Amount of Uncertainty in the Methods Utilized to Design Drilled Shaft Foundations

In 2001, load and resistance factor design (LRFD) for deep foundations was required by the American Association of State Highway and Transportation Officials (AASHTO). Following implementation of LRFD, localized calibration of resistance factors using data from the states of Colorado, Florida, Kansas, Louisiana/Mississippi, Missouri allowed these states to utilize higher resistance factors during design. However, characterizing the uncertainty in the design of DSF, regarding the geotechnical investigation methods and the utilized software programs, higher values of resistance factors may be calibrated to more efficiently design DSF with the same level of reliability. Three test sites within the state of Arkansas, identified as the Siloam Springs Arkansas Test Site (SSATS), the Turrell Arkansas Test Site (TATS), and the Monticello Arkansas Test Site (MATS), were utilized to perform full-scale load tests on DSF. At each site, three geotechnical investigation methods (Arkansas State Highway and Transportation Department [AHTD], Missouri Department of Transportation [MODOT], and the University of Arkansas [UofA]) were utilized to obtained geotechnical parameters. The design of three DSF, at each site, was then performed, and the amount of resistance was predicted, using commercially available software (FB-Deep and SHAFT). At each site, the results obtained from bi-directional load tests were compared with the predicted values and the construction methods and problems (i.e. rock embedment length at the SSATS, collapsed excavation at the TATS, and equipment failure/concrete placement at the MATS) are presented herein. Two site-specific and a geologic-specific calibrations were performed by utilizing the results from the bi-directional load tests that were performed in Arkansas, the Bayesian updating, and the Monte Carlo simulation techniques. For each geotechnical investigation method and for each software program that was utilized during the DSF design, posterior distribution parameters were calculated based on previous calibration databases (i.e. the national database or the Louisiana/Mississippi database). Resulting resistance factor values were calculated for the geologic-specific mixed soils within the state of Arkansas. The calculated resistance factors ranged from 0.57 to 0.80 for total resistance. Furthermore, the FB-Deep software program is recommended in conjunction with the MODOT or UofA geotechnical investigation methods to design of DSF in Arkansas