|Tuesday, September 15th|
Ashlynn Stillwell, University of Illinois, United States
2:00 PM - 2:20 PM
Residential water consumption is understudied globally, especially regarding end uses of water both indoors and outdoors. Many areas of the United States, for example, collect meter-level data on a monthly timescale, and still others lack residential water meters entirely. To better understand end uses of water in the residential environment, additional data collection is needed, along with methods to make sense of data with better temporal resolution. As a demonstration of one possible approach, we installed a custom smart water metering system collecting data on total flow rate, temperature, and pressure at 1-second intervals, based on the data acquisition setup, from a single-family residential home in central Illinois, United States, starting in February 2018. Using one year of 1-second resolution data, we created a method to disaggregate meter-level data into end uses based on event detection as non-zero flow over a duration. Using our algorithm, we then disaggregated concurrent events using derivatives of the total aggregate flow rate and unsupervised machine learning k-means clustering approaches to compare with known end-use signatures for in-home fixtures and appliances. Analysis of the disaggregated data informs an estimate of water end uses, highlighting the dominance of the shower, clothes washer, and faucets in particular. Flow rates for disaggregated end uses reflect differences in fixture/appliance operations compared to manufacturers’ ratings and differences in behavior with human-controlled events (e.g., showers). Results can inform customized water conservation and efficiency recommendations at the household level. The disaggregation algorithm and unsupervised machine learning classification approach are readily transferrable to future residential smart water meter installations with similar temporal resolution data.
Andrea Emilio Rizzoli, IDSIA USI/SUPSI, Switzerland
2:20 PM - 2:40 PM
In this paper we describe an agent-based model that simulates the water consumption behaviour of households exposed to different stimuli and incentives promoting water efficiency and a more sustainable consumption of water. The model includes two main types of agents. The first type represents the water utility, while the second represents the households of the considered urban region. The model of the water utility receives from every household model their daily level of water consumption which is computed on the basis of the household features and the water price, but it also considers how the interaction among the households affects water use, under the assumption that users are stimulated to save more water if similar households in their network of social relationship have a lower average consumption. In the model two processes are modelled: first the diffusion of an app that allows users to be informed about their hourly/daily water consumption and relate their consumption level to the one of similar users, and secondly the process of behavioral change experienced by households exposed to the information provided by the app. The model thus allows to explore how different adoption levels of the app and different sensitivities to the social norm can impact collective behaviour. This is achieved by introducing two information diffusion sub-models that describe the interaction among users and a mechanism to capture also the effect of price incentives. The agent-based model has been implemented using the AnyLogic simulation platform. The model has been applied to data collected in a real world case of 250 households in the Locarno district in Southern Switzerland and validation results are discussed.
Anna Di Mauro, Unicampania
2:40 PM - 3:00 PM
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.
Jia Yi Ng, S'pore Uni of Tech and Design, Singapore University of Technology and Design, Singapore
3:00 PM - 3:20 PM
Population growth, urbanization, and climate-driven changes in rainfall patterns expose cities to increasing flood risks. Stormwater drainage systems are essential to limit such risks during extreme rainfall events. Their design generally relies on simulation-based optimization, where hydrological-hydraulic models are coupled with evolutionary algorithms or other metaheuristics. However, such an approach has high computational requirements owing to the complexity of the simulation models and the large number of iterations required by the evolutionary algorithms. This prevents its application to a large number of rainfall events. Hence, drainage systems are usually optimized with respect to a design storm and may not be robust with respect to different rainfall events. To overcome this issue, we adopt a data-driven emulation approach. First, we generate multiple rainfall events with a copula-based stochastic model, and then select representative rainfall events using clustering. Next, we identify a data-driven dynamic emulator (based on Gaussian Processes) that replaces the high-fidelity hydrological-hydraulic simulation model in the optimization process. The emulator predicts the performance of solutions under the representative rainfall events, gaining speed albeit losing some unneeded accuracy. The framework is applied to a case study in the Nhieu Loc-Thi Nghe watershed, a 33-km2 area located in the central part of Ho Chi Minh City (Vietnam). We demonstrate that this framework finds robust design alternatives with limited computational requirements.
Anna Di Mauro, Unicampania
3:40 PM - 4:00 PM
Over the past decades, technological innovation, globalization, and climate change generated an unprecedented interest towards the impact of human behaviour on resources consumption. Behavioural studies have boosted the attention on water usage in domestic environments as a key element of the urban metabolism, demanding for knowledge about the humans drivers of household water usage. Such information is required to design tailored demand management programs. On this purpose, water consumption information at end-use level (e.g., shower, dishwasher, etc.) can contribute to understanding how and when water is used in residential setting, along with direct implications on estimating future demands, detecting system leaks, identifying consumption patterns, improving users’ awareness, and personalizing costumers’ profiles. Yet, despite technological development and advances in metering systems, only few water end-use consumption data are available to train data driven algorithms for identifying end-use water usage from aggregate data observed at household level. This work presents a case study of data collection of water end-use information in a residential apartment located in Naples (Italy). In order to generate a repository of real end-use consumption, a monitoring system based on Internet of Things (IoT) technology was implemented and installed on the fixtures of a single-family apartment used as pilot site. The IoT system, composed by a flow-meter, a micro-controller and a content management system, is able to automatically detect, collect and store high-resolution water end-use data. The application allowed the generation of over 8 months of disaggregated data, observed at the scale of single end-use fixture. The resulting repository will be released as open data for the scientific community. Furthermore, this work describes the features of the dataset, a preliminary analysis of collected data that exhibits valuable information about users’ consumption behaviours, the potentiality of end-use information to profile users, and highlights future directions for disaggregation via data-driven techniques.
Mario Roidt, Technische UniversitÃ¤t Berlin, Einstein Center Digital Future, Berlin, Dorsch International Consultants GmbH
4:00 PM - 4:20 PM
In 2018, a total 435 TWh of electricity was traded among European countries, creating an active network of virtual water trade. Virtual water of international electricity trade can support multi-scale water resources management strategies, similar to how virtual water embedded in food has revealed the water resources impacts of the food industry. Estimates of electricity-related virtual water transfers are reported in the literature, yet with high uncertainties, primarily due to insufficient or limited data. Recently, ENTSO-E – the European Network of Transmission System Operators – publicly released a large amount of electricity generation, load, and trade data at a high temporal scale (15 min or 1 hour) and the EU’s Joint Research Centre (JRC) released a database with power plant-scale information and cooling water requirements. While such new data open opportunities for better calculation of electricity-related virtual water modelling, several challenges still limit a detailed and Europe-wide analysis of virtual water due to electricity generation. For instance, electricity data with high temporal scale are incomplete, while more reliable electricity data are only available on an annual basis. Also, while electricity generation can be calculated at the sub basin scale, electricity trade can only be calculated at the country scale. The objectives of this work are two-fold. First, we showcase the potential of high-resolution data to support the calculation of electricity’s virtual water, by analysing the most reliable data at three scales: (i) we analyse energy generation and trade and their variation on a high temporal scale; (ii) we calculate virtual water at sub-basin scale, and (iii) we show virtual water trade at the country scale. The current data availability, however, does not allow combining these scales for more reliable results. Second, we identify data requirements and priorities to improve the existing datasets and data reporting regulations to reliably calculate virtual water trade in Europe.
Joseph Kasprzyk, University of Colorado Boulder, USA
4:20 PM - 4:40 PM
The Colorado River spans the United States and Mexico and is an important cultural, economic, and natural resource for 35 - 40 million people. Its complex operating policy is based on the “Law of the River,” which has evolved since the Colorado River Compact in 1922. Operational guidelines were negotiated in 2007 to address shortage reductions and coordinated operations of Lakes Powell and Mead. These interim guidelines – in effect until 2026 – were ultimately agreed on after manually exploring hundreds of alternatives. The Colorado River Basin’s projected water delivery reliability has continued to degrade since 2007, primarily due to a persistent drought causing a lower supply. The magnitude of the future supply-demand imbalance is challenging to predict since the most likely realizations of future water demand and hydrology are unknown, nor are the uncertainties quantifiable. Hence, these future conditions can be described as deeply uncertain. Negotiations for the new 2026 guidelines will need to consider deep uncertainty when searching for and evaluating operational alternatives. This research explores innovative planning approaches that are appropriate for conditions of deep uncertainty and planning in multi-reservoir systems. A Multi-Objective Evolutionary Algorithm (MOEA) is coupled with the Colorado River Simulation System (CRSS) model, built in RiverWare, to generate thousands of new operating policies for Lake Mead. The MOEA-generated policies are then re-simulated across many future water supply and water demand scenarios to test each policy’s performance across a wide range of plausible future conditions. Multiple robust operating policies were identified through applying a satisficing analysis to the set of MOEA-generated policies. The operational similarities between the identified robust policies may shed light on how Lake Mead's operating policy could be formulated to be more robust to a wide range of future hydrologic and water demand conditions. The presentation will use the Colorado River Basin test case to demonstrate the powerful coupling of MOEAs and RiverWare, appropriate for planning in many worldwide complex river basins. Ongoing work includes incorporating Lake Powell into the framework.