Description: The forecasting of urban daily and hourly demand for water resources are the basis for making immediate and tactical decisions by public utilities that provide water supply services to the population. For example, knowing of the hourly water demand during the day (24 hours) gives an opportunity for optimal control over the operation of pumping stations, that is, it saves energy consumption in the process of pumping water. Therefore, the crucial task is to develop methods and models for forecasting of urban water consumption that could properly describe the process of changes in urban water consumption during the day (24 hours) and provide an opportunity to build reliable forecasts of the future hourly water consumption in the city. For the analysis and construction of the model of the time series of urban hourly water consumption the autoregression method is applied.The methods and models for calculating the forecast of urban water consumption are considered, an algorithm for predicting urban hourly water consumption is proposed, based on the resolution of the time series into the basic and residual components and the concept of the same type of days. The presented algorithm was implemented in the form of a software package and underwent a multi-year check in real conditions which confirmed its effectiveness. The average relative error of the hourly forecast of urban water consumption per day does not exceed 5%. The newly developed model (and its software package) for calculating the forecast of urban hourly water consumption can be applied in public utilities which will enable them to optimize their operational and investment decisions.
Keywords: water supply, water consumption, forecast, forecasting, time series, model, statistical analysis, regression, neural network