Ultrasonic Gas Meters: Check Whether Contamination Shifts High-Flow Error
Abstract
Contamination tests on an ultrasonic gas meter can produce different changes at different flow points. Zhou, Li and Xiang measured three G6 meters before exposure, under two laboratory schemes, and after drying. At the highest tested flow of 10 m³/h, the more severe scheme generally led to a positive shift in indication error. Some intermediate observations moved in the opposite direction. This article reconstructs the five original plots from the paper’s tables and explains why the baseline, the individual test stage and the low-flow readings must be kept in the same record. It also identifies which findings need further evidence before they inform a field specification.
Background
An ultrasonic meter's published range can tell a buyer where it is intended to measure. That range alone cannot show how indication error will change after the measuring path has been exposed to contaminants. A test report can answer this question when it distinguishes a meter's initial response, the conditions after each exposure and the response after a defined period of drying or rest. Taking one average across these stages can hide a change at an operating point that is important to the installation.
The question for this comparison is whether a change in measured error is broad across the range or concentrated at one flow. Zhou and colleagues provide the individual readings for three meters. The three samples are valuable for seeing how different responses can occur within one study. They do not provide a distribution for all meter models.
Method and baseline
Before exposure the authors tested samples G6-2683, G6-2716 and G6-2720 at 0.6, 1.0 and 10 m³/h. The initial 10 m³/h errors were −1.47%, −1.14% and −0.93%, respectively. All three started at a negative error at that point. On the two lower flow points listed in Table 3, initial readings were positive. The difference in starting sign is important when interpreting a reported result after exposure.
Figure 1. Baseline results redrawn from source Table 3 and Figure 1.
The paper describes two schemes using laboratory mixtures. In scheme 1 the three sample meters were run through exposure stages followed by a dried check. Scheme 2 used more severe laboratory exposures and distinguishes immediate and dried readings. The redraws below use the published table values. Lines connect sampled values as guides. They do not reconstruct any unmeasured intermediate measurements.
Scheme 1: Read the flow curve for each meter
Figure 2. G6-2683, redrawn from source Tables 3–4 and Figure 2.
Figure 3. G6-2716, redrawn from source Tables 3 and 5 and Figure 3.
Figure 4. G6-2720, redrawn from source Tables 3 and 6 and Figure 4.
The tested meters responded differently. For one exposure under scheme 1, G6-2683 had a 10 m³/h reading of −5.26%, while the same table reports −0.43% for its dried check. The two other samples showed positive 10 m³/h results at some stages and negative ones after the dried check. For a review, preserve the exact sample identity and sequence. Using a single ‘contaminated’ label would remove the distinction between those conditions and obscure whether a measured change persisted after drying.
Scheme 2: Separate immediate and dried results
Figure 5. Comparison at 10 m³/h, redrawn from source Table 3, Tables 7–9 and Figure 5.
Meter | Baseline at 10 m³/h | After Test 8 | After drying and rest | Final change from baseline |
G6-2683 | −1.47% | +3.40% | +2.40% | +3.87 percentage points |
G6-2716 | −1.14% | +4.22% | +2.66% | +3.80 percentage points |
G6-2720 | −0.93% | +5.11% | +3.47% | +4.40 percentage points |
The readings come from Tables 3 and 7–9. The last column is an editorial calculation subtracting the baseline error from the final error. For example, for G6-2720 the final difference is +3.47 − (−0.93) = +4.40 percentage points. This shows that a reading can move closer to zero after drying and still remain far from the starting result. The intermediate points also show reversals, so the experiment does not support a strictly increasing error at every stage.
Lower-flow results deserve separate attention. The same G6-2720 sample had a final reading of +0.17% at 0.6 m³/h, compared with +0.67% at baseline. At 10 m³/h its result moved in the opposite direction. A report that includes only a low-flow result cannot establish how the same sample behaved at maximum flow.
What to put in a comparison report
Label three quantities separately: the initial indication error, the indication error measured at the specified stage, and the difference from the initial value. If the report also discusses recovery, state the stage it follows. For G6-2716 at 10 m³/h, Test 8 and the final rested measurement differ by 1.56 percentage points. That is a reduction from the high reading, while the rested result is still 3.80 percentage points higher than baseline. The two numbers describe different comparisons. A table that labels both simply as a change after testing would be ambiguous.
Keep the flow readings in the same units and preserve the original sign convention. State whether the comparison uses repeated measurements, and ask for the measurement uncertainty where the report will support acceptance. The source’s tables are adequate for the calculations here. A plotted line does not establish the uncertainty at an unmeasured point.
Conclusion
In the study, the high-flow indication error generally shifted positively after the more severe laboratory exposure. Its magnitude depended on the sample and when the reading was taken. For an equipment enquiry or acceptance review, request the tested meter identity, initial readings by flow, the sequence of observations and results after the specified recovery condition. Keep the arithmetic difference in percentage points separate from the percentage indication error.
Limitations and next checks
The study used three instruments and specific controlled laboratory conditions. Its findings do not establish a field contamination threshold, a cleaning procedure, a service interval or a corrective factor for another product. The conditions, repeat results and reporting uncertainty for the proposed device remain important when the acceptance decision depends on the size of the shift.
References and Image Sources