ABSTRACT
As wastewater infrastructure systems approach their design lives, there is a growing demand for maintenance activities to extend the design life and to minimize the potential for loss of function. It is necessary to develop an efficient infrastructure management schemes since not all maintenance and repair projects can be funded (Chouinard 1996). In order to address the problem of making optimal decisions under budgetary constraints, a sewer management technique utilizing demand forecasting techniques is presented in this paper. Accurate forecasting of wastewater demand is dependent on proper identification of wastewater predictor variables. In this paper, predictor variables for wastewater demand are identified. Through the use of artificial neural network (ANN), future wastewater demand is forecasted. Data needed for the development of the sewer demand forecasting models were obtained from the City of Indianapolis and the City of San Diego. The results of the ANN model were compared to traditional statistical methods. The results showed that the ANN model was able to produce better forecasts when compared to multiple regressions.
