Assessing the use of small scale experimental flume models to study debris flow dynamics

Debris flows are powerful, geophysical flows which cause fatalities and damage infrastructure. Understanding these hillslope processes is therefore extremely important but, due to their spontaneous nature and rapid onset, opportunities to directly observe debris flows are rare. In recent decades major advances in understanding debris flow dynamics have been achieved using large scale hillslope debris flumes. Although such models are excellent for understanding the rheology of the bed and flow they cannot be easily configured to investigate macroscale factors such as channel slope and bend curvature. This paper assesses the use of smaller scale models to study debris flow dynamics where the channel configuration (width, depth and curvature) can be readily altered. To achieve this level of flexibility the scale of the model is necessarily reduced so that experiments can be rapidly reconfigured, for varying slope and channel geometries, and the volumes of sediment used in repeat simulations can be readily reproduced. The experiments described here were conducted in a purpose-built, variable slope, ~9 m long flume model. Experiments were carried out at varying slopes: 17° and 25° for both straight and curved channels. Experiments were designed to test the application of the forced vortex equation for superelevation of debris flows in a 0.2 m wide channel. Four channel bends with differing geometries were tested; three radii of curvature - 40, 55 and 70 cm (for a bend angle of 40°) and one radius of curvature of 70 cm for a 20° bend angle. For all runs the grain-size of the debris flow mix consisted of a poorly sorted gravel/sand/clay mix. A combination of direct and video-based observations were used to record debris flow behaviour including variation in superelevation and the average and local velocity of the flow. Straight channel results demonstrate a strong linear relationship between debris flow velocity and channel gradient. In curved channels the magnitude of superelevation was strongly linked to debris flow velocity, radius of curvature and bend angle. Superelevation was shown to increase linearly with debris flow velocity, and was greatest for tighter bend geometries. When compared to observed superelevation, predicted values calculated using the forced vortex equation can substantially overestimate by up to 210% in some circumstances, suggesting that predicted values and back-calculated velocity estimates should be regarded as maxima. Furthermore, the correction factor ';k' in the forced vortex equation calculated from our experimental data suggests that ';k' was not constant but varied two-fold and had a value less than 1.0; contrary to previous observations. However, scaling parameters for geometrically similar debris flows of varying size suggest dynamic similarity may not be consistent over a range of scales. Therefore additional experiments were conducted for the same channel geometry (bend angle) but at three differing scales; and the corresponding behaviour assessed to determine whether debris flow dynamics were scale dependent.