Flow meters: which technology, and what it will not measure

Flow measurement has more technologies than any other process variable, and the selection is usually made on price after the technologies that would have worked have been ruled out by something in the process.

Electromagnetic

A magnetic field across the pipe, electrodes in the wall. A conductive fluid moving through the field induces a voltage proportional to velocity.

No moving parts, no obstruction, no pressure drop, excellent turndown, and largely indifferent to viscosity, density, temperature and profile disturbance. Handles slurries and solids in suspension well.

The single hard requirement is conductivity, usually above about five microsiemens per centimetre. Water based processes are fine; hydrocarbons and deionised water are not.

For a conductive liquid this is nearly always the right answer, and the question is only whether the budget allows it.

Coriolis

The fluid passes through vibrating tubes, and the Coriolis effect twists them by an amount proportional to mass flow.

Measures mass directly rather than volume, and gives density and temperature as well. Very accurate, and indifferent to profile, so straight run is not an issue.

The costs are real: expensive, heavy, a noticeable pressure drop, and sensitive to external vibration which needs mounting attention. Entrained gas degrades it.

Right when mass is genuinely what matters, in custody transfer, batching by weight, or where density is a process variable in its own right.

Vortex

A bluff body in the flow sheds vortices at a frequency proportional to velocity.

Works on liquid, gas and steam, which makes it the usual answer for steam, where most alternatives struggle. No moving parts, tolerates high temperature.

Two limits: it needs a minimum velocity to shed vortices at all, so turndown at the low end is poor, and it needs straight run, typically ten to twenty diameters.

Ultrasonic

Either transit time, comparing pulses sent with and against the flow, or Doppler, reflecting off particles.

Transit time clamp-on is the interesting case: it fits to the outside of an existing pipe with no cutting, no shutdown and no pressure drop. For a retrofit measurement on a line that cannot be broken, it is often the only practical option.

Accuracy is lower than an inline meter, it depends on knowing the pipe wall material and thickness accurately, and it does not like liner or scale.

Turbine

A rotor spun by the flow, pulses counted.

Accurate on clean, low viscosity liquids, and cheap. Also has bearings, which wear, and is damaged by dirt and by running dry.

Best where the fluid is clean and consistent and somebody will maintain it. It is a poor default.

Differential pressure

An orifice plate, venturi or nozzle, with a differential pressure transmitter across it. Flow is proportional to the square root of the differential.

It is everywhere because it has been standard for a century, the primary element is cheap, and every plant already knows how to maintain it.

The square root relationship is its weakness: at a quarter of full flow the differential is one sixteenth, so turndown is around three to one and low flow readings are poor. An orifice plate also imposes a permanent pressure loss, which is energy paid for continuously.

The questions, in the order that eliminates fastest

• Is the fluid conductive? If not, no magmeter.
• Is it steam? Then vortex or differential pressure.
• Do you need mass or density? Then Coriolis.
• Can the line be broken? If not, clamp-on ultrasonic.
• How much straight run is available? This eliminates more options on real installations than anything else.
• What is the lowest flow that must be measured accurately? Check the turndown against it before choosing.