The use of water distribution network (WDN) models is an extended practice [13]. Confidence on decisions taken upon such models depends highly on their accuracy [11]. The parameters uncertainty has to be defined in order to include it in the model. Some of the parameters in a network (e.g. pipes lengths and diameters) can be easily measured and their uncertainty can be calculated on a statistical basis [4]. Demands cannot be measured directly and they have to be estimated using other measurement...
The use of water distribution network (WDN) models is an extended practice [13]. Confidence on decisions taken upon such models depends highly on their accuracy [11]. The parameters uncertainty has to be defined in order to include it in the model. Some of the parameters in a network (e.g. pipes lengths and diameters) can be easily measured and their uncertainty can be calculated on a statistical basis [4]. Demands cannot be measured directly and they have to be estimated using other measurements [10][8]. The uncertainty in the measurements used for that estimation is propagated to the parameters [1]. Besides, demands have their own stochastic nature that induces uncertainty. This paper describes how the pressure measurements are used to infer the uncertainty model in demands for a real network. The real data are treated in order to avoid the effect of boundary conditions. An uncertainty model for demands is calculated to justify the observed behaviour of the measurements. Montecarlo simulations are used for the validation.
Citation
Perez, R., Sanz, G., Cuguero, M.A., Blesa, J., Pep Cugueró Escofet. Parameter uncertainty modelling in water distribution network models. A: Computing and Control for the Water Industry. "CCWI 2015 - Computing and Control for the Water Industry". Leicester: Elsevier, 2015, p. 583-592.