Residential water demand characterization is one of the crucial aspects which still represents a great source of uncertainty in water distribution systems modelling.
In the last two decades many demand models have been proposed, and they either model the demand for end users or for clusters. In the first category there are some examples that use rectangular pulse models or in which water demand variability is reproduced with CDF parameters that were estimated from experimental data.
The downside this kind of models is that they are not easily applicable due to the multiplicity of information they require for the parameters estimation. Indeed, the pursuit of a very fine detail in the demand characterization implies the implementation of higher model complexities and number of parameters that compromises the applicability of such approaches in non-scientific contexts.
Recently an approach [1] based on a mixed distribution (contemplating both null and not null demand) has been proposed to model demand for clustered users.
This approach considers water demand to be combination of a discrete and a continuous random variable, the discrete one is used to model periods of non-null demand.
For these reasons, a Python tool was developed in order to generate congruent synthetic demand scenarios by means of the above-mentioned approach. As input the number of users connected to each node and the shape of the demand pattern is required. The latter is extremely variable and related to the users’ habits, but few relations proposed are implemented to help the user to estimate the peak coefficient. In addition, a simulation period can be specified indicating the number of days that are willing to be represented.
[1] R. Gargano, C. Tricarico, G. del Giudice, and F. Granata, “A stochastic model for daily residential water demand,” Water Sci. Technol. Water Supply, vol. 16, no. 6, pp. 1753–1767, Dec. 2016.