Home > Typical Projects > Operational Forecasting
Development of LG District Smart Water Management System, Shenzhen, China
Author:EwatersPublishDate:2024-05-15


Longgang(LG) District in northeastern Shenzhen covers 388km2 of undulating areas. This is a well developed mixed industrial, commercial, and residential urban area with a population of 4.78 million.

The Longgang District Water Bureau aims to improve water management with the Smart Water Management System, by integrating information resources and online modelling services to provide evaluation results to the various water systems, especially for stormwater & wastewater operation, CSO control, and flood forecasting.

Ewaters, sub-consulting to the project team formed with China Institute of Water Resources and Hydropower, HUAWEI, Power Group China, is responsible for all models development, calibration and development of the real time forecasting and early warning system.

Main Services

l      Model build. Including five stormwater and wastewater network models for sub-areas with 39 reservoirs, 62 rivers, 14 sewage treatment plants,  3 reservoirs and a 1D-2D coupled river network HD& WQ models. Offline models were first built calibrated and with Infoworks ICM and later converted to Sobek for real time simulation on Delft-FEWS.

li      Development of real time forecasting system. Delft-FEWS Platform 2022 was applied to develop the real-time forecasting system. This system integrates data from approximately 35 rain gauges and 450 water level and quality monitoring stations. The data is automatically processed and fed into the models to generate forecasts. These activities were executed in a combination of sequences and parallel processes based on optimized design, allowing for the production of forecast results and reports in a very short time.

The system has been operational for nine months, providing twice-daily forecasts for river and drainage network levels, flow rates, and river water quality for the next 24 hours in about 15 minutes. It also forecasts reservoir levels and flow for the next 72 hours in approximately 3 minutes. Additionally, the system generates hourly rolling forecasts to predict water depth in flood-prone areas for the next 3 hours, completing these in about 6 minutes.

Notably, the system operates automatically without requiring operator intervention, ensuring continuous and reliable forecasting.

Service Delivery and Timeline

Modelling and development of the forecasting system was implemented into 4 phases. In the first phase, Ewaters conducted detailed investigation on the data, management and needs from various stakeholders. and produced an assessment report to confirm project understanding,  requirements and expectations, and a detailed work plan. In phase 2, Offline models were first developed, well calibrated and submitted. In phase 3, Delft-FEWS was applied to develop this complicated operational forecasting system to suit multi sectorial needs. The 4th phase is the completion of system installation, testing, validation, training, and delivery of reports. The delivered system achieved acceptance and client’s approval in December 2023. Currently it is in the warranty phase to further calibrate the models and optimizing the system performance.

Figure 1 The figure above shows reservoir water levels, storage volume, inflow, outflow, river water levels and flows with designed SCADA dynamic display.