The sewer system of St Malo is complex, and includes a number of control mechanisms throughout the network. For this reason, it was a highly suitable site for a study project, started in 2001, with the objective of identifying whether modeling could be useful tool for real time control of sewer networks. A second essential pre-condition for the study was also present at St Malo – the enthusiastic support of local staff.
The St Malo study is planned in two steps:
Step 1 - to simulate real time control in a modeling study. Then, if the case is proved,
Step 2 - to use the model for real-time operational management in the control room.
The European directives relating to wastewater, direct water authorities towards two parallel strategies for the management of urban wet weather effluent. The first addresses the hydraulic parameters and flood prevention; the second focuses on pollution and the quality of the effluent released directly to the receiving water. Modeling has already proved a powerful tool for optimizing wastewater systems within these two constraints, and this study explores extending this success to real time control.
This first step of the study - network modeling - was divided in four parts:
- understanding the existing sewer system in terms of hydraulics and water quality, in order to model and replicate its operation.
- using modeling to identify the management rules to optimize the existing system without any modification.
- proposing modifications to improve efficiency, including modifications of the structure of the sewer networks such as additional storage capacity, or introducing new sources of data such as Radar rainfall measurement to forecast rainfall events.
- finally, the transfer of the model to the operation team.
The sewer system of St Malo
There are two key constraints on sewer operation in St Malo. First, some 400 hectares of the city are former bogs, now dried and built on, but at risk from the exceptional St Malo tides, which can cause backwater flows from the sea to this low ground. To counter this, a 12 m3/s pumping station has been build to evacuate the flow towards the sea. A second constraint stems from the importance of tourism – maintaining good seawater quality is a primary concern for the local authority and for the managers of the sewer system.
The network has four catchments – the main catchment covers the center of the city, and three others cover the periphery. The central sewer system operates by collecting and treating the dry weather flow and the first flush. The main catchment is divided into twelve sub-catchments, and each has an outfall at a device called an interceptor. These collect the dry weather flow and the first wet weather flow to a storage tank and allow the second wet weather flow to pass directly on through the system untreated to the river outfall.
The dry weather flow and first flush water is sent to a central 250 m3 pumping tank and a 6000 m3 storage tank, linked by a sluice. Effluents enter the pumping tank through the inlet-sluice and flow to the storage tank through the link-sluice. When the level in the pumping tank is too high, the link-sluice is closed and effluents are pumped to the storage tank using three 1m3/s capacity pumps. When the whole tank is full, it activates the closing of the inlet-sluice and the closing of the interceptor’s sluices. At the same time each part of the basin is drained off to the WWTP using four pumps, two for each compartment.
To complete the network there are 9 main CSOs and 15 rainfall storage tanks, integrated in the landscape and providing a storage capacity of 200000 m3. The outfall of the network is a treatment plant designed for 122000 PE, with two activated sludge basins able to treat a maximum volume of 2000 m3/h.
Overview of the Saint-Malo system
The tool selected for the network model was InfoWorks CS. Among many other advanced features, the package handles the simulation of real time control very accurately. The model was a subset of the full network, containing 549 nodes, 507 pipes, 30 pumps, 19 sluices and 26 weirs. The total catchment was divided into 12 sub-catchments with each terminated by an interceptor.
The dry weather flow was validated for all the sub-catchments by reference to the level of demand for drinking water.
For validating wet weather performance, there are a long series of measurements between November 1999 and June 2001, and 5 minutes timestep measurements for the level of water over the threshold at each interceptor and for the level of water in the two receiving tanks (pumping and storage). There is also data on the positions of the sluices at each timestep. All the data is transmitted through an optical fiber network to a central command point. The data is completed by rainfall data from a pluviograph located in the main catchment of the city. One pluviograph would not be sufficient to describe very localized summer rainfall events, but it is sufficient to understand homogenous rainfalls.
One of the main problems in the Saint-Malo sewer system is infiltration. The data indicated two types of infiltration:
- rainfall-induced infiltration from the surface water, contributing to flow within hours or days after the end of the rainfall event,
- infiltration from the groundwater reservoir, the effect of which depends on the season.
The model is still being calibrated but the first results are very encouraging, as exemplified by these figures for a 7 day period including 4 days of dry weather and 3 days of wet weather. Results were excellent for 5 interceptors and for the two receiving tanks. The results for the receiving tanks were a validation of the results obtained for each interceptor, as they check that the balance of volume is correct for the total system.
The parameters used for the simulation were:
- the dry weather flow
- a base infiltration flow, constant over the period and specific to the subcatchment
- runoff as defined by the Desbordes model, using a runoff coefficient for each subcatchment reflecting the land use.
Control of sluice positions and pump switch-on and switch-off was simulated according to the actual rules used for the system, and are defined in an RTC file within the model.
Comparison between simulated and actual flow volumes proved good for the first four interceptors and less good for the last. Further examination shows good results when the measured and simulated levels are compared, so the problem may lie with the method of transforming level to flow.
The results for the two storage tanks were also of acceptable accuracy.
Saint-Malo is a very complex site with a lot of control options. The first step of the study successfully achieved both a good representation of the performance of the system, and an accurate simulation and understanding of the effects of the control mechanisms. Specifically: - the model is able to reproduce the behavior of mobile gates and pumps, - good results were obtained, although the calibration is not perfect - the RTC module of InfoWorks CS is really powerful, and will make it possible to optimize the control rules. - Acknowledgements
We want to thank the “Cote d’Emeraude” agency of Generale des Eaux for allowing us to conduct this study of the network and to collect data. Particular thanks go to Max Hardy, Guy Jamet and Hubert Le-Saulnier.
Isabelle Bizien and Frédéric Gogien, Veolia. The results in this article relate to 2002. First presented at Wallingford Software’s 2003 International User Conference