Demystifying Doppler Flowmeters in Sewage Flow Measurement
The measurement of flows in open channels has been an interesting challenge for many decades. Today, increased regulatory pressures and system improvement requirements for open channel conveyance have stepped up the requirements for knowledge of actual flows within conduit systems. For many years, there were very limited options for determining open channel flows.
The generally accepted practice was the use of a level curve verse a flow curve with some primary element. The flow method required major construction and was very expensive. Point velocity flowmeters also became popular and for many years were successfully used as an option to primary elements at a significantly less cost.
Both techniques have significant margins of error but, with the lack of options available, they were the accepted techniques. Recently, area velocity portable flowmeters utilizing Doppler technology for use in open-channel measurement have become in vogue based on their price, general ease of installation and apparent accuracy.
This discussion is from the consumers' standpoint. It includes the various aspects of Doppler technology, a survey of the manufacturer's application of this technology and a toolbox for the educated consumer to make a purchase. By providing a base knowledge of this product area, the purchaser can determine the best product to fit their particular application.
History of Doppler Technology
Sound travels as a wave through some medium. Water is a better conductor of the wave than air as it is more dense. Most physics text books will note that sound travels four times faster in water than air. However, the actual tone depends on the perspective of the reference. Therefore, the tone of a given sound wave will seem to change or "shift" based on the reference point of the listener. Christian J. Doppler documented and developed equations for the shifting of frequencies passing a fixed object. This phenomenon is best illustrated by the passing of a police car, train or ambulance while you are standing still. The sound appears to change tone as the distance increases from your point of reference. This changing of the tone is referred to as the "Doppler Shift."
This shift in tone was discussed in theory but appeared to have little practical application until the twentieth century. Engineers were successful in determining a velocity in a closed channel based on averaging the shift in tones within a pipe. Since the cross section of the flow was fixed, a flow rate was readily determined. However, the application was limited to uniform liquids. In the case of water, air bubbles of uniform size were injected into the flow so the sound could reflect. At first, sewage was not a practical application due to the non-uniform size of the suspended solids. In the late 1980s, select companies had the ability to use averaging algorithms in their equipment to determine a velocity based on averages of shifts over a time average. This was performed by taking a set time (e.g.; one minute) and firing the Doppler once a second, listening for the shifts and taking an average of all the velocities for the given time frame. This technique produced marginal results. Most of the applications were limited to the full pipe situation with very little application to the open-channel world.
Doppler in Open-Channels
The early 1990s brought massive advances in computer technology and software development. These advances were applied to the open-channel arena and products began to appear using Doppler technology.
Two prominent methods divide the market today. One is the Peak Doppler method and the second is the Weighted Average Doppler application. Both methods have advantages and disadvantages, and it definitely depends on the application to determine the best selection. Both applications utilize a fast fourier transform for determining velocity components. Unless you are an electrical engineer, the subject of Fourier Transforms and frequency domains may be intimidating. Therefore, they are discussed here in reference only and described for this application.
Both devices take the sound shifts received from the suspended solids in motion and develop a profile in the frequency domain. The faster the object reflecting the sound moves in the liquid, the greater the phase shift, or change in tone the meter registers. A Fast Fourier analysis of these phase shifts (or tone changes) is made and a distribution pattern is established. Fourier analysis is cumbersome and for a long time not possible on portable-type devices. The advances in micro-PC applications make this possible today. Most dataloggers and RTU/PLC families have the Fast Fourier Transform (FFT) as a pre-set call or function block. Fourier transformation permits these tone changes to be converted to a form understandable on a graph.
Most studies in this area are performed over time. In this case, substitute frequency for time and the chart becomes understandable. This is essentially moving from the time domain to the frequency domain, where one is examining the change in frequency instead of a change in time. Time becomes a constant. This is a snapshot in time for the distribution of frequency shifts. The velocity is determined from these shifts.
The most popular and easiest algorithm to implement is the Peak Doppler. This method takes the highest peak velocity, performs a manufacturer's reduction algorithm and provides an average velocity reading. An example of a reduction algorithm is to, say provide a figure of 70 percent of the peak reading representing an average velocity of the cross section. The highest peak represents the maximum velocity. If the highest recorded velocity is 5.1 ft/s, a reduction factor of 70 percent applied would make the average velocity for calculation 3.57 ft/s (5.1 ft/s x 0.70). The 3.57 ft/s is then used as part of the continuity equation (Q = VA) for determining flow. These devices require minimum software code and less memory than their counterpart. Furthermore, in a fairly clean pipe, with sufficient up and downstream straight runs, a fairly accurate average velocity will be rendered.
Peak Dopplers have significant drawbacks when operated under less than ideal conditions. If the flow distribution is skewed and the sound beam is not of sufficient width, the peak frequency shift to generate the velocity for the algorithm may be missed. Also, if a blockage occurs downstream of the meter and the block acts as an orifice, an unrealistic average velocity will be registered.
This is because there will be a small amount of flow through the "orifice" with most being stagnant or of little movement. The reduction algorithm assumes an ideal velocity profile and will show, as a possibility, more flow than the design capacity of the pipe. Therefore, if 5 percent of the profile is running at 7 ft/s due to the orifice and 95 percent of the profile is running at 0.5 ft/s, the reduction algorithm will not properly reflect the actual profile as it does a simple reduction of one value (peak). It is very important that an ideal or near-ideal velocity profile be produced when utilizing a Peak Doppler meter.
The ability to give equal preference for each sample would provide the user with a better feel for the velocity within a profile, especially for skewed or partially blocked pipes. This is the idea behind a Weighted-Average Doppler meter. For each part of the frequency distribution, the random returns on the signals are analyzed and weighed evenly for the number of samples in a given period. Weighted-Average Dopplers appear to have this advantage and provide a better contribution of the actual profile than a Peak Doppler meter.
However, there are some disadvantages to these meters. The first is that it is a very complex algorithm to code as the array formed can be from some preset design to true chaotic. The random return can also be hampered by the size of the sound beam and its frequency.
There is also an argument concerning the accuracy of almost totally random distribution profiles based on the received signals. Presently, there is no computer program that can truly perform total random analysis. There are many attempts to do this, but most rely on some pre-fixed pattern.
The sample period can also pose a problem as the conditions may dictate variable size samples for more accurate results based on solids content of the pipe being measured. The width of the sound beam is also important since too narrow a beam may miss many of the contributing particles for which the tone shift would significantly contribute to the total velocity profile. Too wide a beam will have echoing effects in small conduits.
In general, testing the various products manufactured under the two methods have proven them to be fairly reliable and reasonably accurate. They provide good temporary monitoring information and can function for many weeks without maintenance.
Independent testing has shown that the specific sensor shape has advantages for unique applications. The electronics packages for all of the units on the market adapt well to the sewer environment (moisture, corrosion, etc.). The unique feature of the doppler flowmeter when observed during several testing periods is the ability for it to fire under mild fouling conditions and maintain a reasonable accuracy. Several documented tests in Philadelphia have qualified this finding. There is a question in using the doppler under large scale testing (i.e.; pipes over 60 inches) simply from lack of testing. However, for pipes under 36 inches, the doppler technology proves to be accurate and repeatable. Ongoing development and research will make dopplers the preferred technology for sewage flow measurement especially for those municipalities where budgets are of concern.