Flow Measurement
Orifice plates, Venturis, pitot tubes, rotameters, vortex, magnetic, ultrasonic and Coriolis meters, and how to choose.
Choosing a flow meter
Flow can be measured from a pressure difference, from the speed of the fluid, or directly from mass. The right meter depends on the fluid (clean liquid, dirty slurry, gas, steam), the pipe size, the accuracy you need and the pressure loss you can afford.
| Meter | Principle | Typical accuracy | Turndown | Pressure loss | Notes |
|---|---|---|---|---|---|
| Orifice plate | differential pressure | 1 to 2 % | 3:1 to 4:1 | high | cheap, standard, clean fluids; needs straight pipe |
| Venturi | differential pressure | 0.5 to 1 % | 4:1 | low | expensive, long; good for large pipes and dirty water |
| Pitot tube | dynamic pressure at a point | 2 to 5 % | 3:1 | very low | air ducts, large pipes; clean fluids |
| Rotameter | float in a tapered tube | 2 to 5 % of full scale | 10:1 | low, constant | local indication; clean fluids |
| Turbine | rotor speed | 0.25 to 1 % | 10:1 to 20:1 | moderate | clean, low-viscosity fluids; bearings wear |
| Vortex | vortices shed from a bluff body | 1 % | 10:1 to 20:1 | moderate | gases, steam, liquids; not for low flow |
| Magnetic | voltage induced by a conductive liquid | 0.2 to 1 % | 50:1 to 100:1 | none | conductive liquids only (water, slurries, acids); no moving parts |
| Ultrasonic | transit time or Doppler | 0.5 to 2 % | 50:1 | none | clamp-on available: no pipe cut; needs full pipe |
| Coriolis | tube vibration gives mass flow directly | 0.1 to 0.5 % | 100:1 | moderate | most accurate, gives density too; costly |
Differential-pressure meters
Orifice plates, Venturis and nozzles all put a restriction in the pipe and measure the pressure difference across it. By Bernoulli and continuity (see Continuity, Bernoulli and the Energy Equation) the flow is proportional to the square root of that difference:
Q = Cd × A₂ × √(2 ΔP ÷ ρ) ÷ √(1 − β⁴)
Because Q ∝ √ΔP, a meter that gives a good signal at full flow gives a weak one at low flow: at 10 percent of full flow the differential is only 1 percent of the full-scale value. That is why orifice meters have a turndown of only 3:1 or 4:1.
Orifice plates
Worked example: an orifice meter on a water line
- 100 mm pipe, 60 mm orifice (β = 0.60), measured differential 25 kPa, Cd = 0.61. The approach factor is 1 ÷ √(1 − 0.130) = 1.072.
- Orifice area 2 827 mm². Q = 0.61 × 1.072 × 0.002827 × √(2 × 25 000 ÷ 998.2) = 13.1 L/s (47.1 m³/h), a pipe velocity of 1.67 m/s.
- The permanent pressure loss is about 63 percent of the differential: 15.7 kPa, paid for every hour the meter runs.
- Turndown: at one fifth of the flow (2.6 L/s) the differential falls to 1.0 kPa; at one tenth (1.3 L/s) it is only 0.25 kPa, within the noise of many transmitters.
Orifice plates are made to ISO 5167 or AGA-3 dimensions and, with the right straight pipe and corner or flange taps, need no calibration. Edge wear rounds the sharp edge and slowly reads low; fouling and a wrongly installed plate read high or low by several percent.
Venturi and nozzle meters
A Venturi uses a smooth converging cone, a throat and a diffuser, so the pressure loss is small and dirty fluids do not collect. The flow equation is the same with Cd ≈ 0.98.
Taking the Venturi from Continuity, Bernoulli and the Energy Equation (100 mm narrowing to 50 mm with a measured differential of 48.6 kPa): Q = 0.98 × 1.033 × 0.001963 × √(2 × 48 600 ÷ 998.2) = 19.6 L/s, within 2 percent of the 20 L/s that produced it, the difference being the discharge coefficient. The permanent loss is only about 10 percent of the differential (4.9 kPa), against 63 percent for the orifice above.
Pitot tubes and averaging probes
A pitot tube faces the flow and measures the stagnation pressure; with a static pressure tap beside it, the difference is the dynamic pressure ½ ρ v², so v = √(2 ΔP ÷ ρ). It reads the speed at one point, so multiply by a profile factor (about 0.9 in a developed turbulent pipe flow) or traverse the pipe at several points.
Worked example: airflow in a duct
- A pitot reads 450 Pa in a 300 mm duct (air ρ = 1.204 kg/m³): v = √(2 × 450 ÷ 1.204) = 27.3 m/s at the center.
- The mean is about 0.9 of that, so Q ≈ 0.9 × 27.3 × 0.0707 = 1.739 m³/s (6 262 m³/h). Pitot tubes are the standard way to balance ventilation systems.
Variable-area, vortex, magnetic, ultrasonic and Coriolis
- Rotameter (variable area): a float rises in a tapering tube until the flow area gives a balance; the flow reads off the scale. Cheap, no power, and the scale is only valid for the fluid it was calibrated for.
- Turbine: a rotor spins at a speed proportional to the flow. Precise and fast, but wears and is damaged by dirt, and needs clean low-viscosity fluids.
- Vortex: a bluff body sheds vortices at a frequency f = St v ÷ d (the Strouhal number St ≈ 0.2). A 20 mm bar at 2 m/s sheds at 20 Hz. Linear, no calibration drift, but needs a minimum flow to shed cleanly.
- Magnetic (electromagnetic): a conductive liquid moving through a magnetic field generates a voltage proportional to its speed, E = k B D v. No obstruction, no moving parts, and the reading does not depend on density or viscosity. In an 80 mm line the usable range (0.5 to 10 m/s) is 9.0 to 181 m³/h. It does not work for oils, gases or very pure water.
- Ultrasonic: sound pulses travel faster with the flow than against it; the time difference gives the speed. Clamp-on versions need no pipe cutting, ideal for checking an existing line.
- Coriolis: a vibrating tube twists in proportion to the mass flow going through it. It measures mass flow directly (and density), unaffected by viscosity or profile; the most accurate and the most expensive.
Installing a meter properly
- Straight pipe: most meters need 10 to 20 pipe diameters straight upstream and 5 downstream, to give a developed velocity profile. Flow straighteners cut that, but a bend, valve or pump just before a meter is the commonest source of error.
- Full pipe: a liquid meter must always run full. Mount it on a rising section or a low point, not at the highest point where air collects.
- Pressure taps: for DP meters, take taps from the side of a horizontal pipe (liquids: below the centerline; gases: above) and use seal pots or purge for dirty liquids.
- Calibrate and check: a periodic check against a reference or a bucket-and-stopwatch test is cheap insurance.
- Remember the unit conversions: flow in m³/h, L/min, gpm or kg/s: the Flow Rate Converter handles it.
Related
- Previous: Pumps, Pump Curves and NPSH. Next: Compressed Air and Gas Flow.
- Pressure, Level and Flow Sensors: signals, wiring and installing the transmitters.
- Pipe Flow and Pressure Drop Calculator.
- Fluid Mechanics and Pipe Flow: the whole guide.
FAQ
Why does an orifice plate need so much straight pipe?
The coefficient is calibrated for a fully developed velocity profile. A bend, valve or reducer upstream distorts the profile and the meter can read several percent wrong. ISO 5167 gives the straight lengths for each fitting.
Why does a magnetic meter not work on oil?
It relies on the liquid being conductive (above about 5 µS/cm). Oils and most hydrocarbons are not; use Coriolis, vortex, turbine or ultrasonic instead.
What is turndown?
The ratio of the highest to the lowest flow over which the meter holds its stated accuracy. A 10:1 meter works from 10 to 100 percent of its maximum.
Which meter for compressed air or steam?
Vortex, thermal mass flow or orifice with temperature and pressure compensation. Gas density changes with pressure and temperature, so volume flow must be corrected to standard conditions.
References
- ISO 5167-1 to -6, Measurement of fluid flow by means of pressure differential devices inserted in circular cross-section conduits running full.
- Miller, R. W., Flow Measurement Engineering Handbook, McGraw-Hill.
- Spitzer, D. W. (ed.), Flow Measurement: Practical Guides for Measurement and Control, ISA.
- ASME MFC-3M (orifice, nozzle and Venturi) and MFC series standards.
Last reviewed: October 2026. Accuracies and discharge coefficients are typical.