PySIMDEUM - An open-source stochastic water demand end-use model

Abstract

Water demand is a crucial input parameter in water distribution system analysis because it can fluctuate over various temporal and spatial scales. In the past, researchers developed stochastic models that can provide realistic consumption patterns for simulations to account for those demand dynamics. Parameters for stochastic models are usually retrieved by fitting these models on smart water meter data. The stochastic demand model SIMDEUM uses an entirely different approach by generating highly realistic water demands based on (country-specific) statistical information only, without the need for measurements. While this approach makes SIMDEUM widely applicable in the water sector, its widespread usage within the community has been hindered due to its software implementation and availability. We produced pySIMDEUM, an open-source and object-oriented implementation of the SIMDEUM software in the popular and freely available programming language Python. The pySIMDEUM software package is not only publicly available for usage within the water field — it is also intended to build the cornerstone of a widespread pySIMDEUM community of active developers. We want to use the WDSA/CCWI conference to address interested researchers or practitioners in the water sector and invite them to contribute to the software package as active part of the pySIMDEUM community. We will show SIMDEUM’s history and past applications, the mathematical approach behind SIMDEUM and pySIMDEUM, where to download and install the pySIMDEUM package, the structure of the program, and a minimal example of how easily pySIMDEUM can be used to generate realistic stochastic water demand patterns from scratch. Furthermore, we will highlight possible future applications of the new pySIMDEUM tool. These applications include automatic parametrisation of pySIMDEUM parameters on smart meter data, coupling stochastic demands directly with hydraulic solvers, or how to enable city-scale stochastic demand simulations.