| dc.description.abstract | Compression index (Cc) of normally consolidated (NC) clays determined by the oedometer experiments is utilized for calculating the consolidation settlement of shallow foundations. The determination of the Cc from the tests takes much more time and so empirical correlations based on clay properties can be a suitable alternative for the prediction of settlement. However, uncertainty in the measurements of input parameters has always been a major concern. Robust optimization is very popular due to its computational tractability for many classes of uncertainty sets and problem types. Therefore, in this research, an innovative method based on robust optimization has been used to investigate the effect of such uncertainties. To achieve these, the results of 433 oedometer tests taken from geotechnical investigation locations in Mazandaran province of Iran have been used. Based on Frobenius norm of the data points, uncertainty definition is presented and examined against the correlation coefficients for several empirical models for predicting Cc value and thus optimum values are determined. The results in compare with previous models indicate the robust method is a better pattern recognition tool for datasets with degrees of uncertainty. The variation of the Cc values with soil properties, namely, water content (?n), initial void ratio (eo), and liquid limit (LL), by considering different value of uncertainties (0, 5 and 10%) was considered and indicated that the effect of eo is more than other two physical parameters (?n and LL). The best model performance during in deterministic valuation and considering uncertainty is further shown. © 2020 The Authors. Published by Pouyan Press. | |
