How to Field-Verify a Temperature Reading in Place
Operations says the reading looks wrong; pulling the sensor means a permit, maybe a shutdown, and an argument. Before any of that, there are four ways to interrogate a temperature point while it stays in service, in rising order of effort. This page is for the technician or engineer who needs a defensible answer today, with the sensor still in the well.
Field-Verify a Temperature Reading in one line: To field-verify a temperature reading without removing the sensor, compare it against an independent measurement in this order: a redundant or adjacent instrument on the same fluid, a reference probe in a nearby test well, a surface measurement corrected for the installation, and finally a transmitter-side signal injection that splits sensor error from electronics error. Agreement within the loop's stated tolerance closes the complaint; disagreement tells you which half to pull.
What You Need
Bring a trusted reference probe with its own readout, a surface probe or infrared thermometer with its emissivity limits understood, a decade box or millivolt source for the injection step, and the loop documentation showing sensor type and range. Know what tolerance the point actually has to meet - a comfort-monitoring point and a reaction-temperature point deserve different levels of rigor.
Compare Against Independent Measurements Nearby
Start with the cheapest evidence: any other instrument measuring the same fluid. An adjacent loop on the same line, a redundant element in the same service, or an inlet-outlet pair with known behavior can all vouch for or against the suspect point in minutes from the historian, before you leave the desk. Be honest about what is comparable - a point upstream of a heater vouches for nothing downstream of it.
If the site installed test thermowells near the measurement, use one: a reference probe in a test well beside the process well is the gold standard of in-place verification, because both sensors see the same fluid at nearly the same spot. Give the reference probe time to reach the process temperature before judging; a deep well and a heavy probe can take surprisingly long to settle.
Use a Corrected Surface Measurement
Where no test well exists, a surface measurement on well-insulated pipe gives a usable sanity check: strip a small patch of insulation, measure the pipe wall with a contact surface probe, and expect the wall to read somewhat below the fluid temperature, with the gap growing as ambient gets colder and insulation gets worse. This will not settle a one-degree dispute, but it reliably answers is this point lying by ten degrees, which is usually the actual question.
An infrared thermometer works for the same check with two cautions: bare shiny pipe will read low unless the emissivity is handled - a paint patch or tape target fixes that, as covered in emissivity in temperature measurement - and the IR spot must be filled by the pipe, not the scenery behind it.
Split Sensor from Electronics with an Injection
If independent measurements disagree with the loop, find out which half is lying without pulling the element. Disconnect the sensor at the transmitter and inject a known signal - a decade box at a table resistance for an RTD, a millivolt source with reference-junction correction for a thermocouple. If the loop reads the injected value correctly, the transmitter and everything downstream are healthy and the sensor or its wiring is the suspect; if it reads wrong, you have a transmitter problem and the procedure in calibrating a temperature transmitter takes over.
A sensor implicated by this split still is not necessarily drifted - lead-wire faults produce the same signature - so check the wiring scheme and lead balance before condemning the element, and when it finally does come out, the ice-point check in bench-testing an RTD gives the definitive verdict.
Verifying the Result and Common Mistakes
Write down what agreed with what, within how much, at what process condition; a verification that lives in one technician's memory has to be repeated by the next one. If the point passed, note the method and close the complaint with the evidence attached. If it failed, you now know which half to schedule work on, which turns a vague pull the sensor job into a specific one.
Common mistakes: comparing against an instrument that is not actually measuring the same fluid, reading shiny pipe with an IR gun and believing the low number, judging a reference probe before it has settled, and skipping the injection step so a healthy sensor gets pulled to fix what was always a transmitter or wiring fault.
Frequently Asked Questions
Can I trust an infrared gun to check a process temperature reading?
For coarse verification, yes, with discipline: measure a high-emissivity surface or add a tape or paint target, get close enough that the spot is filled by the pipe, and expect the surface to read below the fluid temperature on insulated hot lines. It will resolve a ten-degree dispute credibly; it will not adjudicate a one-degree calibration question.
The loop reads wrong but the injected signal reads right - is the sensor bad?
The fault is on the sensor side of the injection point, but that includes the lead wires, terminals, and junction boxes as well as the element. Check lead-wire balance and the connection scheme before replacing anything; a corroded terminal or a collapsed compensation scheme mimics element drift exactly, and it will do the same thing to the new sensor.
Automation services
Need help turning this into a working system?
Merobix integrates SCADA, programs Allen-Bradley and Siemens PLCs, and designs and fabricates industrial control panels.
Meeting requests are reviewed before confirmation.