Evolutionary computing of the compression index of fine-grained soils

Developing simple empirical correlations to predict the compression index of fine-grained soils has been the major concern of many previous studies to enable speedy and economic approach for estimating the consolidation settlement. However, evidence from recent studies showed the necessity to examine these empirical correlations using site-specific results. Latest studies also illustrated the need to develop new site-specific correlations for better accuracy of the prediction of the compression index. Thus, this research has been conducted to examine the accuracy of the available empirical correlations in predicting the compression index of fine-grained soils of Sulaymaniyah Province in north of Iraq and to develop more accurate compression index correlations. One hundred seventy-seven undisturbed samples have been collected and extensively tested in the laboratory to develop a comprehensive database of the compression index and associated soil properties (void ratio, moisture content, liquid limit, plasticity index, and dry density). The developed database has been employed in a comprehensive statistical assessment to evaluate the accuracy of the available compression index correlations. The results of the statistical assessment showed that all the available empirical correlations provide poor prediction of the compression index of north of Iraq fine-grained soils. Consequently, new correlations have been proposed utilizing multi-objective genetic algorithm evolutionary polynomial regression analysis. The new correlations offer better prediction and hence, can be used in routine designs with more confidence for projects within Sulaymaniyah Province. In addition, further examinations of these correlations are required using database from other regions in and outside of Iraq to ensure that the developed correlations can be used with confidence to estimate the compression index for soils outside Sulaymaniyah Province.