Erosion by sliding wear in granular flows: Experiments with realistic contact forces

Abstract:

Debris flow erosion is a powerful and sometimes dominant process in steep channels. Despite its importance, this phenomenon is relatively little studied in the lab. The large drum experiments of Hsu are a notable exception, in which almost-field-scale impact forces were generated at the head of a synthetic debris flow whose properties (grain size, proportion of fines, etc) were varied widely.

A key challenge in these and similar experiments is to explore how erosion rate varies as a function of the scale of the flow (thereby varying inertial stresses, impact forces, etc). The geometrical limitations of most lab experiments, and their short run time, severely limit the scope of such explorations.

We achieve this scale exploration in a set of drum erosion experiments by varying effective gravity across several orders of magnitude (1g, 10g, 100g) in a geotechnical centrifuge. By half-filling our 40cm-diameter drum with dry 2.3mm grains, placing a synthetic rock plate at the back and a glass plate at the front 3cm apart, and rotating the drum at 1-50rpm, we simulate wear in a channelized dry granular flow. In contrast to Hsu’s experiments, we focus on sliding wear erosion at the flow boundary rather than impact/frictional wear at the flow head.

By varying effective gravity from 1g-100g we can tune the pressure exerted by the grains at the boundary without having to change the scale of our apparatus. Using a recently developed depth-averaged, kinetic-energy closure theory for granular flow, we can simultaneously tune the drum rotation rate such that the flow dynamics remain invariant. We can thereby explore how changing the scale of a granular flow, and thus the contact forces of grains on the boundary, controls the rate of rock erosion. Using a small apparatus we can simulate the erosion generated by debris flows several meters deep involving grains up to 10cm in diameter.

Our results suggest that sliding wear is the main erosion process, and are consistent with Archard's equation in which wear rate is proportional to contact force, the speed of sliding, and reciprocal hardness.

In these particular experiments grain-wall sliding appears to dominate over impact and rolling. We study further using 3D grain motion simulations executed with specially developed non-smooth contact dynamics code, allied with analysis of high-speed video.