Zero-Shift Found on an In-Service Orifice DP Flow Loop
A 4-inch orifice meter on a natural-gas line uses a DP transmitter ranged 0-50 inH2O, 4-20 mA, with square-root extraction performed in the DCS (DCS flow span 0-12 MMSCFD). The instrument specification calls for accuracy of +/-0.25% of span. The transmitter was bench-calibrated six months ago and is still in service. You isolate it with the five-valve manifold, vent it, and connect a certified pressure source to the high and low ports with the equalizer closed. Results: 0.00 inH2O -> 4.08 mA; 12.50 inH2O -> 8.08 mA; 25.00 inH2O -> 12.08 mA; 37.50 inH2O -> 16.08 mA; 50.00 inH2O -> 20.08 mA. The line cannot be shut down today. The block valves, equalizer, and vent are all operable.
1. Calculate the zero error and span error in percent of span. Is the transmitter within specification? Which adjustment, if any, is correct, and which adjustments would be wrong?
Read the worked answer
Calculate:
- Output at 0% input = 4.08 mA -> zero error = +0.08 mA
- Output at 100% input = 20.08 mA -> error = +0.08 mA
- Measured span = 20.08 - 4.08 = 16.00 mA -> span error = 0.00%
Convert to % of span: 0.08 mA / 16.00 mA = 0.50% of span.
Compare with tolerance: +/-0.25% of span = +/-0.04 mA. The zero error is twice the limit, so the transmitter is out of tolerance. The span is perfect.
Diagnosis: both endpoints are displaced by the same amount, with no change in gain. That is a pure linear offset (zero shift), not a span/gain error. The intermediate points (8.08 / 12.08 / 16.08 mA) confirm linearity is intact.
Correct action: a zero trim only. Do not perform a full sensor trim (two-point trim): the span is already correct, and re-trimming both endpoints adds risk for no benefit. Do not perform a 4-20 mA output trim either, because the deviation is present in the digital PV, not only at the analog output.
2. Before you trim, what else could produce a 0.5% offset, and how would your diagnosis change if the transmitter zeroed correctly at the ports but still read wrong in service?
Read the worked answer
First, note what the test already proves. Pressure was applied directly at the H and L ports with the equalizer closed and the process isolated, so the transmitter saw a true zero differential. A 0.5% offset under those conditions lives in the transmitter (sensor/electronics drift, a previous trim performed at a different static pressure or temperature, or a stored zero offset from an over-pressure event).
Still rule out before adjusting:
- Verify the pressure standard's certificate and that the reference (deadweight vs. digital gauge) is at the same elevation as the transmitter ports.
- Confirm the transmitter was allowed to stabilize at ambient temperature; a unit brought from a hot skid into shade can drift for 30-60 minutes.
- Check the static-pressure effect: many DP cells have a small zero shift with line pressure. Record whether the as-found data was taken at 0 psig static or at line pressure.
- Record the as-found values on the calibration sheet before touching anything.
How the diagnosis changes: if the unit zeroed perfectly at the ports but reads biased while in service, the error is not in the transmitter. It is in the process connection: condensate or liquid in the high-side impulse line (adds positive head), a partially plugged or sloped-wrong tap, a leaking low-side vent or drain, or a wet leg that has partially vaporized. A zero trim in that case would hide a real installation fault and would be wrong.
3. After trimming, how do you verify the loop, and why does a 0.5% DP error matter more than a 0.5% flow error on this particular loop?
Read the worked answer
Verification:
- Re-apply 0 / 12.5 / 25 / 37.5 / 50 inH2O and confirm within +/-0.1% of span (target better than the +/-0.25% spec, to leave margin).
- Record as-found and as-left values on the calibration record and update the loop folder.
- Return the transmitter to service using the correct sequence: open the equalizer, open the high block, open the low block, close the equalizer. Then confirm the DCS reading against a known process condition.
Why 0.5% of DP is not 0.5% of flow: the DCS extracts the square root, so flow% = sqrt(DP%). The absolute DP offset dominates at low flow.
- At a true DP of 1.0% of span (10% flow), an indicated DP of 1.5% gives sqrt(0.015) = 12.2% flow instead of 10% - a relative error of about 22%.
- At true zero flow, the loop indicates sqrt(0.005) = 7.1% flow. The totalizer keeps counting while the line is actually down.
That phantom flow at zero is the reason the zero trim is not cosmetic, and the reason the DCS low-flow cutoff setting should be reviewed with the process engineer. A span error of the same 0.5% would have been far less significant at the bottom of the range.