Why Your Flow Meter Reads Wrong: A Troubleshooting Guide
10 min read · updated 2026-08-14
When a flow reading is suspect, the transmitter is rarely the culprit — and recalibrating it wastes a shift. These nine causes account for the overwhelming majority of flow measurement errors, each with the specific check that confirms or eliminates it.
Work through them roughly in order. The early entries cost minutes to check and are the most common; the later ones need the line down. Before starting, note the shape of the error — constant offset, proportional error, or error that varies with flow rate — because that alone eliminates most of this list.
1. Square root extraction applied twice — or not at all
Symptom: Reading is correct at 100% flow but badly wrong everywhere else. Double extraction reads high at mid-range; no extraction reads low.
Why it happens: DP across a restriction rises with the square of flow. The square root must be taken exactly once — in the transmitter or in the DCS, never both.
The check: Force 50% DP. A correctly extracted signal reads 70.7% flow. If it reads 50%, extraction is missing; if it reads about 84%, it has been applied twice.
2. Insufficient straight run
Symptom: Consistent bias against a reference meter, often 2-5%, with no other fault evident. Frequently worse after a plant modification.
Why it happens: Elbows, valves and reducers leave a distorted velocity profile and swirl that the meter calibration never accounted for.
The check: Count pipe diameters from the nearest upstream fitting. DP elements often need 10-44 D depending on beta ratio and fitting type. A partially open upstream valve is the worst offender.
3. Flow has dropped into the laminar region
Symptom: Accuracy collapses at low flow, or a vortex meter simply stops indicating.
Why it happens: Inferential meters assume turbulent flow. Below the Reynolds threshold the calibration no longer applies, and vortex meters stop shedding vortices altogether.
The check: Calculate Reynolds number at minimum flow, not design flow — and recheck at the coldest process temperature, since viscosity rises as things cool.
4. Density or reference conditions have changed
Symptom: Mass balance does not close. Two meters on the same line disagree, sometimes by a large factor.
Why it happens: Volumetric meters read actual volume. Converting that to mass or standard volume needs the operating density — and gas density moves with pressure and temperature.
The check: Confirm the density used in the flow computer matches actual conditions, and confirm both meters use the same reference basis. Nm³ at 0 °C and Sm³ at 15 °C differ by over 5%.
5. Impulse lines blocked, leaking, or unequally filled
Symptom: Reading frozen, drifting slowly, or offset at zero flow. Common in slurry and condensing service.
Why it happens: A DP transmitter measures whatever the impulse lines deliver. Plugging, gas pockets in liquid lines, or unequal condensate levels all corrupt the differential.
The check: Close both isolation valves and open the equalising valve — the transmitter should read exactly zero DP. Any offset is an impulse line problem, not a transmitter problem.
6. Wrong internal diameter used in sizing
Symptom: Systematic error of a few percent, present since commissioning and never resolved by recalibration.
Why it happens: Sizing done on nominal pipe size rather than the actual bore. A 4" SCH 40 pipe has a 102.26 mm bore, not 100 mm — enough to shift velocity, Reynolds number and beta ratio.
The check: Verify the internal diameter from the schedule and compare it against what the sizing calculation assumed.
7. Deep turndown beyond the meter’s usable range
Symptom: Good accuracy at high flow, unusable at low flow, and operators no longer trust the reading.
Why it happens: DP flow squares the turndown problem: at 25% flow the DP is only 6.25% of range, so a 0.1% transmitter error becomes a large percentage of the actual measurement.
The check: Compute the DP at minimum flow as a percentage of transmitter span. Below about 10% of span, expect trouble and consider a second transmitter or a different technology.
8. Two-phase flow
Symptom: Noisy, jumping signal that no amount of damping fixes. Often worst at low pressure or after a heat source.
Why it happens: Almost every flow meter assumes a single phase. Gas bubbles in liquid, or condensate in vapour, break the physics the calibration relies on.
The check: Compare the operating pressure against the vapour pressure at operating temperature. For steam, check against the saturation line — a small pressure drop can put you in the wet region.
9. Orifice plate installed backwards or damaged
Symptom: Reading low by 15-25%, immediately after a plate change or plant turnaround.
Why it happens: The sharp upstream edge is what creates the predictable contraction. Reversed, the bevel faces upstream and the discharge coefficient no longer applies. Erosion rounds that edge over time with a similar effect.
The check: Confirm the tab marking faces upstream and the bevel downstream. On any plate more than a few years old in erosive service, inspect the edge before assuming the electronics are at fault.
A note on the order of suspicion
Field experience suggests a rough hierarchy. Configuration errors — square root, density, reference conditions — are the most common and the cheapest to check, because they need nothing but a laptop. Installation problems come next: straight run and impulse lines cause more sustained error than any electronic fault. Sizing errors are third, and they have a distinctive signature: present since day one, never fixed by calibration. Actual instrument failure is last, and genuinely uncommon.
The practical consequence: when someone reports a flow meter reading wrong, the productive first question is not "when was it last calibrated" but "what changed, and what shape is the error."
Calculators referenced: square root extraction · Reynolds number · orifice plate sizing · pipe schedule & bore · actual ↔ standard flow · saturated steam table