Análises determinísticas e probabilísticas de estabilidade em aterros sobre solos de baixa consistência

Abstract

The expansion of urban centers to hitherto uninhabitable areas, which sometimes concentrate regions with soft soils, require additional studies to those practiced in Brazilian geotechnical engineering. The need for embankment and consequently stability analyzes and field and laboratory tests, involve numerous uncertainties, which can be rationalized by means of probabilistic methods. Lemos (2014) studied the nondrained resistance parameter of soft soils by means of field and laboratory tests, besides the characterization tests, in a layer of soft marine clay located in the Brazilian coastal plain. From the results of these tests, with the aid of Slope/W software, the height for a simple embankment was calculated on the foundation soil mentioned above for the safety factor of 1.5, resulting in 0.50 meters in height. From the geotechnical model established in the same study, the influence of the test on the determination of non-drained resistance and total specific gravity, and the consequence on the calculation of deterministic safety factors and reliability indices (β) were verified. It was also verified the effect of the geotechnical model, consideration of stratifications or single soil; the restriction of data use based on the refinement and comparisons of Lemos (2014); besides the choice of the deterministic method of calculation: Morgenstern and Price (M-P), Janbu and Bishop. To estimate reliability indices, the First Order Second Moment (FOSM), Monte Carlo Simulation (SMC) and Point Estimate (EP) methods were used. Finally, with the aid of the FOSM method, the geotechnical parameter that most contributed to the variance of the probability of failure was identified. It has been observed that in general, methods that lead to lower FS, are more likely to rupture. In addition, field trials resulted in greater probabilities of ruin, as the probability calculation was significantly influenced by the variability of the geotechnical parameters. Pre-treatment of the data, to eliminate possible noise, significantly alters the probability of failure. In general, the approximate methods, FOSM and EP, lead to a greater probability of rupture, when compared to SMC. When considering the values of undrained resistance varying with depth, in the layer of the foundation soil under study, a reduction in the probability of rupture is observed when compared to the case of homogeneous soil, the latter being possibly a conservative consideration. The variation of the value of the total specific weight caused practically no probability of rupture for all calculation methods. As a final contribution of this research, it is possible to affirm that the relative reliability analyzes are an excellent tool to supplement the results calculated by deterministic methods.