Clamp-on Ultrasonic Flowmeters
Many ultrasonic flowmeter designs use sensors that are clamped onto the pipe. The concept of clamping a device onto a pipe to determine the flow inside the pipe has a certain appeal. This not only reduces piping changes, but also allows flow measurement using an instrument that has no wetted parts.

Spool Ultrasonic Flowmeters
Some ultrasonic flowmeters designs utilize sensors mounted on a spool piece that is inserted in the pipe. This design requires that the pipe be taken out of service for installation. The spool piece design allows the spool piece, sensors, and transmitter to be calibrated under flowing conditions in a flow laboratory prior to installation.

Probe Ultrasonic Flowmeters
Probe designs use wetted sensors that are in contact with the fluid. Some designs allow the probes to be installed (hot-tapped) onto an existing pipe without taking the existing pipe out of service. Most designs require that the pipe be taken out of service for installation.

Gas, Stack Gas, and Flare Gas Ultrasonic Flowmeters
Ultrasonic flowmeters are typically designed for use in liquid service. Applying this technology to gas service introduces additional sources of uncertainty that include (but not limited to) the effects of pressure, temperature, and composition. To overcome these issues, many gas ultrasonic flowmeters are designed for multi-path operation for gases with well-known properties.

Excerpted from The Consumer Guide to Ultrasonic and Correlation Flowmeters.

In hot condensing vapor service such as steam, the transmitter should be located below a liquid seal.  The liquid seal forms a barrier between the hot vapor and the transmitter that protects the transmitter by not allowing hot vapor to contact the transmitter. 

Typically, the taps are located on the side of the pipe and the impulse tubing is constantly sloped downward to the transmitter.  This configuration allows a liquid seal to form in the impulse tubing without pockets of non-condensable gas that can affect the measurement.  Alternatively, the tap can be located on the top of a horizontal pipe with the impulse tubing rising and then constantly sloping downward to the transmitter to form the seal. 

In addition, the temperature of both impulse tubes should be the same in non-horizontal segments of the impulse tubing runs.  Different impulse tubing temperatures can result in different liquid densities in each seal and can cause measurement error.  The error associated with relatively short impulse tubes will generally be small.  However, the error associated with a flowmeter element located 15-30 meters above its transmitter (typical of older installations) can cause significant measurement error. 

The two impulse tubes are typically routed to the transmitter in parallel with each other.  If they are installed separately, somewhat different routing, supports, and degradation of their insulation can result in measurement error caused by temperature differences between the liquids in the tubing.  As a practical matter, impulse tubing of significant length is typically installed in a tubing bundle with common insulation and heat tracing (if needed).  This tends to reduce the temperature difference between the liquids in the impulse tubing and hence the measurement errors that the difference can cause.  This installation also tends to reduce the installed cost of the impulse tubing. 

The impulse tubes that I inspected were installed together in tubing bundles.  I did note that there were some slight dips in relatively short horizontal runs where someone appeared to have stepped on the impulse tubing.  This is not bad considering the age of the facility.

This article originally appeared in Flow Control magazine. 

1. High temperature
2. High pressure
3. Wiring problem
4. Configuration problem
5. Composition change
6. Steam tracer off
   

High pressure and high temperature should not affect the flow measurement --- presuming that the pressure and temperature limits of the flowmeter have not been violated.  Therefore, Answer A and Answer B should not cause this problem.  In addition, a number of wiring problems (Answer C) and configuration problems (Answer D) that could cause the flow (signal) to be zero can occur. 

Liquid composition changes (Answer E) can also cause this problem to occur.  In general, process conditions whereby the liquid becomes more viscous can cause Reynolds number to fall sufficiently to cause the vortex shedding flowmeter to turn off and "measure" zero flow. 

Surprisingly, turning a steam tracer off (Answer F) can also cause this problem to occur.  In particular, turning off a steam tracer will reduce the liquid temperature and increase its viscosity.  Increasing viscosity reduces Reynolds number that can (as in Answer C) cause the vortex shedding flowmeter to turn off and "measure" zero flow.

Additional Complicating Factors

Liquids are typically assumed to have a constant viscosity at a given temperature.  However, some liquids are non-Newtonian whereby their viscosity changes at a given temperature depending upon how the liquid is handled.  For example, the apparent viscosity of some liquids (such as ketchup) is initially high but tends to fall drastically once flow occurs.  Non-Newtonian fluids are typically more difficult to measure. 

This article originally appeared in Flow Control magazine.