NONLINEAR SETTLEMENT MODELING OF STONE COLUMN REINFORCED SAND FOR RINGWALL FOUNDATIONS BASED ON PLATE LOAD TESTS

Abstract

Settlement control is a critical design consideration for large-diameter oil tank ringwall foundations constructed on granular soils, where stress-dependent stiffness and spatial variability can significantly influence serviceability performance. This study aims to the settlement behavior of stone-column-reinforced dense sand and compares a nonlinear power-law settlement model against a conventional linear baseline using plate load test (PLT) data. Laboratory-scale PLTs were conducted on dense sand (Dr = 80%) with configurations of untreated soil and sand reinforced with one to three stone columns. The results show that stone column inclusion significantly reduces settlement, with values decreasing from 21.3 mm in untreated sand to 9.87 mm, 6.42 mm, and 5.19 mm for one, two, and three-column configurations. The settlement reduction ratios reached approximately 76%, with diminishing incremental improvement as column number increased. The power-law model outperformed the linear baseline, achieving coefficients of determination (R²) of 0.986-0.995 compared to 0.728-0.936, and substantially lower prediction errors across all configurations, a power-law formulation thus provides a better fit to the observed data than a linear assumption.