Why Your Design-Point Uncertainty Is Not Your Real Uncertainty

You run an uncertainty analysis. The result says your meter is within the framework limit. You file the report and move on.

But what if the meter only hits that operating point for a fraction of the year?

The single-point problem

A standard ISO 5168 uncertainty analysis is calculated at a single set of operating conditions: one flow rate, one differential pressure, one density. These are typically the design-point values, or the conditions that were expected when the metering system was engineered.

The result is a single number: the expanded uncertainty at that specific operating point.

This is technically correct. But it is also potentially misleading, because it assumes the meter always operates at that point.

What happens at turndown

For a DP flowmeter, the relationship between flow and differential pressure is quadratic: DP is proportional to the square of the flow rate. This means that at 50% of design flow, the DP is only 25% of the design DP.

Transmitter errors that are specified as a percentage of URL or span are fixed in absolute terms. They do not shrink when the DP signal gets smaller. So as flow decreases, these fixed errors become a larger and larger fraction of the actual reading.

A meter that comfortably meets a 1.5% uncertainty limit at design flow may exceed 5% or even 10% when operating at low turndown. The uncertainty curve is not flat; it rises steeply at low flow.

Other error sources, such as the discharge coefficient, dimensional tolerances, and density, are constant as a percentage of reading. They set the floor of the uncertainty curve. The DP transmitter errors determine how steeply the curve rises above that floor at low flow.

Flow-weighted uncertainty

If the meter spends significant time at low flow, the single design-point uncertainty is not representative of what the meter actually achieved over the year.

The flow-weighted uncertainty accounts for this by computing the uncertainty at each day's average flow rate and then weighting by the flow contribution:

This is the most representative single-number uncertainty for annual totals, which is exactly what EU ETS MRR and fiscal reporting require.

A worked example

Consider a fuel gas orifice meter with these characteristics:

• Design-point uncertainty: ±0.68% (well within the EU ETS Tier 4 limit of ±1.5%)
• DP transmitter accuracy: 0.65% of reading, drift and cal errors on span/URL basis
• Operating pattern: highly variable flow, ranging from 6% to 100% of design across the year

When 365 days of operational flow data are back-tested against the turndown curve, the flow-weighted uncertainty comes out at ±2.35%. That is well above the 1.5% framework limit, even though the design-point analysis says ±0.68%.

The design-point analysis would pass. The flow-weighted analysis would not.

Why this matters for compliance

EU ETS MRR requires operators to determine annual emissions based on activity data. The uncertainty requirement applies to the measurement of that activity data. If the meter's actual operational uncertainty exceeds the tier limit, the operator may not be meeting the applicable tier requirement, even if the design-point analysis says otherwise.

For fiscal and custody transfer metering, the consequences are more direct: the meter may be systematically less accurate than the contract requires, and nobody knows until the flow-weighted analysis is done.

The flow-weighted approach also reveals which days are problematic. In the example above, 68% of days exceeded the framework limit. That is actionable information: it tells you whether the problem is a few extreme outliers or a structural issue with the meter's operating range.

What to do about it

If you suspect your meter spends significant time at low flow, the most useful step is to extract daily average flow data from the historian and back-test it against the turndown curve.

This does not require any physical changes to the metering installation. It is a desk exercise using data you already have.

If the flow-weighted uncertainty exceeds the limit, the options are:

• Reduce the DP transmitter span (re-range to match actual operating DP more closely)
• Install a dual-range DP system (high-range and low-range transmitters)
• Move errors from URL/span basis to reading basis (upgrade to a transmitter with better reading-based accuracy)
• Accept the result and report at the appropriate tier

The point is to know. A meter that looks compliant at the design point but is not compliant in practice is worse than a meter that honestly reports a higher uncertainty, because the first one gives false confidence.