A differential pressure flow transmitter measures pressure drop, but flow through an orifice follows the square root of that pressure drop, so somewhere in the system a square root has to be taken to turn the measurement into flow. The single most important decision when calibrating one is where that extraction happens, because if it happens in two places the low-flow readings go wildly wrong. Once you have decided, the calibration itself is disciplined: you calibrate the differential measurement linearly against known pressures, set a sensible low-flow cutoff, and then verify that the flow output tracks the square-root curve at spaced flow points. Get the extraction location and the cutoff right, and the transmitter reports honest flow across its range.
Calibrate a Square-Root Flow Transmitter in one line: To calibrate a square-root DP flow transmitter, first decide whether the square-root extraction happens in the transmitter or in the DCS, and make sure it happens in exactly one place, never both. Calibrate the differential pressure measurement linearly against known input pressures, set a low-flow cutoff to suppress noisy readings near zero, and then verify the flow output follows the square-root curve by confirming it reads roughly 50, 71, 87, and 100 percent flow at 25, 50, 75, and 100 percent of differential span. Double extraction is the classic error that makes low flow read far too high.
The relationship between flow and differential pressure across an orifice is that flow is proportional to the square root of the pressure drop. Because of that, a differential reading of 25 percent of span does not mean 25 percent flow; it means 50 percent flow, since the square root of a quarter is a half. Somewhere the square root must be applied so that a flow value comes out. The two common places are inside the transmitter, which can be configured to output flow directly with a square-root characterization, or in the control system, which takes a linear differential signal and extracts the root in software. Both are valid, but the system has to do it once and only once.
The failure that ruins more flow measurements than any other is double extraction, where the transmitter is configured for square-root output and the DCS block is also configured to take a square root of that already-rooted signal. The result is a signal that has had the root taken twice, which is the fourth root of the pressure drop, and it reads dramatically high at low flow. At 25 percent of differential the correct flow is 50 percent, but with the root taken twice the reading climbs toward 71 percent, and the error explodes as flow drops toward zero, so the meter can show substantial flow when there is almost none. Before touching a calibrator you must confirm where extraction is configured and disable it everywhere except the one place you intend.
The cleaner convention on modern systems is to leave the transmitter in linear differential output and let the control system take the root, because that keeps the transmitter's raw measurement visible for diagnostics and puts the flow math in one auditable place. But plenty of installations do it in the transmitter, and either choice works as long as it is documented and consistent. The calibration procedure that follows depends on this decision: if extraction is in the DCS, you calibrate and verify the transmitter as a plain linear differential device, and if extraction is in the transmitter, you calibrate the differential linearly but verify the flow output against the square-root curve.
Regardless of where the root is taken, the underlying differential pressure measurement is linear and should be calibrated as such. You apply known pressures to the high and low sides across the transmitter's differential range and confirm that the raw differential reading tracks input in a straight line, setting zero at no differential and span at full differential. This linear calibration is the foundation, because the square-root characterization, wherever it lives, sits on top of an accurate differential measurement. If the differential itself is off, no amount of correct root extraction will make the flow reading right, so you prove the linear measurement first and only then worry about the curve.
The low-flow cutoff exists because the square-root relationship misbehaves near zero. As differential pressure approaches zero the slope of the square-root curve becomes very steep, so a tiny bit of noise or a small zero offset in the differential measurement gets amplified into a large apparent flow. Without a cutoff, a meter sitting at truly zero flow can indicate a wandering few percent of flow just from measurement noise, and integrated over time that phantom flow accumulates into real error on a daily total. The low-flow cutoff forces the flow output to read exactly zero once the differential falls below a chosen threshold, which cleans up the reading and stops noise from accumulating as fictitious throughput.
Choosing the cutoff is a balance. Set it too low and noise still leaks through near zero; set it too high and you blank out genuine low flow that you needed to measure and totalize. The cutoff is usually expressed as a small percentage of flow span and placed just above the noise floor of the differential measurement, which is why calibrating the differential well matters here too, since a quieter zero lets you use a lower cutoff and capture more of the real low-flow range. On custody or allocation service the cutoff has to be chosen deliberately and documented, because it directly affects how much low flow is counted or discarded.
Once the differential is calibrated linearly and the cutoff is set, you verify that the flow output actually follows the square-root curve. The check is to apply differential inputs at 25, 50, 75, and 100 percent of differential span and confirm the flow output reads the square roots of those fractions, which are approximately 50, 71, 87, and 100 percent flow. This is the point where double extraction announces itself immediately: if 25 percent differential produces 50 percent flow you have a single correct extraction, but if it produces something closer to 71 percent you have a second root being taken somewhere and you go back and find it. Spacing the check across the range, rather than just checking the endpoints, is essential because the square-root error is largest at low flow and vanishes at full scale.
It is worth injecting the check at the differential input and reading the flow value at the end of the chain the operators actually see, because that proves the extraction, the scaling, and the engineering-unit conversion all at once. When you inject 25 percent differential and the far end of the system shows 50 percent flow on the correct tag in the correct units, the whole square-root path is confirmed. A cloud SCADA platform such as Merobix lets you watch the flow tag respond as you step the differential input, so the verification happens against the same value that will feed totalization and reporting rather than against an intermediate number that could be re-scaled downstream.
After commissioning, trending both the differential and the derived flow together is one of the best ways to catch a square-root problem that appears later. A meter that suddenly reads high only at low flow, while looking fine at high flow, has the classic double-extraction signature, and that pattern is far easier to recognize on a trend than in a single spot reading. Watching the flow sit cleanly at zero when the differential is at its noise floor also confirms the cutoff is doing its job over time. Because the whole point of the square-root transmitter is an honest flow total, keeping the differential, the flow, and the cutoff behavior visible in the historian is what lets a measurement technician trust the daily volume without pulling the transmitter.
Double extraction happens when the transmitter is configured for square-root flow output and the control system also takes a square root of that already-rooted signal, so the root is effectively taken twice. Because the square-root error is largest near zero, the reading climbs far above the true value at low flow, showing substantial flow when there is almost none. The fix is to confirm extraction happens in exactly one place, either the transmitter or the DCS, and disable it everywhere else.
Either works as long as it happens in exactly one place and is documented. Many modern systems leave the transmitter in linear differential output and take the root in the control system, which keeps the raw differential visible for diagnostics and puts the flow math in one auditable place. Whichever you choose, the differential measurement itself is still calibrated linearly, and only the flow output is verified against the square-root curve.
Near zero differential the square-root curve is very steep, so small amounts of measurement noise or a slight zero offset get amplified into a large apparent flow. Without a cutoff, a meter at true zero flow can indicate a wandering few percent, and integrated over time that phantom flow accumulates into real error on the daily total. The low-flow cutoff forces the flow output to zero once the differential drops below a chosen threshold, cleaning up the reading.
Safety & engineering notice. This article is general educational information, not site-specific engineering, safety, or legal advice, and it does not reflect any particular facility. Standards and regulations (for example OSHA, API, IEC, ISO, NFPA, NIST, and NERC CIP requirements) change and vary by edition, jurisdiction, and application. SCADA and remote monitoring cannot verify physical isolation, atmosphere, lockout/tagout, permit status, or a safe go/no-go decision. Qualified personnel must perform site-specific engineering, hazard analysis, and safety review, and confirm current requirements with the authority having jurisdiction, before acting.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.