Pipe-Failure Forensic (PFF)

Idea

Aging water transmission networks globally suffer from systemic Non-Revenue Water (NRW) losses, frequently exceeding 60% in critically affected regions like Bulgaria. The core issue isn't massive, catastrophic bursts—which are easily visible—but rather the chronic, mid-sized subsurface leaks. These hidden leaks bleed millions of liters over time, slowly destabilizing surrounding infrastructure, yet utilities lack a scalable way to pinpoint them across hundreds of kilometers of extra-urban terrain. Pipe-Failure Forensic (PFF) provides a remote diagnostic model to solve this. Instead of acoustic pinpointing, PFF hunts for the physical footprints of active, prolonged leaks. By algorithmically correlating satellite-derived signs of soil washout, structural asphalt heave, and anomalous moisture retention in vegetation, PFF generates an "Active Leak Confidence Map." This empowers utilities to locate their hardest-to-find transmission failures, transitioning from reactive patching to data-driven network stabilization.

EU space technologies

Detecting mid-sized subsurface leaks requires cross-verifying subtle environmental anomalies. PFF achieves this by fusing the Copernicus constellation into a cumulative evidence chain:

 -Hybrid InSAR Structural Analysis (EGMS + Custom Sentinel-1): We utilize a two-tiered approach to detect ground deformation (subsidence/heave) caused by soil washout. We first leverage Copernicus EGMS for rapid, pre-processed regional screening. Crucially, in remote extra-urban corridors where EGMS data points are sparse or missing, we deploy Custom Sentinel-1 (SLC) Interferometric processing. This targeted fallback guarantees we can generate high-resolution structural footprints of leaks exactly where we need them, eliminating data blind spots.

 -Sentinel-1 (SAR C-Band): We monitor anomalous soil dielectric constants to detect persistent subsurface moisture accumulation along infrastructure corridors, independent of cloud cover.

 -Sentinel-2 (Multispectral): We utilize the Normalized Difference Water Index (NDWI) and Red-Edge proxies. Specifically, we look for "luxury vegetation"—unnatural, localized spurts of lush plant growth or moisture retention that persist even during regional heatwaves or dry spells.The Value: By overlapping structural instability (EGMS) with unseasonal moisture (Sentinel), PFF provides utilities with a robust, high-confidence target map that highlights exactly where underground infrastructure is actively failing.

EU Space for Water

Addressing Challenge 1: Securing equitable and efficient access to water, PFF tackles the root cause of hydraulic pressure loss. Massive NRW rates represent not only a critical loss of a vital resource but also significant exposure to EU regulatory penalties for member states. PFF provides a globally scalable technical prerequisite for network stabilization. By identifying the highest-volume active leaks in transmission backbones, we enable utilities to restore network pressure efficiently. This directly mitigates the need for localized water rationing and ensures that existing water reserves are equitably distributed and safeguarded.

Team

Vilian Badyokova, Project Lead & Geomatics Expert: Geoinformatics specialist with 4+ years applied experience in EU engineering geology and infrastructure stability projects- ensuring PFF bridges the gap between complex remote sensing models and actionable, on-the-ground utility interventions.

Vladimir Toshev, Lead Developer & ML Specialist : Computer Science expert and Data Scientist at the GATE Institute. Specializes in Machine Learning, advanced GIS integration, and Digital Twin spatial modeling, responsible for automating PFF's multi-sensor environmental noise filters.

Assoc. prof. Tanya Vasileva, Scientific Advisor: Associate Professor of Hydrogeology at the Bulgarian Academy of Sciences (BAS). Expert in GIS-based hydrological mapping, providing world-class scientific validation for PFF’s subsurface washout and groundwater correlation models.

Github Repo

https://github.com/vladi-tech/PFF---Pipe-Failure-Forensic/tree/main