CWs for combined sewer overflow treatment (CSO CWs) are vertical flow filters with detention basin and fixedoutflow rate. They receive stochastic loadsinduced by urban runoff and protect natural waters against pollutants and streambed erosion.However, due to the stochastic nature of flows, concentrations and periodicity, optimizing CSO CW design requires a dynamic approach.Computational tools are available but process-based models are difficult to handle [1].Moreover, the absence of user interface in design-oriented tools (e.g. RSF_Sim [2]) demands manual data handling and simulations of multiple designs. Therefore, a new tool called Orage was developed. Orage relies on a core model similar to RSF_Sim.Long-term hydraulics, COD and NH4-N were simulated with good accuracy. Filter material selection and scaling is based on inflow data series and a low number of inputs. The iterative shell calls for simulations repeatedly to (1) optimize hydraulics; (2) select the simplest material which isnecessary to satisfy emission requirements on NH4-N and (3) determine the minimalfilter area at which legislative thresholds can be met. A design is optimized if the maximum of moving average on simulated effluent concentrations (Peak_MA_cc) is at the legislative threshold (NH4N) or below (COD). Fig. 1 shows an example of the iteration process.

According to French standards, constructed wetlands treating combined sewer overflow (CSO CWs) are vertical flow filters with detention basin and outflow limitation. Their purpose is to treat rapid loads of wastewater with stochastic volumes, concentrations and periodicity. The first full-scale CSO CW at Marcy l‘Etoile was monitored to provide in-depth understanding of hydraulics and nitrogen dynamics. Monitoring lasted for three years incl. online equipment. The water content in the media was visualized along the longitudinal section of the filter to follow hydraulics in the fill stage. Tracer tests showed shortcutting at this stage weakening as the filter saturated which tallied with peaks of NH4-N concentrations diminishing at the outflow side. Adverse shortcutting effects can be diminished by minimizing fill time of the media. As for nitrogen dynamics, adsorption capacities showed no difference in the two filter sides, one with a sand-zeolite mixture and the other with pozzolana. An equation was fitted to temperature and adsorbed NH4-N mass measurements to calculate inter-event nitrification. The rate was found to double with every 5.7 °C. The results helped to calibrate the design-support software Orage. Finally, the washout dynamics of NO3-N were analysed to consider the possibility of a second filter stage for denitrification.

Areas similar to free water surface constructed wetlands (FWS CWs) placed between wastewater treatment plants and receiving water bodies, under the perception that they increase water quality. More than 500 systems are in operation with a multitude of configurations and intended outcomes. In order to monitor these areas, research is being carried out to understand the fate of water and conventional pollutants in these systems. To this aim, a FWS CW located in southern France is monitored with traditional grab samples and 24-h flow composite samples. This site has also been instrumented with 6 Ion-Selective Electrodes (ISEs) probes recording continuously ammonium and nitrate concentrations. Because pollutant concentrations are usually low in treated wastewater, sometimes close to quantification limits of laboratory methods, we are developing appropriate methodologies for the management of the probes and the data processing. In this context, we propose a reliable methodology to increase the quality of data from ISE probes. This methodology is based on (i) laboratory experiments for sensor characterization and (ii) field tests. Laboratory experiments allowed characterizing the operating parameters like response time, linearity range, quantification limits, and interferences. Furthermore, for one-year, field tests are led every two weeks to (i) evaluate the required cleaning frequency and (ii) do grab samples analyses that help to validate data from the 6 ISEs. A drift in time appears to be significant for ammonium sensor. An additional experiment is currently monitoring this drift to correct this effect on measurement. This study has confirmed that it is fundamental to understand the technical limitations of the measuring equipment and set appropriate maintenance and calibration methodologies in order to have an accurate interpretation of data. The result is an operating protocol mainly concerning an acceptable cleaning frequency of two weeks, a stronger complementary calibration method using water from the experimental site, an evaluation of the drift and the determination of quantification limits of these ISEs (1 mg/L for ammonium and 0.5 m/L for nitrate). This protocol generates validated data that can be used to study nitrate and ammonium dynamics. In combination with the usual 24-h composite sampling method, it gives a good understanding of the fate of nitrogen within this FWS CW system. An example of data processing will be submitted.

The French Overseas Territories experience big sanitations problems and have to comply with both French and EU regulations. Vertical flow constructed wetland (VFCW) appear well adapted to the context of these regions, but their adaptation to tropical climate requires new guidelines to be defined (area needed, number of filters, species of plants, material to be used …). To this end ATTENTIVE project was build up in the French Antilles with local water offices and supported by the national water authorities. Three different VFCW fed with raw wastewater were built in Martinique and Guadeloupe and are now under monitoring. While the 3 plants are actually in operation, this paper will focus on Taupinière plant (Martinique), sized for 900 p.e., in operation since October 2014. The treatment plant is composed of aunsaturated/saturated vertical flow constructed wetland, receiving raw domestic wastewaters, followed by a simplified trickling filter.The first stage is 0,8m²/p.e. with two filters in parallel (360 m² each), with a 40 cm unsaturated first layer of 2/4 mm gravel, a 15 cm transition layer of 11/22 with intermediate passive aeration pipes and a 60 cm saturated layer at the bottom made of 20/40 pea gravel. The trickling filter(116 m²)is composed of 150 cm of pumice stones. The sludge accumulated at the bottom of the trickling filter is sent back to the first stage.A recirculation loop is implemented on the trickling filter. Before the outlet, effluent passes through UV. The regulationobjectives is to achieve 90% removal for COD, BOD5 and TSS, 80% for TKN, and less than 1000 unit per 100 mL for E.coli and intestinal streptococcus. Nevertheless, the monitoring look after optimal operation to increase TN removal rates as well as the maximum loads that can be treated. The monitoring consists on classical daily composite samples at each treatment stages (30 campaigns) as well as online measurements. The later aim at measuring flows, climatic conditions, and COD, BOD5, and Nitrate (UV/visible analysis) at each treatment stages. Three different loading phases has been implemented from 30 % of the nominal load to 150 %. The paper will present performances of this compact (less than 1 m²/p.e.) treatment system in both dry and rainy season and discuss the optimization of TN removal as well as the footprint reduction with the high loads that can be implemented.

CWs for combined sewer overflow treatment (CSO CWs) are vertical flow filters with detention basin and fixed outflow rate. They receive stochastic loadsinduced by urban runoff and protect natural waters against pollutants and streambed erosion. Thefull-scale site at Marcy l‘Etoile was monitored to gain data about hydrology and to quantify NH4-N adsorption capacities and nitrification rate. The throttled outlet ensuresa uniformflow in the porous media, butonly aftersaturation. Until then,the percolation is focused to the inletzone. As only a partof the filteris water-contacted and detention times are shorter than normal, removal performances are lower. The phenomenon is referred to as shortcutting, a temporary state at commencing load, which might last at low inflow rates. Eighteen TDR probes weredug in the longitudinal section of the filter to follow changes in the water content. This enabled to createan animation of the expansion of saturated area until complete saturation. Furthermore, tracer tests were carried out to signify shortcutting at different stages in the filter (Fig. 1).The filter was fed at the inlet point at a fixed and lowrate until saturation and three fluorescein pulses were dosed withidenticaldelay.The basin was flooded after to follow tracer passage and washout. Results were used to parameterizethemodel-based design-support tool Orage [1]and to suggest an improvementof the outflow limitationstructureto minimize shortcutting.

Constructed wetlands for combined sewer overflow treatment (CSO CWs) are variably saturated vertical flow filters in France. The design-support software Orage aims to facilitate engineering work by optimizing filter area and material sitespecifically which was otherwise encumbered by the stochasticity of CSO flows and concentrations. The optimization process relies on measured or simulated CSO series and a low number of input parameters. The iterative shell calls the core model repetitively. During the process, effluent flows and concentrations are simulated from a range of CW domains and compared to legislative thresholds. The iterative shell was tested both with measured and simulated inflows. First, key parameters of the hydraulic optimization were fixed. Large and underscaled designs are excluded this way and succeeding optimizations for pollutant removal are more efficient. Then, the optimization functions were verified using inflow and available land from an existing CSO CW. At third, the automatizations were used to test model predictions in the function of legislative thresholds. Zeolite-enriched media ensures high NH4-N removal at hydraulic loads exceeding the recommendations of present guidelines, marking clogging as an issue for further research. In summary, the demonstrated simulation experiments verified the optimization approach of the dynamic design tool Orage.

In recent years, unsaturated/saturated vertical flow constructed wetlands(VFCW) treating raw wastewateraregradually considered as apromising solution for their adaptation under various climatic conditions. These facilities provide surface optimization but alsobetter treatment efficiencies compare to classical French VFCW.The main performance improvements are: SS entrapment and carbon consumption for denitrification within saturated layer.When total nitrogen, by nitrification and denitrification,istargeted in the two successive zones, a design compromise has to be fund between nitrification efficiency and available carbon source for denitrification. As performance largely depends on unsaturated and saturated layers depths,a better understanding of their quantitative effects on treatment performance is essential for the adaptation of this system under various installation conditions. The aim of this study was to estimate the influence of these two linked design parameters onN efficiency with a focus onremoval limitations in regard to nitrogen loads. For this purpose, two different pilot-scale experimental configurations were designed: (i) 45 cm of unsaturated and 25 cm of saturated layers and (ii) 55 cm of unsaturated and 35 cm of saturated layers. The mature pilots were operated over 5 months using real wastewater with a feeding/resting period cycle of 3.5/3.5 days with a daily hydraulic load of 0.36 m d-1.In order to vary inlet nitrogen loads, ammonium nitrogen enrichments were added to vary loads from 10 to 40 g N m-2 d-1.24h flow composite samples at inlet and outlet of each pilot were semiweekly collected and analyzed for the following parameters: total suspended solids (TSS), total and dissolved chemical oxygen demand (COD), Kjeldahl nitrogen (TKN), ammonium, nitrate, phosphate and sulfate. Online measurement on a minute time step were done for inlet/outlet flows, oxygen content at three different depths, outlet ammonium and nitrate concentrationsby ion specific probes, and temperature. The paper will present the performance and limitations ofthe two configurations.Dynamics of nitrogen removal processes will be discussed in relation to physicochemical conditions (temperature, oxygen content, hydraulic retention time, carbon sources, etc.).

[Departement_IRSTEA]Ecotechnologies [Departement_IRSTEA]Eaux [TR1_IRSTEA]TED [TR2_IRSTEA]ARCEAU | International audience | In France, 10% of total arable land is equipped with subsurface drainage systems, to control winter and spring waterlogging due to a temporary perched water table. Most of these systems were installed in the1980s and have aged since then and may now need maintenance. Sometimes, the location of the systems is known, but the standard situation in France is that the original as-built master sketches are no longer available. Performance assessment of drainage systems and curative actions are complicated since drain location is unknown. In this article, the authors test the application of a non-destructive drain detection method which consists in water injection at the outfall of the drainage network combined with time-lapse electrical resistivity tomography (ERT) monitoring. To assess the performance of this methodology, which consists in measuring electrical resistivity from electrodes placed at the nodes of a 1.2-m regular mesh, the authors interpreted the signal using a two-step approach. The first step is based on 3D ERT numerical modelling during a scenario of surface infiltration processes (forward modelling followed by geophysical inversion); this step optimizes the ERT method for locating the infiltration at depths below 1m. The second step is the validation of the results obtained by numerical modelling with an experimental data set, using water injection into the drainage network combined with time-lapse ERT monitoring on an experimental field site. The results showed the relevance of time-lapse ERT monitoring on a small agricultural plot for locating the drainage network. The numerical results also showed several limitations of the combined methodology: (i) it is necessary to use an electrode spacing unit less than 1.20m, which does not facilitate investigation on large agriculture plots, (ii) measurements must be taken when resistivity contrast is the strongest between the infiltration area and the soil and (iii) the volume of water needed for injection can limit the extension of the method.

[Departement_IRSTEA]Ecotechnologies [Departement_IRSTEA]Eaux [TR1_IRSTEA]TED [TR2_IRSTEA]ARCEAU | International audience | In France, 10% of total arable land is equipped with subsurface drainage systems, to control winter and spring waterlogging due to a temporary perched water table. Most of these systems were installed in the1980s and have aged since then and may now need maintenance. Sometimes, the location of the systems is known, but the standard situation in France is that the original as-built master sketches are no longer available. Performance assessment of drainage systems and curative actions are complicated since drain location is unknown. In this article, the authors test the application of a non-destructive drain detection method which consists in water injection at the outfall of the drainage network combined with time-lapse electrical resistivity tomography (ERT) monitoring. To assess the performance of this methodology, which consists in measuring electrical resistivity from electrodes placed at the nodes of a 1.2-m regular mesh, the authors interpreted the signal using a two-step approach. The first step is based on 3D ERT numerical modelling during a scenario of surface infiltration processes (forward modelling followed by geophysical inversion); this step optimizes the ERT method for locating the infiltration at depths below 1m. The second step is the validation of the results obtained by numerical modelling with an experimental data set, using water injection into the drainage network combined with time-lapse ERT monitoring on an experimental field site. The results showed the relevance of time-lapse ERT monitoring on a small agricultural plot for locating the drainage network. The numerical results also showed several limitations of the combined methodology: (i) it is necessary to use an electrode spacing unit less than 1.20m, which does not facilitate investigation on large agriculture plots, (ii) measurements must be taken when resistivity contrast is the strongest between the infiltration area and the soil and (iii) the volume of water needed for injection can limit the extension of the method.