Modelling and, as a consequence, decision-making for water distribution networks is ordinarily performed using the deterministic paradigm in which a single set of input conditions gives rise to a single output “truth”.  Reality is not so accommodating, however, and it is readily apparent that significant uncertainties remain in both our knowledge of the condition and the operating constraints of the network.  These uncertainties include variables such as the effective diameter of pipes, characterised by degradation with age and water chemistry, and the quantities of water demanded by consumers.  Traditionally, where these uncertainties have been accommodated in the decision-making process this has been by considering multiple scenarios to model a small number of model states. 

The application of probabilistic modelling for water distribution networks has gained significant traction in the literature in recent years – particularly in the context of decision support systems where stochastic parameter sampling is employed to improve the robustness of the obtained solutions.  Nevertheless, the wide interest in probabilistic modelling has yet to be reflected in the emergence of tools to apply this paradigm. 

This paper introduces VlinderNET a novel tool developed by KWR which seeks to bridge this gap by allowing the user to evaluate and visualize the impact of the manifest uncertainties in the network through the use of probabilistic hydraulic simulation.  VlinderNET permits the specification of complex, cascading Probability Density Functions for the input parameters for a hydraulic simulation.  These PDFs are extensively sampled to produce a wide range of stochastic input variables which are evaluated in a succession of hydraulic simulations which can be parallelized either on a local computer or with cloud support.  The results of the simulations are aggregated and the effects of the uncertain inputs are presented by the tool graphically and spatially both at the component and network level.  The tool further provides an API for third-party applications to integrate the probabilistic paradigm directly into decision support tools in a straightforward and consistent fashion.

How to cite: Morley, M., van Thienen, P., Vertommen, I., and Torello, M.: VlinderNET – a tool for Probabilistic Hydraulic Water Distribution Modelling and Visualization, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-17062, https://doi.org/10.5194/egusphere-egu23-17062, 2023.

A common practice in Dutch and Flemish water utilities is to make masterplans for their network revisions and revise them on a regular basis, approximately every five years. The masterplans represent the ideal redesigns of their networks in terms of company specific objectives and constraints related to existing network infrastructure. The masterplans are used as a guideline when rehabilitating the networks. Among others, one of the objectives is to minimize the residence time, i.e., water age. However, the recurring assessment of water age with traditional methods, within an optimization procedure could take years for convergence for a large network of several thousand nodes. Consequently, the modellers often try to improve the residence time implicitly by minimizing network’s volume via layout optimization and diameter minimization, thus leading to increased velocities in pipes. Recently a graph theory model for estimating water age with satisfying accuracy has been proposed in the literature. The proposed model is estimated to be more than a hundred thousand times faster than the assessment of water age using Epanet, thus enabling the assessment of water age within optimization procedures. This research proposes a novel optimization methodology for simultaneous layout and pipe sizing optimization, employing the proposed graph-based model to explicitly assess the water age objective. To secure the reliability of the optimal solutions the methodology introduces a penalty to limit the size of the branched sections by setting a maximum to the number of customers connected to branched sections. The proposed methodology is applied to a real-world Dutch network. The aim of the research is to compare the optimal designs obtained using implicit (minimizing network’s volume) and new explicit (minimizing maximum water age) approaches.

How to cite: Mitrovic, D., van Laarhoven, K., and Hillebrand, B.: Optimizing a water distribution network design on water age. Comparison between implicit and explicit approaches, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5589, https://doi.org/10.5194/egusphere-egu23-5589, 2023.

Water distribution networks (WDNs) are an important critical infrastructure, but they are increasingly at risk from contamination (WHO, 2014). The causes can be several: chlorination equipment malfunctioning, low pressure, contaminant intrusion in water tank, accidental cross-connection between drinking-water and non-drinking-water, etc... To limit potential threat to public health, it is advisable to install a network of sensors that can monitor water quality in real time and provide information about potential contamination risks. With the proliferation of IoT technologies and low-cost sensors capable of monitoring water quality parameters, it is now possible to implement a real-time monitoring network by overcoming the difficulties associated with biochemical analyses of water samples in a laboratory. Despite the modern technologies, the placement of sensors in the water network is still an open task for researchers. The main sensor placement methodologies use optimization techniques to minimize or maximize either single- or multi-objective functions (Ostfeld et al., 2008), but they require a calibrated model of the network, which is not always available because the calibration process is expensive and time-consuming.

Recently, a novel approach (Santonastaso et al., 2021) based on the use of the topological centrality metric, which does not require hydraulic information and simulations, has been proposed, showing good effectiveness and easy applicability by water utilities to define locations for quality sensors, owing to its simplicity compared to optimization-based approaches.

In this work, different weights such as the length of pipes, the diameter and the water demand were used to improve the performance of the adopted topological approach, as well as to evaluate the impact of the weight, used to compute centrality metrics, in relation to the most used objective functions: number of people exposed to the contaminant; number of detected contamination events; length of contaminated pipes; amount of contaminant consumed by users; detection time of contamination.

References

World Health Organization. (‎2014)‎. Water safety in distribution systems. World Health Organization. https://apps.who.int/iris/handle/10665/204422

Ostfeld A, Uber JG, Salomons E, Berry JW, Hart WE, Phillips CA, Watson JP, Dorini G, Jonkergouw P, Kapelan Z, di Pierro F, Khu ST, Savic D, Eliades D, Polycarpou M, Ghimire SR, Barkdoll BD, Gueli R, Huang JJ, McBean EA, James W, Krause A, Leskovec J, Isovitsch S, Xu J, Guestrin C, VanBriesen J, Small M, Fischbeck P, Preis A, Propato M, Piller O, Trachtman GB, Wu ZY, Walski T (2008) The battle of the water sensor networks (BWSN): a design challenge for engineers and algorithms. J Water Resour Plan Manag 134:556–568. https://doi.org/10.1061/(ASCE)0733-9496(2008)134:6(556)

Santonastaso, G., Di Nardo, A., Creaco, E. et al. Comparison of topological, empirical and optimization-based approaches for locating quality detection points in water distribution networks. Environ Sci Pollut Res 28, 33844–33853 (2021). https://doi.org/10.1007/s11356-020-10519-3

How to cite: Santonastaso, G. F., Di Nardo, A., and Greco, R.: Exploring the performance of topological approach for sensor quality placement in water distribution network, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6650, https://doi.org/10.5194/egusphere-egu23-6650, 2023.

This study presents a data-driven method for detecting pipe bursts in water distribution systems using Long Short-Term Memory (LSTM) neural networks. These types of neural networks are able to process sequential data more effectively than traditional neural networks because they have feedback connections between neurons. The proposed method involves performing one-step ahead predictions about the flow and pressure at different sensor locations in the system, using past time series data along with additional time-related features as inputs. The difference between predictions and actual observations is used to classify bursts and trigger alarms by comparing the errors against a time-varied error threshold. The model is trained using data from burst-free periods in the system. The method was tested using simulated fire hydrant bursts as well as real-world bursts in 8 district metered areas (DMAs) located in the United Kingdom. By harnessing transfer learning, the model can incorporate additional data streams from new sensors, performing well even in data frugal conditions, achieving precision scores of up to 98.1% for the analyzed case studies

How to cite: Glynis, K., Kapelan, Z., Bakker, M., and Taormina, R.: Burst detection in water distribution systems with LSTM, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11190, https://doi.org/10.5194/egusphere-egu23-11190, 2023.

Water distribution network (WDN) is a civil infrastructure for reliable water supply. Among many components in WDN, the pipe delivers the required water demand to users. Pipe bursts, the rupture of pipe wall, cause water losses out of the network and low pressure at the customer’s tap, while its impact varies at different locations. It is important to identify such critical pipes (CPs) and to minimize the failure severity. However, previous CP identification methods are generally complicated and difficult to adopt in practice, highlighting the need for the development of a novel, but practical and simple method. To that end, this study proposes a CP selection approach based on a sensitivity matrix constructed with pipe burst simulation. A sensitivity matrix is constructed by simulating a single pipe failure condition (row) and computing the variation of resulting nodal pressures (column). Then, the summation of the column element’s absolute values is formulated as a new CP index. Finally, the pipe with the maximum CP index value is defined as the most critical pipe. Moreover, this sensitivity matrix can be visualized by the heatmap, which shows the relative influence by using a color density. CP index is presented as the darkest part in the heatmap. The proposed method is demonstrated in two benchmark networks of different layouts, Hanoi and Mays. Despite its simplicity, the proposed method could identify the source pipes which are generally considered to be critical in the engineering sense.

How to cite: Kim, S. and Jung, D.: Identifying the Critical Pipe in Water Distribution Network: Sensitivity Matrix Approach, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13730, https://doi.org/10.5194/egusphere-egu23-13730, 2023.

Due to the climate change, it is expected that there will be longer dry and hot episodes in the future in Central Europe. As a result, temporary water shortages are to be expected in certain parts of Austria. Due to these changes, it is assumed that the future water demand will increase, caused by a change in consumption behavior and the increase of garden irrigations. Furthermore, an increasing population is expected, which may also lead to a shortage, especially in small supply areas.

For water supply companies it is especially important to know the future change in water demand in order to be able to adapt to these changes. Therefore, a water demand forecasting model is derived in this study. In a first step, this study analyses the change in water demand in recent years and the relationship between water demand and climate indices. Furthermore, the change of the future water demand depending on the different climate change scenarios (RCP 2.6, RCP4.5 and RCP8.5) is estimated. For this purpose, different modeling approaches (e.g. multiple linear regression, random forest,…) were tested and a suitable approach is selected. The water demand forecasting model is trained and tested with water demand and weather data reaching back several years. To estimate the future change in water demand, the model is applied to the climate projections and the change between the selected reference period and the two future periods is calculated. The change in demographic development is considered in the last step.

So far, we found that for the selected study site peak water demand will increase by an average between 1.5% and 5.5%, depending on the different climate change scenario for the period 2051-2070 compared with the reference period (2001-2020). It was also determined that demographic development is responsible for the majority of the increase in water demand.

Acknowledgements: The presented research is funded by the Federal Ministry for Agriculture, Forestry, Regions and Water Management of the Republic of Austria

How to cite: Stelzl, A. and Fuchs-Hanusch, D.: Predicting future water demand in Austria due to climate change and demographic development, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14124, https://doi.org/10.5194/egusphere-egu23-14124, 2023.

As could be seen in recent years, ensuring the water supply-demand balance is a topic of increasing concern to supply companies facing the threat of increased demand scenarios resulting from long-term effects due to climate change. Especially demand peaks of multiple hours during the day or persisting demand peaks of several days caused by prolongued dry periods and more heat days throughout the summer force water suppliers to more efficiently control and manage their resources. Being able to take proactive and informed decisions through reliable short-term probabilistic forecasts is therefore crucial in this context.

This research proposes two probabilistic deep learning architectures based on long short-term memory (LSTM) networks to forecast hourly water demand up to 10 days in advance. Both models processes different temporal sequences of data, including past observations of water demand and regressors as well as future regressors with different time lengths. The models encode long-term historic information of the water demand and features, including historic meteorological information, and simultaneously incorporate short-term future information on calender- and weather features using statistically optimized point forecasts (DWD MOSMIX) of the latter. Through implementing the models in an autoregressive manner, the output is fed back into itself at each step and predictions are made conditioned on the previous one to account for correct path dependency between consecutive hours. This way the model produces multi-step-ahead forecasts of variable length by using future information together with the historic context.

In a case study of central Germany, the performance of the proposed deep learning models was compared to a Lasso estimated high-dimensional time series model and a conventional AR(p) model. Results indicate the potential of the proposed approach of using weather forecasts in short-term water demand prediction especially for lead times larger than 24 hours.

How to cite: Johnen, G., Kley-Holsteg, J., and Ziel, F.: Incorporating Weather Forecasts into Short-Term Water Demand Prediction using Probabilistic Deep Learning with Long Short-Term Memory Networks, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5731, https://doi.org/10.5194/egusphere-egu23-5731, 2023.

More than 243000 Hm3/year of drinking water is used in Europe. The popularity of metallic coagulants is due to their comparatively low cost, high availability, and efficiency in removing turbidity and color and sometimes helped by flocculants as polyacrylamides, starch or PolyDadmacs. Water & wastewater treatment drives the market to reach $14.7 Billion by 2024 in this kind of reagents. In this project, a synthetic coagulant will be replaced by a single improve multifunctional organic polymer based on natural plant extracts as a three-step treatment procedure, encompassing coagulation, flocculation and neutralization of pH.

The main drawback of synthetics coagulants is the negative impact on human health and the environment. There are corrosive reagents that contribute undesirable elements such as metals , chlorides or sulphates to drinking water. The sludge generated is known as “alum sludge”, which is the most common residual from water treatment plants. They can cause a deterioration of the pipeline network and produce a waste that finally end in soils or landfills. Moreover, the sludge contains a 7-17% of aluminum concentration, which is mainly used in the agriculture and can be adsorbed, finally, by plants.  Thus, there are two ways to reduce aluminum concentration, one is the efficiency of the coagulation process and reduction of complementary reagents and the other is the substitution of this reagents by a natural one.

Natural coagulants are an alternative of aluminum or iron salts and avoid dissolved aluminum control as required by Directive (UE) 2020/2184, as well as lower costs in complementary reagents. Complementary flocculants as polyacrylamides are limited by the World Health Organization to 0.5 μg/L and are considered harmful to human health. The sludge obtained with the natural coagulant provide organic matter and adsorbs phosphorus so can be used as an active substrate and once saturated, as an agricultural substrate, thus participating in the concept of circular economy.

The main objective of Safe T Water project is to validate a new innovative and environmentally friendly technology in two drinking water treatment plants (DWTP) located in Spain. The first one it is located in Valencia, with a daily production of 48,000 m3 and 750,000 inhabitants and the second one in Madrid, with a daily production of almost 700,000 m3 and supplying 3.2 million inhabitants representing a hard and soft water qualities.

There is a first stage of natural coagulant production in a batch, focused on the start-up and continuous production and manufacturing the product necessary to feed both pilot scale treatment plants. This batch has a production of 4000  Kg/month.

The natural coagulant is evaluated in a 6 m3/h pilot plant flow rate , consisting of a homogenization tank and a coagulation-flocculation and lamellar settling stage followed by a  sand filtration. A Full-scale implementation phase including the validation of the new technology through real drinking water facility is going to reproduce the outcomes.

The comparison between both coagulants must be made under the same conditions, establishing the effectiveness of the natural one as a real and sustainable alternative and this provides to Safe T Water Project a relevant role.

How to cite: Año Soto, M., Almenar, P., and Macián, J.: SAFE T WATER : an eco-sustainable technology to replace aluminum salts with natural coagulants., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6670, https://doi.org/10.5194/egusphere-egu23-6670, 2023.

The Building-Information Modelling (BIM) of hydraulic engineering structures introduces new opportunities for analysis. It is the digital core of the automation of design, construction, and operation processes in water management. Managing the communication of BIM with other hydraulic engineering specific platforms is a relevant current field of research.

Site characteristics, interconnected urban infrastructure and construction methods significantly influence the Infrastructure Engineering and Water Resource Management design process. Digitization and BIM development including its accompanying technologies create the needed prerequisites for hydraulic engineering facilities to be designed and constructed in parallel where any influences from the site influence the design process in real time .

The present study illustrates such a technology development implemented and tested on an example from the practice - a project for a river correction in a settlement. The approach takes advantage of the ability to automate the workflow and communication between Civil 3D and HEC-RAS using Dynamo for Civil 3D. The procedure makes it possible to generate data from design options to fill in a desired sets of parameters. Experiments are being made to create a Deep Learning model -replacement of HEC-RAS for the verification of necessary changes in BIM, as imposed by the general development of the project in the parts roads and sewage system in real time.

The application of Deep Learning techniques requires large volumes of data. The results show that BIM and its automation create prerequisites for using Deep Learning more often. Herein, the possibility of blunders is avoided as such a volume of data would be difficult to obtain manually.

How to cite: Mavrova-Guirguinova, M.: The Digital Era in Hydraulic Engineering Comes with Applications of Artificial Intelligence, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8870, https://doi.org/10.5194/egusphere-egu23-8870, 2023.

Metamodels reproduce the response surface of physics-based models while significantly reducing simulation times. Such techniques are widely employed in water distribution system analysis since they enable the application of computationally expensive methods in designing, controlling, and optimizing water networks. Recent works proposed graph neural networks as candidates for metamodels. These models bear inductive biases as one can draw analogies between links and nodes in the graph with the pipes and junctions. This implies that new metamodels using this approach can be applied to an unseen water network topology without re-training. However, there is no evidence of the transferability properties of those metamodels so far. This work introduces Simplicial Convolutional Networks (SCNs), which offer the potential of developing transferable metamodels that can generalize across different systems.  We test the suitability of SCNs to estimate pipe flowrates and nodal pressures emulating steady-state EPANET simulations. We compare the accuracy of SCN metamodels against graph neural networks on several benchmark water networks available in the literature. Moreover, we show that SCNs are able to generalize better than graph neural networks by evaluating and measuring the performance of the metamodel in an unseen setting.

How to cite: Kerimov, B., Tscheikner-Gratl, F., and Taormina, R.: An EPANET metamodel based on Simplicial Convolutional Networks, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16156, https://doi.org/10.5194/egusphere-egu23-16156, 2023.

The management of urban wastewater systems and the associated modelling of these systems has become indispensable in today's world. In order for these models to represent reality as accurately as possible, a reliable calibration is essential. Water level data is used as a standard, but due to expensive sensors and harsh conditions in the sewer, data can only be collected at a few key points of the system. One novel solution, that has experienced an upswing in recent years, is collecting data using low-cost temperature sensors. Two sensors are needed; one is placed in the stream; the other is placed at the crest of the weir. In the case of dry weather, the sensor measures the air phase, whereas, in the case of Combined Sewer Overflow (CSO), the discharged storm and wastewater is measured. The start and end of a CSO event can be determined via the merging of measured temperature values in both points of the overflow structure. Due to this method, the duration of CSO events in a sewer system can be detected.

In this work, the potential benefits of this novel method for model calibration are assessed. Therefore, autocalibration runs with water level data and fictional temperature data were carried out via OSTRICH for a SWMM model located in Berlin. Furthermore, calibration runs with a different number of measuring sites were performed, to evaluate the amount of necessary measuring sites for a reliable calibration. In order to be able to compare the different approaches, a calibration period of 19 events was first required for the respective datatype. Next, a validation period which consisted of 18 events was carried out and evaluated by the R² of three water level measuring sites for both approaches to ensure comparability. It was revealed that the calibration with duration data based on temperature sensors was able to achieve results as good as the conventional approach using water level data. Due to low spatial distribution of the measuring sites in the model, it could not be finally answered if more measuring sites would yield to even better results. However, already with one measuring site, promising calibration outcomes could be achieved and thus, offers an alternative for water utilities and practitioners.

How to cite: Schütz, P., Gutierrez, O., Busquets, S., Gunkel, M., and Caradot, N.: The use of a low-cost monitoring dataset for sewer model autocalibration, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9159, https://doi.org/10.5194/egusphere-egu23-9159, 2023.

Urban residential water uses entail energy consumption and associated carbon emissions. Reducing residential water uses thus can simultaneously save water and energy and help reduce carbon emissions. However, residential water uses are strongly affected by the choices of household appliances and fixtures, and water use behaviors. In this study, we first conducted a household water use survey in Shanghai, the largest city in China, to understand residential water use behaviors in different seasons. A two-stage stochastic optimization model is developed to optimize water and energy conservation decisions so as to minimize the expected total cost of water and water-related energy uses and maximize carbon emission reduction. Data collected through questionnaire surveys are used to parameterize the optimization model. Water and energy conservation choices are categorized into long- and short-term decisions. The results show that in Shanghai residential water uses has a strong impact on urban carbon emission reduction. Typical temperature range in different seasons strongly affects the effectiveness of short-term conservation actions. The results support a subsidy policy for water-saving appliances that can incentivize citizens in water-saving. Model results are useful for exploring the water-energy-carbon nexus of urban households considering seasonal factors

How to cite: Zhou, J. and Zhu, T.: Optimization of Water-Energy-Carbon Nexus in Urban Residential Water Uses for Shanghai, China, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15728, https://doi.org/10.5194/egusphere-egu23-15728, 2023.