Abstract
The study explores the effects of three different lignocellulose fiber-reinforced (jute, coir, and water hyacinth [WH]) soils on the desiccation potential of compacted clayey silt soil. A new model was developed using artificial neural networks (ANN) for estimating cracking in soil reinforced with different fibers as a function of suction and water content. The program for ANN was developed in house using C++. Before model development, suction and water content were simultaneously monitored for 105 days along with the crack intensity factor (CIF). After model development, relative significance of each parameter (suction and water content) on the corresponding CIF was estimated. Adding lignocellulose fibers significantly increased the water retention capacity in the soil and reduced the CIF significantly as compared to unreinforced soil (almost half the amount). Obtained ANN models were efficient in predicting the CIF. The CIF is inversely proportional to water content and directly proportional to suction. The CIF value in bare soil, jute, and WH composites primarily depends on suction. Because of the increased water retention capacity of coir, the value of CIF depends equally on both suction and water content values. The log normal distribution of CIF was found in soil–jute composites.