Modern flowmeters, especially those used for custody transfer, claim very high accuracy and repeatability. Unfortunately, even when correctly installed and maintained, flow technology rarely performs up to vendor statements. As such, users must leverage proven methodologies and best practices to quantify “real-world” accuracy and repeatability.

Real-World Sources of Error[1]
Accuracy is a combination of bias (consistent error) and repeatability (random error). Standards, including those provided by bodies such as the American Gas Association (www.aga.org), International Organization for Standards (www.iso.org), and American Society of Mechanical Engineers (www.asme.org), or guidelines from agencies such as Measurement Canada (http://mc.ic.gc.ca/), describe how to correctly install and maintain a flowmeter to minimize bias error.2 This, of course, is critical in custody applications, as even a very small bias error can, over time, significantly favor either the buyer or seller. Repeatability is also important, as it allows users to: minimize disagreements between billing and check meters; identify operational problems such as leaks; and improve process control

By definition, a correctly installed and maintained flowmeter should suffer from zero bias error, so accuracy is repeatability. To quantify, suppliers will typically submit their devices to a third-party laboratory for testing. Unfortunately, this third-party laboratory test data is not useful in comparing flowmeters or predicting real-world performance. Installed or “real world” accuracy and repeatability generally do not equal lab test data, since few flowmeters are installed in laboratory conditions (i.e., rigidly controlled ambient temperature; low static line pressures; calibration before every test).

Ambient Temperature Variation
Many flowmeters — certainly most used in custody transfer — are installed outdoors. In some climates, ambient temperatures can easily vary ±25 C or more from calibration temperature. While standards bodies require users to compensate for fluid temperature variation, ambient temperature variations can have an even larger effect, especially with older, analog transmitters. These effects can be easily simulated on the bench by blowing warm air over a transmitter and watching its output change.

Figure 1. Flow Accuracy & Repeatability Error Sources

With orifice meters, ambient temperature variation affects the differential pressure transmitter and the static pressure transmitter. With linear meters, ambient temperature variation will affect any analog components in the flowmeter, as well as the static pressure transmitter used for pressure compensation. Even if actual line pressure is constant, the output of the static pressure transmitter — and hence the calculated mass flow — will vary with changes in ambient temperature.

High Static Line Pressures
With orifice meters, a high or changing line pressure can significantly influence the DP transmitter used to infer flow. To simulate this effect on the bench, the user can apply a small differential pressure across the transmitter. Then, add several hundred pounds of additional static pressure to both sides of the transmitter. In theory, the measured differential pressure should not change. In reality, it does.

Drift/Stability
The output of any analog component will vary over time. As with the ambient temperature effect described above, this can affect all flow technologies. Smart transmitters are more stable than older, analog transmitters or transducers. Within regulatory and contractual restrictions, a more stable transmitter will allow the user to obtain equivalent accuracy and repeatability when calibrated less frequently. An inferior device must be calibrated more frequently to maintain acceptable performance.

Quantifying Real-World Error

Figure 2. Typical Smart Transmitter Characterization NOT Optimized for Flow

Reputable suppliers publish specifications, which allow the user to calculate and predict the impact of error sources and other real-world effects on installed flow accuracy and repeatability. Published specifications can be used to calculate flow error caused by the DP transmitter in an orifice meter installation. The results at 100 percent flow, under typical installed conditions, are shown in Figure 1. In this example “Reference Accuracy” is a trivial component of total transmitter installed error. Two DP transmitters with identical 0.075 percent reference accuracy can provide dramatically different installed accuracy and repeatability. If a supplier does not publish specifications for real-world effects, this does not mean that their products are immune to these effects — usually, the reverse is true.

Although the example of an orifice meter is used below for illustration, it is important to note that installed accuracy is worse than laboratory accuracy for all flow technologies. Table 1 shows the impact of this error at lower flowrates. These errors apply to accuracy and repeatability and are fixed over the entire DP range (DP Éø flow2). Since DP declines twice as fast as flow, small errors at 100 percent — and small differences in transmitter accuracy — are magnified at lower flowrates. Improvements in transmitter accuracy, which seem trivial at full-scale flows, show dramatic effects at lower flows.

Table 1. Impact of DP Transmitter Error at Lower Flows

Flow Rate (SCFM)
1,000
750
500
250
125
DP
100
50
25
6.25
1.56
“Better” .075%
0.09%
0.16%
0.37%
1.48%
5.9%
“Worse” .075%
0.21%
0.38%
0.85%
3.40%
13.6%
Analog
0.65%
1.16%
2.60%
10.40%
41.6%

Low-Flow Characterization
Reputable suppliers minimize real-world effects by characterizing (or footprinting) their transmitters over a broad operating range. Completely different from calibration, characterizing a transmitter involves exposing it to a range of conditions — in particular, ambient temperature — and observing the impact on the measurement. Better smart transmitters include a built-in temperature sensor to measure ambient temperature. During operation in the field, observed ambient temperature variations can be automatically compensated in the microprocessor. Figure 2 shows a typical DP transmitter characterization, in which the impact of varying ambient temperature on the transmitter’s output is measured.

Best Practices:
Flow Accuracy & Repeatability

1. In designing any flow measurement system it is important to determine mutual accuracy and repeatability expectations over the range of flow (turndown). In custody-transfer applications, the key stakeholders are, naturally, the buyer and seller. In noncustody applications, the key stakeholders are the users of the data that the flowmeter provides. This determination of mutual performance expectations should be made by individuals who have a clear understanding of all of the costs of measurement disputes caused by poor accuracy and repeatability.
2. The second step is to quantify uncontrollable operating conditions. For a flow measurement, these can include:

  • Expected ambient temperature variation
  • Maximum static line pressure
  • Line pressure and temperature variation
  • Maximum allowable pressure loss
  • Minimum flow that needs to be measured (turndown)


3. Next, define hardware and installation and maintenance procedures to ensure that the measurement provides the required installed performance under the expected (uncontrollable) operating conditions. When using DP flowmeters, remember to specify that the DP transmitters be characterized for flow. This will dramatically improve flow turndown, as noted earlier.

While the first and second steps involve gathering data, the third step may require calculations and/or testing.

Suppliers normally characterize transmitters using reference inputs evenly distributed over the transmitter’s DP range. The benefit of this approach is that it allows transmitters to be used in a variety of differential pressure applications, including flow, level, pressure drop across a filter, etc. Unfortunately, while it maximizes flexibility, evenly distributed characterization is not optimized for flow applications for two reasons:

  • Half the characterization points are negative, hence not useful in the vast majority of (unidirectional) flow applications.
  • The positive points, while evenly distributed by DP, are clustered towards the high flows due to the square-root relationship between DP and flow.

The result of this clustering is that no characterization is performed between zero percent and 45 percent flow. This is, as noted previously, the flow range where DP transmitter accuracy is most important, since transmitter uncertainty is usually dominant only at low flows.

To avoid this problem, users should instead ask their supplier to provide DP flow transmitters that are “characterized for flow.” As shown in Figure 3, for these transmitters the supplier will apply the majority of the characterization effort to low-flow conditions, where the DP transmitter accuracy has the greatest impact on flowmeter performance. This will provide significantly better real-world performance at low flows, resulting in dramatically better flowmeter turndown. For example, a DP transmitter which is characterized for flow can double flow turndown in a given installation for a given installed accuracy and repeatability.

Figure 3. Typical Smart Transmitter Characterization Optimized for Flow

While ambient temperature variation is most prevalent in outdoor applications, line pressure variation can also be important. Newer, smart transmitter designs offer built-in compensation for line pressure variations.

About the Author
Mark Menezes, P.E., has 15 years of experience in industrial automation, including seven years with control systems and loop controllers and eight years with measurement. He has a degree in Chemical Engineering from University of Toronto and an MBA from York University’s Schulich School of Business. Currently, Mr. Menezes manages Emerson Process Management’s measurement business in Canada. He can be reached at Mark.Menezes@emersonprocess.com or 905 795-0299.

For More Information: www.rosemount.com


References
1. Menezes, “Calculating and Optimizing Repeatability of Natural Gas Flow Measurements,” Pipeline & Gas Journal, July 2001.
2. Miller, R.W., Flow Measurement Engineering Handbook, McGraw-Hill, 1996.


This article was adapted from a paper presented earlier this year at the Canadian Hydrocarbon School of Measurement.