Ultrasonic Flowmeters

I think the OP refers to flare gas, not flue gas. This is the flammable gas sent to a flare for safe release to the atmosphere, not exhaust gas from a furnace:

Flare gas flow measurement is especially challenging due to the high turndown requirements and the variable gas composition. Here is a sampling of different flowmeters and why they don't work particularly well for flare gas:

• Orifice/venturi/Pitot tube: calibration affected by gas density (variable), plus poor turndown due to square-root characterization
• Vortex: low-flow cutoff makes the meter read zero at low flow rates (i.e. poor turndown ratio)
• Thermal mass: calibration affected by specific heat of the gas (variable)
• Turbine: potentially high temperatures, also affected by swirl

At the oil refinery where I used to work, we tried a thermal mass meter on the main flare line, with extremely poor results. If you can put a thermal mass flowmeter on each line feeding the flare header (where you have a more consistent composition to calibrate each flowmeter) and then sum up all the individual line measurements to achieve a total, you might get a workable solution. The economics of this approach of course varies with the number of process streams dumping into the flare header.

There are two basic types of ultrasonic flowmeter: Doppler and Transit-Time. Both types of ultrasonic flowmeter work by "pinging" a sound wave pulse into the fluid flow -- it's what they do with the result that differs. Doppler flowmeters detect echoes off of any particulate matter or bubbles in a liquid flow stream, the frequency shift of the echoed wave compared to the original pulse indicating the velocity of the object. Transit-time flowmeters use pairs of transducers sending sound wave pulses back and forth diagonal to the flow direction, comparing the time-of-flight for the sound wave going upstream (against the flow) versus downstream (with the flow). The difference in time is directly proportional to fluid velocity. Only transit-time ultrasonic is suitable for gas flows; Doppler can't detect a flowing gas because there's nothing in the gas stream big enough to *reflect* a sound wave.

Now, on to your question about clamp-on meters: there are inexpensive, and (as always) you generally get what you pay for. One of the chief problems with clamp-on meters is poor acoustic coupling between the ultrasonic transducer and the fluid stream. Even if you manage to get good coupling between the transducer and the pipe wall (which involves careful placement of the transducers plus a gel to enhance conduction of the sound wave), there is the problem of "ringing around the pipe" where the sound waves may ripple around the circumference of the pipe to the second transducer rather than go through the fluid inside the pipe.

Insertion-style ultrasonic meters have their transducers embedded in the flowtube itself, so that the tube wall is neither a barrier to the sound nor an unintentional acoustical path for the sound to skirt the flow. With gas flows, you need all the help you can get achieving good acoustical coupling with the gas, which is one reason why your vendor is saying the insertion type is more suitable.

There's another good reason why a clamp-on ultrasonic meter would be unsuitable for flare service (even if you could get a strong ultrasonic signal through the gas) and that's because clamp-on meters are single-path devices. In order to get good accuracy measuring gas flows through a very wide range of Reynolds number (i.e. excellent turndown) you need to take transit-time measurements of velocity through multiple paths in the pipe. This is called a "multi-path" or "multi-chord" ultrasonic flowmeter, now becoming popular for measuring the flow of natural gas according to the AGA9 standard in America. As you might guess, these are very expensive -- the latest quote I've heard is $2000 US per inch pipe diameter. Here's what one looks like:

In addition to the need for multi-path measurement, you'll also need to measure the flowing pressure and temperature of the flare gas, because just measuring velocity doesn't tell you how much gas quantity (in mass units of standard volumetric units) you've got flowing through the pipe. This requires more sensors and a flow computer, all of which raises the cost.

An alternative to ultrasonic for flare flow measurement is the relatively new technology of optical flowmeters. The claims made for these meters are impressive (turndown ratios of 1000:1 or better!), and the technology itself is intriguing. Here is one such product, designed specifically for flare gas flow measurement:

http://www.photon-control.com/opticalflowmeters.html

Clamp-on ultrasonic flow meters are not used in flare gas applications for several reasons:

Clamp-on transit-time meters operate based on Snell's law (refraction). Sound waves will change direction as they transit from one medium to another as a function of the soundspeed (C) in these different mediums, and the sine (SIN) of the angle that the ultrasonic signal is introduced to the medium:

Cpipe/SINpipe = Ctransducer/SINtransducer = Cgas/SINgas

So, if the signal is introduced to the pipe at an angle of 45 degrees to the pipe wall (for example), the angle that the signal travels through the pipe wall will be something other than 45 degrees (where the pipe soundspeed is different that the transducer soundspeed). Also, the angle will most certainly be very different through the gas, as the soundspeed of the gas will be considerably lower than either the transducers, or the pipe material.

Example soundspeeds:

CS pipe soundspeed ~3200 m/s

Gas soundspeed ~400 m/s

Transducer soundspeed ~2450 m/s

A typical soundspeed for flared gas in a plant environment may be around the 400 m/s range. However, this soundspeed can vary dramatically depending on the gas composition, and on the temperature and pressure. It may be 350 m/s at one moment, or 450 m/s at another, depending on upsets in the plant.

If you crunch the numbers into the equation above, you will arrive at angles through the process (the gas) that will be around maybe 5 - 6 degrees.

So, the first problem you are faced with is that there will be very little axial separation of the upstream and the downstream transducers (the transducers will be almost directly opposite each other, but on opposite sides of the pipe). To appreciate what this means, imagine the transducers were exactly opposite each other; there would be zero time-differential between the upstream transmission and the downstream transmission (at any flow velocity up to infinity). The smaller the time-differential, the greater the inaccuracy. The interrogation angle with wetted (inserted) transducers is typically 45 degrees (sometimes 60 degrees, or even higher), so the time-differential is much much greater. And, there is NO change in the signal angle due to refraction.

The second problem is signal attenuation. With clamp-on flow meters, the signal attenuation just getting through the pipe wall is huge. The amount of sound energy left, after transiting through the first pipe wall, is about 0.007% of what the initial energy was. Square that value to see what energy is left due to the second pipe wall, after the signal has transited across the gas. If the density of the gas increases, this helps us. The soundspeed increases, and the angle widens. However, the density does not increase enough unless the pressure increases dramatically (the last thing you want in a flare application). That is why reputable clamp-on gas meter manufacturers specify minimum gas pressure requirements, and will not use them in flare applications. Enough said.

The third problem is blow by. At extreme velocities (flare applications during plant upsets and emergency flares), the ultrasonic signal can be pushed past or blown by the receiving transducer; and never heard from again (pardon the pun). That is another reason for increasing the angle. The greater the separation of the transducers, the bigger the angle becomes and the greater the tolerance to this blow by effect. However, this is a balancing act between improving the tolerance to blow by, and not creating such a long measurement path that the signal attenuates to nothing.

Multi-path meters:

Multi-path transit-time meter designs can sometimes be used for flare applications. However, most of the traditional multi-path meter manufacturers have minimum pressure requirements for their meters (usually a couple of hundred PSI or greater), which eliminates them as candidates for flare measurement. But, with a well designed meter, a single path (in some cases two paths may be required - one for low velocity, and one for high) ultrasonic meter is very capable of providing good measurement accuracy that will meet regulatory compliance at both low velocities (no flare) and extreme velocities (up to 120 m/s). Most jurisdictions require +/- 5% measurement inaccuracy for flares.

Bill Pesklevits

Procon Systems Inc.