Multi-Position Gas Meter Test Benches: Why the Same Meter Can Give Different Results

Multi-position gas meter verification equipment allows several gas meters connected in series to be verified simultaneously, improving testing efficiency. Position-dependent differences can, however, affect the accuracy of verification results. Drawing on practical experience, the authors combined theoretical analysis and experiments to investigate three sources of position-dependent differences: pressure loss, meter error characteristics, and flow-field distribution. The findings identify measures that may reduce these differences and improve the reliability of gas meter verification in series.

Keywords: gas meter; verification equipment; position spread; accuracy

Natural gas plays an important role in the energy system. As China's cities and gas distribution networks expanded, demand for gas meters and the corresponding verification workload increased. Connecting meters in series allows a multi-position bench to test several units in one run. The authors observed that the same meter could give different verification results at different bench positions. Such differences can affect how a meter's metrological performance is assessed. The study investigates their causes and possible remedies to inform the use and design of multi-position benches.

The six-position sonic-nozzle bench used in the study had a flow range of 0.04–65 m³/h and a relative expanded uncertainty of Uᵣ = 0.36% (k = 2). It comprised a reference sonic-nozzle system, temperature and pressure instruments, piping and a drive system, mounting positions and a sampling system, and a computer control system. When the ratio of static pressure at the nozzle throat to upstream stagnation pressure reaches its critical value, mass flow through the nozzle reaches its maximum for the given upstream conditions. Lowering the back-pressure ratio further leaves the nozzle flow unchanged. The sampling system records the volumes accumulated by the meter under test and by the nozzle reference. The indication error is obtained by comparing the meter's volume with the reference volume corrected for temperature and pressure. The test meters were class 1.5 G25 ultrasonic gas meters with a flow range of 0.25–40 m³/h.

Can one gas meter produce different indication errors when mounted at different positions on the same serial verification bench? The study asks how pressure loss, meter error characteristics and inlet flow distribution contribute to the observed spread, and whether a piping change can reduce it under the tested conditions.

The authors combined an inlet pressure and flow analysis with two experimental comparisons. A six-position sonic-nozzle bench was used to compare indication errors across positions for six class 1.5 meters. A separate three-position bench was used to compare the same ultrasonic meter at 40 m³/h before and after adding a straight pipe section, while holding the stated environmental conditions unchanged. The three mechanisms and the respective test data are presented below in the order of the source paper.

Pressure loss is the mean pressure drop between a gas meter's inlet and outlet at a specified flow rate. The paper relates it to friction within the meter, including friction associated with its flow passages and moving components. The authors cite the then-applicable Jiangsu regulation JJG (Su) 206–2018, which specified a maximum pressure loss of 300 Pa for meters with 16 m³/h ≤ Qₘₐₓ ≤ 65 m³/h. They estimate that the inlet-pressure difference between the first and sixth positions of a six-position bench may exceed 1,000 Pa and affect the comparison.

The analysis treats the meter inlets as being at the same elevation and approximates air as an ideal gas. The paper applies the following Bernoulli relation at the inlets:

p + ½ρv² + ρgh = C　(1)

p = inlet pressure (Pa); ρ = gas density (kg/m³); v = velocity (m/s); g = gravitational acceleration (m/s²); h = inlet elevation (m); C = constant.

v = √[2(C − p − ρgh)/ρ]　(2)

In the authors' analysis, with C treated as constant, Equation (2) associates lower inlet pressure p with higher inlet velocity v. For a fixed cross-sectional area s, Q = v × s, so changes in velocity change the local volumetric flow rate. The authors argue that pressure losses in meters connected in series can therefore alter the flow conditions at different positions and affect their verification results. They propose reducing friction inside the meter, improving inter-position piping and assessing the cumulative pressure loss when setting an allowable position difference and choosing the number of test positions.

Meters can differ in indication error, stability and repeatability because of differences in manufacture. The authors identify such differences as an important contributor to the results obtained at different bench positions. They define the position spread Δ as follows:

Δ = |Eₘₐₓ − Eₘᵢₙ|　(3)

Eₘₐₓ and Eₘᵢₙ are the highest and lowest indication errors measured for the same test meter across the bench positions.

Using the criterion cited in the paper, the position spread was limited to one quarter of the numerical accuracy class of the meter under test. The authors tested six class 1.5 meters under the same environmental conditions on the six-position bench, applying a threshold of 0.375%. The data are reproduced in Table 1.

Two of the six meters exceeded the 0.375% threshold used in the study. This demonstrates that the measured position spread can differ among meters tested on the same bench. The authors recommend improving meter stability and reliability. To assess position consistency on a measurement standard, they recommend selecting a test meter with stable metrological characteristics.

Bends are common in gas meter test benches and field installations. The paper explains that velocity and centrifugal-force differences across a bend can generate secondary flow, increasing disturbance and reducing the uniformity of the velocity profile. This can influence results across bench positions, particularly for meters sensitive to inlet flow profiles, such as ultrasonic and thermal gas meters.

To investigate the effect of an added straight pipe section, the authors used a three-position sonic-nozzle bench incorporating bends. It had a flow range of 0.04–65 m³/h and a relative expanded uncertainty of Uᵣ = 0.34% (k = 2). They tested the same ultrasonic meter, serial number 35180824128, at 40 m³/h with and without the added section while keeping environmental conditions unchanged. The reported meter stability test result was 0.23%. See original Table 3 for the measurements.

After the straight section was added, the measured position spread fell from 0.70% to 0.28%. The authors interpret the result as evidence that the added straight section improved the flow distribution and position consistency in their setup. They suggest straight lengths of 8D–15D, where D is the pipe's internal diameter, and propose flow guides where available space is limited. They also recommend reducing the number of bends in the bench piping.

The six-position results show that the choice of test meter matters when assessing bench position consistency. Two meters exceeded the study's 0.375% criterion, while the other four did not. On the separate three-position bench, adding a straight section reduced the spread from 0.70% to 0.28% for the tested meter, a reduction of 0.42 percentage points. Taken together, the experiments support checking both meter stability and the inlet piping arrangement when diagnosing a position-dependent result. The straight-section comparison covers one meter at one flow point.

Drawing on practical verification experience, theoretical analysis and experiments, the paper examines the causes of position-dependent differences and potential remedies in terms of pressure loss, meter error characteristics and flow-field distribution. The authors intend these findings to help verification laboratories and gas meter manufacturers improve the accuracy and reliability of multi-position testing.

A follow-up study could repeat the straight-section comparison across several flow points, meter technologies and inlet configurations. Recording position-specific pressure and temperature, repeated readings and measurement uncertainty would help separate the effect of piping from meter variability and establish how far the observed improvement transfers to other benches.

Equations (1)–(3) reproduce the source analysis. In an actual bench, losses between positions must be accounted for before using the Bernoulli relation to predict a velocity difference. The study’s historical criteria, sample readings and straight-pipe lengths should be interpreted within the original test context.

Xu Z., Zhao Z., Zhou Y. Analysis of Causes and Countermeasures for Position Differences in Multi-Position Gas Meter Verification Equipment [translated title]. 2022, 32(2), pp. 34–36. DOI record.

Editorial cover photo: Betty Carrell, U.S. Department of Energy. U.S. Government Work per source record. Cropped for cover layout with title overlay. The photograph illustrates engineering work.