Connect what already exists
GIS, asset records, failures, work orders, pressure and flow, acoustic sensors, weather, soil, topography and water-quality data.
Smart Water Infrastructure
Water infrastructure decisions, supported by data.
Use existing asset, operational, environmental and sensor data to identify higher-risk assets, detect leaks earlier, assess sewer condition and support water quality operations.
80%failures detected in the top 20% of prioritised pipesQuarterly validation
80%predicted locations within 200 m of actual failuresCompared with actual failures
95%reported leak classification accuracyAcoustic sensing analytics
Asset RiskHigh-Risk Pipe SegmentPriority for field review
Leak EventAnomalous SignalReview acoustic signal
9Mpipes analysed
525,000 kmwater network analysed
700K+historical failure records
160+leaks detected
7,000+ MLwater loss avoided
$19M+estimated avoided loss value
Proven in the field
Results tied to actual failures, field investigations and operational use.
Tap or hover over a result to see the validation context recorded in the project material.
ASSET RISK20% → ~80%
Prioritise a smaller part of the network while capturing most observed failures.
Quarterly validation against actual failures
LOCATION80% within 200 m
Narrow the search area before inspection or sensor deployment.
Prediction compared with actual locations
LEAKS160+ / 7,000+ ML
Field deployment translated detection into measurable water-loss reduction.
Greater Sydney operational programme
WATER QUALITY~90%
Multi-day raw-water forecasting across normal and extreme conditions.
Validated in NSW and Victoria
Results validated against field and historical evidence
Focus inspection and maintenance where the evidence says it matters most.
Water utilities already collect GIS, asset, failure, pressure, flow, sensor, weather, and water quality data. The challenge is turning these separate datasets into a clear view of where attention is needed first.
The work has been evaluated against actual failures, field investigations, sensor results and operational water-quality observations. The aim is not simply to produce a score, but to narrow the search area and support practical decisions under limited inspection and maintenance budgets.
Integrated Water Data View
From network-level patterns to individual assets and events.
Review asset risk, anomalies, trends, key factors, and field findings in one place, with a common evidence base for engineering and operational teams.
GIS risk mapsRisk & uncertaintyAnomalies & alertsAsset prioritisationScenario simulationModel updates
DATA · GIS + sensors + history
ANALYSIS · risk + forecast + anomaly
VALIDATION · failures + field evidence
Core Capabilities
Four areas where data analytics can support day-to-day and long-term water management.
Data → Prediction → Decision → Evidence
Keep existing utility systems. Add a decision layer that learns from new evidence.
Find risk, anomalies and trends
Use task-appropriate spatial, time-series, survival, anomaly and risk models to rank assets or forecast conditions.
Turn analysis into priorities
Support inspection, sensor placement, planned maintenance, renewal, dosing and operational scenario assessment.
Validate and update
Use actual failures, inspections, repairs, sensor findings and water-quality observations to review and update the analysis.
Case Studies
Projects across drinking water, wastewater, leakage, water quality, and asset management.
Sydney · Leak Detection & Failure Prevention
Use risk prediction to guide sensor placement, field checks, and planned maintenance.
Using 20 years of historical data, pipe risk analysis, and acoustic monitoring, 270 sensors were deployed in priority areas. Sensor data, asset records, risk scores, and field investigations were brought together in a single operational view.
160+leaks detected
7,000+ MLwater loss avoided
95%reported leak classification accuracy
$19M+estimated avoided loss value
The acoustic sensing and analytics approach was progressively field-tested and incorporated into Sydney Water's business-as-usual monitoring and maintenance processes.
Drinking Water Assets
Water Main Failure Prediction
80 / 20~80% of failures in top 20% prioritised pipes200 m~80% of predicted locations within 200 m
Use asset, environmental, historical, and operational data to estimate pipe-level risk and support inspection, maintenance, and renewal prioritisation.
Validated every three months against actual failures · ~30 utilities · 9 million pipes · 525,000 km of network
Wastewater Assets
Sewer Rising Main Failure Prediction
50–60%future failures identified from top 20% high-risk pipes
Use pipe, soil, topography, waterway, flow, and chemical dosing data to assess likely failure risk and failure type.
Model trained and validated using 200+ historical failures · project material estimates failure reduction up to 30% initially and 50% over three years
Digital Twin
Gravity Sewer Corrosion & End-of-Service Life
<8%EOSL model residual error50%+severe corrosion cases in top 20% risk assets
Combine sensing, microbial data, condition information, and predictive modelling to estimate corrosion progression and remaining service life.
Prediction linked with in-situ sensing, smart coupons and physical condition evidence
Climate Resilience
Dynamic Prediction of Raw Water Quality in Catchments
~90%overall prediction accuracy
Use upstream water quality and historical and forecast weather data to provide multi-day forecasts for treatment planning.
Validated across NSW and Victorian catchments · project material states the software is used in business-as-usual operations
Distribution Water Quality
Woronora & Macarthur Water Quality Prediction
Field validatedcompliance data + mobile total-chlorine analysers
Predict water age, total chlorine, and key operating conditions, with GIS-based views to support daily operations and scenario assessment.
Supports daily or hourly operational adjustment, chlorine dosing and what-if scenario assessment
Research Translation
Long-Term Water Data Science Capability
Research and applied work spanning predictive maintenance, failure modelling, anomaly detection, water quality, sensing, and visual analytics.
Built through long-term research and industry collaboration
Research Base
Selected publications behind the work.
Water infrastructure research spanning failure prediction, leak detection, sewer condition, and water quality from 2015 to 2025.
Water main failure
Leak detection
Sewer condition
Water quality
Advanced Model - A Digital Innovation for Water Quality Prediction and Dynamic Optimisation of Water Treatment
A. Mohiuddin, Y. Wang, H. Tian, Y. Wang, B. Blayney
Ozwater'25Use of Machine Learning and Robotic Sensing to Target Renewals in Concrete Gravity Sewers
B. Stephen, M. Kacprzak, B. Li, T. Guo, Y. Wang, V. Viswanathan, S. Kodagoda, K. Thiyagarajan, D. Vitanage
OzWater'23Prediction and Prevention of Sewer Rising Main Failure
J. Xu, D. Vitanage, M. Kacprzak, I. Chowdhury, M. Valix, D. Fletcher, T. Guo, J. V. Miro, D. Gamini, Y. Wang
OzWater'22Failure Prediction for Large-scale Water Pipe Networks Using GNN and Temporal Failure Series
S. Liang, Z. Li, B. Liang, Y. Ding, Y. Wang, F. Chen
CIKM 2021Critical and Small Pipe Prediction within 200m of Failure
B. Liang, J. Xu, Z. Li, S. Liang, Y. Wang, F. Chen, D. Vitanage, R. Nikoloska
OzWater'21Advances in Leak Prevention to Minimise Unaccounted Water
R. Nikoloska, D. Vitanage, L. Bykerk, J. V. Miro, B. Liang, J. Xu, Y. Wang
OzWater'21A Data Driven Approach for Leak Detection with Smart Sensors
B. Liang, S. Verma, J. Xu, S. Liang, Z. Li, Y. Wang, F. Chen
ICARCV 2020Predicting Water Quality for Whole Woronora Delivery Network with Sparse Samples
B. Liang, Z. Li, R. Taib, G. Mathews, Y. Wang, S. Lu, F. Chen, T. Hua, R. Ius, A. Peters, D. Vitanage, C. Doolan
ICDM 2019View more publications +
2023
Machine learning prediction to improve operational resilience in managing disinfection for Macarthur Water Supply System. Ozwater'23.
2021
Water Quality Prediction in a Chloraminated System for Drought Resilience. Ozwater'21.
2020
Long-Term Pipeline Failure Prediction Using Nonparametric Survival Analysis. ECML PKDD 2020.
2020
Data-Driven Water Quality Prediction in Chloraminated Systems. Water e-Journal.
2019
Utilizing machine learning to prevent water main breaks by understanding pipeline failure drivers. Ozwater'19.
2018
Domain Knowledge in Predictive Maintenance for Water Pipe Failures. Springer.
2018
Long-Term RNN: Predicting Hazard Function for Proactive Maintenance of Water Mains. CIKM 2018.
2018
Corrosion Prediction on Sewer Networks with Sparse Monitoring Sites: A Case Study. PAKDD 2018.
2017
Predictive analytics toolkit for H2S estimation and sewer corrosion. OzWater 2017.
2016
Improved Data-Driven Reticulation Watermain Failure Prediction. OzWater 2016.
2015
Prioritising Water Pipes for Condition Assessment with Data Analytics. OzWater 2015.
2015
Data Driven Water Pipe Failure Prediction: A Bayesian Nonparametric Approach. CIKM 2015.
Applied in Industry
From research and field trials to operational use.
Several solutions have been trialled or adopted in routine monitoring, maintenance, and operational workflows, with recognition from water industry and technology awards.
2018Australian Museum Eureka Prize
Excellence in Data Science
Excellence in Data Science
2022AWA R&D Excellence Award (NSW)
UTS Medal for Research Impact
UTS Medal for Research Impact
2024NSW iAwards
Sustainability & Environmental Solution
Sustainability & Environmental Solution
2025Australian Financial Review AI Awards
Sustainability Category Recognition
Sustainability Category Recognition
About
Applied water analytics built around real infrastructure decisions.
Smart Water Infrastructure brings together applied research and industry work across asset failure prediction, leak detection, sewer condition, and water quality management.
The focus is on using existing utility data, validating results against field evidence, and turning analysis into practical support for inspection, maintenance, renewal, and operations.
01 Utility data
02 Field validation
03 Operational decisions
Sydney WaterSA WaterUnitywaterGreater Western WaterQueensland Urban UtilitiesWaterNSWMelbourne WaterHong Kong WSDUKWIRKWR
Contact
About the Work ↗
Interested in applying these methods to your water network?
For industry collaboration, pilot studies, applied research, or technology translation, please contact:
yang dot wang at uts dot edu dot au
zhidong dot li at uts dot edu