JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 120 Sewer Flow Measurement sediment and debris deposition, turbulence, confined space/hazardous conditions is- sues, access, variable pipe slope along a reach resulting from differential settlement of individual pipes, and different pipe sizes. Nevertheless, continually increasing environmental concerns and the need to more optimally manage stormwater and wastewater flows have increased the need to accurately monitor flows in storm, sani- tary andcombined sewers.These concerns are not new, e.g., North Rhine-Westphalia, Germany, issued a decree that the most important detention facilities of the com- bined sewer network were to be equipped with continuous monitoring devices more than 20 years ago (Weyand, 1996). Fortunately, as the need for measurement devices capable of high accuracy has increased new and improved measurement techniques also have been developed over the last 20 years. This chapter attempts to summarize the accuracy, advantages, and disadvantages of the available techniques. There are basically two basic types of flow measurement techniques: (1) those that rely on a relation between stage and discharge, e.g., Manning’s equation and flumes; and (2) those that estimate average velocity by acoustic or electromagnetic means and multiply this by the cross-sectional area obtained through a depth measurement device and known conduit geometry. Most of these devices have two parts: (1) a primary device that directly interacts with or controls the flowing water; and (2) a secondary device for measuring water depth (Church et al., 1999). This chapter focuses on the general characteristics of the various measurement techniques of types 1 and 2 and does not provide a direct comparison of the com- mercially available equipment for flow measurement in sewers that apply the various measurement techniques. Due to the limited space available and the limited num- ber of independent evaluations of measurement equipment, a proper comparison of the equipment cannot be done here. Further, it is not the purpose of this chapter to advocate or criticize any particular device, but rather to give the readers basic information on the measurement techniques to aid in the selection of the appropriate technique. When purchasing equipment readers should carefully review the liter- ature provided by the manufacturers, discuss experience with the equipment with professional colleagues, and apply the time-honoured principle of caveat emptor (let the buyer beware). 2.2.1.1 Purposes of Flow Monitoring There many reasons for flow monitoring, among the most common are: (1) Real-time control (RTC) of the sewer system. Existing large sewers can be controlled by gates, e.g., to increase storage capacity and prevent overburdening of treatment plants (Curling et al., 2003). RTC also can optimize treatment plant operation to ensure consent standards are met and to minimize the total pollutant load reaching the environment (Watt and Jeffries, 1996) or improve plant efficiency in order to provide capacity for future sewer extensions (Anon., 1996). JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 Introduction 121 (2) Sewerage system operational considerations. Information about storage and dis- charge conditions can, for example, be the basis for optimizing cleaning and maintenance work (Weyand, 1996). (3) In regional sewerage networks, flow monitoring can equitably allocate costs among communities. (4) Compliance with regulatory requirements. (5) Provide data for calibration and verification of numerical models (Baughen and Eadon, 1983). (6) Identify inflow and infiltration (I/I) problems. (7) Performance evaluations of pumps and hydraulic structures (e.g., overflow structures). Items (1)–(4) typically require long-term, effectively permanent, monitoring, whereas items (5)–(7) generally require only temporary monitoring. On the basis of 10 years of sewer monitoring experience in Germany, Weyand (1996) made two very important, related observations: (1) it is important to start the planning of monitoring systems with the formulation of necessary demands; and (2) experience shows that the requirements of monitoring systems rise with their use. Thus, it is quite possible that sites that originally were established for a temporary study may become long-term sites, and careful selection of equipment and sites is necessary. 2.2.1.2 Equipment Selection Considerations Huth (1998) prepared a list of considerations for selection of flow measurement equipment. Six of his eight issues are: (1) Know the relative strengths and weaknesses of the available equipment. (2) Buy the level of accuracy required for the application. For example, high accu- racy is needed for RTC, cost allocation, and model calibration and verification; whereas lesser accuracy may be required for I/I studies, basic sewerage system operation, and performance evaluations of hydraulic structures. However, it must be remembered that requirements of monitoring systems rise with their use. (3) Know your flow rate. Sanitary sewers may have fairly constant flows, whereas storm and combined sewers have wider flow ranges and require equipment that is accurate over a wide range of flow conditions. Curling et al. (2003) stressed the importance of having high accuracy over the full range of flows noting that varying accuracy will increase variation in modelling results and lead to poor understanding of problem sites, improper estimation of capacity, and improper allocation of capital improvement funds. JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 122 Sewer Flow Measurement (4) Learn what is in the water. Debris may clog some equipment (flumes) and reduce the performance of other equipment (acoustic transducers) requiring frequent maintenance, also high particulate loads may affect the ability of sound waves to penetrate the flow. (5) Location, location, location (discussed in detail in Section 2.2.1.3). (6) Make sure there is power. Similarly,Church et al. (1999) noted that selection of the most appropriate method for collection of accurate flow data that are representative of a particular site re- quires knowledge of the flow regime(s), range of flow rate and depth, rapidity of flow changes, channel geometry, and the capabilities and accuracies of the methods available for measuring flow. 2.2.1.3 Monitoring Locations The importance of proper site selection cannot be overstated (Church et al., 1999). Most of the flow measurement techniques described in this chapter work best at sites where fully developed, uniform, open channel flow not subject to backwater effects is present composing optimal hydraulic conditions. Fully developed, uniform open channel flow usually requires many diameters of straight, uniform, undisturbed pipe upstream and downstream of the measurement location. For example, Johnson (1995) notes that the British Standard 1042 recommends that upstream from the measurement point a straight length of pipe equal to 30 to 50 diameters is sufficient depending on the type of turbulence causing device, whereas downstream from the measurement point 5 diameters of straight pipe should be present. Shorter sections of straight pipe could affect flow measurement accuracy. The accuracy of some methods also may decrease due to backwater effects and transitions from open-channel to pressurized pipe-full flow. Practical considerations may makeit necessary to place a monitor at a location with nonoptimal hydraulic conditions. For example, important locations, such as over- flows, bifurcations, and known flooding points, may require individual monitoring irrespective of hydraulic conditions (Baughen and Eadon, 1983). Borders between communities may require monitoring irrespective of hydraulic conditions for ‘po- litical reasons’ in cost allocation. Accurate model calibration and verification may require monitoring of each subcatchment (Baughen and Eadon, 1983). Finally, local constraints such as accessibility, power supply, and nonhydraulic goals of monitoring may also necessitate using nonhydraulically optimal sites. For example, monitoring locations might be selected for ease of pollutant sampling regardless of hydraulic conditions, as was the case of combined sewer monitoring in the Chicago, USA, area reported by Waite et al. (2002). Some of the techniques discussed in this chapter are better at measuring flow under nonhydraulically optimal conditions than others. Thus, once the monitoring JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 Introduction 123 locations are selected the following questions (after Church et al., 1999) must be considered: (1) Is the flowmeasuring techniqueapplicable to the flowand channel characteristics at the site? (2) Is the flow measuring technique capable of measuring the full range of flows? (3) Will the flow measurements be of sufficient accuracy to meet the objectives of the study? 2.2.1.4 Characteristics of Ideal Sewer Flow Measurement Equipment In order to deal with the complex hydraulic environment of sewer systems, Wenzel (1975) recommended that the ideal device for flow measurement should have the following characteristics: (1) capability to operate under both open channel and full flow conditions; (2) a known accuracy throughout the range of measurement; (3) a minimum disturbance to the flow or reduction in pipe capacity; (4) a minimum of field maintenance; (5) compatibility with real-time remote data transmission; (6) reasonable construction and installation costs. Drake (1994) further suggested that the equipment must provide reliable and accurate level and/or flow measurements within dynamic conditions, withstand a corrosive environment, overcome turbulence, and resist entanglement with floating matter. 2.2.1.5 Quality Assurance and Quality Control For any flow monitoring, but particularly for sewer flow, detailed quality assurance and quality control (QA/QC) programmes are necessary. Church et al. (1999) de- scribe in detail the key components of a QA/QC programme for flow monitoring, and their main QA/QC components are summarized as follows: (1) Frequent and routine site visits by trained/experienced personnel to maintain equipment and keep the site clean. (2) Redundant methods for measuring flow. JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 124 Sewer Flow Measurement (3) Technical training of project personnel. Weyand (1996) also stressed that it is necessary to have specially trained and qualified staff for operating and calibrat- ing the sewer flow meters. (4) Frequent review by project personnel of data collected. Weyand (1996) also noted that data quality must be continually checked to detect equipment malfunctions. (5) Quality audits, in the form of periodic internal reviews. (6) Quality audits, in the form of periodic external reviews. Church et al. (1999) noted that frequent calibration of equipment is necessary because of the difficult monitoring environment, and that the difficultiesof measuring in this environment result in a high probability of incomplete record, even when stations are well maintained and properly calibrated. 2.2.2 MANNING’S EQUATION The simplest form of stage-discharge relation is obtained by assuming that Manning’s equation is valid for the selected monitoring location. Using Manning’s equation discharge, Q, is calculated as Q = 1 n A(h)R(h) 2/3 S 1/2 (2.2.1) where n is Manning’s roughness coefficient, A(h) is the cross-sectional area of flow, R(h) is the hydraulic radius of the flow, h is the depth (or pressure head for full-pipe flow) of flow, and S is the energy slope of the flow. In this technique, h is measured using a pressure transducer or bubbler system, A and R are calculated as a function of h from the known conduit geometry, S is approximated as the pipe slope, and n is estimated from standard tables on the basis of pipe material and condition. Soroko (1973) noted that Manning’s equation may be appropriate for discharge calculation in channels with a straight course of at least 61 m, preferably longer, the course being free of rapids, abrupt falls, and sudden contractions or expansions. The primary advantage of this technique is that only a stage measurement device is needed to estimate flow. The primary disadvantages of this technique are that proper estimates of S and n are difficult to obtain. For steady, uniform flow in a channel as specified by Soroko (1973) the bed slope equals the energy slope, how- ever, for unsteady, nonuniform flow common in storm and combined sewers the bed slope and energy slope diverge. Further, even in cases where the bed slope approxi- mates the energy slope well, determination of the bed slope is difficult. Most often the bed slope is estimated from design plans, but this can be substantially different from the actual pipe slope. For example, Melching and Yen (1986) compared ‘as built’ measurements of pipe slopes between manholes with the slope indicated on JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 Manning’s Equation 125 the plans for 80 storm sewers in Tempe, (AZ, USA) and found a standard construc- tion error of 0.0008 m/m. Given that slopes of these sewers ranged from 0.001 to 0.0055 m/m, the standard construction error represented a substantial portion of the design slope in this case. Even when the pipe slope between manholes at the mea- surement location has been measured in the field there may be inaccuracies in the estimated slope because differential settlement and/or sag of pipes in the measure- ment reach between manholescausethe measured slope to not be representative of the energy slope. Determination of Manning’s n from tables for sewer pipes is at best a ‘guessti- mate’ (Soroko, 1973). Slime, debris, deposition, and decay of the pipes may cause Manning’s n for a pipe to be substantially different from values in the standard tables. Further, in pipes Manning’s n is a function of depth not a constant. Lanfear and Coll (1978) state that at depths of 5 to 70 % of pipe diameter, n is 20 to 30 % higher than the value for full pipe flow obtained from standard tables, failure to account for this phenomena will cause flows to be overestimated by more than 20 %. For improved estimates of n, Wright (1991) recommended that Camp’s dis- tribution of n as a function of depth (Figure 2.2.1) be used to adjust the value of Manning’s n. Wright (1991) presented the results of a field study for 22 sites in Grand Rapids (MI, USA) that illustrates the accuracy of the typical application of Manning’s equa- tion for estimation of flow in storm sewers. The pipes ranged in diameter from 45 to 243 cm, and in 5 of the 22 cases slopes were estimated fromfield measurements while the remaining slopes were based on design drawings. The actual flow rate was esti- mated using a hand-held electromagnetic velocity meter to measure the maximum 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 n/ / n(full) h/ / D 1 1.1 1.2 1.3 1.4 Figure 2.2.1 Camp’s normalized distribution of Manning’s n versus relative depth in a circular section JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 126 Sewer Flow Measurement velocity and assuming that the mean velocity was 0.9 times the maximum velocity. At the 22 sites discharge was measured an average of nine times, and the measured discharge was used to calculate values of S 1/2 /n for each measurement. Average val- ues of S 1/2 /n were determined for each site, and compared with the values of S 1/2 /n for each site estimated from field conditions including the variation of Manning’s n with flow depth (as would be done in the typical application of Manning’s equation). The mean percent error was 28.9 %. In 50 % of the cases the errors were greater than 25 %, and in 27 % of the cases errors were greater than 50 %. Wright (1991) also presented a case for sewers in Mobile, (AL, USA) that illustrated the even poorer results obtained with Manning’s equation in pipes subject to surcharge and backwa- ter. If a site is subject of surcharge and backwater, two stage gauges should be used, and the water-surface slope should be used to approximate the energy slope. This approach is rarely applied in practice. Lanfear and Coll (1978) found that a ‘fitted’ Manning’s equation, calibrated by a single flow measurement, provided good agreement with observed flow data and eliminated the need to measure slope. A single discharge measurement is used to calculate S 1/2 /n, which then is applied to all other flows. This approach was illus- trated for multiple flow measurements in three 122-cm diameter brick sewers with ‘as built’ slopes between 0.00044 and 0.0144 in Washington (DC, USA). Lanfear and Coll (1978) stated that if S 1/2 /n is determined for those flows of most concern, most of the error caused by variable Manning’s n is eliminated. This implies that if a wide range of flows are of interest, the value of S 1/2 /n may need to be cali- brated throughout the flow range. Marsalek (1973) stated that under conditions of unsteady, nonuniform flow in pipes, Manning’s equation underestimates flows in the rising stage and overestimates flows in the falling stage. Finally, Alley (1977) reported that the accuracy of the Manning’s equation technique is, at best, about 15 to 20 %. 2.2.3 FLUMES Flumes have been used to measure open channel flows in small streams, irrigation canals, water and wastewater treatment plants, and sewers for more than 50 years. Flumes are flow-constriction structures that control the flow hydraulics such that flow is directly related to head (Church et al., 1999). The most common type of flume constricts the flow such that critical flow results somewhere in the constricted section, which results in a unique relation between head and discharge as detailed later. These flumes are known as critical-flow flumes. Flumes work best at sites where the potential for surcharge, full-pipe pressurized flow, and backwater effects are expected to be negligible. Flume measurements are reliable for both uniform and nonuniform flow unless the sewer becomes surcharged (Parr et al., 1981). Baughen and Eadon (1983) noted that flumes give misleading results if they are surcharged and this condition is not suspected. JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 Flumes 127 The flow computation principle applied for critical-flow flumes may be derived as follows (Wenzel, 1975). The energy conservation equation is applied between a reference section 1 located immediately upstream of the flow constriction (flume) and section 2 is located in the constriction a distance L downstream from section (1) resulting in: h 1 + α 1 Q 2 2gA 2 1 + z 1 = h 2 + α 2 Q 2 2gA 2 2 + z 2 + h L (2.2.2) where α is the kinetic energy correction factor, g is the acceleration of gravity, z is the vertical distance from some datum, and h L is the head loss between sections 1 and 2. In the application of Equation (2.2.2) the following assumptions are made: (1) steady flow; (2) hydrostatic pressure distribution at section 1; (3) small slope such that the flow depth h approximately equals the vertical component of depth; and (4) two- and three-dimensional effects are negligible or accounted for as coefficients or energy loss terms (Wenzel, 1975). Equation (2.2.2) can be solved for discharge if all other terms are measured or evaluated as follows: Q =  2g(h 1 − h 2 + LS 0 − h L ) α 2 A 2 2 − α 1 A 2 1  1/2 (2.2.3) where S 0 is the bed slope. If open channel flow is present and A 2 is sufficiently small, critical flow will occur at some point in the constriction. If section 2 is defined as the point of critical flow the following relation is derived from the fact that the Froude number equals 1 for critical flow: Q 2 B 2 gA 2 = 1 (2.2.4) where B 2 is the width of the free surface at section 2. Substitution of Equation (2.2.4) into Equation (2.2.3) and using known relations between A, h, and B, the discharge can be implicitly determined by measuring only h 1 and evaluating h L , since all other terms are known. The head loss can be determined from boundary layer theory (Wenzel, 1975), but typically the relation between flow and discharge for a flume is determined by laboratory ratings. The Palmer–Bowlus flume was first proposed in the 1930s (Palmer and Bowlus, 1936), was extensively tested in the 1950s (Wells and Gotaas, 1958), and has be- come the most commonly used critical-flow flume in sewer systems. Palmer–Bowlus flumes have low head loss and can be installed in manholes where there is a stan- dard, straight-through design, or they can be installed in the half section of the sewer conduit (Soroko, 1973). Palmer–Bowlus flumes can be permanently installed, or be portable devices which can be inserted in the downstream pipe of a manhole using JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 128 Sewer Flow Measurement a pneumatic seal (Baughen and Eadon, 1983). Wells and Gotaas (1958) extensive laboratory experiments on the Palmer–Bowlus flume indicated that accuracy within 3 % of the theoretical discharge is readily attainable at depths up to 0.9D (where D is the upstream pipe diameter) for flumes installed in circular conduits. However, Hunter et al. (1991) indicated that Palmer–Bowlus flumes typically are inaccurate at depths greater than 0.75D. Figures 2.2.2 and 2.2.3 show two standardized trapezoidal Palmer–Bowlus flume sections for which a rating table is presented in Ludwig and Parkhurst (1974). Ludwig and Parkhurst (1974) noted that it is believed that the typical trapezoidal sections offer advantages regarding flow range and the provision of more accurate mea- surements at low flow values. Figure 2.2.4 shows a standardized rectangular throat Palmer–Bowlus flume for which a rating table is presented in Ludwig and Parkhurst (1974). Ludwig and Parkhurst (1974) noted that a value of D/10 represents a desir- able rise in the base of rectangular flumes installed within circular conduits. Standard Palmer–Bowlus flumes only have a stage measurement device in the approach sec- tion. To measure pressurized full-pipe flow, pressure should be measured at both sections 1 and 2 (approach and throat, respectively). Flumes with such two pressure sensor designs are known as Venturi flumes, which are discussed in the following paragraphs. 1 2 D/2 D/10 hc hu D Figure 2.2.2 Standardized Palmer–Bowlus trapezoidal flume with a bottom width of one-half of the pipe diameter JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 Flumes 129 1 2 D/3 D/10 hc hu D Figure 2.2.3 Standardized Palmer–Bowlus trapezoidal flume with a bottom width of one-third of the pipe diameter hu hc D B T t = D/10 Figure 2.2.4 Standardized Palmer–Bowlus rectangular flume [...]... transition slopes of 1:6 for the entrance and exit sections of the constriction Eight 25- and 30-cm models were tested in two laboratories and it was found that this flume was capable of accurately measuring flows under conditions of free and submerged, forward and reverse, free-surface flow, and forward and reverse pressurized full-pipe flow Hager (1989) proposed and tested a substantially different critical-flow... upstream and three downstream Soroko (1973) and Doney (1999a) list the following advantages of electromagnetic flow meters: high accuracy, the ability to easily handle fluids with high solids content, extremely wide range including reverse flows, obstructionless flow path, minimal pressure loss, small straight pipe requirements, and low maintenance, i.e it is unaffected by grease on electrodes and silt and. .. 0.9 m) The MPB was tested in the laboratory in 30.5- and 45.7-cm diameter pipes and in the field in a 122-cm diameter pipe The field tests involved controlled tests with hydrant flow and storm data For the hydrant flow, once each flow had stabilized, it was measured by both tracer dilution and acoustic meter; these discharges were in close agreement and also agreed closely with the MPB flume calibration... transducer and those far away causes a distortion in the return signal frequency spectrum that is difficult to resolve and typically results in the nearby, slower particles disproportionally affecting the measurement, which then is biased low (Metcalf and Edelh¨ user, a 1997) 2.2.5.3 Independent Evaluation of Doppler Area–Velocity Flow Meters Watt and Jefferies (1996) reported the results of extensive field and. .. This involved laboratory and field checks of 6 monitors that had been deployed in the field and evaluation of 59 monitors that had been used for short-term field studies The results of Watt and Jefferies (1996) are summarized in this section Field calibration of the depth sensors was carried out during each visit to the six monitoring locations Zero drift was found to be frequent, and in two of the six cases... high and three of the five were detected during comparisons with data from nearby sites at which flow discrepancies of up to 200 l/s were found No specific reasons for the poor quality of JWBK117-2.2 JWBK117-Quevauviller October 10, 2006 20:18 Char Count= 0 Sewer Flow Measurement 136 data at these sites could be found, and the quality assurance checks were insufficient to prevent such data errors Watt and. .. profiles and backwater effects in a pipe Valentin (1981) also made an early application of an electromagnetic flow meter to free-surface flow in a pipe and found that for depths between 0.5D and 0.8D errors ranged from −4 to −8 % and for depths between 0.8D and D the error decreases from −6 to 0 % Doney (1999a) described a practical electromagnetic flow meter for use in partially filled pipes This instrument... trapezoidal, and triangular throat shapes B¨ rzs¨ nyi (1982) developed a Venturi flume involving a o o lateral constriction of the channel In free, open-channel flow and in pressurized full-pipe flow the meter operates with accuracy better than ±5 %, and in submerged open-channel flow the accuracy is estimated at ±8 % Hunter et al (1991) proposed a Venturi flume with a truncated circular throat shape and transition... 1999) The varying shapes that Venturi flumes can take is documented in the literature Kilpatrick and Kaehrle (1986) designed and calibrated a modified Palmer–Bowlus (MPB) type flume for both open channel and pressurized full-pipe flow The modification involved the use of a longer flume with flatter side slopes and a greater floor thickness as well as adding a pressure sensor in the throat The flume was designed... submergence by backwater, and rapid installation in running water He also noted that the disadvantage of this flume is that debris could get caught on the cylinder and clogging could result in sewers that carry appreciable debris loads Laboratory experiments with this flume found that a diameter ratio of 0.3 between the pipe and the cylinder seemed optimal in terms of approaching flow stability and discharge capacity, . turbulence, and resist entanglement with floating matter. 2.2.1.5 Quality Assurance and Quality Control For any flow monitoring, but particularly for sewer flow, detailed quality assurance and quality. calibration and verification may require monitoring of each subcatchment (Baughen and Eadon, 1 983 ). Finally, local constraints such as accessibility, power supply, and nonhydraulic goals of monitoring may. JWBK117-Quevauviller October 10, 2006 20: 18 Char Count= 0 1 28 Sewer Flow Measurement a pneumatic seal (Baughen and Eadon, 1 983 ). Wells and Gotaas (19 58) extensive laboratory experiments on the