Summary: Poster presentacion
Introduction
Water scarcity is a pressing global issue impacting between 2 and 3 billion people
due to unsustainable consumption and development practices (Corazza, 2023).
Italy, in particular, is notable for its high-water consumption, with an annual extraction
exceeding 30 billion cubic meters and a per capita consumption of 220 liters per day,
which is significantly higher than the European average of 123 liters (EURISPES,
2023). The issue is further compounded by the high rate of water loss in distribution
networks. For instance, in regions such as Marche, there is a notable 34.3% loss
(EurEau, 2021), highlighting the urgent need for effective water management
solutions.
Objective
Our research focuses on evaluating GPR as a non-destructive geophysical
technique for the early detection of leaks in underground water distribution networks.
Conventional methods, such as pressure/flow monitoring and acoustic detection,
often face limitations in early leak identification and lack sufficient resolution in urban
environments (Cheung & Lai, 2019; Negm et al., 2023). In contrast, GPR has shown
potential for superior precision and detail in locating and assessing leaks. We aim to
confirm this by evaluating GPR through both laboratory experiments and numerical
models using electromagnetic wave simulations (FDTD) (Gamal et al., 2023).
Methodology
Our research is conducted in three distinct phases:
1. Data Collection
The first phase involves creating a real-world experimental database. We developed
a "laboratory" scale model outdoors to incorporate urban noise effects. This model
includes water pipelines buried in a surrounding medium (sand) with known
electromagnetic properties. Key parameters include the dimensions and distribution
of underground infrastructure and the electromagnetic properties of the terrain. We
collected 2D and 3D GPR data under controlled conditions across three
experimental scenarios:
Scenario A: Dry soil with three pipes of different diameters, no leakage.
Scenario B: Medium moisture soil with three pipes of different diameters, medium
water injection (1 liter per minute for 2 minutes).
Scenario C: Full moisture soil with three pipes of different diameters, higher water
injection (5 liters per minute for 2 minutes, with repetitions).
2. Processing
The second phase involves processing the real data and constructing numerical
simulation models of GPR responses. Real data processing follows conventional
methods to generate interpretable GPR profiles. We identified scenarios with the
highest contrast—dry and fully wet soil—as references for numerical model
construction. The numerical models replicated experimental tests using
electromagnetic wave simulation tools. This involved:
Input Parameters: Extracting geometry and electromagnetic properties from
reference models, including:
Geometric extraction of leak coordinates (wet soil case).
Estimating the dielectric distribution affected by leakage.
Extracting antenna input parameters.
Output Parameters: Running numerical simulations using Reflexw software, which
utilizes finite difference time domain (FDTD) models to simulate electromagnetic
wave propagation. Data was adjusted with Radan7 software to replicate signal
attenuation and interaction with buried objects.
3. Results and Validation
The final phase produced GPR profiles and a GPR cube from synthetic data, which
were compared with experimental results to validate the models.
Results
Qualitative Comparison: We compared GPR images generated from real and
synthetic data, focusing on waveforms over the leakage area in the central
profile. This allowed us to qualitatively assess signal behavior in response to
water leaks in pipelines.
Quantitative Comparison: Pearson correlation coefficients were calculated to
measure the similarity between real and synthetic signals, providing a
quantitative assessment of the model's accuracy.
Discussion
Our analysis includes:
Centerline and GPR Cube: Images were generated from experimental
models with real GPR data and numerical models with synthetic data at
different depths (Z=0.24m, 0.30m). These images provide insights into the
GPR's effectiveness in detecting leaks in both dry and wet soil scenarios.
Models Comparison: Graphs compare results from qualitative analyses of
centerline waveforms and quantitative Pearson correlations. These
comparisons highlight the model's performance in different leakage
scenarios.
Conclusions
Experimental Data: Our study confirms that experimental data from both
controlled and real-world conditions effectively evaluate GPR’s response to
water leaks in buried pipelines.
Synthetic Data: Synthetic data validation shows that numerical models
accurately simulate GPR performance, providing reliable predictions for
various leakage scenarios.
Overall Effectiveness: This research demonstrates that GPR is a highly
effective technique for early leak detection in underground water distribution
networks, offering substantial improvements in infrastructure management
and maintenance.
