|Budget Amount *help
¥3,600,000 (Direct Cost: ¥3,600,000)
Fiscal Year 2000: ¥1,600,000 (Direct Cost: ¥1,600,000)
Fiscal Year 1999: ¥2,000,000 (Direct Cost: ¥2,000,000)
A pile is often modeled as a beam on nonlinear elastic foundation so as to include the effect of subgrade reaction. In design codes, several methods of determining the coefficients of subgrade reaction are available. However, the accuracy of the coefficients thus obtained is not so high. The existing methods may not be satisfactory when the performance-based design procedure is introduced.
For the purpose of improving the accuracy, it is desirable to conduct the in-situ experiment of a pile. In recent years, therefore, it has been tried to evaluate the coefficients of subgrade reaction based on the results of in-situ experiments. However, since the relationship between the subgrade reaction and the deflection of a pile is nonlinear, it is quite difficult, if not impossible, to obtain the coefficients of subgrade reaction analytically from the experimental results. The coefficients are evaluated by trial and error in most cases.
Against this background, the numerical procedure for computing the deflection of a loaded pile has been established first in the present study. In this procedure, the finite element method with two-node beam elements is applied and the iterative method based on the Newton-Rahpson technique in incorporated. Based on this numerical procedure, the back-analysis procedure to determine the coefficients of subgrade reaction is proposed. In this back-analysis, the coefficients are evaluated by minimizing the difference between experimental data and computed values. In the proposed back-analysis, the accuracy of the coefficients of subgrade reaction for the lower layer of the multi-layer foundation is very sensitive to the error involved in the experimental data. From the viewpoint of the bending rigidity of a pile and the way the experimental data are handled, the strategies of improving the accuracy of the back-analysis are discussed.