Flash floods with high sediment concentrations are common natural events
in mountain rivers, and the vulnerability associated these events are
expected increased in the future. Numerical models are fundamental tools to design strategies aimed at reducing potential damages caused by flooding. However, their implementation is far from trivial, since multiple factors control the dynamics of the flow, especially in mountainous regions, where the morphology of the river channels and the sediment concentrations play a significant role on the flood propagation. In the Andes mountains in South America. Additional factors, such as El Ni\~no-Southern Oscillation (ENSO) effects, generate rapid floods with high concentrations of sediment on steep slopes. In this region, a large number of flash floods have been recorded with catastrophic consequences. The main objective of this investigation is to gain fundamental insights on the effects of high sediment concentrations on the propagation of floods in an Andean watershed. We develop a 2D finite-volume numerical model of the non-linear shallow water equations over natural terrains and complex geometries. We expand the efficient model developed by \citeA{Guerra2014}, to incorporate the effects of the sediment load on the dynamics of the flow, considering also the changes on the rheology to account for the additional stresses produced by the sediments. We perform simulations of flows with different sediment concentrations in the two main tributaries of the \textit{Quebrada de Ram\'on} watershed, an Andean catchment located in central Chile, where part of the city occupies the lower section of the river basin, at the foothills of the Andes mountain range. Our simulations show that the most important effects on the flood propagation are observed in the range of concentrations from $0\%$ to $20\%$. By comparing simulations with clear water and $60\%$ of sediment concentration, the maximum flow depth at different locations along the channel became $38\%$ deeper and total flooded area $76\%$ bigger. The results also show that other hydrodynamic variables, such as the arrival time to the peak flow or the shape of the hydrograph, are not significantly affected by the sediment concentration. In future research we will perform multiple simulations to focus on developing a real-time automated system of hazard prediction due to floods, generated by different meteorological conditions.