Swell and Shrinkage Percentages for Various Soil Types and their Prediction from Intrinsic Soil Properties

Abstract

Earthwork construction involves excavation, hauling, placing and compaction of soil, gravel, and other materials that exist on the soil surface. The soil volume varies depending whether the soil is bank, loose or compacted material. Therefore, the final generated volume of earthwork should be adjusted for volume changes during the above states by applying shrinkage and swell-correction percentages. Estimating these correction factors by engineering expertise or selecting predetermined tables without extensive knowledge of the local soils has proven to be costly and may be misleading. Accordingly, the current study was initiated to develop the database of swell/ shrinkage percentages for various soil materials and link them to other soil properties. Standard procedures were applied for determining soil density at different states along with the physical, chemical and geotechnical properties for soils obtained from 39 surveyed projects within and on the outskirts of Erbil city. The obtained data were subjected to different statistical analysis and the results indicated that the swell percentage ranged from 36.10 -55.7% for clays, 18.40 – 69.20% for silts and 11.90 – 54.5% for gravels. Shrinkage percentage ranged from 9.20 – 16.5% for clays, 4.40 – 20.20% for silts and 0.80 - 23.5% for gravel. Overall, within each group, the swell percent was superior to the shrinkage percent. Additionally, the swell percent was characterized by having a higher coefficient of variation compared to that of shrinkage percent. The in situ soil density and clay content have emerged to be the most effective soil properties for predicting swell percent. On the other hand, the influential variables for predicting shrinkage percent were in situ soil density and the maximum dry density. The mean absolute error of prediction of swell and shrinkage percentages were 6.79 and 1.17 respectively, indicating that shrinkage percent  can be predicted more accurately compared with swell percent.