LFRD Calibration of Pullout Resistance Factors Using Survival Analysis for Soil Nails in North Texas

Abstract

Reliability-based analysis of the calibration of pullout resistance factors (φPO) for soil nails are presented in this paper. To conduct the calibration, a total of 47 verification tests conducted in cohesive soil within the Lyndon B. Johnson (LBJ) Express construction project in North Dallas Texas were collected. Three failure criteria, i.e., creep test data, test curve evaluation, and load-elongation curve were selected to estimate the nominal bond resistance between soil and grout. As a result, 25 tests met at least one of the failure criteria. In addition, 45 existing test results from the National Cooperative Highway Research Program (NCHRP) Report 701 in cohesive soil were incorporated to the test database and resulted in a total of 92 soil nail tests including 22 nonfailed tests. These 92 tests were separated into three databases for analysis and calibration purposes. Statistical analysis was conducted on these databases to evaluate the predicted (Rp) to measured (Rm) nominal bond resistances and normal and lognormal distributions were conservatively fitted to the lower tail of the failed test data. A new survival analysis approach was proposed to analyze the resistance distribution, which allows for incorporation of the nonfailed verification testing results. Utilizing these distributions, calibration of φPO was conducted by load and resistance factor design (LRFD) reliability analysis with the strength limit state and Monte Carlo simulations. A target reliability index (βT) of 2.33 and load factors (γQ) of 1.0, 1.35, 1.5, 1.6, and 1.75 were selected for the calibration as a result of the redundancy and loading conditions commonly found in soil nail walls (SNWs). From this analysis, 65φPO values were calibrated and ranged from 0.42 to 1.49, allowing engineers to select a φPO value appropriate to the SNW characterizations and conditions in cohesive soil.