Projections of municipal water demand with ANN and SD models under climate change

点击翻译按钮获取中文摘要

Water utilities across North America have observed declining residential water consumption over the past three**decades. Because they typically bill for water use on a per-unit-volume basis, such reductions in use have**financial implications. Further, water rates are usually set years ahead of time, so that inaccurate monthly to**annual predictions compound the financial effects of reduced water use. On shorter time scales of hours to**weeks, water utilities also make operational decisions to meet municipal demands. Short term forecasts help to**plan maintenance and operating schedules for system infrastructure and inform decisions about water levels**and drawdowns for reservoirs, and they tend to be dominated by weather. Thus, accurate prediction of**municipal water use is important for both short-term (operational) and long-term (planning) aspects of urban**water management.**This project will 1) provide forecasting models for both daily (operational) and weekly (planning) water**demands for EPCOR in Edmonton, Alberta, and 2) incorporate the weekly demands into a unique decision**support tool for long-term municipal water management. The two forecasting models will be developed using**Artificial Neural Networks (ANN), which have been applied widely for water demand projections. ANNs offer**excellent predictive ability and are relatively straightforward to develop and validate. Next, the planning**version of the ANN model will be used to revise an existing municipal water management model, called the**Calgary Water Management Model (CWMM), which will be set up for Edmonton. The CWMM can be used to**support municipal planning horizons of years to decades and is particularly useful for assessing the effects of**population growth, climate change and the effects of conservation policies and technological change over the**long term.