When the Transmitter and the Gauge Disagree

The argument has been going on in pump rooms for as long as there have been transmitters. The operator trusts the gauge because it is on the pipe and has been there for twenty years. The technician trusts the transmitter because it was calibrated last spring. The engineer trusts neither and asks for a third reading. All three are right to be suspicious, and the argument is settled not by seniority but by understanding where a four psi difference can come from and then measuring against something better than either.

Differences that are physics

Two instruments reading the same pressure at different heights read different pressures, because the column of water between them weighs something. Water adds 0.433 psi for every foot of elevation, so a transmitter mounted six feet above the gauge tap, on a liquid-filled impulse line, reads about 2.6 psi lower than the gauge, and that is correct. The SCADA number should be corrected for it, in the transmitter's zero or in the scaling, and the correction should be written on the loop sheet so the next person does not calibrate it out. Temperature matters too: a gauge in the sun on a summer afternoon reads differently from one in a cool building, and a transmitter's specification includes a temperature effect that is usually small but not zero.

Differences that are accuracy

A gauge has an accuracy class, and the class is a percentage of the full range, not of the reading. A common industrial gauge in the 2.5 percent class on a 0 to 160 psi dial is allowed to be off by 4 psi anywhere on the dial, and by more once it has spent a decade on a pipe with a pump cycling against it. A transmitter of ordinary quality is specified at a fraction of a percent of span. Two instruments can therefore disagree by several psi with both inside their specifications, and the disagreement is not a fault in either. It is a statement that the gauge was never a reference. If the plant needs a gauge that can check a transmitter, it needs a test gauge in the 0.25 percent class or a pressure calibrator, kept in a case, not on a pipe.

Differences that are errors

Scaling

The transmitter is ranged 0 to 200 psi and SCADA scales the signal as 0 to 160, or the raw count range in the controller is set for a different card. The transmitter is right, the number on the screen is wrong, and a loop check with a simulated 12 mA shows it in a minute.

Trapped air

A liquid impulse line with air in it reads low and unsteady, and the reading changes when the line is tapped. Bleed it. A gas line with condensate in it has the opposite problem.

Zero drift

A transmitter that has crept a psi or two off zero over a couple of years reads that error at every pressure. Vent it to atmosphere and see what it says.

A damaged gauge

A gauge that has been overpressured, frozen, or pounded by pulsation may read anything. A gauge that does not return to zero when vented is done.

Isolation valves

A partly closed or plugged root valve makes a reading sluggish and, with a leak past it, wrong. Compare the response to a pump start.

Pulsation

A gauge with no snubber reads the average of a needle that is bouncing; the transmitter with damping reads something else. Neither is lying.

Settling it

1. 1

   Write down both readings as found

   Before anyone bleeds, adjusts, or taps anything, record the gauge, the transmitter's local display, the SCADA value, and the time. This is the record that tells you what the plant has been seeing.

2. 2

   Work out the elevation

   Measure the height difference between the two taps, or between the tap and the transmitter, and compute the correction. Much of the argument often ends here.

3. 3

   Check the scaling

   Compare the transmitter's configured range with the controller and SCADA scaling. Inject a known current at the loop and confirm the screen shows the expected value.

4. 4

   Bleed the lines

   Bleed the impulse lines and watch both readings. A jump is trapped air or condensate, and the reading after bleeding is the one to keep.

5. 5

   Apply the reference

   Connect the test gauge or calibrator at a tee on the same tap, and compare all three. The reference decides. Adjust only the instrument that is wrong, and only after the as-found is recorded.

6. 6

   Record the result

   As-found, as-left, the reference used and its calibration date, and the elevation correction, on the loop sheet where it will be seen next time.

The same argument with level

Level has the same disagreements with different names. A hydrostatic transmitter at the bottom of a tank and a sight glass on the side differ by where the transmitter's diaphragm sits relative to the zero the operators expect, and by the specific gravity if the liquid is not water. A radar unit reads distance to the surface and depends on the reference height entered at commissioning; a wrong tank height makes every reading wrong by the same amount. A float switch and a transmitter disagree by the float's own travel and the hysteresis of its switch. In each case the procedure is the same: record as-found, work out the geometry, check the scaling, then apply a reference such as a tape or a known fill.

What the operators actually need

Operators do not need the gauge and the screen to agree to the decimal. They need to know which one to act on and why the other differs. A note on the screen, or on the gauge, that says the transmitter reads 2.6 psi low by elevation and that the gauge is a rough local indication, ends the argument in the pump room for good. A plant that has written that down for every loop where it matters has done something most plants never do: it has decided what its instruments mean.