Laboratory Investigation and Neural Network Modeling of Engineering Properties of Marine Clay Treated With Portland Cement

Abstract

The amount of municipal waste is significant in Macau, among which marine clay makes a significant proportion. Due to marine clay’s high natural moisture content it is not suitable to be used for engineering purposes. This study aims to provide a preliminary experimental investigation of the engineering properties of marine clay when stabilized with cement. Laboratory experiments are conducted to investigate the compaction properties of the stabilized soil. Results show that with the addition of cement, the Optimum Moisture Content (OMC) increases and the Maximum Dry Unit Weight (MDUW) decreases. Furthermore, strength tests such as unconfined compressive and unconsolidated undrained triaxial tests were conducted to explore the engineering properties of the stabilized mixture. It is apparent that strength is a function of curing duration. As the curing time increases, the strength increases due to the hydration processes of Portland cement. Results also show that the highest strength was achieved when the remolded water content of the mixture is on the dry side of OMC. Artificial neural networks were developed to predict the compaction properties of the cement stabilized marine clay such as OMC and MDUW. Furthermore, the strength of the stabilized mixture was also predicted. The ‘best’ architecture of the neural network was investigated in this study.