The project ResiWater aims to better prepare water utilities to crisis management by restoring their system back to normal situation as fast as possible. The network resilience is enhanced by developing a new secure sensor network and model-based assessments for mitigating water distribution system failures. <br />Keywords: <br />Critical Infrastructure; Disaster; Cyber and terrorist attack; Security; Resilience; Sensors and sensor network; Simulation and resilience training; Key performance indicator.
Water Supply Systems (WDSs) are critical infrastructures that may fail to distribute drinking water of adequate quantity and quality. Given their distributed topology and interconnectivity with other infrastructure systems, WDSs are exposed to a variety of risks including cyber and terrorist attacks, natural disasters and widespread technical failures. This research project considers three case studies representing important scenarios of civil security. It concerns partial or complete failure of water distribution, water quality deterioration and cascade effects. <br /> <br />The detection of faults and the capacity to return quickly to a normal state after failures and interruption of services are essential for water utilities. The ResiWater project aims to improve the following three aspects for better network security and enhanced resilience: prevention, surveillance and response of water distribution systems facing the major threats. <br /> <br />The good prevention and level of required preparation go as far as system analysis from a hydraulic point of view, with a focus on vulnerability assessment, system robustness and resilience. The monitoring includes the continuous recording of the real system state with a network of high-performance sensors for fast detection of water quality deterioration or system breakdown. To serve in case of connectivity loss, modelling tools should be more robust for solving numerically difficult problems (or poorly conditioned) than those of the normal state. Knowing the degree of uncertainty of the model results, which depends on the level and on the quality of the instrumentation, the modelling error but also the parameter uncertainty, is a strong request for the operator decision-making.
In the ResiWater project, the following solutions have been investigated: innovative water quality and flow sensors and secure sensor networks, enhanced self-learning monitoring and event detection methods, robust hydraulic simulation methods and training simulator, tools for assessing WDS vulnerability, resilience and robustness and decision support for design.
The extended usage of low-cost flowmeters and spectrometers have been studied. Additionally, a broad-spectrum biosensor prototype has been performed that reacts immediately to toxic substances in water. Innovative and more sophisticated sensor systems have been also investigated in terms of their potential to increase the resilience of WDS.
Enhanced self-learning monitoring and event detection methods, which can be applied to all listening sensors in the network have been developed. Thanks to spatial and temporal aggregation, the solution is able to recognize and adapt to changing operation modes (short-term and long-term) of the WDS, while reducing the false positive rate.
A robust pressure-driven solution for hydraulic modelling was achieved that was tested to large disconnected components. An Applied Mathematics PhD thesis has focused on parameter uncertainty propagation and reduced-order modelling. The calculation engine for robust modelling has been integrated in a training simulator for learning new avoidance strategy in case of extreme events.
Tools for the assessment of global vulnerability and resilience have been developed. The tools have been evaluated by means of real-world use cases. Additionally, experiments were conducted at TZW (Dresden, Germany) on the small-scale test network and also on the private water network at CEA/DAM (Gramat, France).
Highlights and final achievements are:
- A new vulnerability resilience framework has been proposed for ranking the crisis case studies defined by the project end users. As a result, several resilience key performance indicators (rKPIs) are suggested for assessing the absorptive and adaptive resiliencies.
- Experience was gained for designing a secure and integrated sensor network. The prototype AquaBioTox has been improved for achieving a toxicity biosensor that is automatized and long-term stable. Moreover, the potential of a new generation of online spectroscopic sensors has been studied. Finally, a new protocol has been developed for determining the wet area on a flow meter site.
- A self-learning software solution has been delivered for alarm generation that is based on the dynamic and incremental principal component analysis. A multi-step approach that comprises temporal segmentation and spatial clustering is proposed for monitoring water quality within the distribution network.
- A training simulator has been developed for learning from a robust pressure-driven solution in case of extreme events.
- The uncertainty quantification for the water quantity and quality model outputs is enabled thanks to polynomial chaos and reduced-order modelling.
- A crisis cost analysis was combined with a survey that was conducted to assess the consumers' willingness to pay for two projected resilience programs on a cyberattack on one end-user network.
- The costs valuation was constructed on real crisis cases from the three end users.
- A survey was made to estimate the consumer willingness to pay considering two resilience programs.
- Recommendations for extending the influence of laws and regulations (European and national) on crisis management. considering the four phases of crisis (prevention, response, recovery and learning phase).
- Cyber risk analysis of IT Infrastructure was conducted for two of the Water Utilities.
The ResiWater project partners have devised new ways to manage networks to minimize the impacts of significant contamination or failure of water distribution. Through various crisis scenarios, they designed an innovative sensor network, a method for detecting abnormal events, a training simulator and key performance indicators. Thus, the most relevant means to be implemented are identified to enable managers to react more quickly and effectively. In addition, recommendations to extend the scope of laws and regulations, cyber risk analysis and willingness to pay for more resilience, cost-benefit analyses have been developed to propose more resilient but also sustainable and economical solutions.
The ResiWater project has therefore made it possible to develop solutions and tools to increase resilience following severe dysfunctions to be treated. The solutions envisaged are based on the current physical structure of the network and do not call it into question. ResiWater tools and performance in terms of resilience and security could be improved by considering network reconfiguration and the location of new hydraulic equipment. The solution sought could aim at greater structural adaptability of the network to cope with more frequent natural disasters and anthropogenic risks. This resilience by design will have to be complemented by the implementation of an improved and adaptive control system such as remote control of actuators in the network.
It also seems interesting to be able to consider an integrated and holistic approach to cascade effects, artificial intelligence and big data, as well as virtual reality, augmented reality and serious game. Finally, sustainable development is a high priority for future generations.
This industrial research project has resulted in the production of more than 10 papers in international and national peer-reviewed journals and more than 24 presentations at international conferences.
The software tools of the partners Irstea and 3S Consult were supplemented with robust pressure-driven hydraulic models and a criticality calculation. In addition, 3S Consult has developed a resilience training simulator and Irstea has developed the hydraulic state uncertainty calculation. Finally, in order to demonstrate case studies, the Porteau software was interfaced to the DEMOCRITE software platform (project #ANR-13-SECU-0007-01; www.anr-democrite.fr/), which makes it possible to represent vulnerability maps.
Partners IOSB and IGB will transfer the automated biosensor prototype (AquabioTox) to other water utilities. There is a strong demand from water utilities for automated and robust biosensors. The measuring system will be marketed by a sensor company.
ResiWater's partners, particularly those working on sensor development, sensor network design, measurement and anomaly detection (TZW, Fraunhofer IOSB, Fraunhofer IGB, ENGEES/ICUBE) will aim to register patents on their innovative sensor systems. All partners will endeavor to use the results obtained by the project for additional publications and as background knowledge in subsequent research projects.
Fundamental research work in Applied Mathematics has been carried out for the modelling of decomposed systems under conditions of insufficient pressure with particular consideration given to the operation of pumps and control devices. In addition, the uncertainty about the model results was quantified.
Water is a fundamental resource for human and economical welfare and modern society depends on complex, interconnected infrastructures to provide safe water to consumers. Water Distribution Systems (WDSs) are constantly exposed to deliberate or accidental contaminations or may undergo a partially or full system collapse. This could be caused by terrorist attacks, cascade effects, major technical accidents or natural disasters.
The project ResiWater aims to develop tools to prepare water utilities for crisis management and enhance their resilience with regards to three specific case studies: collapse of WDS, water quality deterioration and cascade effects between water, energy and IT infrastructures. For the realization five main steps were defined: specification of critical case studies, design of integrated and secure sensor networks, development of a self-learning module for abnormal event detection, development of robust hydraulic and water quality simulation tools for modelling of extreme events, and decision support tools for improving resilience of WDSs.
The French-German cooperative research project consists of end users (BWB in Germany, CUS and VEDIF in France), technical and socio-economic research institutions (Fraunhofer IOSB, Fraunhofer IGB, TZW, CEA, Irstea, ENGEES) and industrial partners on French and German sides (VEDIF, 3S Consult and Pretherm).
It ideally combines top-level research with the practical needs of water supply utilities. Among the main expected results, two simulation software tools are planned to be extended for crisis management and preparedness: those of partners Irstea and 3S Consult. The three water utilities will benefit from the outputs, training and decision support tools.
Monsieur Olivier PILLER (Irstea Institut national de recherche en sciences et technologies pour l'environnement et l'agriculture)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Irstea Irstea Institut national de recherche en sciences et technologies pour l'environnement et l'agriculture
ENGEES ICUBE et GESTE Ecole nationale du génie de l'eau et de l'environnement de Strasbourg UMR GESTE et UMR ICube
CUS Service des Eaux de la Communauté Urbaine de Strasbourg
VEOLIA EAU D'ILE DE FRANCE SNC
CEA Commissariat à l'énergie atomique et aux énergies alternatives
Help of the ANR 991,577 euros
Beginning and duration of the scientific project: March 2015 - 36 Months