|dc.description.abstract||The occurrence of debris flows in Norway is a challenge to settlement and infrastructure. Due to a high mobility, that causes long run-outs, they can lead to severe damages. This constitute a large cost to the Norwegian Water and Energy Directorate (NWED), the Norwegian Public Roads Administration (NPRA) and the Norwegian National Rail Administration (NNRA). A predicted increase in precipitation during this century will most likely increase the frequency of debris flows. Thus, the need for accurate run-out predictions will increase. Simpler empirical methods have been used by operators, but there is little knowledge of applying numerical methods to Norwegian debris flows.
The aim of this research is to model Norwegian debris flows by numerical and empirical models, in order to determine the best prediction tool/tools to deal with Norwegian debris flows. RAMMS, DAN-3D and FLO-2D are the assessed numerical models, whereas the NGI-model, the Energy line model and UBCDflow are the considered empirical run-out prediction tools. Five debris flows were selected and described for this purpose. Each case was back-analysed by the numerical models, through calibration of the volume. The sensitivity related to the Voellmy coefficients is considered for RAMMS and DAN-3D, by a simple analysis on two debris flows. The µ coefficient of the depositional area associated with each case is back-calculated. The run-out distances of the debris flows were predicted with the empirical models, and compared to the numerical model results.
The numerical models required different release volumes to back-analyse sufficient run-outs of the cases. Changes in the volume did affect the models outputs differently. RAMMS and FLO-2D seem to be most sensitive to volume changes, although the effect on the run-out distance was rather small. RAMMS and FLO-2D simulations showed more spreading in the depositional area. This is possibly due to the characteristics of their development site. DAN-3D did in general model larger velocities, and is more applicable to less channelized debris flows than RAMMS and FLO-2D. RAMMS modelled suitable velocities, whereas FLO-2D modelled some below the range. RAMMS is considered as the most user-friendly model.
No major differences were identified between RAMMS and DAN-3D in the sensitivity analyses. The run-out distance is most sensitive to changes in the Voellmy friction coefficient, µ, while the modelled maximum velocity is most sensitive to the Voellmy turbulent coefficient, ζ. The sensitivity was slightly affected by the applied release volume. Back-calculated µ coefficients were sensitive to the slope angle of the lower flow path.
The results showed that the numerical models provided just as reasonable results as the empirical models. However, based on this research it may seem like none of the models by itself is a sufficient tool. A possible solution to debris flow hazard prediction in Norway might be a combination of two methods. The results suggest that RAMMS and the NGI-model perhaps are an applicable combination. Nevertheless, DAN-3D might, in some cases, also be a commendable tool.||