Comparison of Flow Measurement Techniques 139
Sydney, and in a 1 m wide rectangular flume at the River Hydraulics and Hydrology
Section of the Civil Engineering Research Institute of the Hokkaido Development
Bureau in Japan. Many of these tests were done over long periods involving a range of
flows, and for all these tests the average error between the ADFM flow measurement
and those of the laboratory rating or dye was less than 2 %.
The ADFM has been found to be highly accurate in a wide range of tests without
at-site calibration. The disadvantage of the ADFM is moderately high cost (three to
four times that of Doppler AVFMs). When choosing between AVFMs and the ADFM
the potential cost of inaccurate measurements should be weighed against the extra
cost of the ADFM. One case where high accuracy was sought was the Thames
Tideway Study of the large combined sewers from the London, UK, area that drain
into the Thames River for which 18 ADFMs were deployed and have provided high
accuracy at sites with very complex hydraulics (Curling et al., 2003).
2.2.7 COMPARISON OF FLOW MEASUREMENT
TECHNIQUES
In 1995, the US Geological Survey (USGS) in cooperation with the Federal Highway
Administration outfitted a 61 m length of straight, 137 cm diameter, 0.2 % slope,
concrete storm-sewer pipe in Madison (WI, USA) with multiple instruments for
the purpose of comparing these instruments. The details and results of this study
are briefly summarized in Church et al. (1999). However, additional details of this
field test were obtained as a written communication from D.W. Owens of the USGS
Wisconsin Water Science Center (D.W. Owens, personal communication, 1998).
Because this field test involved three of the previously discussed flow measurement
techniques it is presented as a separate section.
Owens (D.W. Owens, personal communication, 1998) reported that the test site
had the following characteristics that are typical of storm sewer locations where
discharge monitoring may be desired:
(1) The concrete pipe sections had settled different depths creating pipe joints
that acted as minor controls during lower flow conditions. As the water level
increased, the smaller controls were drowned out.
(2) The flow conditions at the site change rapidly because of the small drainage area
(77.7 ha), high amount of impervious surface, and intense summer rainstorms.
(3) Access to the pipe is limited creating a hazardous condition when the pipe is
flowing.
(4) Standard discharge measurements are nearly impossible to collect because of
the access and rapidly changing flow.
He also noted that the site was subject to relatively minor sediment loads.
140 Sewer Flow Measurement
The standard for judging the accuracy of the flows obtained from the various
measurement techniques was a MPB flume (Kilpatrick and Kaehrle, 1986) that had
been rated using 243 dye-dilution flow measurements over a flow range of 0.057 to
2.32 m3/s (hereafter referred to as ‘the measurement standard’). The average percent-
age error in the dye dilution discharge calculations was estimated as 4 % with a range
from 1 to 14 % (D.W. Owens, personal communication, 1998). Fifty runoff events
were monitored during a 6-month period and the resulting hydrographs and total
storm runoff volumes obtained with the flow measurement techniques were com-
pared with those obtained with the measurement standard. The flow measurement
techniques evaluated included:
(1) Critical-flow flumes in the form of the theoretical rating for the MPB flume.
(2) Manning’s equation applied at three locations in the study pipe.
(3) AVFMs–Automated Data Systems (ADS), ISCO 4250, and American Sigma
950 Doppler AVFMs and Marsh–McBirney Flow-Tote electromagnetic AVFM.
Data were collected at 1 min intervals for all meters except the ADS meter for which
a 2.5 min interval was used. The meters were placed in series in the pipe.
The comparisons of the measured hydrographs revealed the following (D.W.
Owens, personal communication, 1998):
(1) The hydrographs obtained from the AVFMs are noisier than hydrographs ob-
tained with the measurement standard. Inspection of the data showed that this
resulted from erratic velocity measurements.
(2) The AVFMs had periodic velocity dropouts wherein the velocity measurement
dropped down to a value that was much lower than the previous and following
measurements.
(3) At higher flows (>0.4 m3/s), the Doppler AVFMs tended to underestimate the
flow. At lower stages, the Doppler signal tended to work better. These results
are indicative of the range bias for deeper flows that is common for the Doppler
AVFMs.
(4) The electromagnetic AVFM tended to be the closest to the measurement stan-
dard. Furthermore, the electromagnetic velocity measurements displayed less
noise than the Doppler measurements.
(5) The theoretical discharge for the MPB flume closely matched the measurement
standard.
(6) The Manning equation technique produced mixed results based on monitoring
location in the pipe.
Box plots were made of the percentage differences between the results of the
various techniques/equipment and the measurement standard for the total storm
Conclusions and Perspectives 141
runoff volume (Church et al., 1999). Table 2.2.1 was prepared using the same data
used to prepare Figure 7 in Church et al. (1999), which was provided by D.W. Owens
of the USGS Wisconsin Water Science Center. The comparison of the total storm
volumes in Table 2.2.1 yielded the following results:
(1) The electromagnetic AVFM yielded the best overall results with a median error
of 0.4 % and an interquartile range of −9.4 to 4.4 %.
(2) The theoretical rating of the MPB flume also yielded good results with a median
error of +10.8 % and an interquartile range of 2.7 to 17.9 %.
(3) All uncorrected Doppler AVFMs underestimated total storm volumes with me-
dian errors ranging from −6.6 to −28.8 % and mean errors ranging from −10.1
to −30.5 %. Again an indicator of the range bias for deeper flows.
(4) One of the Manning’s equation sites was affected by backwater resulting in a
median error of nearly 100 %. Another Manning’s equation site was affected by
drawdown resulting in 25 % of the storms having underestimates greater than
30 %. The final Manning’s equation site was not affected by either backwater
or drawdown and had a median error of 24.4 % and an interquartile range of
−0.4 to 36.8 %.
It is difficult to derive general results from measurement comparisons at one site,
but Church et al. (1999) raised two important conclusions from this study. The data
clearly indicate the need to calibrate the flow measurement device using measure-
ments obtained with an independent method. Further, although flow measurement
techniques can be adjusted using verification data to minimize bias, the very large
uncertainty in flow measurements exhibited by some of the flow measurement tech-
niques is likely to remain after the adjustments.
2.2.8 CONCLUSIONS AND PERSPECTIVES
Many methods are available for measurement of flow in sewerage systems. Flumes
have been available since the 1930s, and electromagnetic and acoustic methods for
velocity measurement have been used since the 1970s and 1980s, respectively. During
these long periods of use, manufacturers and users have fine-tuned the equipment so
that reliable measurements may be obtained in real-time by telephone line or radio
transmission. If real-time data are desired, users must pay special attention to the
accessibility of the site to power and phone lines or radio transmission to a central
station.
All the flow measurement equipment is capable of yielding accurate discharges for
the appropriate hydraulic conditions (although the range of appropriate conditions
for Manning’s equation is quite limited). Flumes, electromagnetic flow meters and
ADFMs have been found to yield high accuracy (within ±5 %) for a wide range
of flow conditions. However, flumes and electromagnetic meters may be difficult
Table 2.2.1 Summary of storm volume errors in per cent relative to the rated modified Palmer–Bowlus flume for various flow measurement techniques/
equipment applied in a 137 cm diameter in Madison (WI, USA). (Data provided by D.W. Owens, USGS Wisconsin Water Science Center)
Technique/Equipment Mean Median Flow weight (mean) Min. 25th Percentile 75th Percentile Max. No. of storms
Modified Palmer–Bowlus flume
theoretical rating
10.1 10.8 10.2 −20.5 2.7 17.9 23.9 50
Electomagnetic AVFM/Marsh–McBirney
Flow-Tote
−2.2 0.4 0.2 −30.1 −9.4 4.4 24.0 43
Doppler AVFM/American Sigma 950 −30.5 −28.8 −28.0 −58.6 −36.9 −26.0 −8.6 42
Doppler AVFM/ISCO 4250 −10.1 −6.6 −12.4 −27.8 −19.9 −4.4 11.8 33
Doppler AVFM/Automated Data Systems −18.5 −19.0 −22.1 −38.8 −27.7 −14.0 −13.8 29
Manning’s equation/Location 1 −11.6 −0.6 −18.5 −69.8 −30.5 8.0 16.8 50
Manning’s equation/Location 2 18.8 24.4 8.4 −30.1 −0.4 36.8 64.8 48
Manning’s equation/Location 3 99.5 99.3 86.5 34.0 73.0 121.7 155.2 48
References 143
to install at some locations. ADFMs are easier to install, but are more costly than
acoustic Doppler area–velocity flow meters. Thus, users must consider site condi-
tions, cost, use of the data, and desired accuracy when selecting the appropriate flow
meter for the project at hand.
Probes for measuring dissolved oxygen concentration, conductivity and temper-
ature in real-time are commonly used in treatment plants and stream systems. Their
use in sewerage systems has been limited due to the possibility of damage by debris
in the confined space of the sewer pipe and the difficulty to keep the probes clean in
the harsh sewer environment. Other probes for measuring nutrients and other chem-
ical constituents in real-time are in development. As these probes are improved,
development of ways to use them in sewerage systems could be very valuable and
is encouraged as a topic of future research and development.
REFERENCES
Alley, W.M. (1977) Guide for Collection, Analysis, and Use of Urban Stormwater Data: A Con-
ference Report, Easton, MD, 28 Novembar–3 December 1976. ASCE.
Anon. (1996) World Water Environ. Eng.,April, 36.
Baughen, A.J. and Eadon, A.R. (1983) J. Inst. Water Poll. Cont.,82(1), 77–86.
B¨orzs¨onyi, A. (1982) Advances in Hydrometry, IAHS Publ. No. 134, 19–23.
Church, P.E., Granato, G.E. and Owens, D.W. (1999) Basic Requirements for Collecting,
Documenting, and Reporting Precipitation and Stormwater-Flow Measurements. US Geo-
logical Survey Open-File Report 99–255.
Curling, T., Leafe, M., and Metcalfe, M. (2003) Avoiding the Pitfalls of Dynamic Hydraulic Con-
ditions with Real-Time Data. Available online at http://www.mgdinc.com/pdfs/Thames%20
Tideway%20Hydraulic%20Results.pdf.
Day, T.J. (1996) Water Eng. Manage.,143(4), 22–24.
Diskin, M. (1977) J. Irrig. Drain. Div.,ASCE,102(IR3), 383–387.
Doney, B. (1999a) Water Eng. Manage.,146(11), 32–34.
Doney, B. (1999b) Water Eng. Manage.,146(11), 11–12.
Drake, T. (1994) Water Eng. Manage.,141(12), 34.
Hager, W.H. (1989) J. Irrig. Drain.,ASCE,114(3), 520–534.
Hughes, A.W., Longair, I.M., Ashley, R.M. and Kirby, K. (1996) Water Sci. Technol.,33(1), 1–12.
Hunter, R.M., Hunt, W.A., and Cunningham, A.B. (1991) Water Environ. Technol., 3(2), 47–51.
Huth, S. (1998) Water Technol.,21, 78–80.
Johnson, E.H. (1995) Water SA,21(2), 131–138.
Kilpatrick, F.A. and Kaehrle, W.R. (1986) Trans. Res. Rec.,1073, 1–9.
Lanfear, K.J. and Coll, J.J. (1978) Water Sewage Works,125(3), 68–69.
Ludwig, R.G. and Parkhurst, J.D. (1974) J. WPCF,46(12), 2764–2769.
Marsalek, J. (1973) Instrumentation for Field Studies of Urban Runoff. Research Program for the
Abatement of Municipal Pollution Under the Provisions of the Canada-Ontario Agreement on
Great Lakes Water Quality, Ontario Ministry of the Environment, Project 73-3-12.
Melching, C. S. and Yen, B.C. (1986) Slope influence on storm sewer risk. In Stochastic and Risk
Analysis in Hydraulic Engineering, B. C. Yen, ed. Water Resources Publications, Littleton, CO,
pp. 79–89.