Using AI sound recognition for smarter water discharge
Prototype developed that automatically distinguishes between clean water and contaminated water using acoustic signals. Foundation established for cost-efficient, autonomous control of discharge sluices.
Format
3 Phased
Work Type
Field Work & Remote
Focus
Infrastructure
Waterschap Drents Overijsselse Delta manages an extensive network of waterways, pumping stations and discharge structures across the northern Netherlands. With growing pressure on water quality and efficient discharge management, the organisation was exploring smarter, more automated solutions that could work within the existing infrastructure rather than replacing it.
The core question was both technical and strategic: could a sound-based system be reliable enough to operate sluices autonomously, without any significant mechanical modifications to pipes or structures? That kind of expertise is rarely found in-house at a water authority.
The challenge
The organisation needed to determine whether AI-driven sound recognition could reliably distinguish between clean water and water carrying silt or sand, and use those signals to control sluices automatically. The system had to operate around the clock under variable acoustic conditions, with no invasive changes to the existing infrastructure.
Blackbear's role
The assignment called for a specialist with an uncommon combination of Machine Learning expertise and hands-on experience in operational or embedded environments. Blackbear helped sharpen the scope, translated the technical and operational requirements into a clear brief, and matched the right professional to the project, someone equally capable of handling the AI challenge and the realities of a field-based technical environment. The phased structure, running from feasibility through to a working prototype, kept the engagement manageable and results-focused throughout.
Result
The project delivered a working prototype that analyses acoustic signals within discharge processes and automatically controls sluices based on detected water composition. Detection accuracy and viscosity estimation were both validated on-site. The system provides a concrete foundation for broader rollout and answers the fundamental question of whether sound-based detection is technically and economically viable at sewage and wastewater scale.
Alien @ WDO Delta
Adviseur Verandermanagement
