An “all pipes” hydraulic model of a drinking water distribution system was constructed with a bottom-up approach of demand allocation. This means that each individual home is represented by one demand node with its own stochastic water demand pattern. These water demand patterns were constructed with the end-use model SIMDEUM. A sensitivity test with respect to the resulting residence times was performed for several model parameters: time step, spatial aggregation, spatial correlation, demand pattern and number of simulation runs. The bottom-up approach of demand allocation was also compared to the conventional top-down approach, i.e. a single demand multiplier pattern is allocated to all demand nodes with the base demand to account for the average water demand on that node. The models were compared to measured flows and residence times in a small network. The study showed that the bottom-up approach leads to realistic water demand patterns and residence times, without the need for any flow measurements. The stochastic approach of hydraulic modelling, with a 15 minute time step, some spatial aggregation and 10 simulation runs, gives insight into the variability of residence times as an added feature beyond the conventional way of modelling.
Keywords: demand allocation, stochastic modelling, water quality modelling