Pitot Tube

What a Pitot Tube Is and What Is Being Decided

The hardware is a probe and a pressure connection. A classic pitot static tube is an L-shaped probe, typically 316 stainless steel 8 or 10 mm in section, with a rounded impact nose containing the total pressure port and a circumferential ring of static holes a few diameters downstream where the flow has re-established; the two signals leave through separate legs to a manometer or a differential pressure transmitter. An averaging pitot primary is a straight insertion bar spanning the duct, carrying a column of total pressure ports facing upstream positioned on an equal-area grid and one or more downstream static ports, so a single DP signal represents the duct's mean velocity rather than one point. Both are primary elements in the metrological sense: they create the measurement physics, and a secondary instrument reads it.

What is being decided is therefore a system, not a fitting. First, the quantity: a velocity map or a single-point velocity for fan and stack work, which wants the traverse tube and the velocity-area method, or a continuous volumetric or mass flow reading for process monitoring and custody-adjacent balancing, which wants the averaging primary and its flow coefficient. Second, the velocity window: dynamic pressure in pascals equals one half rho v squared, so air at 10 meters per second produces about 60 pascals, at 25 meters per second about 375 pascals, and the transmitter chosen for a 40 meter per second duct will be blind at the bottom of the same duct unless the range is split or the low end is designed first. Third, the installation reality: straight run available upstream, duct size, whether the line can be depressurized, whether a flange or a compression fitting with isolating valve carries the insertion, and how the blockage correction of the inserted bar is accounted. Fourth, the signal chain: the pressure transmitter range and turndown, square-root extraction in the transmitter or the historian, density compensation from a pressure transmitter and temperature input when gas conditions move, and totalization when energy accounting is the point.

The competitive frame keeps the selection honest. Against the orifice plate flowmeter, the pitot primary wins on permanent pressure loss, an averaging bar's blockage loss is a few percent of an orifice's, and on installed cost in large ducts; the orifice wins on standardization depth, custody-grade uncertainty, and dense liquids. Against the electromagnetic flowmeter, the pitot wins on non-conductive media, partly filled and very large lines, and price; the electromagnetic wins on accuracy, roughly 0.5 percent of reading versus several percent for a field-installed averaging bar, and on slurries. Against the ultrasonic flowmeter, the pitot wins on cost and on dirty-gas tolerance in many duties; the ultrasonic flowmeter wins on non-intrusive accuracy and no penetration for clamp-on types. Against the thermal mass flowmeter, the pitot wins on immunity to composition changes in moderate-velocity gas; the thermal mass flowmeter wins at low velocities where DP physics starves and on direct mass reading in clean dry gas. Against the vortex flowmeter, the pitot wins below the vortex Reynolds floor and in large ducts; the vortex flowmeter wins with clean liquids and steam in pipes at medium velocity. And against the variable area flowmeter, the pitot wins with continuous transmission and large sizes; the rotameter wins on local indication economics in small clean lines.

One positioning note: this page is the single-product page for the pitot primary itself. The orifice plate flowmeter family, the differential pressure transmitter that reads any DP primary, and the transmitter families for gauge and absolute pressure each have their own pages; here the assumption is a velocity measurement made by a probe, and every number that follows is chosen for that world.

Velocity to Pressure: the Physics and Its Limits

The governing relation is one line. Dynamic pressure equals one half density times velocity squared; solve for velocity and the instrument takes a square root, so the signal compresses at the bottom of the range and expands at the top. Air at standard density, 1.2 kilograms per cubic meter, at 5 meters per second gives about 15 pascals, at 10 about 60, at 20 about 240, at 40 about 960, and at 60 about 2,160 pascals; water at 2 meters per second gives about 2,000 pascals, which is why liquid pitot work reads millibars while air work reads tenths. The immediate consequence is the turndown shape: a pitot system honest from 10 to 60 meters per second spans roughly 36 to 2,160 pascals, a 60 to 1 turn in velocity that is only a 36 to 1 turn in pressure, and the bottom sixth of the velocity range lives in the transmitter's noise floor unless the range is engineered for it.

The corrections separate engineering-grade measurement from a reading. Compressibility: above roughly 60 meters per second in gas, or whenever the dynamic pressure exceeds a few percent of absolute static pressure, the incompressible relation under-reads and the isentropic correction applies; for air this starts between Mach 0.1 and 0.3, and stack and fan duty sits safely below it while compressed-air nozzle duty may not. Yaw sensitivity: a standard pitot static nose tolerates a few degrees of yaw before the total port under-reads; where duct swirl is real, a Kiel shield around the impact port holds coefficient to beyond 30 degrees of misalignment, at some loss of simplicity. Density: the velocity equation wants actual density, from line pressure and temperature for gas, so energy-balance installations pair the DP signal with a pressure transmitter reading and a temperature input and compute mass flow in the controller; ignoring a 20 percent density shift creates a 10 percent velocity error and a 20 percent mass error, the kind that quietly voids an energy audit. Reynolds and blockage: the tube coefficient holds near 0.99 across the practical Reynolds range for ISO 3966 geometry, but an inserted averaging bar blocks flow area, and the blockage correction, a few tenths of a percent for slim bars in large ducts, grows to several percent for fat bars in small ducts and belongs in the calculation, not in optimism.

The velocity-area method is how a single tube maps a whole duct. ISO 3966 prescribes the traverse: a grid of measurement points on equal-area centroids, commonly a log-linear or Tchebycheff layout, a minimum of 20 points for circular ducts by the standard's schedules, with the mean of point velocities weighted by area giving duct flow. The method's discipline is what makes a pitot traverse a reference measurement: velocity profile measured, not assumed; each point's dynamic pressure read to a resolution the grid deserves, which argues for a micromanometer or a low-range DP transmitter with 0.1 pascal display; and repeatability checked by a second traverse when the audit will be disputed. A permanent monitoring installation that cannot traverse approximates the same physics with the averaging bar, whose port layout bakes the equal-area weighting into hardware and whose calibration constant folds the residual profile error into one coefficient.

Uncertainty arithmetic closes the chapter, because it is where pitot systems are won or lost. The system error combines the primary coefficient error, roughly 1 percent for a manufactured averaging bar with a stated coefficient, the DP measurement error, which at a square-root characteristic costs half its percentage in velocity terms but dominates at range bottom, and the density computation error. A field-realistic budget lands at 2 to 5 percent of reading for a well-installed averaging system across its honest range, and at 1 to 2 percent for a clean traverse with a good micromanometer. That is audit-grade for ventilation balancing and combustion air supervision, and it is not custody transfer, and the RFQ should state which world the installation lives in before any vendor number is believed.

Types, Construction, and Duty Boundaries

Types divide by measurement intent. The single-point pitot static tube, the L-shaped classic, serves traverses, spot checks, and reference audits; it is the ISO 3966 instrument and the calibration lab's own tool. The averaging multiport bar, one straight probe across the duct axis, serves continuous monitoring in ducts and large pipes to ASME MFC-12M practice; it trades a slightly softer coefficient story for a permanent signal with no moving parts. The Kiel probe serves swirling and distorted flows where a standard nose would misread. The S-type pitot, two facing tubes side by side, serves stack testing to EPA Method 2, where its robust large ports survive soot and its higher coefficient, typically around 0.84, is accepted with its calibration. Double-averaging or opposite-port geometries exist for the worst profiles. Each type answers one profile reality, and choosing by price rather than by profile is how monitoring installations drift.

Construction follows service severity. Stainless steel 316 is the standard material for air, flue gas to about 400 deg C at the probe, and clean water duty; higher alloys, Hastelloy or Inconel equivalents, serve acidic stack gas and refinery duty; PTFE or PVC sheathed probes serve scrubber-adjacent corrosive air. The impact and static ports are drilled, so particulate tolerance is finite: dust loading above a few grams per cubic meter plugs small ports and argues for the S-type geometry, purge air, or a different technology. Mountings ladder from a compression fitting on a low-pressure duct, to a threaded boss with isolating valve for insertion under moderate pressure, to a full flanged insertion with a packing gland and a retractable handle for hot-tap installation and removal under line pressure; the flange is the only honest answer above roughly 2 bar or on gas lines that cannot be opened. The impulse lines to the transmitter complete the construction: short, sloped so condensate drains back, and on gas service with a drip pot on the low side, because a line that holds liquid is a manometer the calculation did not include.

Duty boundaries, sentence by sentence. HVAC and combustion air ducting, large rectangular and circular, velocities 5 to 40 meters per second: the averaging bar owns the duty, low loss, low cost, one penetration. Stack and flue monitoring with particulate: the S-type for testing, the purged averaging bar for continuous, and a calibrated relationship to a reference method in the permit. Compressed air auditing: the pitot traverse with a low-range DP instrument is the standard energy-audit method, permanent monitoring favors a thermal mass flowmeter where the gas is clean and dry. Large water and treated-effluent lines where magmeter cost scales badly: insertion averaging primaries compete, with the electromagnetic flowmeter winning where accuracy class matters and partly filled pipes disqualify both. Low air velocity below roughly 5 meters per second: DP physics starves, and the thermal mass flowmeter or an anemometer class owns the duty. Clean liquid and steam in process pipes with accuracy demands: the orifice plate flowmeter or coriolis and vortex families own measurement, the pitot is absent. Pulsating flow: pitot systems average badly under strong pulsation, an orifice or vortex installation with damping is the safer answer. Every boundary here is a physics statement, and Chapter 5 turns each into an RFQ decision.

Key Specification Parameters

The decode table is the datasheet skeleton. Every row is an RFQ line, and the two right columns are what each row decides and what it costs to leave blank.

The coefficient row is the one that separates measurement from indication. A manufactured averaging primary carries a discharge or flow coefficient from its maker, anchored to ASME MFC-12M-style testing, and the uncertainty quoted against that coefficient is only as good as the velocity profile the bar actually sees; a bar calibrated on a fully developed profile in a lab duct and then installed three diameters downstream of an elbow is a different instrument until a site calibration reconciles it. The RFQ line asks for the coefficient, its basis, and the straight-run assumption behind it: upstream 10 diameters and downstream 5 is the classical default, maker tables sometimes accept less with stated penalty, and a vendor with no stated straight-run basis has quietly assigned the error to the buyer.

The velocity range row is the one that prevents the blind instrument. State minimum and maximum velocity with the operating density, and require the DP range and the transmitter accuracy at both ends: a system specified only at maximum flow will be quoted a range that reads the top decile beautifully and the bottom third not at all, and square-root extraction makes the crime invisible on the datasheet. Where the velocity window is wide, split-range transmitters or a dual-range averaging installation are the honest answers, and they cost less than a monitoring program that cannot see turndown.

The pairing row prices the system truthfully. The primary is a few hundred dollars of machined steel; the measurement lives in the differential pressure transmitter, whose low-range units, 0 to 100 or 0 to 250 pascals with 0.075 percent class spans, carry the accuracy budget, and in the density chain, where a pressure transmitter and a temperature input feed the computation. Budget the DP cell first, the density inputs second, and the probe third, and the quotation stack will match the physics. Impulse plumbing belongs on the same drawing: slopes, drip pots, isolating valves, and a five-valve manifold for zero checks, the same instrument manifold discipline any DP installation uses, because a trapped bubble in one leg is a bias nobody's certificate covers.

The certificate stack is the delivery ticket: material certification to EN 10204 for alloy probes, a calibration certificate stating the coefficient and its basis for every manufactured primary, Ex certification where the atmosphere is classified, and for stack-testing duty the reference-method paperwork the permit names. A quote missing the stack is not cheaper, it is later.

Selection Decision Factors and RFQ Language

The decision sequence runs from the flow profile outward, because every later step prices the earlier one.

1. Characterize the duct or pipe: size and shape, straight run available upstream and downstream, fittings and fans that set the profile, and from these, whether a traverse, an averaging bar, or a Kiel or S-type geometry is the honest instrument.
2. Write the fluid window: medium and particulate loading, density range from pressure and temperature, velocity minimum and maximum with units, and the accuracy class the decision actually needs, balancing grade or process grade or energy audit grade.
3. Size the primary: insertion length to the centroid for spot probes, duct-width bar for averaging, blockage correction computed, material for chemistry and temperature.
4. Engineer the DP range: convert the velocity window to pascals at operating density, choose transmitter span so both ends sit inside the accuracy claim, and decide square-root extraction in the cell or the historian.
5. Fix the installation: mounting class by line pressure, hot-tap valving where the line cannot open, impulse routing with slopes and drip pots, and the five-valve zero-check manifold.
6. Specify the computation: density compensation inputs, mass or standard-volume output, totalization for energy accounting, and the alarm thresholds the monitoring exists to feed.
7. Close with the commercial lines: coefficient basis and calibration certificate, straight-run assumptions, blockage correction stated, spares for the ports and gland, and a site calibration or verification traverse in year one.

The cost comparison is installed cost against loss and accuracy. A flanged averaging installation on a 1-meter duct prices a fraction of a full-bore electromagnetic flowmeter of the same size and almost nothing in permanent pressure loss against an orifice run; a coriolis or turbine flowmeter is not in the conversation at that diameter. The honest comparison at small diameters inverts: below roughly DN100 in clean liquid, full-bore meters price competitively and outclass the pitot in accuracy, which is why the pitot family's share concentrates in large ducts and pipes. Energy-audit duty prices the traverse: a two-day ISO traverse with a micromanometer against a permanently installed bar plus one verification traverse, and the answer depends on how often the question is asked. Table the decision on cost per credible reading, not cost per fitting.

The RFQ language that makes quotes comparable, written as the datasheet lines a vendor must answer with numbers: primary type and port layout; insertion or bar length against duct dimension; material and maximum temperature; coefficient value, its calibration basis, and the straight-run assumption behind it; blockage correction for the stated duct; velocity or flow range with the DP range at operating density; mounting class and pressure rating with hot-tap provision; impulse connection size; transmitter pairing proposed with accuracy class at range ends; computation scope, density inputs and outputs; certificates, material, calibration, and Ex where the site requires. A quotation answering those lines is a measurement system; anything else is a probe with a price.

Installation, Standards, Failure Modes, and Maintenance

Installation follows the profile and the plumbing. Mount in the longest straight run available, upstream of the disturbance by the classical 10 diameters and downstream by 5 where the duct allows, and where it does not, write the deviation into the site calibration rather than hoping. Orient the total ports squarely upstream, verify with a yaw check on commissioning, and clock averaging bars across the duct axis on the equal-area line the port layout assumes. Plumb impulse lines short and sloped, drip pots on the low legs of gas service, and a five-valve manifold at the transmitter for in-service zero; a zero that cannot be checked is a calibration that expires silently. On hot-tap installations, the isolating valve, the packing gland, and the insertion sequence under pressure follow the valve maker's procedure, and the retraction tooling belongs in the stores list, not improvised at the next outage.

The standards frame names the method and the acceptance. ISO 3966 governs the velocity-area method with pitot static tubes, grid layouts, and the correction arithmetic; ASME MFC-12M governs multiport averaging primaries, their coefficients and installation effects; EPA Method 2 governs S-type stack traverses in the United States permit world, with its port counts and acceptability criteria; AMCA and ASHRAE test codes reference pitot traverses as the reference for fan air performance. The measurement chain leans on the DP instrument standards of the differential pressure transmitter family, and installation quality leans on the same impulse-plumbing discipline that category teaches. None of this is exotic; the failure mode is a monitoring report that cites none of it and cannot survive an audit.

Failure modes arrive in a predictable order, and each has a specification answer. Port plugging shows as drift, then freeze, of the reading on dusty service: the answer is the S-type or purged geometry for loading above a few grams per cubic meter, periodic blowback, and a differential across the primary checked against the zero baseline. Impulse-line liquid shows as a step bias after weather changes: the answer is slope, drip pots, and heat tracing on outdoor gas legs. Zero drift in the DP cell shows as a nonzero reading at known no-flow: the answer is the five-valve zero check on a schedule, monthly at commissioning frequency, quarterly once stable. Profile change, a new elbow, a changed damper, a fouled filter, shows as a slow shift that the instrument cannot see: the answer is the verification traverse at year one and after any ductwork change, reconciling the bar's coefficient to the new reality. Erosion of the nose on wet particulate service shows as coefficient change: the answer is alloy selection and the calibration certificate renewed when the probe is replaced. Each failure is cheap to prevent at specification and expensive to diagnose after the energy report is signed.

Maintenance is a calendar of small acts. Quarterly: zero check at the manifold, impulse lines drained, ports visually inspected where access allows. Annually: probe withdrawal and inspection on flanged units, port cleaning, gland packing adjustment, and the coefficient paperwork reviewed against the year's verification. On the verification traverse, run the ISO grid, compare against the continuous system, and either confirm the coefficient or write the site factor into the historian with the date. Keep one spare probe per critical service in stores; a bent bar is not a repair, it is a coefficient change wearing a disguise. Log everything against the tag, because the pitot system's uncertainty claim is a chain of dated evidence, and the chain is the calibration.

The commissioning record closes the loop: as-installed dimensions, insertion depth and orientation, straight-run description with the nearest disturbances named, DP range and transmitter zero, computed coefficient and density basis, and the first traverse result against the continuous reading. One page per installation, kept with the loop drawings. Plants that keep it defend their numbers in any audit; plants that do not re-argue the physics every year.