Level Loop Checks Out but Reads Low When the Tank Fills
During commissioning of open tank T-201, a DP transmitter measures level over a working range of 10 ft of water. Range is 0-120 inH2O = 0-100% of the DCS span. The high port connects to the lower tap; the low port is vented to atmosphere. To clear a new pipe support, the transmitter was mounted on a bracket 8 in above the lower tap; the high-side impulse line rises from the tap and is liquid-full. Documents are clean: the transmitter is ranged 0-120 inH2O with a valid bench calibration certificate, and the DCS analog input is scaled 0-120 inH2O. The loop check was signed off after injecting 4/12/20 mA at the marshalling cabinet and seeing 0/50/100% on the DCS. During the wet check, the sight glass reads 5.0 ft (60 in) above the lower tap, the DCS reads 43.3%, and the transmitter local display reads 52.0 inH2O.
1. Explain the discrepancy quantitatively. Why did the loop check and the bench calibration both pass?
Read the worked answer
Convert the DCS reading to engineering units: 43.3% of 120 inH2O = 52.0 inH2O, exactly matching the transmitter's local display. So the transmitter and the DCS agree with each other - the signal path is healthy.
Compare with the truth: the sight glass shows 5.0 ft above the lower tap = 60.0 inH2O of head above that tap.
Bias: 60.0 - 52.0 = 8.0 inH2O, exactly the height of the transmitter above the lower tap.
Why: the impulse line is liquid-full, so the pressure at the transmitter equals the pressure at the tap minus the static column from the tap up to the transmitter. Level head is measured from the tap, so every reading is low by 8 inH2O of head, which is 8/120 = 6.7% of span. At a verified empty tank the transmitter would see -8.0 inH2O and indicate about -6.7%, which would also hold the low-level alarm on.
Why the checks passed: the loop check verified only the electrical path - 4/12/20 mA injected at the cabinet and read back at the DCS. The bench certificate verified the transmitter against applied differential pressure. Neither test includes the physical installation: the transmitter is doing exactly what it was ranged to do, but the process connection imposes a -8 inH2O bias at zero level. The loop folder had no check tying the transmitter mounting height to the tap elevations.
2. What is the correct configuration fix, how do you verify it, and why is a blind field zero trim with an empty tank the wrong move?
Read the worked answer
Correct fix - zero elevation (suppression): set LRV = -8.0 inH2O and URV = 112.0 inH2O. The span stays at 120 inH2O. Check that the transmitter model permits a negative LRV within its sensor limits (8 inH2O is trivial for a DP cell). The DCS scaling is unchanged - 4 mA must correspond to true 0% level and 20 mA to true 100%.
Check the math: at a true 60 in level, the DP seen by the transmitter is 60 - 8 = 52 inH2O, so output = (52 - (-8)) / 120 = 50.0%. Correct.
Alternative: relocate the transmitter to tap level or below, which removes the bias without re-ranging - but that means re-piping and a new bracket, a longer job.
Verification: two points minimum. At the sight glass 5.0 ft, the DCS must read 50.0%. At a verified empty tank, 0.0%. Record as-found and as-left LRV/URV.
Why a blind empty-tank zero trim is wrong: with the tank empty the transmitter legitimately sees -8 inH2O, and the true 0% must be verified against a known reference, not assumed. Zeroing to whatever the transmitter happens to read at that moment is precisely how this error survived the original commissioning - it produces a plausible-looking 0% while the span reference is wrong. If a field zero is used, it must be done with the level independently verified and then confirmed at a second known point.
Consequence if left uncorrected: the high-level alarm at 90% would annunciate at a true 96.7%, and at an empty tank the indicated level of about -6.7% keeps the low-level alarm asserted - nuisance trips and a missed high-level alarm.
3. What would have caught this, and what do you add to the commissioning and loop-check procedure?
Read the worked answer
What would have caught it: a wet check (actual fill against a known reference) done before the loop was signed off, or an installation walk-down comparing the as-built mounting height to the datasheet and the tap elevations. The instrument was correct; the installation was not.
Add to the procedure:
- Installation verification, before calibration sign-off: confirm tap elevations against the P&ID/datasheet, transmitter mounting height relative to the taps, impulse line slope and liquid fill, and vent/drain location. Record the elevation difference on the loop sheet.
- Calculated bias check: compute the expected zero-elevation/suppression from the physical dimensions and compare with the transmitter's configured LRV. Any mismatch is a hold point.
- Two-stage check: the electrical loop check (4/12/20 mA at the marshalling cabinet) plus a process-variable check at a known condition. For level: a verified empty or known-level reading before the wet check, then a wet check against the sight glass at two levels.
- Full-range verification: confirm the variable responds at both extremes with the correct direction, not just the midpoint.
- Documentation: as-found and as-left LRV/URV, tap elevations, and the reference used for each verification point, in the loop folder.
Close out: update the loop folder and as-built drawings with the new range, re-run the alarm and trip verification at the corrected scaling, and note the finding in the commissioning punch list with the root cause (mounting height changed to clear a support, no MOC against the installation drawing).