Reliability Analysis of Drilled Shaft Resistance: A Case Study of the Circunvalacion Norte Viaduct Project-Paper

Paper

Pdf Summary

The study evaluates drilled-shaft resistance for Costa Rica’s Circunvalación Norte viaduct, which is supported by 426 shafts and 58 pile caps. Using results from 138 pressuremeter tests, the authors compare conventional deterministic design with two reliability-based methods: a random-variable (RV) model and a spatially correlated random-field (RF) model. Both probabilistic approaches use Monte Carlo simulation; the RF model represents soil variability through a Karhunen–Loève expansion.<br /><br />Deterministic calculations generally indicate more feasible designs because they use average soil properties and understate the risk associated with variability. For example, a 15 m-long, 1.0 m-diameter shaft has a calculated failure probability of 0.004 deterministically, compared with 0.33 using RV and 0.19 using RF. Overall, RV typically predicts higher failure probabilities than RF because it assigns one soil-strength value to the entire shaft, while RF models spatial variation and the averaging of local differences along the shaft.<br /><br />The RF sensitivity study found that longer horizontal correlation lengths slightly reduce failure probability; differences between the tested correlation lengths were about 5%. Mean resistance estimates converged to similar values after roughly 10,000 simulations.<br /><br />The authors conclude that deterministic methods may lead to unsafe designs when soil variability is substantial. Probabilistic reliability analysis provides a more informative basis for assessing shaft performance, with RF offering a more realistic representation of spatially variable soils.

Keywords

drilled-shaft

reliability analysis

pressuremeter tests

Monte Carlo simulation

random variable model

random field model

spatial correlation

failure probability

soil variability

Karhunen-Loève expansion