Modeling of the Geotechnical Properties of Soils Improved with Quarry Dust at Different Compaction Energies

Abstract

The compaction and unconfined compressive strength (UCS) properties of soil-quarry dust mixtures were investigated at different compaction energies for 10 different soil samples of low to medium plasticity. The quarry dust content of the soil was varied from 0 – 40% by weight of the soil, and the soil-mixtures were subjected to Atterberg limit tests, compaction tests, and UCS tests. The results show that plasticity properties were reduced with increasing quarry dust content, while maximum dry unit weight (MDUW) increased with both quarry dust content and compaction energy. The optimum moisture content (OMC) was generally reduced with increasing quarry dust content, but the trend varied with different samples. The UCS values showed no clear trend in the samples studied, and the result suggested that addition of quarry dust does not necessarily increase the UCS of all soil samples but can lead to loss of strength in some soils. Several statistical and machine learning equations were developed for the prediction of MDUW, OMC, and UCS for soil-quarry dust mixtures using artificial neural network (ANN) and multiple linear regression analysis (MLRA). Using the coefficient of determination (R2) as a basis, it was observed that ANN made better predictions than MLRA.

Keywords

• soil-quarry dust mixture
• compaction
• unconfined compressive strength
• artificial neural network
• multiple linear regression analysis