Every orifice measurement has an uncertainty, and a custody contract usually specifies a tolerance the measurement must stay within, so a measurement engineer needs a way to add up where the uncertainty comes from and check the total against that tolerance. That accounting is the uncertainty budget: a structured list of every input to the flow calculation, the uncertainty of each, how strongly each input influences the flow through its sensitivity coefficient, and the combination of all of them into a single overall figure. This guide works through building an orifice uncertainty budget the way AGA 3 and ISO 5167 frame it, combining the component uncertainties through their sensitivity coefficients by root-sum-square, and shows which terms dominate at low versus high beta and differential. It closes with how a SCADA platform can compute a live per-meter uncertainty so operators know when a station drifts outside its custody tolerance.
Orifice Uncertainty Budget in one line: An orifice meter uncertainty budget lists every input to the flow calculation, the discharge coefficient, the expansibility factor, the bore and pipe diameters, the differential pressure, the static pressure, the temperature, and the density, assigns each an uncertainty, multiplies each by its sensitivity coefficient that says how strongly it moves the flow, and combines the results by root-sum-square into an overall flow uncertainty. Which terms dominate depends on the operating point: the discharge coefficient and diameters matter across the board, while the differential pressure term grows large at low differential and the bore-to-pipe terms shift with beta. A SCADA platform can compute this live per meter so a station drifting outside its custody tolerance is visible.
The starting point of the budget is the list of inputs that the orifice flow equation depends on, because each is a source of uncertainty. The main terms are the discharge coefficient, the expansibility factor, the orifice bore diameter, the meter pipe diameter, the differential pressure across the plate, the static pressure, the flowing temperature, and the density of the gas. Each of these is known only to within some uncertainty: the discharge coefficient carries the uncertainty of the Reader-Harris/Gallagher correlation itself, the diameters carry measurement and thermal-expansion uncertainty, the transmitters carry their calibration uncertainty, and the density carries the uncertainty of whatever method or measurement produced it.
An uncertainty on an input only matters to the extent the flow depends on that input, and that dependence is the sensitivity coefficient. A sensitivity coefficient answers the question: if this input changes by one percent, by what percentage does the computed flow change? Some inputs have a sensitivity of one, meaning a one percent error in them causes a one percent error in flow, while others are amplified or damped. The classic example is the differential pressure and the density: because the flow depends on the square root of the differential pressure and the density, a one percent error in either moves the flow by only about half a percent, so their sensitivity is roughly one half. The bore diameter, by contrast, has a sensitivity larger than one because the flow depends on the bore area, which depends on the diameter squared, so a small diameter error is magnified.
The bore and pipe diameters interact through the beta ratio, and their sensitivity coefficients change with beta, which is why the same diameter uncertainty contributes differently at different beta ratios. At high beta the pipe diameter's influence grows relative to low beta, so a fixed uncertainty in the pipe measurement matters more on a large-beta plate. Getting the sensitivity coefficients right is the technical heart of the budget, because they translate each raw input uncertainty into its actual contribution to the flow uncertainty, and an input with a large uncertainty but a small sensitivity may matter less than an input with a modest uncertainty and a large sensitivity.
Once each input's uncertainty has been multiplied by its sensitivity coefficient to give that input's contribution to the flow uncertainty, the contributions have to be combined into one overall number. They are not simply added. Because the individual uncertainties are largely independent and can fall in either direction, adding them would overstate the total by assuming every one happens to be at its worst in the same direction at once, which is extremely unlikely. Instead the contributions are combined by root-sum-square: square each contribution, add the squares, and take the square root of the sum. This is the standard way independent uncertainties combine, and it is what AGA 3 and ISO 5167 prescribe for the orifice flow uncertainty.
The root-sum-square has an important practical consequence: the largest contributor dominates the total far more than the small ones. Because the contributions are squared before adding, a term that is twice as large as another contributes four times as much to the sum, so the overall uncertainty is driven mainly by the one or two largest terms, while the many small terms barely move it. This means the way to improve a meter's uncertainty is to attack its dominant term, and the way to understand a meter's uncertainty is to identify which term dominates, rather than treating every input as equally worth chasing.
Which term dominates depends on the operating point, which is why a single budget number is not the whole story. The differential pressure contributes through its sensitivity of about one half, but the transmitter's uncertainty is usually expressed relative to its calibrated span, so at low differential, near the bottom of the transmitter's range, the differential term becomes a large fraction and can dominate the budget, whereas at high differential it shrinks. The discharge coefficient's own uncertainty is a floor that is always present. The diameter terms shift with beta. So a meter running at high beta and low differential has a very different budget than the same meter at low beta and high differential, and building the budget at the actual operating point is what makes it meaningful.
The fact that the dominant term shifts with the operating point is exactly why a live uncertainty estimate is more useful than a single nameplate figure. A budget computed once at a design condition tells you the uncertainty the meter can achieve when it is running where it was designed to run, but it says nothing about the hours when the meter is running light, at a low differential where the transmitter term balloons, or outside the valid Reynolds range of the discharge coefficient. Because the inputs to the budget, the differential pressure, the static pressure, the temperature, the density, and the resulting Reynolds number, are all quantities the flow computer already has, the uncertainty can be recomputed continuously from the current operating point.
A cloud SCADA platform such as Merobix can carry the fixed parts of the budget for each meter, the calibrated uncertainties of the transmitters, the diameter uncertainties, and the discharge coefficient uncertainty, and combine them live with the moving operating point to produce a per-meter uncertainty that tracks in real time. Trended alongside the flow, this shows the operator not just what the meter is measuring but how well it is measuring it right now. A meter that spends its overnight low-flow hours at a differential so low that its uncertainty triples is a meter whose custody tolerance may be quietly violated for part of every day, and that is invisible in the flow reading but obvious in a live uncertainty trend.
This turns the uncertainty budget from a document produced at commissioning into an operational signal. When the live uncertainty for a meter rises above its custody tolerance, whether because the flow has dropped into the transmitter's noisy low range, the Reynolds number has fallen below the discharge coefficient's valid floor, or a density input has become unreliable, the platform can flag it so the operator knows the measurement is no longer defensible at contract quality. Because the budget is computed per meter from that meter's own operating point and calibrated inputs, the alert is specific and actionable, pointing to the actual dominant term rather than a generic warning, and letting the operator address the cause, retune the range, restrict the turndown, or fix the input, before the out-of-tolerance measurement accumulates into a disputable volume.
A sensitivity coefficient expresses how strongly the computed flow responds to a change in one input, answering how much the flow changes for a one percent change in that input. It converts each input's raw uncertainty into its actual contribution to the flow uncertainty. For example, because flow depends on the square root of differential pressure and density, those inputs have a sensitivity of about one half, while the bore diameter has a sensitivity greater than one because flow depends on the bore area, which goes as the diameter squared.
Because the individual input uncertainties are largely independent and can err in either direction, so it is very unlikely they all reach their worst case in the same direction at once. Simply adding them would assume that improbable coincidence and overstate the total. Root-sum-square, squaring each contribution, summing, and taking the square root, is the standard way independent uncertainties combine, and it has the effect that the one or two largest terms dominate the total while the many small ones barely matter.
The differential pressure term usually dominates at low differential. The transmitter's uncertainty is typically expressed relative to its calibrated span, so near the bottom of the range a fixed span uncertainty becomes a large fraction of the small reading, and even after the one-half sensitivity for the square-root dependence it can outweigh the discharge coefficient and diameter terms. At high differential the same term shrinks. This is why a meter running light overnight can have a far larger uncertainty than the same meter at high flow.
Primary references from the standards bodies and regulators that define this topic:
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.