Flow Indication Low After Transmitter Swap
A 6-inch orifice meter measures 0–500 gpm of water into a process header. The DP transmitter was originally calibrated 0–100 inH2O with square-root extraction performed in the transmitter, so its 4–20 mA output was linear with flow. During a shutdown the transmitter failed and was replaced with a spare calibrated 0–100 inH2O but left in linear (no square-root) mode. The DCS flow-input block has its square-root function disabled - it was turned off years ago because 'the transmitter already does it.' Two weeks later operators report the flow indication looks low at high rates and the plant mass balance no longer closes. A portable clamp-on ultrasonic meter reads 250 gpm while the DCS shows 125 gpm at that same moment. Production wants to keep running until the next outage, six weeks away.
1. Explain quantitatively why the DCS reads 125 gpm when the true flow is 250 gpm, and predict what the DCS would show at a true flow of 400 gpm.
Read the worked answer
Why it reads low
With a square-root-extracting transmitter removed and the DCS square root disabled, the DCS is displaying the raw DP fraction as if it were a flow fraction.
- True flow fraction = 250/500 = 0.50
- DP fraction = 0.50² = 0.25 → 25 inH2O (linear output ≈ 8.0 mA)
- DCS displays 0.25 × 500 = 125 gpm ✔ matches the ultrasonic reading.
So the indicated value follows I = F²/500, where F is true flow in gpm.
Prediction at 400 gpm
- DP fraction = (400/500)² = 0.64 → 64 inH2O → 14.24 mA
- DCS displays 0.64 × 500 = 320 gpm
The error is proportionate, not constant: the indication is correct only at 0 and at full scale. At 100 gpm true, the DCS would read 20 gpm. This is why the mass balance 'sort of' closes at high rates and falls apart at turndown.
2. The flow controller is in automatic with a setpoint of 400 gpm. What is the actual flow, and what is the practical consequence for the process and for a low-flow setpoint of 100 gpm?
Read the worked answer
The controller drives the valve until the indicated value equals setpoint, i.e. F²/500 = SP.
- SP = 400 gpm → F = √(400 × 500) = √200,000 ≈ 447 gpm (about 12% overfeed)
- SP = 100 gpm → F = √(100 × 500) = √50,000 ≈ 224 gpm (about 124% overfeed)
The consequence is not a display nuisance - the loop is closed on a corrupted measurement, so the error is fed back and amplified at low rates. Any ratio control, totalizer, or batch dose built on this signal is wrong in the same direction. At low setpoints the process is being overfed by more than double while the console shows it on target, which is a credible route to a downstream upset or relief scenario.
3. What is the correct fix that does not require an outage, and what precautions and follow-on checks are required?
Read the worked answer
Fix
Re-enable the square-root function in the DCS flow-input block and leave the replacement transmitter in linear mode. The two square roots must never both be active or both be absent - exactly one must be in the path.
Precautions during the change
- Put the flow controller in MANUAL and hold the current valve output; the PV will jump as soon as the block changes (at 250 gpm it goes 125 → 250 gpm).
- Warn operations and the board operator before the change, and record the before/after values.
- Verify the DCS block is actually the only square root - confirm the transmitter's own config is linear.
Follow-on checks
- Confirm zero and span on the new transmitter (4 mA at zero DP, 20 mA at 100 inH2O).
- Re-check controller tuning - the loop gain in engineering units has changed by roughly a factor of two over the operating range.
- Fix the totalizer, ratio stations, and any historian or advanced-control tag that consumes the same point.
- Check configuration control: the root cause is that the original design intent (sqrt in the transmitter) was not documented, so the spare was configured differently with no cross-check. Add the square-root location to the instrument datasheet and to the loop check sheet.