Field simulation of settlement analysis for shallow foundation using cone penetration data

This paper deals with the settlement analysis of a vertically loaded strip footing by the use of two-dimensional random field finite element method. Total settlement and rotation of the footing have been calculated by elasto-plastic solution in finite element model. Deformation modulus of soil E-d was assigned conforming to 2D Gaussian random field using Karhunen-Loeve series expansion. Spatial variability of E-d was represented by the horizontal and vertical correlation lengths (theta(vE) and theta(hE)). CPTu database from Adapazari, Turkey has been employed to estimate the correlation lengths. Soil profile was modelled with 4 layers in accordance with random field from in-situ test results, and correlation lengths were assigned to each different layer. Random field realizations were produced by MATLAB code, and probability density functions (PDF) of maximum total settlement and rotation of the footing were constructed using Monte Carlo Simulations by iterative solutions of each realization with finite element analysis. Probabilities of failure (P-f) by settlement and rotation were calculated from PDFs. Initially, the analysis was carried out using average values of each horizontal and vertical correlation lengths that were assigned to every soil layer. Subsequently, analyses were iterated with maximum and minimum values of the correlation lengths in order that both the effect of horizontal and vertical spatial variability can be considered. 1000 calculations were performed for the 5 analysis models, with 200 random field realizations for each model. The effect of varying theta(vE) and theta(hE) on total settlement and rotation of the footing has also been investigated.

Automated summary

This paper discusses settlement analysis of a vertically loaded strip footing using 2D random field finite element method. Elasto-plastic solution and deformation modulus of soil were considered with spatial variability, and CPTu database and Monte Carlo simulations were utilized.