Effect of Number of Soil Strength Measurements on Reliability of Spread Footing Designs

Abstract

One of the primary objectives of using load and resistance factor design (LRFD) is to produce designs that consistently reach some established target reliability. Achieving this objective is challenging in geotechnical applications because variability and uncertainty in design input parameters vary substantially from site to site and with different quantities of site characterization. Analyses were performed to simulate spread footing design by using different numbers of undrained shear strength measurements. Results revealed that the likelihood was high of over- or underestimating both the mean value of a design parameter and the uncertainty in the mean value when small numbers of measurements were considered. The analyses also revealed that LRFD methods that prescribed constant resistance factors frequently produced underreliable designs when small numbers of measurements were considered. The percentage of underreliable designs decreased when increasing numbers of measurements were used. However, this decreasing percentage of underreliable designs was accompanied by an increasing percentage of overreliable designs rather than an increasing percentage of designs that achieved the target probability of failure within practical limits. In contrast, LRFD methods prescribing resistance factors that depended on uncertainty in design parameters tended to produce a consistently low percentage of underreliable designs when more than five measurements were used. As greater numbers of measurements were considered, such methods produced increasing percentages of designs that practically achieved the target probability of failure and decreasing percentages of overreliable designs.