Flow Meter Accuracy vs Repeatability: Understanding the Difference

When evaluating a flow metre for your industrial process, the datasheet is the primary technical reference. Yet two specifications cause endless confusion: accuracy and repeatability. They sound similar, they both relate to measurement quality—but they describe fundamentally different performance characteristics.

This guide clarifies the distinction, explains how to read datasheets correctly, and shows why the difference matters in real-world applications.

Accuracy: How Close to the True Value

Accuracy is how close the metre's reading is to the actual, true flow rate. In technical terms, it measures systematic error—a bias toward over- or under-reading that is consistent and measurable.

Consider a Coriolis metre with ±0.1% accuracy, measuring crude oil at 100 bbl/day. The true flow is 100 bbl/day, but the metre might report anything from 99.9 to 100.1 bbl/day, depending on the systematic error in that particular instrument.

How Accuracy is Expressed

Manufacturers specify accuracy in three formats. Understanding which applies to your metre is essential.

• ±% of Reading: Error scales with the flow rate. A metre with ±0.5% of reading at 50 bbl/day has ±0.25 bbl/day uncertainty. At 100 bbl/day, the uncertainty is ±0.5 bbl/day. Coriolis and electromagnetic metres typically use this.
• ±% of Full Scale: Error is fixed regardless of flow rate. A metre with 0–100 bbl/day range and ±2% of full scale has ±2 bbl/day uncertainty at any flow—10 bbl/day, 50 bbl/day, or 100 bbl/day. Vortex and ultrasonic metres often use this.
• ±% of Rate: A less common variant of "% of reading", often used in custody transfer applications to distinguish the measurement basis (volume rate vs. mass rate).

The choice of spec has enormous practical impact. At 10% flow (10 bbl/day on a 100 bbl/day metre), a ±0.5% of full-scale spec yields ±2% actual error, whereas ±0.5% of reading gives only ±0.05% error.

Repeatability: How Close Successive Measurements Are

Repeatability is how consistently the metre reports the same flow under identical conditions. In technical terms, it measures random error—the scatter or noise in repeated measurements.

If you measure the same flow 10 times under identical conditions, repeatability describes how tightly the 10 readings cluster around their average. A high-repeatability metre produces tight clustering; a low-repeatability metre shows wider scatter.

Why Repeatability is Often Better Than Accuracy

A counterintuitive but universal pattern: repeatability is almost always better (smaller) than accuracy. For example:

• Coriolis: ±0.2% accuracy, ±0.05% repeatability
• Electromagnetic: ±0.5% accuracy, ±0.1% repeatability
• Vortex: ±0.75% accuracy, ±0.1% repeatability

This happens because systematic error (accuracy) includes fixed biases that are hard to eliminate. Random error (repeatability) can be reduced through signal averaging and filtering, making modern metres very reproducible despite imperfect accuracy.

Why the Distinction Matters

Custody Transfer and Billing

When you're measuring product transfer between two parties and charging by volume or mass, accuracy is paramount. If your metre is biased 1% high, you're systematically overcharging customers (or undercharging, if biased low). Repeatability is irrelevant; you need accuracy certified and verified before installation.

Regulatory authorities (like the UK's Weights and Measures body) mandate accuracy verification for custody transfer. Repeatability alone is insufficient.

Process Control and Feedback

In process control applications—where you're measuring flow into a reactor, blending tank, or distillation column—repeatability often matters more than accuracy. The control system uses flow feedback to adjust pump speed, valve position, or temperature. If the metre noise (poor repeatability) is high, the controller oscillates and wastes energy.

A systematic +2% bias is often acceptable in process control (the process self-corrects). Random ±5% noise is devastating (constant control hunting).

Blending and Batching

Blending applications (paint, fuel, chemical formulation) need both accuracy and repeatability. You must hit the target blend ratio precisely (accuracy requirement) and consistently (repeatability requirement). A biased metre produces off-spec product; a noisy metre produces inconsistent batches.

Reading Datasheets: Practical Examples

Example 1: Vortex Metre (±0.75% of Full Scale)

Specification: ±0.75% of full scale. Pipe size: DN50 (2 inches). Maximum range: 0–100 m³/h.

Actual error at 10% flow (10 m³/h):

• Error = 0.75% × 100 m³/h = ±0.75 m³/h
• As % of actual flow: 0.75 / 10 = ±7.5%

The metre is terrible at low flow because full-scale error is fixed. At 100% flow, accuracy is ±0.75%. At 10% flow, it's ±7.5%.

Example 2: Coriolis Metre (±0.1% of Reading)

Specification: ±0.1% of reading. Same range: 0–100 m³/h.

Actual error at 10% flow (10 m³/h):

• Error = 0.1% × 10 m³/h = ±0.01 m³/h

The Coriolis metre maintains ±0.1% accuracy across the entire range. At 100 m³/h, error is ±0.1 m³/h. At 10 m³/h, it's ±0.01 m³/h.

Which Metre Wins?

For applications below 50% flow: Coriolis. For custody transfer or blending where accuracy is critical: Coriolis. For high-flow process monitoring on a budget: Vortex is acceptable if flow stays above 50%.

Environmental Factors: Temperature and Pressure Effects

Stated accuracy and repeatability assume stable operating conditions. Real-world pipes are not stable. Temperature swings, pressure fluctuations, and vibration all degrade performance.

Temperature Compensation

Coriolis metres have excellent temperature stability because mass flow is mass-based (independent of density). Electromagnetic metres require temperature and pressure compensation for accurate density correction. If compensation is poor or absent, errors grow.

High-quality electromagnetic metres include integrated pressure and temperature sensors. Budget models may not, causing ±1–2% additional error in variable-temperature streams.

Vibration and Pulsation

Coriolis metres are sensitive to mechanical vibration because the measurement is based on micro-scale tube deflection. In noisy mechanical environments (centrifugal compressors, reciprocating pumps), Coriolis repeatability can degrade by 0.2–0.5%.

Electromagnetic metres are immune to vibration but sensitive to pulsating flow. Vortex metres perform poorly with high pulsation. Ultrasonic metres (clamp-on) are robust to both.

Accuracy and Repeatability Across Metre Types

Here is a reference table of typical accuracy and repeatability specifications for common flow metre technologies, measured under ideal conditions:

Note: These are typical ranges under ideal conditions. Real-world performance depends on installation, fluid properties, and maintenance.

Linearity, Hysteresis, and Zero Stability

Datasheets sometimes include additional terms that refine accuracy:

Linearity

Accuracy (as specified) assumes linearity across the full operating range. Some metres drift at the extremes. If a metre's datasheet claims ±0.5% accuracy but specifies ±1% accuracy below 20% flow, linearity is poor. Always check the footnotes.

Hysteresis

Some metres read differently depending on whether flow is increasing or decreasing. This is hysteresis. In custody transfer, this can cause discrepancies if the metre is used bidirectionally. Check datasheets for hysteresis statements; premium metres specify <±0.05%.

Zero Stability (Zero Drift)

When no flow is passing, an ideal metre should read zero. In practice, electronic drift causes meters to drift. A metre with zero drift of ±0.2% over one year means the zero point drifts by 0.2% of full scale annually. In low-flow applications, zero drift is critical.

Common Misconceptions

"Repeatability Shows How Accurate a Metre Really Is"

False. A metre with ±0.05% repeatability could have ±2% accuracy if it has a fixed calibration offset. Repeatability describes noise; accuracy describes bias. Both matter.

"All Metres From the Same Manufacturer Are Equally Accurate"

False. Accuracy degrades with age, maintenance, temperature exposure, and installation. A 10-year-old Coriolis metre may drift to ±0.3% even if it left the factory at ±0.1%. Periodic calibration is essential for custody transfer applications.

"If My Application Only Needs ±2% Accuracy, Any Metre Will Do"

Not necessarily. You also need to verify repeatability to ensure the control system remains stable. And you must understand whether the datasheet spec is "% of reading" or "% of full scale" to avoid nasty surprises at low flow.

Practical Selection Guidance

For custody transfer or billing:

• Demand accuracy better than ±0.5% (preferably ±0.2% or better)
• Insist on "% of reading" not "% of full scale"
• Plan for periodic calibration verification every 2–3 years
• Coriolis is the de facto standard; electromagnetic is acceptable for water

For process control:

• Prioritise repeatability (<±0.2% if possible) over absolute accuracy
• Accept ±1–2% accuracy if repeatability is excellent
• Vortex, electromagnetic, and turbine all work well
• Avoid metres with poor zero stability in low-flow applications

For blending and batching:

• Demand both accuracy (<±0.5%) and repeatability (<±0.1%)
• Use "% of reading" specs
• Coriolis or high-quality electromagnetic are preferred
• Test with your actual fluid before committing

Summary

Accuracy is how close the reading is to the true value (systematic error). Repeatability is how consistently the metre reads under identical conditions (random error). Both are necessary; neither alone is sufficient.

Custody transfer demands accuracy. Process control demands repeatability. Blending demands both. Always read the datasheet carefully to understand which performance spec is quoted, at what percentage of full scale, and under what environmental conditions.