Analysis and estimation models of residential water demands from hourly monitored data
Résumé
Hourly monitored data of several residential and industrial water meters installed in one drinking water system in the South of France are analyzed to identify dynamical features and properties. Small residential customer water consumption may be very irregular with no evident time periodicity. As average water consumption increases, time periodic patterns emerge at the daily and weekly scale. A general autoregressive integrated moving average model is proposed to mimic water use. Exogenous variables are also introduced to account for influence on water consumption due to seasonal trends, holidays, maximum daily temperature, rainfull and evapuration index.