11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 J.Smits 1, M.R. Moens 2 *, M. Klootwijk 3, H. van Vliet 4 1 Delft University of Technology, Stevinweg 1, room 4.84, 26GA Delft, the Netherlands, email: jairsmits@gmail.com 2 ARCADIS, Utopialaan 4-48, 52 BA s-hertogenbosch, the Netherlands, email: m.r.moens@arcadis.nl 3 Municipality of Breda, Claudius Prinsenlaan 1, 48 DC Breda, the Netherlands, email: m.klootwijk@breda.nl 4 Delft University of Technology, the Netherlands, email: h.vandervliet@student.tudelft.nl *Corresponding author, e-mail m.r.moens@arcadis.nl ABSTRACT The existing control system for pumping stations is at the end of its life span in the municipality of Breda (the Netherlands). For this reason in 23 a decision was made to replace the system by a new control system, which is equipped with the most modern information and communication technologies. A field laboratory has been build for training purposes of the new system as well as to allow research in overflow measurements and flow-measurements in (partly-filled) sewers. There are many ways to regulate the flow. Experiments can be conducted with groundwater and wastewater. Using this field laboratory four different types of flow meters are tested in the range 15-45 m3/h and compared with a fully calibrated flow meter in the pressure pipeline connected to the pump. Several types of tests have been carried out to determine the stabilization time and to test several flow meters for partly and fully filled sewers. It is found that some flow meters do not operate well under different circumstances. The accuracy of the flow meters is mostly about 5-1%, also in case of partly-filled sewers. The experiments have also shown that, despite measuring at the best locations and under optimal conditions, the accuracy remains 5-1%. KEYWORDS Field laboratory; Flow-measurement; Overflow measurements; Sewer; Stabilization time pag 1, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 INTRODUCTION The EU Water Framework Directive (WFD) was adopted in 2. Since then the water quality of open water and the treatment of sewerage water has great attention by many municipalities in the Netherlands. The municipalities have great interest in knowing the performance of their sewer system and whether there is a need to improve the performance. Monitoring the sewer system plays an important role in this. In Breda (the Netherlands), the existing control system for the pumping stations was at the end of its life span. As a consequence the maintenance became more costly and it was uncertain if the requirements set by the WFD could be met. For this reason, the municipality of Breda decided in 23 to modernize their sewer system by developing a new control system. This control system consists of the most modern information and communication techniques and will store large quantity of data from several sources of information. The new control system should give a clear overview of the performance of the sewer system and should lead to fewer combined sewer overflow events and a more optimized use of storage. A field laboratory has been build for training purposes of the new control system as well as to allow research in overflow measurements and flow-measurements in (partly-filled) sewers. There are many ways to regulate the flow. Experiments can be conducted with groundwater and real wastewater. In this paper the field laboratory and its capabilities are described in more detail. The paper also presents the findings on the stabilization time for flow in range of 15-45 m 3 /h and presents some first results of tests conducted with different types of flow measurements devices using groundwater only. Figure 1. Schematics of the field laboratory. Four different flow meters are installed for testing. : electromagnetic flow meter (factory-calibrated) 1: ultrasonic flow meter (clamp-on) 2: electromagnetic flow meter (mouse) 3: doppler/ultrasonic flow meter 4. radar flow meter (contact-less) pag 2, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 To get insight in the best accuracy you can get under field conditions the possibilty was created to install the measurement devices at different locations. There was also direct access to the data of the calibrated flow meter. Field laboratory The field laboratory consists of three main parts: the pumping station, the overflow device and the sewer pipeline with two manholes, see figure 1. The manhole at the end, the inflow manhole, receives the pumped water and the one in the middle of the sewer is to install monitoring devices like flow-meters. Water can be pumped from the pumping station through the pressure tubes to the overflow device or to inflow-manhole. By gravity, water will flow through the sewer to the overflow device and eventually to the pumping station. Pumping station. Two pumps are installed in the pumping station (1.5 x 1.5 x 3.3 m). The small pump of 1.3 kw can pump water from the pumping station only to the overflow device. The large pump (9. kw) can pump water to the inflow manhole as well. Above each pump a calibrated electromagnetic flow meter is installed. These devices can measure the discharge with an accuracy of +/-.5%. The pressure tube connected to the large pump has a diameter of 315 mm above the ground and a diameter of 4 mm below the ground. The diameter of the pressure tube connected to the small pump is 1mm. Water can enter the pumping station through a 6 mm diameter concrete sewer pipe. Overflow device. A concrete wall is located in the middle of the overflow device (3. x 3. x 3. m). On top of the wall, structures of any design can be placed to represent overflow devices from the field. Downstream of the wall, a gate is located. The water can only pass the wall if the gate is closed. The gate must be partially opened to create a partially filled sewer, upstream of the overflow device. The small pump can pump water to either side of the wall, the large pump can only pump water upstream the wall. The water can enter through a pressure tube or through the sewer. Sewer pipeline. The sewer pipeline has a length of 5 m and a diameter of 6 mm. At the upstream end of the pipeline, the inflow manhole (1.5 x 1,5 x,8 m) is located. A second manhole with the same dimensions is located 15 meter downstream of the inflow manhole. Flow meters can be installed and connected to the control system in this manhole. Electromagnetic flow-meter. For the tests described in this paper, the electromagnetic flow-meter, which is located above the large pump, is used for reference. The flow-meter was calibrated in the manufacturers factory. The uncertainty of the meter is.5-2% for flow rates smaller than 1 m 3 /h and.5% for flow rates larger than 1 m3/h. METHODS Prior to tests with flow measurements from various flow-meters, the stabilization time for different flow-rates is determined. For comparison of flow measurements with the calibrated pag 3, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 electromagnetic flow meters, it is important to know the stabilization time at a given flow rate. If the flow fluctuates, measurements cannot be compared. Fluctuations in flow can be explained as follows: When the pump setting is changed in order to create a higher discharge, the water level in the pumping station drops. As a consequence, the pump characteristic changes. The water level in the pumping station will increase thereafter because it takes some time for the water to return to the pumping station. Again the pump characteristic changes and results in a different discharge. It takes some time before these fluctuations are dampened. When the fluctuations are dampened, the stabilization time is known. Two tests were performed to determine the stabilization time: - test 1: fully developed sewer, increasing and decreasing flow-rates in the range of 22-32 m 3 /h; - test 2: partly filled sewer (< 5%), decreasing flow-rates in the range of 18-4 m 3 /h. The pump-setting (revolutions per minute) was increased every ten minutes by twenty rpm (~2.5 m 3 /h) for test 1. First the rpm-setting was decreased from 15 to 7 rpm at day 1. The second day, the rpm-setting was increased from 7 to 15 rpm. The rpm-setting was decreased from 15 to 7 rpm for the test 2. It was chosen to first let the rpm-setting decrease and then increase, because at some prior tests unexpected jumps in discharge occurred for the test with increasing rpm-settings. This jump did not occur at a test with decreasing rpm-settings. Instability during the filling process of the pressure tube may cause the jump in discharge with increasing rpm-settings. The pressure tube is filled more quickly if the pump is initially set at maximum rpm-setting. Once the pressure tube is completely filled, the actual test can start and instabilities are mostly avoided. In test 1, the discharge per rpm-setting of the series with decreasing rpm-settings and of the series with increasing rpm-setting are compared. For both series the moving average and the bandwidth of 1% is calculated. For this test, the moving average is calculated by incorporating every new measurement point to the total sum of measurements and then divide by the total number of measurements upon that measurement point. It is assumed that a stable situation has occurred if the moving average differs less than.2 m 3 /h over a period of 2 seconds. The stabilization time was computed in a different way for test 2. The moving average of test 1 is greatly influenced by the magnitude of instabilities which occur right after a change in pump-setting. The moving average in test 2 is computed as the average of the following ten measurement points. As a consequence, however, for the last ten measurement points of each rpm-setting, the moving average will come more closely to the actual measurement point. These moving averages are to be disregarded when determining the stabilization time. The stabilization time is determined by checking whether the moving average is not ever increasing or decreasing, if a sinusoidal-trend is distinguishable and if measured fluctuations are not too big (fluctuations must be below.5 m 3 /h). After the stabilization times for the two tests were determined, two tests with four different flow-meters were performed. At first three flow-meters were installed at different locations in the system. The first is an ultrasonic flow meter which is clamped on the pressure tube, see location 1 in figure 1. A mouse flow-meter, using the principle of Faraday s law to measure pag 4, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 the flow, is installed at location 2 and finally a Doppler flow-meter is installed at location 3. At the time that the measurements took place, logging the measurements was not possible. Therefore measurements are read and noted down on paper every ten minutes. After these tests another flow meter (radar) is installed at location 4. The first test with the flow-meters was performed in two days (3-5-27, 1-6-27). On the first day the rpm setting was increased and on the second day it was decreased. The gate in the wall at the overflow device was closed so that a fully developed sewer was created upstream of the overflow device. The second test with the flow-meters was performed on one day (25-6-27). The gate in the wall at the overflow device was opened so that the sewer was partially filled. Unfortunately there was little time to perform the test. Therefore the pump setting was decreased every 2 minutes to a setting at which the electromagnetic flow-meter measured a discharge of 1 m 3 /h less. Figure 2. Result of first test to determine the stabilization time. The sewer was fully filled during the test and the pump setting was decreased every ten minutes by 2 rpm from 13 rpm to 1 rpm and increased back to 13 rpm. pag 5, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 Figure 3. Discharge at pump setting 14 rpm. The left series is the discharge during the series with decreasing pump settings. The middle series represent the discharge during the series with increasing pump setting and the right series is the discharge after the accidental decrease of pump setting. RESULTS AND DISCUSSION Firstly the results from the tests to determine the stabilization time are discussed. Secondly the results from the tests with four flow-meters are given. Figure 2 shows the discharge over time of the first test. The sewer was fully filled during this test. By accident the pump setting was decreased instead of increased at t = 135 s and later there was a break of half an hour. Figure 3 shows an example of the discharge at the same pump-setting from the decreasing pump-setting series and from the increasing pump-setting series. Clearly, the average pag 6, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
Discharge (m3/h) 11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 discharge is higher in the increasing series, but a stable situation is only reached (after 2 s) in the decreasing series. In general, the discharge was higher in the increasing pump-setting series than in the decreasing pump-setting series. A stable situation is more often reached within ten minutes after increasing the pump-setting than after decreasing the pump-setting. No evident relationship could be found between pump-setting and stabilization time. From the test with a partially filled sewer it is concluded that a stable situation is reached within ten minutes at discharges below 3 m 3 /h, but is seldom reached for discharges higher than 3 m 3 /h. Flow measurements by four flow meters 4, 35, 3, 25, 2, 15, 1, 5, Electromagnetic flowmeter Doppler flowmeter Ultrasonic flowmeter Mouse' flowmeter', 7 75 8 85 9 95 1 15 11 115 12 125 13 135 14 145 15 Pump setting (rpm) Figure 4. Test result with four different flow meters; a (factory calibrated) electromagnetic -, Doppler -, ultrasonic - and mouse flow meter from a series with increasing pump settings and a fully filled sewer. In the first phase of the experimenets measurements from three different types of flow meters have been compared with the measurements from the factory-calibrated electromagnetic flow meter. Figure 4 and figure 5 show the results of the measurements by the four flow meters during the series where the pump setting was respectively increased and decreased and when the sewer was fully filled. The Doppler measurements below 2 m 3 /h (,2 m/sec) are very unstable and inaccurate. At a discharge of more than 3 m 3 /h, the Doppler meter constantly overestimates the discharge. The ultrasonic flow meter is the most accurate flow meter and the mouse flow meter is mostly underestimating the discharge. Figure 6 shows the results for partially filled sewers. Figure 7 shows the percentage of filling during this test. As the discharge decreases, water level in the sewer increases. The Doppler meter was able to measure flows lower than 2 m 3 /h. Again the ultrasonic flow meter was pag 7, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
Discharge (m3/h) 11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 very accurate and the mouse flow meter generally underestimates the discharge and becomes less accurate at lower discharges. Flow measurements by four flow meters 4, 35, 3, 25, 2, 15, 1, 5, Electromagnetic flowmeter Doppler flowmeter Ultrasonic flowmeter Mouse' flowmeter', 7 75 8 85 9 95 1 15 11 115 12 125 13 135 14 145 15 Pump setting (rpm) Figure 5. Test result with four different flow meters; an electromagnetic -, a Doppler -, ultrasonic - and a mouse flow meter from a series with decreasing pump settings and a fully filled sewer. pag 8, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 Figure 6. Test result with four different flow meters; an electromagnetic -, a Doppler -, ultrasonic - and a mouse flow meter from a series with decreasing pump settings and a partially filled sewer. Figure 7. Filling of the sewer during the test with three different flow meters and a series of decreasing pump settings. A summary of the measurements results of the different flow meters is given in table 1. In case the sewer pipe was fully filled, the accuracy is in the order of what is claimed by the manufacturer. The accuracy is slightly lower for the mouse flow meter in case of partially filled sewers, but can still be considered accurate enough in case the measurements are used for dimensioning and benchmarking the sewers. In case measurements are needed for calculations of water balances and for calibrations of computer models, then the accuracies pag 9, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 found for the mouse flow meter are somewhat low. Note that the ultrasonic flow meter measures the flow in the pressure tube, which is constantly totally filled. This is the reason that the results of fully and partially filled sewers pipes do not differ that much. Table 1. Summary of measurements results of three different flow meters. Type Location Accuracy According to manufacturer At 1% filling of sewer pipe At partially filled sewer pipe Doppler sewer unknown ~6-7% ~5% Mouse sewer ~5% ~2-4% ~8% Ultrasonic pressure tube ~1% ~.5-1% ~1% In the second phase of the experiments another type of flow-meter, which measures water surface velocity by using radar-detection, was tested (at 27-11-27). The waterlevel is measured by ultrasonic waves. A great advantage of this type of meter is that there is no contact with the waste water and is thus less sensitive for pollution and needs less maintenance. The test was conducted for a partially filled sewer and for a fully filled sewer. At full-pipe flow the radar-meter switches to another measuring principle to measure the velocity. Prior to the test, water was pumped at the highest pump-setting for about 45 minutes to get a stabilized flow. After that the pump capacity has been lowered four times with 25 rpm. After this lowering there was a break of fifteen minutes. From each interval of ten minutes the average has been taken and compared with the calibrated Krohne-flow meter. The results of this experiment are presented in table 2. The test was repeated for a fully filled sewer. Due to computer problems the number of readings is limited. The handwritten results show an accuracy no more than 15% (at 1% filling, this particular radar flow-meter measuring principle is based on electromagnetic induction). pag 1, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
Flow (m3/h) 11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 radar-flow meter vs calibrated Krohne-meter 35 3 25 2 15 1 5 Krohne Radar gem Radar gem Krohne 11: 11:3 236,57 12: Time 12:3 13: 13:3 Figure 8. Discharge during the experiment with the radar flow-meter and a partially filled sewer. The averages of the Krohne and Radar are given by gem Radar and gem Krohne. Table 2. Summary of measurements results of radar-flowmeter Type Location Accuracy According to manufacturer At 1% filling of sewer pipe At partially filled sewer pipe Radar sewer unknown > 15% ~5-15% CONCLUSIONS The field laboratory built by the municipality of Breda allows many different kinds of tests. A large range in flow rates can be created by which flow-meters and water level-meters can be tested. The meters can be placed in manholes, sewers and overflow devices. Testing different meters in a single test allows defining the accuracy of the meters under different flow conditions. A stable flow-rate occurs less frequent when the flow-rate is large. Once the flow is (more or less) stable, measurements from different meters can be compared. When conducting tests with large flow settings it takes longer than 1 minutes before the flow becomes stable. Four different flow-meters were tested in three different tests. The ultrasonic flow-meter, which is clamped on the pressure tube, has the highest accuracy of less than 1%. For fully filled sewers, the Doppler flow meter (<5% accuracy) performed better than the mouse flow-meter (<1% accuracy) and the radar flow-meter (< 15% accuracy). The accuracy of the three meters was around 1% for partially filled sewers. The Doppler-meter did not give good results at a velocity less than.2 m/sec. The experiments have shown that despite being free to choose the best location and conditions for the measurement device and despite having access to the data of the calibrated pag 11, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,
11 th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 28 flow-meter, the accuracy remains 5-1%. Further research is necessary to get insight in the accuracy of in-situ calibration of measurement devices. It is to be expected that the accuracy will further decrease because of the difficult circumstances in sewer systems to perform an in-situ calibration using the velocity-profile. Measuring the flow in partly-filled sewers with an accuracy higher than 5-1% is a step forward towards modelkalibartion using (mobile) flow meters. ACKNOWLEDGEMENT We are very grateful to the staff of the sewerage department at the field laboratory in Breda. We also like to thank the following companies and organizations for the funding of the field laboratory: the municipality of Breda, Stichting RIONED, ARCADIS, Bergschenhoek Civiele Techniek, Flow-Tronic, IMD, Imtech and ITT Flygt. pag 12, J.Smits, M.R. Moens, M. Klootwijk, H. van Vliet,