A methodology for synthetic household water consumption data generation

Abstract

In the smart cities context, real-time knowledge of residential water consumption has become increasingly important, especially given the fast evolution of sensors, ICT and the production of big, high-resolution data coming from the urban environment. A variety of reasons often leads to the creation of continuity gaps in these data series, thus making the need for a methodology that produces reliable and realistic synthetic data urgent. In this article, we present a methodology that generates synthetic household water consumption data; we showcase it in two case studies, Skiathos, Greece and Sosnowiec, Poland, which exhibit significant differences in water consumption patterns. The methodology captures the stochasticity of daily residential water use. Algorithm validation is implemented through the comparison of various metrics for actual and generated data; this way, we show that the suggested approach is capable of adequately simulating water consumption in both micro- and macro-time scale. © 2017 Elsevier Ltd