Square root extraction is the step that converts a differential pressure signal into a flow rate, because in a differential pressure meter the flow is proportional to the square root of the pressure drop, not to the pressure drop directly. It sounds like a small piece of math, but where and how many times it is applied is a genuine field problem, since applying it twice by accident, called double-rooting, quietly corrupts the flow reading. This guide explains why the square-root relationship exists, where the extraction should happen, and how to avoid the classic double-root mistake.
Square Root Extraction in one line: Square root extraction is the conversion of a measured differential pressure into flow, reflecting the physics that DP-based flow rises with the square of flow so flow equals a constant times the square root of DP. The extraction must happen exactly once in the signal chain, and applying it at both the transmitter and the flow computer, known as double-rooting, produces a badly distorted reading.
In an orifice, venturi, or other differential-pressure meter, the fluid speeds up through the restriction and its pressure falls, and the physics ties that pressure drop to the square of the velocity. Because flow rate is velocity times area, the differential pressure ends up proportional to the square of the flow, which means to recover flow you take the square root of the differential pressure. This is the fundamental reason DP flow measurement is nonlinear.
The square-root shape has a practical consequence at the low end of the range. Near zero flow the differential pressure becomes very small, and taking the square root of a tiny, noisy signal amplifies uncertainty, so accuracy degrades badly at low flow. This is why DP meters have limited turndown and why a low-flow cutoff is commonly applied to force the reading to zero below a threshold, preventing noise near zero from registering as phantom flow.
It also means a linear-looking differential signal is not a flow signal until the root is taken. A 4-to-20 milliamp differential pressure transmitter output, if not already linearized, represents pressure, and treating it as if it were flow overstates flow at the low end and understates it at the high end. Somewhere in the chain, the root has to be applied, and knowing where is the key to getting it right.
Modern DP transmitters can be configured for a linear output, meaning they report differential pressure straight, or for a square-root output, meaning they perform the extraction internally and output a signal already proportional to flow. Flow computers and SCADA systems can also perform the extraction. The rule is that it must be done exactly once, so the whole chain has to agree on which device owns the step and every other device must leave the signal alone.
In modern custody measurement the strong preference is to send the transmitter's linear differential pressure to the flow computer and let the flow computer do the full calculation, including the square root, because the flow computer also needs live static pressure, temperature, and gas properties to compute a corrected volume. Extracting the root at the transmitter and then feeding that into a flow computer that expects raw differential pressure is a recipe for error. Keeping the transmitter linear and letting the flow computer own the math keeps the chain unambiguous.
The reason older installations sometimes extracted at the transmitter was to give simple downstream devices a ready-to-use flow signal, which made sense when the receiving equipment could not do the math. Today that legacy habit is the source of confusion, because a transmitter left in square-root mode from a previous life can silently corrupt a new flow-computer setup that assumes linear input. Confirming the transmitter's output mode is part of a careful commissioning.
Double-rooting happens when the square root is applied twice, most often when a transmitter is set to square-root output and the flow computer or SCADA is also configured to extract the root. The result is a reading that is heavily distorted, wildly wrong at the extremes of the range while sometimes looking deceptively reasonable near mid-scale, which is exactly what makes it hard to catch. The totals accumulate on a bad curve and the error propagates into reports.
The mirror-image failure is no extraction at all, where a linear differential signal is treated as flow because everyone assumed some other device took the root. This overstates and understates flow across the range in the opposite way and, like double-rooting, produces plausible-looking numbers that are quietly wrong. Both failures share a root cause: unclear ownership of the extraction step across the signal chain.
For cloud SCADA, the platform usually receives an already-computed flow from the flow computer, so a double-root committed upstream is invisible unless you look for its signature. Trending the raw differential pressure alongside the reported flow, and comparing behavior at low versus high flow against expectation, can expose an extraction error, and comparing a meter's totals against a downstream meter or tank movement flags the systematic bias. A platform that lets an operator overlay these signals turns an otherwise buried math error into something you can see and correct at the source.
In a differential pressure meter the fluid accelerates through a restriction and the pressure drop rises with the square of the velocity, and since flow is proportional to velocity, the differential pressure is proportional to the square of flow. Inverting that relationship means flow equals a constant times the square root of the differential pressure. This is why DP flow measurement is inherently nonlinear and loses accuracy at low flow.
Double-rooting is applying the square root extraction twice, typically when a transmitter is set to square-root output and the flow computer also extracts the root. The result is a badly distorted flow reading that is very wrong at the ends of the range and can look reasonable in the middle, making it hard to detect. The fix is to ensure the root is taken exactly once by only one device in the chain.
In modern custody measurement the transmitter is usually left in linear mode, sending raw differential pressure to the flow computer, which then performs the full calculation including the square root along with static pressure, temperature, and gas properties. This keeps the extraction in one place and avoids ambiguity. Extracting at the transmitter is a legacy practice that can cause double-rooting if a downstream device also expects to do it.
Merobix reads your field devices into a cloud SCADA - the real thing behind these terms, live in days from any browser.