Flow measurement

In short

Flow measurement determines the quantity of a fluid passing a point per unit time — volumetric (m³/s, LPM, m³/h) or mass (kg/s, t/h) — and is the most commercially sensitive measurement in process plants because fluid is bought, sold and taxed by it. The physics ranges from differential-pressure (DP) devices based on Bernoulli's equation to electromagnetic, ultrasonic, Coriolis and vortex-shedding principles. It is a guaranteed scoring area in GATE (IN and EC/EE papers) and in university Sensors/Industrial Instrumentation papers, usually through orifice sizing, discharge-coefficient and square-root-relationship numericals. Indian employers — Reliance, IOCL, BPCL, ONGC, GAIL, NTPC, Tata Steel, L&T, Emerson, Yokogawa, Endress+Hauser, ABB, Honeywell, Forbes Marshall — test it in technical rounds for instrumentation, commissioning and custody-transfer roles.

Explained

Fundamentals and classification

• Volumetric flow rate Q = A·v (m³/s); mass flow rate ṁ = ρ·Q (kg/s). For gases, volumetric flow must always be stated at reference conditions (Nm³/h at 0 °C, 1.013 bar, or Sm³/h at 15 °C) because density varies with pressure and temperature.
• Reynolds number Re = ρvD/μ decides the flow regime: laminar below ~2,000, transition 2,000–4,000, turbulent above ~4,000. Most meters are calibrated for turbulent flow; a laminar reading on a turbine or DP meter is wrong.
• Velocity profile matters: turbulent flow has a flat profile, laminar flow a parabolic one with centreline velocity = 2 × mean velocity. Insertion and point-velocity meters must correct for this.
• Classification by principle:
  1. Differential pressure (inferential): orifice, venturi, flow nozzle, pitot tube, Annubar, Dall tube, elbow meter, target meter.
  2. Variable area: rotameter (constant ΔP, variable area — the inverse of an orifice).
  3. Positive displacement: nutating disc, oval gear, lobed impeller, piston, diaphragm (household gas meters).
  4. Velocity/inferential rotating: turbine, propeller, anemometer, cup type.
  5. Electrical/modern: electromagnetic, ultrasonic (transit-time, Doppler), vortex shedding, thermal mass, Coriolis mass.
  6. Open channel: weirs (rectangular, V-notch), Parshall flume, channel flow by Manning's equation.
• Direct versus inferred mass flow: only Coriolis and thermal meters measure mass directly; everything else needs density compensation (pressure and temperature inputs) to report mass.

Differential-pressure meters — theory and hardware

• Derivation outline (examined frequently): write Bernoulli between upstream tap (1) and throat (2) for incompressible, frictionless, horizontal flow — p₁/ρ + v₁²/2 = p₂/ρ + v₂²/2 — and continuity A₁v₁ = A₂v₂. Eliminating v₁ gives the ideal throat velocity, then multiply by A₂ for ideal discharge and by the discharge coefficient C_d for the actual one. The factor 1/√(1 − β⁴) is the velocity-of-approach factor E.
• Beta ratio β = d/D (throat or bore diameter ÷ pipe diameter), normally 0.2–0.75. Low β gives good ΔP and accuracy but large permanent pressure loss; high β is the reverse.
• Orifice plate: a thin plate with a sharp-edged concentric bore, clamped between flanges. C_d ≈ 0.60–0.62 (vena-contracta/flange taps). Cheapest, best-documented, no moving parts; but permanent head loss is high (40–80% of ΔP), the sharp edge wears and rangeability is only about 3:1 or 4:1.
  • Variants: concentric (clean liquids/gases), eccentric and segmental (bore offset to the bottom for slurries, or to the top for gas-bearing liquids), quadrant-edge/conical-entrance (low Reynolds, viscous fluids), integral orifice (small lines).
  • Tap types: flange taps (25 mm each side), corner taps, vena-contracta taps, pipe taps (2½D and 8D).
  • Vent hole (above bore) for liquid service carrying gas; drain hole (below bore) for gas service carrying condensate.
• Venturimeter: converging cone (~21° included), short throat, long diverging cone (5–15°). C_d ≈ 0.95–0.99, permanent loss only ~10% of ΔP, handles some solids — but it is bulky, costly and needs a long straight run. Preferred for large water mains and power-station circuits.
• Flow nozzle: between the two — C_d ≈ 0.95–0.98, loss ~30–50% of ΔP, compact, erosion-resistant, standard for high-velocity/high-temperature steam flow.
• Pitot tube: measures stagnation minus static pressure, giving point velocity v = √(2ΔP/ρ); the pitot-static (Prandtl) tube has both ports in one probe. Low cost, negligible obstruction, but very small ΔP and sensitive to misalignment and dirt. The Annubar/averaging pitot has multiple ports across the diameter and reports an averaged velocity.
• Secondary element: a DP transmitter (capacitance or piezoresistive diaphragm) with a 3-valve or 5-valve manifold, impulse lines, and 4–20 mA/HART output.
• Square-root extraction: since Q ∝ √ΔP, the transmitter or DCS must take the square root. At 10% of range the ΔP is only 1% of span, so errors blow up at low flow — hence the poor rangeability. Modern smart transmitters do the extraction digitally with a low-flow cut-off.
• Installation rules: DP taps horizontal or slightly below for liquids (impulse lines sloping down to the transmitter, self-venting), above the line for gases (sloping up), with seal pots or diaphragm seals for steam and corrosive service. Straight run typically 10–20D upstream and 5D downstream, or use a flow straightener/conditioner.
• Expansibility (expansion) factor Y or ε (<1) corrects compressible gas flow for density change across the restriction; it is 1 for liquids.

Variable-area, positive-displacement and turbine meters

• Rotameter: a tapered transparent tube (glass or metal with a magnetic follower) with a float; the float rises until the annular area gives a buoyancy–drag–weight balance, so ΔP across the float stays constant and area varies with flow. Reading is nearly linear, rangeability about 10:1, accuracy 1–2% of full scale. It must be mounted vertically with upward flow; it cannot be used on opaque dirty fluids in the glass version. Floats are shaped (spherical, plumb-bob, viscosity-immune) to reduce viscosity sensitivity. Calibration is density-specific — a water-calibrated rotameter on another liquid needs a √(ρ_cal/ρ_act) correction.
• Positive displacement meters: trap and count discrete volumes, so they are true totalisers — used for domestic water and LPG/gas billing, lube oil and petrol-pump dispensing. High accuracy (0.2–0.5%), good at high viscosity, no straight-run requirement; but moving parts, slip/leakage at low flow, cannot take solids, cause pressure drop.
• Turbine meter: a bladed rotor spins proportionally to velocity; a magnetic pick-up gives pulses, so output frequency ∝ flow. Defined by the K-factor (pulses per litre). Accuracy 0.25–0.5%, rangeability 10:1 or better, excellent repeatability — hence standard for hydrocarbon custody transfer with meter provers. Needs clean, lubricating, non-corrosive fluid and straight runs; bearings wear and over-speed on slug flow.

Modern electrical meters

• Electromagnetic (magnetic) flowmeter: applies Faraday's law — a conductive liquid moving through a magnetic field induces E = B·L·v across electrodes, so voltage ∝ velocity. Fully unobstructed bore, no pressure drop, linear, bidirectional, handles slurries, pulps, sewage, acids. Requires conductivity above roughly 5 µS/cm, so it cannot measure hydrocarbons, oils, demineralised water or gases. Pulsed-DC excitation removes electrode polarisation and noise; the pipe needs a liner (PTFE, rubber, ceramic) and earthing rings. Accuracy 0.2–0.5% of rate. Dominant in Indian water supply, ETP/STP and chemical plants.
• Ultrasonic transit-time: two transducer pairs send pulses diagonally with and against the flow; the time difference Δt is proportional to velocity. Needs a clean, acoustically homogeneous liquid; clamp-on versions are non-invasive and ideal for retrofit and portable surveys. Doppler type measures the frequency shift from reflecting particles or bubbles — suited to slurries and dirty liquids. Multi-path ultrasonic meters are now accepted for gas custody transfer.
• Vortex shedding meter: a bluff body sheds alternating vortices (von Kármán street) at frequency f = St·v/d, with Strouhal number St ≈ 0.2 roughly constant over a wide Re band; a piezo sensor counts them. Linear, no moving parts, rangeability 20:1, good for steam, gas and clean liquids; but needs a minimum Reynolds number (about 10,000–20,000), so it fails at low flow, and it is sensitive to pipe vibration and two-phase flow.
• Thermal mass flowmeter: either measures the power needed to keep a heated element at constant temperature difference, or the temperature rise of the stream for a fixed heat input; output is mass flow and depends on the gas's specific heat. Excellent for low gas flows, purge and leak detection.
• Coriolis mass flowmeter: fluid passes through vibrating U- or straight tubes; the Coriolis force twists the tubes and the phase/time shift between inlet and outlet sensors is directly proportional to mass flow, while the tube's resonant frequency simultaneously gives density (hence also concentration and volumetric flow). Accuracy up to 0.1%, no straight-run requirement, independent of conductivity, viscosity and profile. Costliest option, limited to moderate line sizes, affected by entrained gas and external vibration. Now the reference for chemical dosing and liquid custody transfer.

Open-channel and special methods

• Weirs: flow is inferred from head over the crest. Rectangular weir Q ∝ H^1.5; triangular (V-notch) weir Q ∝ H^2.5 — better at low flows because head changes more sharply. Used on canal outlets and ETP outfalls.
• Parshall flume: a shaped venturi-like channel, self-cleaning, low head loss — the standard for municipal sewage and irrigation measurement.
• Dilution/tracer methods (salt or dye injection) and ultrasonic open-channel meters measure flows where no structure can be built.
• Anemometers: vane, cup and hot-wire types for air velocity in ducts and HVAC balancing; the pitot traverse (log-linear or equal-area method) gives duct average velocity.
• Two-phase and multiphase flow (oil-water-gas in upstream petroleum) uses combined gamma densitometers with venturis — accuracy is far lower and is an active field.

Performance terms, standards and Indian practice

• Accuracy stated as % of rate (true over the range) is far better than % of full scale (error grows in proportion at low flow) — a standard interview trap.
• Rangeability/turndown = max ÷ min measurable flow. Orifice ~4:1, rotameter ~10:1, turbine/magnetic ~20:1 to 30:1, vortex ~20:1, Coriolis up to 100:1.
• Repeatability matters more than absolute accuracy for closed-loop control; absolute accuracy and traceability matter for billing.
• Standards: ISO 5167 (parts 1–6) is the governing international standard for orifice plates, nozzles, venturis, cone and wedge meters, and the IS series on differential-pressure flow measurement is aligned with it. AGA Report No. 3 / API MPMS Chapter 14.3 governs natural gas orifice metering and API MPMS generally governs petroleum custody transfer; ASME PTC 19.5 covers flow measurement in performance tests. IS 14615 (parts 1–4) adopts ISO 5167 for pipe-flow measurement by DP devices, and IS 9108 covers open-channel measurement by thin-plate weirs and notches (IS 1192 is the velocity–area method) — several older IS numbers in this area have been revised or withdrawn, so always verify the current edition number and year on the BIS portal before quoting it in a specification.
• Legal metrology: flow meters used for trade (petrol pump dispensers, water and gas billing meters) are regulated under the Legal Metrology Act, 2009 and its Rules, requiring model approval, initial verification and periodic stamping by the State Legal Metrology department. Petrol pump dispensing units are verified and sealed; tampering is an offence under that Act.
• Other Indian regulation: CPCB/SPCB consent conditions require continuous effluent flow metering with online data transmission at many red-category plants; CGWA permits for groundwater abstraction require flow meters on borewells; CEA/CERC and PNGRB regimes require metering on gas pipelines and generating stations. CPCB specifications have been revised periodically — confirm the current circular for the exact device and telemetry requirement.
• Calibration: by gravimetric or volumetric test rigs, master meters, pipe provers (ball/piston provers) and bell provers for gas; traceability through NABL-accredited laboratories to CSIR-NPL, New Delhi, as the national standard.

Key points

• Volumetric vs mass flow — Q in m³/h depends on density; ṁ in kg/h does not, so gas flows are always quoted at reference conditions (Nm³/h).
• DP meters — infer flow from a restriction; output Q ∝ √ΔP, so square-root extraction and poor low-flow accuracy are inherent.
• Beta ratio β = d/D; a low β gives a larger, more measurable ΔP but a larger permanent pressure loss.
• Discharge coefficient C_d — actual ÷ ideal flow; ≈0.61 orifice, ≈0.96–0.98 nozzle, ≈0.98 venturi.
• Permanent pressure loss ranking — orifice (highest) > nozzle > venturi > Dall tube > magnetic/ultrasonic (effectively zero).
• Rotameter — constant pressure drop, variable area; near-linear scale, must be vertical with upward flow, calibration tied to fluid density.
• Positive displacement meter — counts trapped volumes; a true totaliser, best for viscous liquids and billing, worst with solids.
• Turbine meter — frequency output and a K-factor in pulses/litre; the traditional hydrocarbon custody-transfer meter.
• Magnetic flowmeter — Faraday's law, E = BLv; obstruction-free and slurry-tolerant but needs conductivity (~>5 µS/cm), so useless on oils and gases.
• Ultrasonic — transit-time for clean liquids, Doppler for dirty ones; clamp-on versions give non-invasive retrofit measurement.
• Vortex meter — f = St·v/d with St ≈ 0.2; linear and robust but needs high Reynolds number and dislikes vibration.
• Coriolis meter — only common meter giving true mass flow plus density simultaneously; highest accuracy (~0.1%) and highest cost.
• Thermal mass meter — uses heat transfer and gas specific heat; best for low gas flows and leak/purge detection.
• Pitot tube — gives point velocity from stagnation minus static head; the Annubar averages several ports for a mean velocity.
• Weirs — rectangular Q ∝ H^1.5, V-notch Q ∝ H^2.5; the V-notch resolves small flows better.
• Reynolds number and straight run — all inferential meters assume a developed turbulent profile; 10–20D upstream or a flow conditioner is mandatory.
• % of rate beats % of span — always ask which basis an accuracy figure uses before comparing meters.
• Standards — ISO 5167 for DP primary elements, AGA 3/API MPMS for gas and petroleum metering, Legal Metrology Act, 2009 for trade metering in India.

Formulas and examples

Example 1 — Orifice plate on a water line. Water (ρ = 1,000 kg/m³) flows in a 100 mm pipe through a 50 mm concentric orifice, C_d = 0.61, measured ΔP = 40 kPa. Find Q and ṁ.

• β = 50/100 = 0.5 → β⁴ = 0.0625 → 1 − β⁴ = 0.9375 → E = 1/√0.9375 = 1.0328.
• A₂ = (π/4)(0.05)² = 1.9635 × 10⁻³ m².
• √(2ΔP/ρ) = √(2 × 40,000/1,000) = √80 = 8.944 m/s.
• Q = 0.61 × 1.0328 × 1.9635 × 10⁻³ × 8.944 = 1.106 × 10⁻² m³/s ≈ 11.07 L/s ≈ 39.8 m³/h.
• ṁ = 1,000 × 0.01106 = 11.07 kg/s. Pipe velocity = 0.01106/((π/4)(0.1)²) = 0.01106/7.854 × 10⁻³ = 1.41 m/s.

Example 2 — Venturimeter with a mercury manometer. A venturimeter, 200 mm inlet × 100 mm throat, C_d = 0.98, carries water; mercury deflection h = 180 mm (ρ_Hg = 13,600 kg/m³).

• ΔP = (13,600 − 1,000) × 9.81 × 0.18 = 12,600 × 9.81 × 0.18 = 22,250 Pa.
• β = 0.5 → E = 1.0328; A₂ = (π/4)(0.1)² = 7.854 × 10⁻³ m².
• √(2 × 22,250/1,000) = √44.50 = 6.671 m/s.
• Q = 0.98 × 1.0328 × 7.854 × 10⁻³ × 6.671 = 0.05304 m³/s ≈ 53.0 L/s ≈ 190.9 m³/h.
• Permanent loss on a venturi ≈ 10% of 22,250 Pa ≈ 2.2 kPa; an orifice on the same duty would lose roughly 13–18 kPa — the pumping-cost argument for a venturi.

Example 3 — Square-root extraction and low-flow error. A DP transmitter is ranged 0–2,500 mmH₂O for 0–100 m³/h.

• At ΔP = 625 mmH₂O: %DP = 25% → flow = √0.25 × 100 = 50 m³/h.
• At ΔP = 100 mmH₂O: %DP = 4% → flow = √0.04 × 100 = 20 m³/h.
• Sensitivity check: a ±1% of span DP error (±25 mmH₂O) at 20% flow (ΔP = 100 mmH₂O) gives flow between √(0.03)×100 = 17.3 and √(0.05)×100 = 22.4 m³/h, i.e. about −13% to +12% of reading — which is exactly why orifice turndown is limited to roughly 4:1.

Example 4 — Magnetic flowmeter output. A magmeter of bore 80 mm, flux density B = 0.05 T, produces E = 1.6 mV.

• v = E/(B·D) = 1.6 × 10⁻³/(0.05 × 0.08) = 1.6 × 10⁻³/4 × 10⁻³ = 0.4 m/s.
• A = (π/4)(0.08)² = 5.027 × 10⁻³ m² → Q = 0.4 × 5.027 × 10⁻³ = 2.011 × 10⁻³ m³/s ≈ 7.24 m³/h ≈ 120.6 LPM.

Example 5 — Vortex meter and K-factor. A vortex meter in a 100 mm line has a bluff body d = 25 mm, St = 0.2, and outputs 48 Hz.

• v = f·d/St = 48 × 0.025/0.2 = 6.0 m/s.
• A = 7.854 × 10⁻³ m² → Q = 6.0 × 7.854 × 10⁻³ = 0.04712 m³/s = 169.6 m³/h.
• K-factor = f/Q = 48/0.04712 = 1,019 pulses per m³ ≈ 1.02 pulses per litre.
• Re check for water at 20 °C (ν = 1.0 × 10⁻⁶ m²/s): Re = 6.0 × 0.1/1 × 10⁻⁶ = 6 × 10⁵ — comfortably above the ~20,000 minimum, so the meter is in its linear band.

Where it is used

• Refineries and petrochemicals (Reliance Jamnagar, IOCL, BPCL, HPCL): thousands of orifice-plus-DP loops for process control and turbine/Coriolis skids with provers for custody transfer and excise accounting.
• Water and wastewater (municipal corporations, Jal Jeevan Mission, Namami Gange, STPs/ETPs): electromagnetic meters on pumping mains and district metering areas, Parshall flumes and weirs on channels, and CPCB-mandated online effluent flow telemetry.
• Power plants (NTPC, state gencos): flow nozzles and venturis for feedwater and condensate, ultrasonic for cooling water, with ASME performance-test flow measurement for heat-rate guarantees.
• Gas transmission and city gas (GAIL, IGL, Mahanagar Gas, PNGRB-regulated entities): AGA-3 orifice runs, multi-path ultrasonic meters and diaphragm/PD meters at domestic connections, with flow computers correcting to Nm³.
• Pharma, food and dairy (Amul, Sun Pharma, Nestlé India): Coriolis and hygienic magnetic meters with tri-clamp fittings for recipe dosing, CIP flow verification and GMP batch records.
• Job roles: instrumentation engineer (meter sizing and specification sheets), commissioning and loop-checking engineer, calibration/NABL laboratory technician, DCS/SCADA engineer configuring square-root and totalisation, and OEM application/sales engineer at Emerson, Yokogawa, ABB, Endress+Hauser, Krohne and Forbes Marshall.

Interview questions

• Q: Why is the relationship between flow and differential pressure non-linear? A: Bernoulli's equation gives ΔP proportional to v², so Q ∝ √ΔP; the transmitter or DCS must extract the square root to get a linear flow signal.
• Q: Why does an orifice plate have such poor rangeability? A: Because at 10% flow the ΔP is only 1% of span, so a small DP error becomes a large flow error — practical turndown is about 4:1.
• Q: Which DP element has the lowest permanent pressure loss and why? A: The venturimeter (or Dall tube), because the long gradual diverging cone recovers most of the velocity head instead of dissipating it in turbulent eddies.
• Q: Why can't an electromagnetic flowmeter measure petrol or diesel? A: It needs the fluid to be electrically conductive (roughly above 5 µS/cm) to generate and conduct the induced emf; hydrocarbons are insulators.
• Q: Which meters give true mass flow without density compensation? A: Coriolis (from the Coriolis force on vibrating tubes) and thermal mass meters (from heat transfer and specific heat); all others measure volume and need density input.
• Q: Why must a rotameter be installed vertically with upward flow? A: The float's position results from a balance of gravity, buoyancy and drag, so the measurement depends on the weight acting along the tube axis against upward flow.
• Q: What is the beta ratio and how does choosing it involve a trade-off? A: β = bore/pipe diameter; a small β gives a larger, easier-to-measure ΔP but a much higher permanent pressure loss and pumping cost.
• Q: Where would you drill the vent and drain holes in an orifice plate? A: A vent hole at the top for liquid service carrying gas, and a drain hole at the bottom for gas service carrying condensate, so nothing accumulates against the plate.
• Q: Why does a vortex meter have a low-flow cut-off? A: Vortex shedding is only stable and the Strouhal number only constant above a minimum Reynolds number (about 10,000–20,000); below it the output is non-linear or absent.
• Q: What is the difference between accuracy in % of rate and % of full scale? A: % of rate error scales with the actual reading, so it stays proportionally constant; % of full scale is a fixed absolute error that becomes a huge percentage of reading at low flow.
• Q: How would you measure flow in an existing pipe you cannot shut down? A: With a clamp-on ultrasonic transit-time meter, which needs no process penetration — a Doppler clamp-on if the liquid is dirty or aerated.
• Q: Why do gas flows have to be stated in Nm³/h rather than m³/h? A: Gas volume changes with pressure and temperature, so volumetric flow is meaningless unless corrected to standard reference conditions by a flow computer using the gas law.
• Q: What governs a petrol-pump dispenser's measurement legally in India? A: The Legal Metrology Act, 2009 and its Rules — the meter needs model approval, initial verification and periodic stamping/sealing by the State Legal Metrology department.

Common mistakes

• Forgetting the velocity-of-approach factor 1/√(1 − β⁴), or using the pipe area instead of the throat/bore area, in DP calculations.
• Using the manometer deflection directly as head of the flowing fluid instead of converting with (ρ_m/ρ_f − 1).
• Taking the square root of the DP signal twice — once in the smart transmitter and again in the DCS — producing a badly wrong flow reading on a live plant.
• Specifying a magnetic flowmeter for a non-conductive fluid, or a turbine meter for a dirty or non-lubricating one.
• Ignoring the upstream straight-run requirement and installing a meter immediately after an elbow, valve or pump, then blaming the meter for the error.
• Quoting volumetric gas flow without reference conditions, or mixing LPM, m³/h and Nm³/h in the same calculation.
• Comparing a 1% full-scale rotameter with a 1% of rate Coriolis meter as if their accuracies were equal.
• Mounting liquid impulse lines so they trap gas (or gas lines so they trap condensate), giving a steady, undiagnosed zero offset.

Related topics

Pressure measurement · Transducers and sensors basics · Level measurement · Temperature measurement · Final control elements and control valves · Process control loops and DCS/SCADA