Diaphragm Gas Meters: Center an Error Curve Without Narrowing Its Span
Abstract
When a diaphragm gas meter gives errors that are consistently offset, a transmission-ratio adjustment can help center the curve. Quan, Tang and Liang examined that approach for wide-range household meters and checked it using measurements at ten characteristic flow points. This article reconstructs the study’s distinction between error-curve position, high-flow span and weighted mean error. The reported G4.0-B3 sample retained a span of 0.73 percentage points after calibration, with weighted results of 0.003% and 0.19% for two different point sets. For buyers and laboratories, the case shows why the measurement points, weights, and before-and-after records all matter.
Background
A meter can show a consistent offset across its operating range while still displaying the same pattern of variation with flow. Quan, Tang and Liang investigated a production adjustment for wide-range diaphragm meters. The adjustment uses a different transmission gear pair. The resulting change is treated in the paper as a nearly uniform translation of the relative-error curve. Where that description holds for the meter under test, a change in the mean position can be considered separately from the high-flow span. The distinction affects which evidence a production team should collect before selecting an adjustment.
Research question
Can two characteristic points supply a practical centering target while a fuller set of measurements verifies the result? The source answers this for the G4.0 examples in its own test program. Its Figure 1 uses a G4.0-B1 example to explain the shape of the curve. The later validation in Figure 3 uses a G4.0-B3 sample. The difference in model designation matters because the transition flow and the pattern at the lowest test points differ between the two examples.
Figure 1. English redraw of Quan et al., original Figure 1, p. B29. Peak and maximum-flow errors are reported in the text. Other marker ordinates are approximate readings from the published plot.
Calibration method
The general weighted mean error in source Equation (7) is the sum of the products of point weights and relative errors divided by the sum of the weights. Equation (5) defines the high-flow curve opening, εh = Emax − Emin. For the author's simplified two-point example, qmid is 0.2 times qmax. The weights in Equation (14) reduce to 2/3 for qmax and 1/3 for qmid. A weighted result is therefore meaningful only when it names the flow points and the weighting procedure.
In the study's notation, setting the two-point weighted mean close to zero and preserving εh gives the target high-flow errors Eqmax,target = −εh/3 and Eqmid,target = 2εh/3 (Equations 23–24). The target shift ΔE is the target error minus the observed error at either point (Equations 26–29). The paper maps that calculated shift to the nearest available transmission gear pair. A post-adjustment measurement is required by the paper's process because a gear pair implements a discrete shift. Its decision diagram separates meters eligible for adjustment from those outside the company’s internal screening bounds.
Figure 2. English redraw of source Figure 2, p. B31. The decision thresholds belong to the example process.
Results and discussion
The G4.0-B3 validation sample had qmax of 6.000 m³/h and a transition flow of 0.250 m³/h. Before adjustment, the authors measured +1.55% at qmid and +0.82% at qmax. Subtracting gives the reported span of 0.73 percentage points. Their calculated target shift was approximately −1.0630 percentage points. The nearest listed choice was gear pair 4 at −1.0526 percentage points. The arithmetic below retains the source's distinctions between a measured point, a recommended gear pair, and a summary of multiple measurements.
Quantity | Source value | Interpretation |
Maximum flow qmax | 6.000 m³/h | G4.0-B3 example |
Transition flow qt | 0.250 m³/h | Separates the study’s flow ranges |
Errors at qmid and qmax before adjustment | +1.55% and +0.82% | Two source measurements |
Curve span before and after adjustment | 0.73 percentage points | Reported span in the study |
Calculated translation ΔE | −1.0630 percentage points | Nominal calibration target |
Gear pair 4 translation | −1.0526 percentage points | Nearest choice in the source gear table |
Weighted result, two points | 0.003% | Post-adjustment result for qmid and qmax |
Weighted result, ten points | 0.19% | Post-adjustment result for a different point set |
Figure 3. English redraw of source Figure 3, p. B32. Marker values read from the plot are approximate.
The two weighted results differ because the calculations use different point sets. The wider check can reveal individual errors or a low-flow pattern that the two-point target does not describe. Figure 3 is useful for that check because both curves are on the same axes. A buyer comparing calibration reports should ask whether the labs used the same test flows, reference conditions, order of adjustments and weighting method. Reporting a single mean or span alone makes it difficult to tell whether the underlying results are comparable.
Conclusion
For this sample, the transmission change moved the relative-error curve toward zero while leaving the reported high-flow span at 0.73 percentage points. The two-point calculation gives an adjustment target and the broader series provides evidence about the outcome. For a diaphragm meter selection review, provide the expected minimum and maximum flow and ask for a report with the error at each applicable test point, the weighting basis and the final adjustment record.
Limitations and outlook
The paper demonstrates an approach on specified wide-range household meters. A separate validation would be needed for a different transmission design, representative sampling, or an independent uncertainty estimate. The figures show examples and screening criteria reported by the researchers. The applicable acceptance requirements should be established for the destination market and meter class.
Notes
Percent errors and percentage-point differences serve different purposes in this article. The figure markers correspond to the characteristic operating points of the source. Where the published graphs are used to read a value, treat the point as approximate and use the text or table for exact reported values.
References and Image Sources