Leakage detection and localization in water distribution systems: A model invalidation approach

Abstract

Model-based methodologies can assist in addressing the challenging problem of leakage detection and localization in water distribution systems. However, this is not trivial due to inherent non-linearities and parametric uncertainties. Most importantly, due to the small number of available sensor measurements compared to the number of system states, the inverse problem for estimating leakages is highly under-determined. In this work, the utilization of a priori available information about the system is proposed to formulate a hydraulic model of the system in its non-linear form in which uncertainties are modeled by intervals defined by a lower and upper bound. A novel optimization-based methodology then utilizes pressure and flow measurements to perform leakage detection through model-invalidation. A modification of the optimization algorithm is activated in the case of a detection to refine possible leak locations and retain only the ones that can be explained by the interval model and available measurements from multiple time-steps. The proposed methodology is demonstrated on a benchmark network and evaluated using a leakage diagnosis benchmark dataset.

Robust Data-Driven Leak Localization in Water Distribution Networks Using Pressure Measurements and Topological Information

“…Finally, the clustering depicted in Figure 3b is defined by Equation (17), which is based on the common resistance path explained in Section 3. These two last clusters that only require topological information could be used in the leak localization method at the cluster level defined in Equation (10). It is important to emphasize that the clustering based on the resistance common path, proposed in this paper and depicted in Figure 3b, resembles the clustering based on the actual leak effect in the network (given by the model) depicted in Figure 3c much more than the clustering based in the hydraulic distance depicted in Figure 3a.…”

“…where θ is a normalization factor. Then, θ j can be interpreted as a likelihood index, and the leak localization at cluster level defined in (10) can be formulated at node level as:…”

“…[8] uses leak signature space, Ref. [9] analyzes the sensitivity matrix and residuals, and [10] uses pressure and flow measurements to perform leakage detection through model-invalidation. On the other hand, data-driven methods analyze the monitoring data, combining tools such as artificial intelligence (e.g., classifiers [11][12][13][14] or artificial neural networks [15,16]).…”

Data-driven and model-based framework for smart water grid anomaly detection and localization

“…To consider the uncertainty of model parameters, such as demand uncertainty, sensor noise, and leak sizes, Soldevila et al (2017) reported a method for leak localization based on the Bayesian classifier, which was also tested on the same academic dataset as Ponce et al (2014). More recently, Vrachimis et al (2021) presented leakage detection and localization via model-invalidation, by which the uncertainties of the model parameters including pipe roughness and nodal demand are emulated by the prescribed upper and lower bounds. A 'health' or none-leak condition can be established by the model simulations using the parameters within the bounds.…”

The Role of Prior Assumptions for Leak Localisation in Water Networks with Uncertainties

“…The role of prior assumptions for leak localisation in water networks with uncertainties 3 with the sensitivities of the pressure measurements to the leak flow from every possible leak location (Casillas et al, 2013). Scenario falsification (Goulet et al, 2013) and model invalidation (Vrachimis et al, 2021) deal with uncertainty by assuming uncertainty bounds, but both methods assume the existence of only a single leak.…”