Evaluating uncertainty in debris flood modelling for the design of a steep built channel

Abstract

Hydrogeomorphic hazards are natural hazards that involve the mobilization, transport, and deposition of mixtures of water and debris or sediment and can take the form of floods, debris floods and debris flows. Such hazards occur in a continuum with varying size and concentration of entrained sediment and debris. Within this continuum, debris floods occur when large volumes of water in a creek or river entrain the gravel, cobbles, and boulders on the channel bed; also known as “full bed mobilization”. Debris floods have transient behaviour over the duration of the event with pulses of sediment laden (i.e., boulders and woody debris) flow and more diluted (i.e., water-like) flows. There is limited guidance in available literature on the hydraulic modelling of debris floods, in particular, Type 2 debris floods (i.e., diluted debris flows). There are several fluid dynamics models that could be used, and several rheologies that can be used to parametrize the flows, however, the complexity of real debris flood behaviour generally needs to be simplified to an equivalent fluid rheology in practice. Following heavy, prolonged rainfall in southwestern British Columbia, Canada, in mid-November 2021, several road and railway crossings were damaged by hydrogeomorphic hazards and erosion. These events highlight the need for the design and construction of bridge crossings able to withstand hydrogeomorphic hazards for transportation network resiliency. The modelling work described in this study was in support of the design of an armoured channel for a site that was impacted by a debris flood in November 2021. The proposed crossing is a steep and complex channel geometry, with channel slopes between 8 and 35%. Estimates of, flow depths, velocities, and shear stresses, were required for design. To capture the full effects of the steep and complex geometry of the proposed channel, debris floods were modelled in both two-and three-dimension using HEC-RAS and FLOW-3D, respectively. To provide conservative but realistic design values, multiple debris flood scenarios were modelled with the intention of capturing the range of transient behaviour expected over the duration of a debris flood and evaluate the uncertainties in the model parameterization. The debris flood model parameterization included a high and low mobility Bingham rheological parameters set (e.g. viscosity and yield stress) and modelling the flood as laminar or turbulent flow. The higher mobility turbulent flows are more representative of a flood condition that has a lower sediment concentration during the later stages of a debris flood event, while the lower mobility laminar cases are expected to be more representative of surge fronts with a higher sediment concentration. Different debris flood cases provided critical design values for different parameters. Generally, the low mobility laminar flow was the most conservative for flow depth. Modelled velocity and shear stress were not only dependent on the debris flood case, but varied within the channel sections between the two and three-dimensional results. Design values were proposed using a percentile of the amalgamated results of all debris flood cases modelled to capture the variation of the modelled cases.