A methodology to incorporate load tests into the reliability-based design of deep foundations for the serviceability limit state

The evaluation of the serviceability performance of foundations is important for the geotechnical design. This thesis provides framework to perform reliability analysis for the serviceability design of foundations and implements it on a case study for design of tall towers in Saudi Arabia. First, different prediction models for axial displacement such as t-z analysis and finite element models were examined. Then, a reliability model is proposed that captures both the epistemic and aleatory uncertainties in the predicted settlement. Furthermore, since there are few if any formulations for reliability analysis for serviceability limit state in current LRFD design codes, the accepted level of risk of a serviceability failure is evaluated on the basis of risk and decision analysis. In addition, a methodology is proposed to utilize the proof load tests in the field to update the probability of failure based on the Bayesian technique. Finally, a framework for planning for a proof load test program is provided using Monte Carlo simulations. The results shows the importance of the underlying assumptions in the settlement prediction models. Also, the case study shows that the probability of serviceability failure is reduced when there is an increase in the soil stiffness. Moreover, the results show the significance of obtaining a load-displacement curve from the field and how it contributes to reducing the epistemic uncertainty. The importance of defining the amount of deformation that is problematic to the structure is stressed.