How a Water Utility Group Achieves Cost Reduction and Efficiency Improvement Through Drainage Network Monitoring
Executive Summary: This case study details how Yicai Technology implemented a comprehensive drainage pipe network monitoring system covering 28.6 km² and 136 km of pipelines. The project achieved a data completeness rate of 98.7%, reduced false alarms from 18% to 4.6%, and shortened waterlogging response times from 45 minutes to under 8 minutes.
Introduction: Data as Core Levers for Pipe Network Monitoring
The urban drainage pipe network is one of the most concealed and difficult-to-operate-and-maintain systems in municipal infrastructure. Compared with conventional roads, buildings, and power facilities, drainage pipe networks remain underground for long periods, characterized by deep-buried sections, numerous nodes, complex materials, and dispersed ownership.
Once blockage, damage, cross-connection, or siltation occurs, issues are often not discovered until rainfall-induced ponding, road waterlogging, or sewage overflow happens. With the increasing frequency of extreme heavy rainfall events, urban flood-prevention management has placed higher requirements on real-time sensing. Data on liquid level, flow, rainfall, pump-station operation, and gate status have shifted from “auxiliary records” to being the “basis for dispatch.”
"Project acceptance data showed that after system commissioning, data completeness reached 98.7%; the false-alarm rate decreased from 18% to 4.6%; and the response time for waterlogging in key areas was shortened from about 45 minutes to less than 8 minutes."
Project Background: Triple Pressure on Old Urban Areas
The service area is located in an old urban zone developed at an early date, featuring a complex drainage regime with both separated and combined stormwater/wastewater pipes. Pipe diameters range from DN300 to DN1800, utilizing materials such as concrete, FRP sand, HDPE, cast iron, and brick-built drains. Field surveys identified 4,860 manholes, 1,920 stormwater inlets, 4 pump stations, and 7 critical interception nodes.
1. Localized Ponding
Records show 12 obvious waterlogging spots in 2023, with max depth 0.6m and traffic interruptions lasting >2 hours.
2. Sewage Overflow
Combined sewer overflow occurred at two nodes during a June 2023 event (138mm/24hr), causing resident complaints.
3. Lack of Transparency
Dispatch relied on field patrols and citizen reports due to missing continuous data on pumps, gates, and levels.
Prior to the project, existing systems lacked unified data standards. The construction objectives were defined as establishing continuous monitoring at key points and combining this with hydraulic models to achieve visible operating status, early abnormal warning, traceable dispatch commands, and quantifiable flood-prevention response.
Technical Challenges: Underground Complexity
The primary challenge was not device installation, but deviations between underground data and field conditions. Preliminary surveys found approximately 11% of pipe sections had ledger information deviations. Cross-connections, silted sections, and structural risks required on-site reconfirmation before determining monitoring locations.
- Liquid Level vs. Flow Conversion: Unlike closed pressurized pipelines, drainage networks experience transitions between free-surface and full-pipe flow. Siltation and backwater surcharge distort the relationship between level and flow capacity.
- Harsh Construction Environment: Manholes are confined, humid, and deep (up to 4.5m). Installations must avoid obstructions, reflections, and sediment deposition.
- Power and Communication Constraints: Municipal power is often unavailable nearby. The solution utilized solar panels with lithium batteries (>15-day endurance) and mixed NB-IoT/4G/LoRa networking with edge gateways to mitigate signal attenuation.
Solution: Multi-Parameter Coordinated Sensing
The project established a coordinated sensing system comprising four main categories of devices:
| Device Type | Quantity | Key Specifications & Use Case |
|---|---|---|
| Liquid-Level Meters | 128 sets | 80 GHz radar, ±1 cm accuracy, IP68 rated. Suitable for high humidity and foam interference. |
| Flow Meters | 86 sets | 12 Doppler open-channel (trunks); 74 ultrasonic/level-area (branches/manholes). |
| Rain Gauges | 22 sets | Tipping-bucket, 0.2 mm resolution, 1-minute acquisition for time alignment. |
| Video & Auxiliary | 15 + 12 sets | Video for debris/blockage; Conductivity/turbidity probes for overflow identification. |
At the platform level, multi-layer thresholds were set. Liquid-level alarms triggered at 85%, 90%, and 95% design fullness. Flow alarms used historical baselines (e.g., >30% increase in 30 mins). Pump-station dispatch logic jointly assessed level, flow, and rainfall to push pre-drawdown recommendations, transforming single-device alarms into comprehensive scenario judgments.
Implementation Process: Closed-Loop Advancement
The 180-day implementation period was divided into three phases:
- Survey & Location Design (60 days): Segmented verification of 136 km using laser rangefinders and CCTV. Locations prioritized historical ponding points and pump-station forebays.
- Device Installation (75 days): Adopted “no water shutoff, no road closure” approach. Each location received a QR-code asset record. Stainless-steel screens were added for open channels, and vibration-damping brackets for arterial roads.
- Commissioning & Calibration (45 days): Device readings compared against manual measurements. Hydraulic models were calibrated using three 2023 rainfall events (2-year, 5-year, and 10-year return periods), reducing peak liquid-level prediction error from 22 to 9 minutes.
Data Quality & O&M Mechanism
To ensure long-term effectiveness, strict data-quality rules were established, including sudden change detection and correlation verification. For example, if a level meter rose >30cm in 10 mins without corresponding upstream rain or adjacent sensor changes, it was flagged as a suspected anomaly.
Tiered Maintenance Plan
- Quarterly: Cleaning and accuracy checks for routine level meters and rain gauges.
- Monthly: Inspection of screens and flow-guide plates for open-channel flow meters.
- Biweekly: Lens fouling and angle misalignment checks for video devices.
- Pre-Flood Season: Comprehensive patrol removing sediment and assessing battery health.
Fault Response SLA: 4 hours (arterial roads), 8 hours (branch roads), 24 hours (remote channels).
Operating Effects: Real-Time Data Changes Dispatch Rhythm
After a 90-day trial operation, the system demonstrated significant improvements in flood prevention capabilities.
Performance Metrics
- Data Completeness: 98.7%
- False Alarm Rate: 4.6% (down from 18%)
- Response Time: < 8 mins (down from 45 mins)
- Overflow Events: -41%
Case Study: Sept 2024 Heavy Rain
During a 68mm rainfall event, the platform issued an orange warning 36 minutes in advance based on level/flow trends. Pre-drawdown operations started pumps early, lowering forebay levels by 0.35m. Consequently, max trunk pipe level stayed below overflow threshold (93%), and downstream ponding depth was limited to 0.28m (vs 0.6m in similar 2023 events).
The combined liquid-level and flow warning provided 25–35 minutes more lead time than relying solely on road ponding video, creating a critical operating window for proactive dispatch.
Outlook: Extending to Resilient City Governance
The value of this monitoring system lies in incorporating underground pipe network status into a quantifiable, warnable, and reviewable management closed loop. It shifts management from viewing device status to viewing pipe-network operation processes.
For the next phase, Yicai Technology recommends expanding in three directions:
- Structural Safety Monitoring: Adding settlement, deformation, leakage, and harmful-gas sensors in high-risk sections.
- Water-Quality Management: Integrating turbidity, conductivity, and ammonia-nitrogen analysis at discharge outlets for environmental supervision.
- Model Predictive Dispatch: Using digital twin platforms to simulate pipe-network status and compare dispatch plans for the next 1–3 hours.
Improving future drainage and flood-prevention management requires continuous optimization among monitoring accuracy, model credibility, O&M responsibility, and cross-departmental coordination.
