Laboratory Evaluation and Neural Network Modeling of Treated Macau Marine Clay

Abstract

The marine clay is generally regarded as poor materials for urban applications because of its high settlement and instability. In this study, the soft marine clay is mixing with cementitious material such as cement and lime in different ratios, so that the geotechnical characteristics of blends improved. A series of laboratory experiments are conducted to verify the enhanced performance of treated marine clay. The optimum moisture content (OMC) along with maximum dry unit weight obtained from Standard Proctor compaction test and Harvard miniature test. Unconsolidated undrained test (UUT) and unconfined compression test (UCT) are implemented to find out the compressive strength of samples. Blends selected from the dry side of the compaction curve own larger compressive strength than the others, which means an appropriate amount of water is enough for blends mixing. The results of laboratory tests are utilized to establish neural network models to predict engineering properties such as, compressive strength, optimum moisture content, and maximum dry density of treated marine clay. The correlation between the additive contents as well as curing days and compressive strength has been approved according to the test results. The properties such as maximum dry density, water contents, and additive contents are used as inputs to predict the compressive strengths. The predicted consequences of neural networks are well fitted with laboratory test results.