Key points are not available for this paper at this time.
Population growth and climate change have increased the demand for freshwater resources. In Iran, the average per capita freshwater consumption is approximately 5 % to 85 % higher than the global average. In addition, high levels of water loss and inefficient use of drinking water emphasize the need to reduce reliance on these resources. One solution is the use of non-drinking water distribution networks (WDNs), which are often designed based on current conditions but should also be optimized for future scenarios. This study investigates the existing non-drinking WDN at the University of Isfahan and determines an optimal operation strategy considering future water demand. In other words, a new approach is proposed to overcome the limitation of climate-influenced and population increasing water demand value by prediction them. For this purpose, an optimization model is equipped with data-driven based water demand prediction model for proper pump schedules considering the limitation of full life-cycle-cost formulation. Here, the operation of the network’s pumps is optimized using a Binary Genetic Algorithm (BGA), which determines their on/off schedules based on electricity costs and pump depreciation. In addition, water demand is predicted for the next five years using an Artificial Neural Network (ANN), based on historical consumption data (2013–2017). Results show that energy consumption can be reduced by 19.77 % in summer and 37.5 % in winter using the proposed method. Furthermore, the best ANN model leads to an R² value of 0.89 (training) and 0.85 (testing/validation), indicating strong predictive performance.
Najarzadegan et al. (Tue,) studied this question.