinetic energy is continuous across the phase transition but appears stronger after ts. As a result of this work we will explore the hypothesis that the liquid-solid phase transition, observed in terms of mean velocities and granular temperatures is best modelled as smooth transition.This work was funded by Portuguese Foundation for Science and Technology (FCT) through Project PTDC / ECI-EGS / 29835/2017 - POCI-01-0145-FEDER-029835, financed by FEDER funds through COMPETE2020, by National funds through FCT, IP. and partially funded by FEDER Project by the FCT Project RECI/ECM-HID/0371/2012.
This research aims to experimentally characterize cooling, clogging and jamming of a dry granular flow in a chute partially obstructed by a stopping wall with two slits adjacent to the side walls. We ensemble‐average velocities, determined with Particle Tracking Velocimetry, and its fluctuations, to compute mean flow and granular temperature fields. Full chute‐wide jamming is triggered by the formation of stable arch‐like clogging structures in front of the slits. The statistical distribution of the clogging instant is not heavy‐tailed, which indicates that clogging occurs only when the flow through the slits is liquid‐like. An upstream‐progressing jamming wave eventually forms, similar to that observed in fully obstructed chutes. There is no triple point anywhere, since the flow cools down to a granular liquid before jamming. We identified three main stages of jamming wave propagation. The initial buildup is characterized by high values of the upstream Froude number, slow progression, and transformation of kinetic into potential energy. This occurs with significant granular head losses, as particles attempt to flow over the jam. In the second stage, accretion becomes dominant, characterized by smaller head losses and, consequently, higher values of the jamming wave celerity. In the third stage, the jamming wave propagates against a cooler but faster flow that pushes against the jam, slightly increasing the wave strength. Accretion is the main mechanism of jammed mass increase, justifying a further increase of wave acceleration. The macroscopic aspects of the jamming wave dynamics can be described, as a first approximation, by shallow‐water theory.