Use of Rhodamine WT tracer experiments to check and validate in-situ flow meters M. Lepot, G. Lipeme Kouyi and J.-L. Bertrand-Krajewski Université de Lyon, INSA Lyon, LGCIE, 34 avenue des arts, F-69621 Villeurbanne cedex, France. (E-mail:
[email protected]) Abstract. Flow monitoring in urban drainage systems is widely used for the assessment of both water quantity and quality. Data derived from flow meters in field or laboratory conditions should be systematically checked and validated to avoid errors of measurement. After a brief reminder of the principles of flow measurement by means of instantaneous fluorescent dye tracer experiments (Rhodamine WT), the details of the method are presented. Uncertainty is calculated using the two standard methods (law of propagation of uncertainties and Monte Carlo simulations). This method enables to have an independent and accurate checking of discharge values in sewer systems, using a simple variance test. Two tests to validate the method have been carried out: Rhodamine WT tracer vs. electromagnetic flow meter and Rhodamine WT tracing vs. salt tracing. First applications of the method in combined sewer systems are presented in the paper, aiming to check in situ flow meters under various conditions. Keywords. Tracer experiments; Rhodamine WT; flow meters; combined sewers; data validation. INTRODUCTION The increasing need for knowledge on the transfer of water and pollutants in urban hydrology and legislation requirements have led researchers and operators to install an increasing number of flow meters in sewer systems. In spite of technological advances, flow meters located in free surface flows provide, in some cases, wrong measurements, particularly for low (less than 10 cm) and high (higher than 40 cm) water levels in complex geometries. Measurement results have to be controlled by independent methods. The use of fluorescent tracers for flow rate measurement is not new, however there are few routine applications (Jozja, 2008; Orand and Colon, 1993). This paper presents an application of on-line fluorescent Rhodamine WT tracer experiments in sewer pipes with uncertainty assessment (Lepot, 2012). MATERIALS AND METHOD Tracer experiments were carried out with the following equipment: a fluorimeter (Turner Designs, SCUFA), an ultrasonic water level sensor (Mobrey, MSP422), a data logger (Allborn, Allmeno V2290-5), Turner Designs Rhodamine WT commercial (21.33 % w/w) or diluted solutions, adjustable laboratory pipettes for tracer injection (Nichiryo, Nichipet EX 0,5-10, 10-100 and 100-1000 µL) and a laptop. Two field applications have been carried out in combined sewers: i) the first one in a man entry trunk sewer in the centre of Greater Lyon, with a double sidewalk: two measurement
points along the sewer have been used. The upstream point (Quai Arloing) is located immediately downstream a pumping station; the downstream point (Quai des Etroits) is located approximately 3.7 km downstream the first one. Discharges were ranged between 250 and 550 L/s during experiments. ii) the second site in a 500 mm circular sewer in Genas, East of Greater Lyon, France. Tracer experiments were used to establish a rating curve for water depths lower than 10 cm. Method In the laboratory. Two successive calibrations of the fluorimeter are recommended: the first one according to the manufacturer protocol with three points, and the second one using 14 standard solutions with concentrations from 0 to 130 g/m3. Uncertainties of the injected volumes are evaluated. Calculation and optimization of the pair (concentration, volume). The mass to be injected is calculated according to uncertainties, hydraulic conditions (velocity, water depth, wet area) and the desired maximum concentration in the measurement section. Data acquisition. Rhodamine WT concentrations are measured in situ by the fluorimeter with a time step dt = 1 s in the measurement section. The time series is the raw time series. Calibration correction. This step estimates the true Rhodamine WT concentration (corrected time series) from the raw time series by using the inverse of the calibration function. Semi-automatic processing of the corrected signal. The corrected time series is analysed to detect and remove background fluorescence, noise and possible artefacts. Calculation of flow and its standard uncertainty. They are calculated for each injection. Standard uncertainty is estimated by the law of propagation of uncertainties and Monte Carlo simulations. In case of K successive injections, the flow is equal to the mean of flows of each injection and the uncertainty is calculated from uncertainties for each flow and with the Student formula. Checking of the tested flow meter. The flow meter to be tested is validated if the uncertainty of the difference between QRHODAMINE,F and QTEST is lower than the uncertainty of this difference. VALIDATION OF THE METHOD Two tests allowed validating the above method by comparing its results with results provided by known robust methods: the first one with an electromagnetic flow meter in laboratory conditions and the second one in situ with traditional salt tracer experiments. FIRST CASES OF APPLICATION Checking flow meters along a combined trunk sewer The tests were carried out during two consecutive mornings at the Greater Lyon sites Quai Arloing and Quai des Etroits (Fig. 1). Upstream (Quai Arloing), the in situ flow meter, despite the narrow range of observed flows, alternatively overestimate and underestimate the true flows. The measurement site is located immediately downstream a pumping station and the
flow meter integrates over 6 min recording time steps the highly varying flows due to starts and stops of the pumps. The 6 min averaged values given by the flow meter are consistent with the mean flow obtained with Rhodamine WT tracer experiments: the difference between mean flow values (5 L/s) is less than its expanded uncertainty (60 L/s). Downstream (Quai des Etroits), the flow meter systematically underestimate the discharges measured by Rhodamine WT tracer experiments, with a bias increasing with the discharge value: from 18.8 % for 270 L/s to 54.6 % for 350 L/s (Fig. 1). The red straight line, used to correct the values given by the flow meter, has been validated by two methods: daily cumulated volumes and Computational Fluid Dynamics simulations.
Figure 1. Comparison of QRHODAMINE,F (y-axis) and QTEST (x-axis) for upstream (left) and downstream (right) flow meters. Rating curve for small water levels The second application was carried out in the Genas 500 mm circular pipe. The operator had initially installed autonomous water level measurement sensors (IJINUS, M011501A) at some points of interest in the sewer system. Rating curves Q = f(h) were then fitted for each point of interest by using additional measurement campaigns using flow meters (Hydreka, Mainstream 4). From the campaign data sets including water level h, flow velocity U and discharge Q, it appeared that, for water levels lower than 10 cm corresponding to most frequent dry weather conditions, the rating curves were not compatible with the calibrated Manning Strickler formula which was correct for uniform and steady flows with higher water levels. The hypothesis was that, for low water levels, the flow meters did not deliver accurate results. Rhodamine WT tracer experiments were then carried out. Figure 2 shows the results obtained for only one point of interest (they are similar for the other points). The right graph is a zoom of the left one for water levels lower than 10 cm. The initial rating curve (dashed line) was derived from the field campaign data set (black dots). Tracer experiments gave the pairs of values (h, Q) shown as circles: they are clearly above the points measured by means of the in situ flow meter. Taken the tracer experiments results into account, a new calibrated Manning Strickler formula was established, shown as the continuous line in Figure 2.
Figure 2. Use of Rhodamine WT tracer experiment to establish a rating curve. CONCLUSION Rhodamine WT tracer experiments in sewers has shown satisfactory results. Developing an appropriate methodology including uncertainty assessment and replicated injections provides an independent and accurate method for the verification of in situ flow meters and interesting perspectives regarding existing techniques: small injection volumes (opportunity to work with high flow rates), tracer almost absent in the effluents, short time step for data acquisition and immediate in situ data processing and quality control. However it also shows disadvantages: lack of commercial standard solutions, irreducible uncertainty of the Rhodamine WT commercial solution, replicated injections requires time which may be too long during storm events with highly variable flows. Beyond simple checking of flow meters, the method allows calibrating flow meters by establish relationships QTRACEREXPERIMENT = f (QFLOWMETER), specific for each site and each sensor to correct the data provided by the in situ flow meter. REFERENCES Jozja N. (2008). Importance de la composante analytique dans la fiabilité de l’interprétation d’un traçage (Importance of the analytical component in the reliability of the interpretation of a tracer expriment). Actes du colloque « Hydrogéologie et karst au travers des travaux de Michel Lepiller ». Orléans, France, 16-17 mai 2008, 207-218. (in French). Lepot M. (2012). Mesurage en continu des flux polluants en MES et DCO en réseau d’assainissement. PhD thesis: INSA Lyon, France, 257 p. (in French). Orand A., Colon M. (1993). La fluorimétrie appliquée à la mesure de débit en milieu torrentiel (Fluorimetry applied to flow measurement in supercritical flows). Revue des Sciences de l’Eau, 6(2): 195-208. (in French).
