Name: BRENDA DA SILVA ALTOÉ
Type: MSc dissertation
Publication date: 23/08/2018
Advisor:
Name |
Role |
|---|---|
| PATRÍCIO JOSÉ MOREIRA PIRES | Advisor * |
Examining board:
| Name |
Role |
|---|---|
| PATRÍCIO JOSÉ MOREIRA PIRES | Advisor * |
Summary: This work presents a study about the behaviour continuous helix piles, tested with static loads,
which have been executed in an industrial construction in Mato Grosso do Sul. Such study was
based on STP field tests, performed on a terrain composed of alluvial sediments, with irregular
layers with sandy, silt and loamy composition. The study consists in two approaches,
deterministic and probabilistic, being that the deterministic, aims to compare safety factors (FS)
estimated by five semi-empirical methods: Aoki-Velloso (1975) - Velloso-Lopes (2002); Aoki-
Velloso (1975) modified by Monteiro (1997); Décourt-Quaresma (1978) Décourt (1996);
Alonso (1996) and Antunes-Cabral (1996), with values obtained by extrapolation of the loadsettlement
curves, drawn from the static loads, by the methods: Van der Veen (1953); Van der
Veen (1953) modified by Aoki (1976) and NBR 6122 (ABNT, 2010). In this field, the mean
and standard deviation of the errors of the FS values obtained by semi-empirical methods in
relation to the ones obtained via extrapolation are presented. Décourt-Quaresma (1978)
Décourt (1996) method was the most conservator one, presenting the lowest error, both related
to the mean and the standard deviation of FS. For analyses with extrapolation, Van de Veen
(1953) - Aoki (1976) method was the one that generated the lowest error, both in the mean
value of FS and in its standard deviation. The probabilistic analysis took into consideration the
soundings with radius of 100m, 150m and 200m around each tested pile, to statistically estimate
the values of the mean and the variance of the safety factor in these regions by the semiempirical
methods. The Kolgomorov-Smirnov test confirmed Gaussian distribution for the FS,
allowing the use of Morlá Catalán and Cornell (1976) methodology to calculate the
realiableness indexes and the rupture probabilities. Décourt-Quaresma (1978) Décourt (1996)
method was the most conservator one, with the lowest 􀴤𝐹􀴤􀴤𝑆􀴤 values and the highest ruin
probability. Larger sample sizes not necessarily resulted in more accurate estimates of the FS
probability distribution, although in most of the cases the FS error has decreased both in mean
and standard deviation with the increase of the sampling area. Besides, the ruin probability
tends to be slightly sensitive to variations in the sampling areas.
