Mike Metcalf - MGD Technologies Inc.; Tracy Vermeyen - Bureau of Reclamation, Water Resources Research Laboratory; Steve Melavic, John Fields - Bureau of Reclamation, Mid-Pacific RegionMike Metcalf - MGD Technologies Inc.; Tracy Vermeyen - Bureau of Reclamation, Water Resources Research Laboratory; Steve Melavic, John Fields - Bureau of Reclamation, Mid-Pacific Region

Introduction

A new type of flowmeter, the ADFM Velocity Profiler^TM (Profiler) was demonstrated on March 3 and 4, 1997 at the Bureau of Reclamation's Water Resources Research Laboratory in Denver, Colorado. The demonstration was organized by Tracy Vermeyen of the Bureau of Reclamation and Mike Metcalf of MGD Technologies, the manufacturer of the Profiler. A subsequent field test of the instrument was conducted in the San Luis Drain near Los Banos, CA, on March 14, 1997. John Fields and Steve Melavic of the Bureau of Reclamation's Mid-Pacific Regional Office were responsible for organizing this test. This report documents the results of these tests.

Figure 1. Cross section view of typical Profiler application. This figure shows the spatial relationship of the depth cells and the profiles relative to the transducer housing.

The Instrument

Figure 1 shows a typical Profiler installation for measuring open channel flow in a pipe. A transducer assembly is mounted on the invert of a pipe or channel. Piezoelectric ceramics emit short pulses along narrow acoustic beams pointing in different directions. Echoes of these pulses are backscattered from material suspended in the flow. As this material has motion relative to the transducer, the echoes are Doppler shifted in frequency. Measurement of this frequency shift enables the calculation of the flow speed. A fifth transducer is mounted in the center of the transducer assembly and is used to measure the depth.

The Profiler divides the return signal into discrete regular intervals which correspond to different depths in the flow. Velocity is calculated from the frequency shift measured in each interval. The result is a profile, or linear distribution of velocities, along the direction of the beam. Each of the small black circles in Figure 1 represent an individual velocity measurement in a small volume known as a depth cell.

The directions of the velocity profiles in Figure 1 are based on the geometry of the Profiler's transducer assembly. Figure 2 shows a side view of the transducer assembly. The profiles shown in Figure 1 are generated from velocity data measured by an upstream and downstream beam pair. The data from one beam pair are averaged to generate Profile No. 1, and a beam pair on the opposite side of the transducer assembly generates Profile No. 2.

Figure 2. Side view of the Profiler transducer assembly and its beam geometry.

Since Doppler measurements are directional, only the component of velocity along the direction of the transmit and receive signal is measured, as illustrated in Figure 2. Narrow acoustic beams are required to accurately determine the horizontal velocity from the measured component. The narrow acoustic beams of the Profiler insure that this measurement is accurate. Also, the range-gate times are short and the depth cells occupy a small volume - cylinders approximately 5 cm (2 in.) long and 5 centimeters in diameter. These small sample volumes insures that the velocity measurements are truly representative of that portion of the flow and potential bias in the return energy spectrum due to range dependent variables is avoided. The result is a very precise measurement of the vertical and transverse distribution of flow velocities.

The velocity data from the two profiles are entered into an algorithm to determine a mathematical description of the flow velocities throughout the entire cross-section of the flow. The algorithm fits the velocity data to the basis functions of a parametric model. The parametirc model is used to predict flow velocities at points throughout the flow. The resulting velocity distribution is integrated over the cross-sectional area to determine the discharge.

The key benefit to this approach is that the system will operate accurately under variable hydraulic conditions. As hydraulic conditions change, the change will manifest itself in the distribution of velocity throughout the depth of flow. As the Profiler is measuring the velocity distribution directly, it can adapt to changes in the hydraulics, and generate a flow pattern that is representative of the new hydraulic conditions, insuring an accurate estimate of flow rate.

Test Procedure

The Profiler was first tested in two sites at the Bureau of Reclamation's Water Resources Research Laboratory in Denver: a 4-ft-wide flume, and a 12-ft-wide rectangular channel. In both lab sites, the system was installed on the bottom of the flume, centered with the transducer's long axis aligned with the flume's axis (flow direction). No in-situ calibration or rating was performed.

4 ft flume tests - The test began with a flow depth of approximately 4 ft. The depth was increased after about one hour to around 7 ft. Profiler flow measurements were then compared with the venturi meter flows to check for accuracy and repeatability.

12 ft channel tests - The test began with a depth of flow of approximately 2 ft. The depth was decreased after about one hour to about 1.2 ft. Profiler flow measurements were then compared with the venturi meter flows to check for accuracy and repeatability.

The venturi meter flows were determined using a mercury manometer to measured the pressure differential across the venturi. This manometer was manually read several times during the tests. Once set up, the flow was held constant as it was controlled by an active feedback control system. The laboratory venturi meter was calibrated prior to the Profiler demonstration. A weigh tank facility was used to calibrate the venturi meter and the uncertainty in the venturi meter measurement was within +0.8 percent of the weigh tank flow rate.

Following the laboratory tests, the Profiler was placed in a "live" channel - the San Luis Drain located near Los Banos, CA. This channel is an irrigation drain for part of the San Joaquin Valley. The channel is trapezoidal in shape, with an 8 ft bottom width and 2:1 side slopes.

The Profiler was placed on an aluminum strap about 44 ft in length. Hinges were placed on the strap so that the 8 ft. center piece would lay flat on the channel bottom and the rest of the strap would conform to the side slopes of the channel. The channel was in normal operation during the installation with a flow depth of approximately 7.2 ft.

Results

Figure 3 shows the results from the 4 ft flume tests. The round symbols represent the Profiler data and the square symbols are the spot readings from the venturi meter. As shown in figure 3, the Profiler data agrees very well with the venturi meter readings.

At the beginning of the data record there is some scatter to the Profiler data, as the flow into the flume had just been set and the depth in the flume was still equilibrating. After the first few points, the flow rate readings and depth readings become steady. The Profiler measured an average flow rate of 6.86 ft³/s during the initial depth of flow of 4 ft (after depth equilibration), compared to 6.98 ft³/s for the venturi readings during the same time period; a difference of -1.72%. We also see a change in the Profiler flow rate measurement after the level was increased to 7 ft at around 10:00 a.m. The Profiler over predicts the flow because one pair of acoustic beams intersects the walls of the flume. Consequently, the average velocity is skewed higher because it is measured near the middle of the flume where velocities are larger than near the wall.

Figure 3. Profiler and venturi flow rates measured in the 4 ft flume are plotted as a function of time. The Profiler's depth reading is also plotted to illustrate the effect of fluctuating water surface level on the discharge measurement accuracy.

The spacing on the x-axis is irregular because several sampling schemes, in which parameters such as bin size and sampling interval, were varied. In most cases, changing these parameters did not affect the overall accuracy of the flow rate measurement.

Figure 4 shows the results from the 12 ft channel test. The round symbols are the Profiler data and the square symbols are the spot readings from the venturi meter. For this test, the Profiler data were averaged over a variety of time periods. Whereas Figure 3 is a plot of "raw" flow data, where each point corresponds to a individual measurement separated by an interval varying from one to five minutes. The first seven ADFM measurements were collected with an one minute averaging interval. The average of these seven measurements is 10.48 ft³/sec. The venturi meter reading was 10.26 ft³/sec, and this flow remained constant for the remainder of the test. The next five ADFM measurements were collected with an averaging interval of 5 minutes. The average of these five measurements is 10.29 ft³/sec. The next six ADFM measurements were collected with an averaging interval of 2 minutes. The average of these seven measurements is 10.43 ft³/sec. This test demonstrates how Profiler measurements becomes more precise as the number of measurements averaged together were increased. The ability to average hundreds of flow measurements over a short period results in a very precise flow measurement. This is illustrated by comparing the average of the all eighteen Profiler measurements with the two venturi meter readings. The Profiler average was +0.97% different from the venturi meter average. The last ADFM measurement was collected with an averaging interval of 2 minutes at a depth of 1.2 ft. The measured flow rate was 10.0 ft³/sec. This measurement was made with a 10:1 width to depth ratio. This demonstrates the Profiler's unique capability to measure flows in wide, shallow channels.

Figure 4. Profiler and venturi flow rates measured in the 12 ft channel are plotted as a function of time. The Profiler's depth reading is also plotted. Initially the Profiler's discharge measurements were variable, but with time the accuracy improved to within 1 percent of the venturi-measured discharge.

Figure 5 contains a plot of the flow rate, depth, and average velocity measured in the San Luis Drain near Los Banos, CA on March 14, 1997. All three measured parameters were steady over the test period. Variations in the flow rate correlate with variations in the average velocity, as the depth remains fairly constant. The average flow rate over the duration of the test was 91.1 ft³/s. Flow rates of approximately 80 and 86 ft³/s were measured using traditional stream gaging methods. Stream gaging velocities were measured with a Marsh-McBirney Flo-Mate, which is an electromagnetic velocity meter.

Figure 5. Flow rate, depth, and average velocity plotted as a function of time. The average flow rate measured in the San Luis Drain was 91.1 cfs with a standard deviation of 2.0 percent.

The first discharge measurement was obtained by manually measuring velocities at 0.2 and 0.8 of the depth measured from the water surface, at regular intervals across the channel. These velocities were used to compute an average velocity for a particular section of the channel. Multiplying this average velocity by the cross-sectional area of the individual section gives a flow rate for that section. The section flow rates are summed to determine the total flow rate for the channel.

The second discharge measurement was obtained by making a single velocity measurement at a height above the bottom of 0.6 of full depth, at regular intervals across the channel. This value was used as the average velocity in that section to compute a flow rate for that section. Again, section flow rates are summed to determine the total flow rate in the channel. Note: This velocity measurement should have been measured at 0.6 of the depth measured from the surface not the bottom.

It should be mentioned that there was some concern over the accuracy of the manual velocity measurements. The relationship between the velocities measured at the various depths was not as expected. In particular, the velocity value measured at a height above the bottom of 0.8 of full depth, in some cases, was lower than anticipated.

Conclusions

• For laboratory tests, the Profiler measured flow rate with an accuracy of approximately 1.7% in the 4 ft. flume and 1.0% in the 12 ft. channel. The Profiler was able to accurately measure flow rate even with a width to depth aspect ratio of 10:1.
• This test demonstrated that the Profiler does not require an in situ calibration or rating Profiler to make an accurate flow measurement. Accurate flow measurements were attained without any special consideration to the installation, aside from placing the Profiler in the middle of the flow and aligning it with the direction of flow.
• The Profiler was successfully installed in a "live" channel, without interrupting the flow. Flow rates measured were repeatable and within roughly 10% of a traditional stream gaging measurement. However, there was some concern that the stream gaging measurement might not be accurate, as some velocities appeared lower than anticipated. This would lower the flow rate estimate generated by the stream gaging method.
• This demonstration was useful in showing the ability of the ADFM Velocity Profiler to measure flow rates in a variety of conditions with a minimal amount of time required to install and setup the instrument. This new technology has the potential to provide flow measurement in areas where traditional discharge measurement devices are impractical. It also could be a valuable tool for calibrations of existing flow measurement structures and for research studies that require velocity profile measurements.
• This instrument can accurately measure detailed velocity profiles in an open channel which can be used for engineering studies. For example, the ADFM can be used to measure velocity profiles which describe flow into and around strutures such as fish screens or fish ladders.

Related links

MGD Technologies Inc. is a professional firm specializing in the assessment of the condition and performance of underground utility systems. The services are provided by highly trained individuals, experienced in the innovative application of integrated, advanced technologies. MGD's ADFM measures the flow velocity at discrete points throughout the depth of flow.
MGD Technologies Inc.
9815 Carroll Canyon Road, Suite 200
San Diego CA 92131 USA
Tel. 619 695 9225
Fax 619 695 6890