Accuracy of ultrasonic flow meter

[llamallama]

All,

I need to know the factors effecting the accuracy of a clamp-on ultrasonic flow meter. I recently purchased one to do some in-plant potable water conservation work. It was an Economy-bay model and the tech support is non-existent, but it was what I could sneak in under the budget. It seems to be repeatable, and I believe it could be set up to be fairly accurate. Also, the pipe is new installation, so buildup of scaling or corrosion isn't likely. And the problem persists on several installation locations. The problem is this:

The meter asks for various parameters which I have entered as accurately as I can. From that information the unit calculates a spacing for the installation of the transducers (which I set with calipers). Based on all that it outputs flow rate along with the ratio of the measured speed of sound in the fluid to the theoretical. This ratio is fairly high and the instructions say to check the transducer installation if that occurs. My question is, should I accept the flow rate as accurate with the measured speed of sound 85-90% of the theoretical, or adjust the transducer spacing until the theoretical and measured speed of sound align? Or adjust some other parameter. I'm assuming the meter calculates the length of the travel path based on the user input.

Input parameters I've double checked as correct: Pipe diameter (4.50"), Pipe wall thickness (0.12"), Pipe material (Stainless), Liquid (water 20C), Transducer style (TS-2 from the nameplate), Transducer setup (V arrangement), Pipe Lining (None).

Output: sound velocity (1251 m/s), flow rate (49.xx gpm), signal strength out of 99.9 (88.9, 89.1), signal quality out of 100% (94%).

Any help would be appreciated.

 

[danw2]

To your question about factors affecting ultrasonic flow measurement:

The flow profile affects the transducer energy path through the medium, so the accuracy of the measurement is influenced by upstream turbulence (control valve, intrusions), piping configuration (elbows and Tees), straight run upstream/downstream and Reynolds number. Siemens claims a proprietary means of compensating for various piping anomalies, but no compensation for being downstream of a control valve.

The accuracy is affected changes in the speed of sound in the fluid as fluid temperature changes or differs from one fluid to another. If the medium is always ground temperature water then you're good, but if the fluid itself changes or the temperature changes, you need to ascertain the fluid's sonic velocity at that temperature.

Zero shift depends on manufacturer implementation;; transit time is not necessarily self zeroing like a magmeter. zero offset can be significant.

Mounting the transducer on a pipe seam will affect its signal and the resulting accuracy.

 

[llamallama]

Thanks for the reply. I can see there's a lot to these meters. I guess my best bet is to follow the install dimensions suggested by the meter and assume that the disparity in the calculated SOS vs the theoretical is to be expected?

I did get a chance to stop flows in the line. I would assume at that point (V=0) the calculated SOS in water shouldn't be affected by the surrounding piping configuration. The SOS was still the same as with the flows running at 50-150gpm (about 1250 m/s). I'm not over a seam, and the pipe dimensions are pretty close to published values. My guess is the meter calculated transducer spacing isn't right (the meter may be basing the calculated spacing on a different type of transducer head?) or the meter isn't accounting for the pipe wall correctly.

I'm not an instrumentation guy, so I have a limited knowledge of instrumentation. Does your reference to "zero shift" mean the calculation could read low, but the comparison would still be accurate? My only concern is that the meter instructions do say that if the difference is greater than 3% the installation is probably incorrect, and (1465/1250 ) = 117%. The meter also makes this comparison and shows the 117% delta.

Because the readings are consistent, I'm rooting for this little guy to be something I can use if I can figure out the accuracy issue.