A City Pipe Network Project: Practical Application Case of Drainage Network Monitoring

城市管网 2026-11-19 5 min read
A City Pipe Network Project: Practical Application Case of Drainage Network Monitoring
Urban Drainage Network Level and Flow Monitoring and Flood-Prevention Dispatching Case Study: Digital Transformation of a 28.

A City Pipe Network Project: Practical Application Case of Drainage Network Monitoring

Digital Transformation of a 28.6-Kilometer Drainage Network in a City's Core District

Urban Flood Prevention IoT Monitoring Data-Driven Dispatching
Overview of urban drainage network monitoring system covering 28.6km pipes and pumping stations

The drainage network monitoring project in the core district of a certain city covers approximately 12.4 square kilometers, connecting 28.6 kilometers of stormwater, sewage, and combined sewer networks. It deploys 168 level, flow, and open-channel monitoring terminals, linked with 6 rainfall monitoring stations, 4 pumping stations, and 12 gates.

Led by the local drainage company with Yicai Technology participating in terminal deployment and platform integration, the system was commissioned for the 2024 flood season. During multiple heavy rainfall events, it formed a continuous data chain around water level, flow, rainfall, and operations, reducing flooding spots from 12 historically to 5 and shortening emergency response time from about 38 minutes to 12 minutes.

This case demonstrates that the value of drainage network monitoring lies not only in “seeing” underground water levels, but more importantly in shifting flood-prevention dispatching from experience-based judgment to data-supported decisions through coordinated analysis.

I. Project Background: Shifting from “Experience-Based” to “Data-Based” Dispatching

Historically, drainage management relied on manual patrols. The district, built relatively early, featured large variations in pipe diameters and combined sewers prone to cross-connections, tree root intrusion, and sludge accumulation. Preliminary surveys identified 28.6 km of network (18.2 km stormwater, 6.4 km sewage, 3.6 km combined), involving 346 manholes and 12 flood-prone risk points.

During the 2022 flood season, three intense rainfall events caused road flooding at 12 locations, with a maximum depth of 42 cm. Three locations were low-lying underpasses where drainage took over 4 hours. Lacking real-time data, dispatchers could only assess pressure based on pumping station current and field feedback, making advance pre-release difficult.

In 2023, a renovation project launched to establish monitoring at trunk pipes, flood-prone points, and key gates, integrating with weather forecasts and GIS. Objectives included online level/flow monitoring, three-level warning thresholds, and system commissioning before the 2024 flood season.

II. Technical Challenges: Communication, Power, Dredging, and Data Quality

Underground Environment

Manholes are typically 700–1,000 mm in diameter and 3–6 m deep, with humidity near 95% and corrosive gases like hydrogen sulfide. Equipment must balance explosion protection, corrosion resistance, and fall protection.

Installation Constraints

About 32% of sites were in narrow lanes (<3m wide) inaccessible to large vehicles. 47% had high traffic, compressing construction windows to 10 PM – 4 AM. Night work was essential for transport and underground operations.

Flow monitoring faced complexity due to sediment, debris, and surface fluctuations causing reading drift. Small slopes and low velocities made single-gauge flow calculation difficult. Level data was affected by sediment accumulation and radar beam obstructions; without baseline correction, sediment buildup could be misinterpreted as rising water levels.

Communication signals attenuated significantly underground (4G below -110 dBm at some points). Power supply was constrained by lack of mains electricity and solar shading. The project needed to balance monitoring density, battery life, and data reliability during frequent rainstorm reporting.

III. Solution: Integrated Sensing and Linkage

The solution adopts a four-layer architecture: sensing—transmission—platform—dispatching.

  • Level Monitoring: 126 radar and pressure gauges (0–5m range, ±5mm–±1cm accuracy) installed at trunk pipes, forebays, storage ponds, and pumping station inlets.
  • Flow Monitoring: 32 partially full pipe ultrasonic/electromagnetic meters (±3% FS accuracy) and 10 open-channel meters at outlets and box culverts.
  • Rainfall Assessment: 6 stations using tipping-bucket and radar gauges (0.2mm resolution).
  • Pump & Gate Integration: Data from 4 pumping stations (current, voltage, status) and 12 gates (opening, limit status).

Communication & Power Strategy

Dual-Mode Comms: NB-IoT and 4G. Normal interval: 5 mins; Rainfall warning mode: 1 min. Key stations maintain second-level acquisition.

Solar Power: 10–20W panels with 120Ah lithium batteries, targeting ≥15 days autonomy during cloudy/rainy weather.

The platform integrates GIS topology, hydraulic models, and threshold rules. Data quality control flags sudden changes or mismatches to prevent single-sensor failures from affecting dispatching decisions.

IV. Implementation Process

The project spanned five stages from October 2023 to May 2024. After reviewing 168 candidate sites, 126 level points, 32 flow points, and 10 open-channel points were confirmed.

326 Night Work Sessions
48 Manholes Dredged
137 Comparison Tests

Installation involved mounting brackets, fall-protection nets, and sealing openings. To address weak signals, antenna heights were adjusted and 4G boosters added at 21 sites. Commissioning included manual zero-point calibration and flow meter rating via weir/flume comparisons.

Before the 2024 flood season, live tests for rainfall >50mm/h achieved a 98.6% online rate and <90s data ingestion delay. Acceptance focused on anti-sedimentation capability, low-flow stability, and alarm closed-loop workflows.

V. Data Support: Thresholds and Algorithm Calibration

Water level warnings are set at three tiers: Attention (2.5m), Warning (3.0m), and Danger (3.6m). Pumping station forebay levels trigger pre-release (>3.2m) or emergency forced drainage (>4.0m).

Flow validation uses joint checks: if level rises but flow doesn't increase synchronously, or deviates >15% from historical periods, the system flags potential sedimentation or equipment anomalies.

Hydraulic Model Performance: Calibrated on 42 rainfall events in 2023, achieving a peak flow error of ~9.4% and Nash-Sutcliffe efficiency coefficient of 0.81.

During the July 18, 2024 event (68.4mm 1-hour max rainfall), District B trunk pipe levels rose from 1.24m to 3.84m. The platform issued a surcharge risk warning 42 minutes in advance. Within 28 minutes of rainfall onset, level rise rates exceeded 0.12m/10min, triggering automatic dispatching recommendations.

VI. Flood-Prevention Dispatching Actions

The platform converts data into actions. Three hours before forecasted heavy rain (>30mm/h) with high trunk levels (>2.0m), gates are opened for pre-release to lower levels by 0.5–0.8m. If external river levels exceed 3.5m, backflow-prone gates are closed. Rapid forebay level rise (>0.15m/10min) triggers additional pump units.

July 18 Event Response Results

  • 3 pumping stations started 1.5 hours prior; water level pre-lowered by 0.63m.
  • 2 backflow-prone gates closed; 4 mobile pump trucks deployed.
  • Dynamic adjustment: One station switched from 2 to 3 units; gate opening increased from 40% to 65%.
  • Peak level reduced by 0.31m compared to un-dispatched model scenarios.

Consequently, flooding spots decreased from 12 (in similar 2022 rain) to 5. Maximum flooding depth dropped from 42cm to 24cm, and average drainage time shortened from 4.6 hours to 1.8 hours. No low-lying passage experienced traffic interruption for more than 30 minutes.

VII. Final Results: Operational Efficiency

Anomaly Detection & Repair

Detected 38 level anomalies, 21 flow anomalies, and 14 mismatch events. 28 were confirmed as physical issues (sediment, roots, damage). Post-repair, conveyance capacity increased by ~22%, and peak levels at historical points dropped by 0.3–0.5m.

System Reliability & Cost

Platform data completeness: 98.3%; Terminal online rate: 99.1%; Warning accuracy: 86%. Annual inspection/emergency repair costs decreased by ~30%. Emergency response time shortened from 38 to 12 minutes.

Manual patrols were optimized from 120 to 78 times per week, focusing on alarm points. For the drainage company, this data provides a basis for prioritizing future renovations, such as dredging sections with long-term high levels but low flow.

VIII. Conclusion and Outlook: Toward Resilient Networks

This case advances monitoring from "data collection" to "flood-prevention coordination." Future plans include adding AI-based video surveillance for flooding recognition, supplementing water quality/manhole tilt monitoring, and integrating all assets into a unified digital twin dispatching system.

Modeling will improve by introducing longer-duration rainfall and groundwater impacts for better simulation of combined sewer overflows. Warning information will be shared with municipal roads, transportation, and emergency departments to coordinate road closures and personnel evacuation.

Experience shows that level and flow monitoring must be advanced simultaneously with dredging, repair, and pumping capacity improvement. Otherwise, threshold alarms alone cannot change flooding outcomes.

The focus shifts from "how many sites" to "whether data supports decision-making." Integrated analysis of level, flow, rainfall, and video becomes the foundation for urban flood resilience, providing a reference path for other cities: establish sensing at high-risk nodes, form closed-loop dispatching via platform linkage, and convert data into executable capability.

🔧 Recommended Products

Based on this article, we recommend the following Ecolor products:


24h Fast Response

Need Help Choosing the Right Product?

Our engineering team is ready to provide free technical consultation and quotes

Share this article: