Reliability-based approach for the design of spread footings on granular soils -

Abstract

Uncertainties in geotechnical engineering have a major effect on the overall design process. However, these uncertainties are still partly integrated in the standard codes of practice, and some geotechnical problems are still handled in a deterministic fashion. Reliability-based design methodologies have been used by several codes such as the Canadian National Building Code, AASHTO (1997) and Eurocode7 for Ultimate Limit State designs. However, these codes still apply deterministic approaches for Serviceability Limit State designs. In the first part of this study, the existing hyperbolic load-settlement relationship proposed by Akbas and Kulhawy (2009), for the calculation of settlements will be updated based on a new set of datapoints of settlement versus loads compiled by Akbas (2007). Akbas and Kulhawy's (2009) model was derived by fitting a hyperbolic function to a set of 125 load settlement curves of footings on granular soils comprised mainly of footings having widths less than 1.0m. As such, the applicability of this model to practical size footings is questionable. Based on the above, the derived model was updated using a set of load versus settlement datapoints of full scale footings using a first order second moment bayesian technique. In the second part, a new load-settlement relationship that uses a power function will be proposed. This entails fitting a power model to the set of 125 full load settlement curves utilized previously by Akbas and Kulhawy (2009). This model will be later updated using the same set of datapoints of load versus settlement that were previously utilized to update the hyperbolic model. Once the normalized load-settlement models (hyperbolic and power) have been established, reliability based analyses will be conducted on traditional Meyerhoff (1965) and Burland and Burbidge (1984) models and on the normalized load-settlement relationships. Afterwards, SLS resistance factors will be proposed for all methods, in order to maintain acceptable levels of reliability. The SLS res