HS5.11

Water utilities around the globe are facing an extraordinary demand for the secure supply of potable water as a result of population growth, urbanisation and climate change. New knowledge, technologies and systems-based approaches are urgently needed to adaptively optimise resource capacity, operations and assets utilisation during a time of escalating environmental, regulatory and financial pressures.
This presentation summarises fundamental mathematical and engineering challenges for the design, optimisation and control of next generation water supply networks. These networks dynamically change connectivity (topology), hydraulic conditions and operational objectives. The design and control of dynamically adaptive water supply networks aims to improve pressure management, resilience, efficiency, incident management and sustainability.
The current ability of complex water networks to dynamically adapt their connectivity, operational conditions and application objectives is extremely limited. Water supply networks are operated as disjointed (or loosely coupled) sub-systems that have evolved over many years. The operational practice of sub-dividing water supply networks into small discrete areas, District Metered Areas (DMAs), has been successfully implemented by the UK water industry to reduce leakage in excess of 30% in the last 25 years. A DMA has a fixed network topology with permanent boundaries, typically a single inlet and it includes between 1,000 and 3,000 customer connections. By closing boundary valves to form small metered areas, the natural redundancy of connectivity and supply within large looped networks is severely reduced; thus affecting operational resilience, water quality and energy losses. Consequently, the implementation of DMAs has introduced operational constraints that affect both consumers and utilities. Furthermore, these constraints are beginning to inflict financial penalties upon water utilities through recently introduced performance indicators.
The dynamically adaptive sectorisation and configurability of water supply networks that we have pioneered combines the benefits of the traditional DMA approach for leakage management with the advantages of substantially improved resilience and considerably enhanced management of pressure, energy, failure incidents and water quality. This operational method includes the replacement of a subset of kept-shut boundary and control valves with self-powered network controllers with varying modulation functions. The controllers modify the network connectivity and continuously monitor and control the hydrodynamic conditions. To enable this new approach, we have been developing novel monitoring, modelling and control methods and technologies. We have been extensively evaluating these in operational networks. 
In this presentation, we summarise design-for-control strategies to improve the pressure management and resilience of sectorized water distribution networks (WDN). We formulate the mathematical optimization problems and describe solution methods for the resulting large-scale non-linear (NLP) and multi-objective mixed-integer non-linear programs. We also discuss analytical and engineering challenges for the scalable implementation of these methods. 

How to cite: Stoianov, I., Pecci, F., and Ulusoy, A.-J.: Water Supply Networks with Dynamically Adaptive Connectivity and Hydraulic Conditions: Design and Control, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-8049, https://doi.org/10.5194/egusphere-egu22-8049, 2022.

In recent years, researchers have developed intrusion detection tools to improve the security of Water Systems in the presence of novel and sophisticated cyber-attacks. Nevertheless, such tools lack prescriptive analytics conceived to determine the best course—that is, how to control a system undergoing an attack. Here, we fill in this gap by presenting a numerical modelling framework based on the coupling of DHALSIM with a Deep Q-Learning agent.  The former is a simulation environment providing a high-fidelity representation of both hydraulic processes and ICT devices, while the latter is an optimal control algorithm tasked with the problem of curbing the impact of cyber-physical attacks by re-operating key hydraulic devices, such as pumps and valves. Their integration is made feasible thanks to a new feature implemented in DHALSIM—the stepwise simulation—allowed by the addition of a new simulator wrapper, namely Epynet. The evaluation phase is run considering the system in normal operating conditions and undergoing cyber-attacks. Our results show a good behaviour in terms of Demand Satisfaction Ratio. In particular, the control agent manages to satisfy the customers demand, without overflowing the tanks in the system.  To our knowledge, these results are the first in the unexplored area of attack recovery in water distribution systems.

How to cite: Murillo, A., Salaorni, D., Taormina, R., and Galelli, S.: Optimal real-time control of water distribution systems undergoing cyber-physical attacks, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-11444, https://doi.org/10.5194/egusphere-egu22-11444, 2022.

Due to population growth, urbanization, and climate change, it is nowadays necessary to go for an ever-more adequate management of water resource in order to satisfy current and future demand. In this regard, an accurate estimation of water consumption is helpful for the implementation of strategies aimed at developing efficient water systems [1]–[2]. Strategic assessments are often carried out with the support of predictive or descriptive demand models (e.g. [3]). However, when no observed data are available, these models have to be parameterized according to predefined parameters distributions (e.g. probability distribution of duration, volume, flow rate of each end use), but the availability of this kind of information derived by field observation is rather limited.

The current study aimed at exploring the characteristics of water consumption at nine households – different in terms of occupancy rate and end-uses – located north of Amsterdam (The Netherlands), in which smart monitoring of water consumption at 1-s temporal resolution with 0.1 L/pulse accuracy started in 2019. The aggregate water consumption observed at each household was automatically disaggregated into individual end-use events, which were then manually classified by expert analysts based on the responses of water use questionnaires subjected to household occupants. Specifically, more than 64,000 events registered over about 445 days of monitoring were labelled in five categories of indoor water use: dishwasher, washing machine, faucets, shower/bathtub, and toilet.

Statistical analyses were then conducted for each household in order to evaluate: i) the daily per capita end-use water consumption; ii) the end-use parameter values (i.e., duration, volume, flow rate, per capita daily frequency) and their main statistical properties such as mean, variance, and probability distributions. On the one hand, the results confirmed that, on average, the largest components of the daily residential water consumption were related to the use of showers/bathtubs and toilets (43 and 30 L/person/day, respectively), followed by washing machines, faucets, and dishwashers (16, 14, and 3 L/person/day). On the other hand, the largest average volumes per event were tied to showers/bathtubs and washing machines (64 and 63 L/use), while the highest average frequency of use was observed for faucets and toilets (14 and 4 uses/person/day). Moreover, different parameter distributions were estimated, depending on the end-use and the parameter considered.

References

[1]    K. Aksela and M. Aksela. “Demand estimation with automated meter reading in a distribution network”, Journal of Water Resources Planning and Management, vol. 137, no. 5, September 2011, pp. 456–467.

[2]    S. H. A. Koop, S. H. P. Clevers, E. J. M. Blokker, and S. Brouwer, S. "Public attitudes towards Digital Water Meters for households" Sustainability, vol. 13, no. 11, June 2021, 6440.

[3]    E. J. M. Blokker, J. H. G. Vreeburg, and J. C. van Dijk, “Simulating residential water demand with a stochastic end-use model”, Journal of Water Resources Planning and Management, vol. 136, no. 1, January 2010, pp. 19–26.

How to cite: Mazzoni, F., Blokker, M., Alvisi, S., and Franchini, M.: Evaluating residential water consumption at high spatio-temporal level of detail: a Dutch case study, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-5700, https://doi.org/10.5194/egusphere-egu22-5700, 2022.

As could be seen in recent years, the impact of climate change is already detectable in water demand patterns and results in new challenges for the water supply sector. Demand peaks caused by changing climate conditions such as longer dry periods force water suppliers to a more efficient control and management of their assets and water resources to avert supply shortages. Especially demand peaks of multiple hours during the day or persisting demand peaks of several days and weeks threaten the supply demand-balance. By utilizing accurate forecasts of the expected water demand, suppliers are enabled to better prepare their assets for such extreme conditions.

To adapt to the consequences of changing hydro-climatic and demand conditions, this research proposes a water demand forecasting model to predict such extreme demand conditions caused by climate change for the short- to mid-term range. Here, a special emphasis is put on modelling the impact of weather variables on the water consumption caused by climate change. Those effects are complex, non-linear and multidimensional in nature and therefore challenging to model. Focusing on the practical usage, the forecasting model is appropriate for real-time application providing accurate forecasts coupled with a high interpretability. This allows the quantification of the ongoing effects of climate change and enables a better consideration of the underlying uncertainty.

Our case study uses real data on district level from two regions in West and Central Germany. To appropriately account for the practical need of varying forecast schemes, historical demand and weather data are used at quarter-hourly, hourly as well as daily resolution.

Multiple linear, non-linear and stacked models tailored to the forecasting purpose and the varying horizons are implemented with a clear focus on interpretability and forecasting accuracy. To model the underlying uncertainty, complete probablistic forecasts are proposed. Model assessment takes place by utilizing appropriate metrics as the MAE, CRPS or energy score.

How to cite: Johnen, G., Kley-Holsteg, J., Niemann, A., and Ziel, F.: Probabilistic water demand forecasting focussing on the impact of climate change and the quantification of uncertainties in the short- and mid-term, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-10025, https://doi.org/10.5194/egusphere-egu22-10025, 2022.

Digital water meters are increasingly installed in water distribution networks providing detailed information about the water consumption in households at a high temporal resolution (e.g., ranging from seconds to daily readings). While the benefit on household scale is well described in literature (e.g., scarcity billing, awareness raising, leakage detection in domestic installations), recent research is also investigating the potential of digital water meters for an accurate fault management on network scale. In this context, water losses represent a major challenges for the operation of water distribution networks (WDNs), and a timely detection and localisation of water leakages is of greatest interest to reduce these losses. Especially model-based techniques require accurate nodal demands for the numerical simulation of the hydraulic states, which can be obtained for example by using high resolution consumption data.

Therefore, the aim of this work is to first investigate the influence of different temporal resolution of household water consumption data and to define an optimal temporal resolution for the detection and localisation of water leakages. However, power supply (e.g., transmission interval), communication technology (e.g., packet losses), and urban population (e.g., consumer agreement to digital water meters) influence the temporal and spatial quality of data received in a real-word implementation and may differ from the optimal performance. Therefore, different methods are tested to overcome the data gaps caused by data transmission and availability uncertainties. As case study, a real WDN from a pilot project in the city of Klagenfurt is used which is extended by artificial water demand series (temporal resolution varies between 1 min and 24 h) and water leakages. Following, performance of leakage detection (data-based approach) and localisation (model-based approach) in combination with machine learning techniques is evaluated by using detection time and distance between leakage and identified location as selected indicators.

The first results showed that the temporal resolution of consumption data influences the applicable methods for an efficient leakage detection and localisation. For example, water consumption data with a temporal resolution of 15 min allow an accurate mapping of consumption fluctuations, therefore the difference between inflow and measured values is very well suited to identify leakages. In contrast, using the same technique for 24 h consumption data (e.g., difference from inflow and daily mean value), the morning and evening peak would also be indicated as a possible leakage and thus requiring different approaches.

How to cite: Oberascher, M., Halm, A., Ullrich, T., and Sitzenfrei, R.: Analysing the influence of different temporal resolutions of water consumption data for leakage detection and localisation, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-7399, https://doi.org/10.5194/egusphere-egu22-7399, 2022.

The reliability of water distribution networks (WDN) can be defined as the ability of the system to meet water demands under both normal and abnormal conditions (Bao and Mays, 1990). The measure of reliability is influenced by many factors: possible failure of one or more components, unusual high demand, connectivity of the network, poor water quality, etc. Although reliability of WDN was originally defined by Goulter (1987) and Walters (1988), measuring reliability is an open challenge for researchers. Consequently, there is no established measure of WDN reliability.

In recent years, many studies have attempted to mathematically define the reliability of water distribution networks and several synthetic indices are provided as proxy of reliability measures. The objective of this paper is to investigate the suitability of two of these metrics: one based on surplus of power head and other   on Shannon’s informational entropy for the assessment of reliability of partitioned water distribution networks (WDNs). The creation of permanent DMAs involves permanently an alteration of the WDN by closing multiple lines at the same time, and, therefore, it is a more severe test than those commonly used in the scientific literature where a little or no disruption occurs to the operation of the WDN (e.g., segment isolation or demand amplification). In addition, the two metrics were compared with other known indicators of hydraulic performance to determine which of them is better suited to evaluate the reliability of water network partitioning. For this purpose, a medium-sized water distribution network in South of Italy was used as case study and the hydraulic simulations were performed with a pressure-driven approach using EPANET2.2 software.

References

Bao, Y., Mays, L.W. (1990). "Model for water distribution system reliability",  J. Hydrual. Eng., 116, 1119-1137

Greco, R., Di Nardo, A., Santonastaso, G.F (2012). "Resilience and entropy as indices of robustness of water distribution networks", J. Hydroinformatics, 14 (3), 761–771.

Gouher, I.C. (1987). "Current and future use of system analysis in water distribution network design", Civ. Engrg. Sys., 4(4), 175-184.

Walters, G. (1988). "Optimal design of pipe networks: a review." Proc.,1st Int. Conf. on Compo and Water Resour., Vol. 2, Computational Mechanics Publications, Southampton, U.K., 21-31.

How to cite: Santonastaso, G. F., Di Nardo, A., and Greco, R.: Reliability metrics for permanent water network partitioning, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-6635, https://doi.org/10.5194/egusphere-egu22-6635, 2022.

Water utilities tackle various problems in planning and operating their systems with complex and computationally expensive hydraulic models, i.e., maximizing system resilience, fault isolation, risk assessment, optimal pump scheduling, or water loss reduction via pressure management. To meet limited computational budgets, engineers employ less resource-intensive surrogate models. Current surrogate models based on artificial neural networks deliver similar accuracies as hydraulic models with lower computational costs. However, they require retraining when applied to an unknown water distribution system, which increases their computational load and limits their general applicability. Recent advancements in graph-based machine learning address these limitations. Graph neural networks (GNNs) naturally connect with the network elements (e.g., pipes and valves with edges, junctions, and tanks with vertices)  of water distribution systems, proving themselves to be a promising candidate for surrogate modeling. Once trained on a specific network to be a surrogate model, GNNs possess inductive biases that allow transferability to an unseen topology. In this work, we adopted a demand-driven simulation of a water distribution system in a graph machine learning setting. We built a synthetic dataset of demand-driven simulation with EPANET, founded on the example of real-world water distribution systems, and trained an attention-based GNN to emulate the hydraulic simulator. The accuracy was evaluated inductively on an unseen larger-sized water distribution network. We observed that the model showed promising transferability results to a larger network without the need for additional re-training on the unseen topology.

How to cite: Kerimov, B., Tscheikner-Gratl, F., and Steffelbauer, D.: Transferable surrogate models based on inductive biases of graph neural networks for water distribution systems, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-7921, https://doi.org/10.5194/egusphere-egu22-7921, 2022.

In this study, we propose and evaluate a Cellular Automata (CA)-based high-resolution hydrological model for an urban digital water information framework. Pluvial flooding in the extreme events and water balance in the non-rainy seasons are usually simulated by different modeling frameworks hampering holistic understandings of the complex water cycle in the urbanized areas. However, for smart water systems such as digital twins or multiverse, street-resolving, high-fidelity water information is required regardless of types of hydrologic events. To provide seamless urban water information on the digital world such as digital twins, Cellular Automata, a rule-based machine learning technique, is adopted and extended to simulate continuous hydrological variables such as inundation depth, infiltration, soil water content, and evapotranspiration in the complex urbanized domain. A proto-type CA model is implemented in the Oncheon-Cheon catchment in Busan, South Korea, which is highly urbanized and vulnerable to pluvial flooding. In the presentation, we discuss advances and challenges in machine learning-based integrated urban water modeling.

How to cite: Noh, S. J., Lee, E., Choi, H., Lee, G., and Kim, S.: Cellular Automata-based high-resolution hydrological modeling for urban digital water information, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-11531, https://doi.org/10.5194/egusphere-egu22-11531, 2022.