Previsão de alcances de rupturas de taludes de Belo Horizonte por meio de retroanálises: uma contribuição para a gestão do risco

Abstract

The objective of this research is to evaluate the applicability of the dynamic analysis methodology to predict the runout of slope failures in Belo Horizonte and to obtain rheological parameters of the studied materials. Five retro-analysis of well documented slope failures were carried out, using the model DAN-W (Dynamic Analysis of Landslides), release 10, developed by Hungr (1995). This computational tool allows the dynamic analysis of landslides using eight different rheological models and with different parameters. The frictional model uses friction angle at the sliding surface and the pore-pressure coefficient ru as rheological parameters. To the Newtonian fluid model, the rheological parameter is the dynamic viscosity. The values of the back-analyzed parameters were defined as the intersection between the observed field values of the sliding mass reaches and those simulated using the models. The compatibility of each model to each slope was evaluated considering the following aspects: if the behavior simulated reflects what was observed in the field; if the simulated curves show logical and defined behavior, and if the results of the retro-analysis are consistent with results obtained in tests carried out or in the bibliography. The selection of the slopes favored those with a representative geology of the area of Belo Horizonte, with well known geometry before and after the failure, and with characterization and strength tests information. The slopes chosen were denominated Ponteio Slope, Planetoides Slope, Taquaril Slope, Engenho Nogueira Slope and Juliana Slope. The rheological parameters were compared with the literature. It was observed that for all slopes back-analyzed, the frictional model showed better results and the Newtonian fluid model did not converge. The rheological parameters obtained in the back-analysis of Ponteio Slope were also used to simulate another failure, occurred in 2005, after a stabilization solution, using bioengineering techniques associated with soil nailing, was implemented in 2003. The purpose of this simulation was to verify the contribution of the stabilization solution in decreasing the extent of debris flow. It was observed that the stabilization solution could not prevent the slope failure, but it decreased the extent of failure to approximately 1/5 of the reach estimated with DAN-W program, assuming no stabilization. The methodology to predict slope failures reaches using dynamic analysis proved applicable to Belo Horizonte slopes, since a backanalysis of a similar case is previously performed and the compatibility criteria of the rheological model are achieved.