Presenter/Author Information

Anna Di Mauro, Unicampania

Keywords

data modelling; water fixture profile; water end-use; time serie

Start Date

15-9-2020 2:40 PM

End Date

15-9-2020 3:00 PM

Abstract

The integration of innovative smart technologies in the water sector made available a great amount of data about water consumption able to support urban water management, improve utilities procedures, and make consumers aware of their use of water. In the recent years, the possibility to obtain water consumption data at end-use level has become of interest in order to perform water demand forecasting and demand side management. In fact, end-use measurements can allow correlating the aggregate household water consumption to the water consumption ascribed to individual end-uses. Thus, the data contributes to understanding how water is used in homes. This allows the utilities to improve water service delivery and management, identify consumption patterns, detect anomalies and losses, improve users’ awareness endorsing sustainable behaviour, and personalize billing profiles. It is straightforward to observe that each end-use consumption is conditioned by human activity, and it changes over time (i.e. daily, weekly, seasonal). Therefore, there is a need to identify new data modelling approaches to manage end-use data and enable intelligent water management. Starting from similar experiences applied in the energy field, the paper shows how water end-use time series can be modelled to profile users. It presents a model for obtaining a parametric water consumption profile able to characterize a household in terms of fixtures usages. The method is tested on a database of real residential end-use measurement, and it combines a statistical approach to extract significant features to instantiate the consumption model, a clustering approach to classify water usages and a regression approach to describe water end-use consumption usages representative of clusters. As results, the proposed approach provides the procedure to obtain a water consumption profile that characterizes the water usage at end-use level showing the importance of disaggregated data and data modelling to identify and profile users’ behaviours.

Stream and Session

false

Share

COinS
 
Sep 15th, 2:40 PM Sep 15th, 3:00 PM

Data modeling of water end-use time series

The integration of innovative smart technologies in the water sector made available a great amount of data about water consumption able to support urban water management, improve utilities procedures, and make consumers aware of their use of water. In the recent years, the possibility to obtain water consumption data at end-use level has become of interest in order to perform water demand forecasting and demand side management. In fact, end-use measurements can allow correlating the aggregate household water consumption to the water consumption ascribed to individual end-uses. Thus, the data contributes to understanding how water is used in homes. This allows the utilities to improve water service delivery and management, identify consumption patterns, detect anomalies and losses, improve users’ awareness endorsing sustainable behaviour, and personalize billing profiles. It is straightforward to observe that each end-use consumption is conditioned by human activity, and it changes over time (i.e. daily, weekly, seasonal). Therefore, there is a need to identify new data modelling approaches to manage end-use data and enable intelligent water management. Starting from similar experiences applied in the energy field, the paper shows how water end-use time series can be modelled to profile users. It presents a model for obtaining a parametric water consumption profile able to characterize a household in terms of fixtures usages. The method is tested on a database of real residential end-use measurement, and it combines a statistical approach to extract significant features to instantiate the consumption model, a clustering approach to classify water usages and a regression approach to describe water end-use consumption usages representative of clusters. As results, the proposed approach provides the procedure to obtain a water consumption profile that characterizes the water usage at end-use level showing the importance of disaggregated data and data modelling to identify and profile users’ behaviours.