Resumo:
Reliability-based design, such as LRFD, aims at meeting desired probability of failure levels for engineered structures. The present work attempts to contribute to this field by analyzing the influence of spatially variable soil/rock strength on the axial resistance uncertainty of single and multiple shafts in group layouts. This includes spatial variability over the individual shaft surfaces, effects of limited data, random measurement errors, and workmanship. A possible correlation between boring data inside or near the footprint of a foundation and the foundation itself is considered. In a geostatistical approach, spatial averaging (upscaling) and a degenerate case of ordinary kriging are applied to develop variance reduction charts and design equations for a series of foundation group layouts (single, double, triple, and quadruple). For the potential situation of an unknown horizontal correlation range at a site, the worst case scenarios are identified and demonstrated in an example problem. Resulting probabilities of failure are applied to the whole foundation (i.e., group) rather than single objects. It is found that a boring at the center of a group footprint can significantly reduce resistance prediction uncertainty, especially under the worst case scenario for unknown horizontal correlation range. In contrast, independent of the presence of a center boring or not, the uncertainty reduction through additional borings becomes small, once four or five borings are available.
