A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Water Engineering and Management

Thesis (M. Sc.) - Asian Institute of Technology, 2000

Water loss is the huge problem for the management of urban water supply system due to the composition of physical and commercial losses in the network. To handle the water loss, it is essential to know the important factors influencing and how to manage on this issue. The main objective of this research is to find out the most important factor affecting on water loss and using different techniques in the water distribution system of the Samut Prakan Branch under the management of the Metropolitan Waterworks Authority. For this analysis, 40 DMAs from the Samut Prakan Branch were selected to collected the data associated with water loss. Correlation matrix, multiple regression analysis (MRA) and artificial neural network (ANN) were studied to analysis the water loss data. Correlation matrix was used to find out the explanatory variables by removing each variable based on multi-collinearity and the significant level. Three different datasets were imported into MRA and ANN models; all variables, explanatory and important variables. Master models from MRA and ANN were selected to control the rate of water loss and sensitivity analysis was performed to discover the most sensitive factor and give recommendation for the different measures. Among the MRA models, semi-log models were better than linear and log-log model. Moreover, ANN model with hyperbolic tangent function has the higher accuracy than the model with rectified linear unit. ANN model gave the better performance than MRA model, with 95% accuracy. Based on the sensitivity indices of each variable, average diameter of valve is the most sensitive variable to give the priority for the reduction of water loss. When the average pipe diameter of valve was reduced, the rate of water loss decrease more from the observed rate. The results from sensitivity analysis can help to policy makers in decision making for the water loss reduction in the Samut Prakan Branch.