Why Where You Put the Probe Decides the Answer

Why this matters

A probe does not report the system. It reports its own sensing element, and the element only knows what it is touching, immersed in, or looking at. Everything else you believe about that number is an inference you made on the probe's behalf. Most of the time the inference is safe and nobody has to think about it. The rest of the time a technician puts a thermal camera on a panel, reads a comfortable number off a terminal that is running hot enough to anneal the lug, and closes the cover. The instrument was fine. The element was not coupled to the thing the tech thought it was coupled to, and nothing on the display says so.

A sibling article covers which point along a gradient to pick and how to judge that against the tolerance of your question (see References). This one is about whether the element is coupled to the quantity you want at all, and whether the point you chose stands for the bulk.

A probe reports its own condition

Write the chain out once and it stops being abstract. The system has a temperature, a pressure, a velocity. That quantity reaches the sensing element through some path: metal-to-metal contact, immersion in a fluid, a column of air, a cone of infrared radiation. The element settles at whatever that path delivers. The display converts the element's state into a number and prints it with no memory of the path.

So every placement question is really two questions. What is the element coupled to, and how strongly? And is that spot representative of the bulk you intend to describe? They fail independently. A perfectly coupled probe in an unrepresentative spot gives you a true number about the wrong place. A badly coupled probe in the perfect spot gives you a false number about the right one.

Before you place a probe anywhere energized or hot

• An infrared or contact survey with the cover off is energized work. It sits behind the gate at 29 CFR 1910.333(a)(1), which requires de-energizing before work on or near live parts unless that is infeasible, and treats a test that can only be done energized as one of those cases. Use the electrical protective equipment 29 CFR 1910.335(a) requires for the exposure, work within the arc-flash boundary and PPE determined by your employer's electrical safety program under NFPA 70E-2021 in the edition that program adopts, and stand out of the line of the panel opening rather than square in front of it.
• Do not solve a placement problem by moving your face closer to an open energized panel. The two compliant answers are an infrared viewport listed for that enclosure, so the cover stays on, or an instrument with a tighter optic used from where you already are.
• Hot surfaces and hot fluids. Contact probes on steam, hot water and flue components come off hot enough to burn; handle by the grip, wear heat-resistant gloves, and set the probe on a non-combustible surface. If you have to disturb thermal insulation to reach pipe metal, treat the insulation as unknown until identified: cutting, tearing or abrading older thermal insulation can release respirable fibres, so wet the cut, use a respirator selected under a 29 CFR 1910.134 program, and stop entirely if the material is suspect asbestos-containing until it is assessed under 29 CFR 1910.1001.
• Anything rotating. A probe going into a duct, plenum or belt guard means the drive is de-energized, locked and blocked under 29 CFR 1910.147 first. That standard is the right one here because the hazard is mechanical; it excludes electrical utilization hazards at 1910.147(a)(1)(ii)(C).

Coupling: immersion, contact and view

Immersion. A stem or a well conducts heat along its own length to whatever is holding it, so a shallowly-inserted probe reads partway between the fluid and the mounting. The working rule of thumb in temperature measurement is immersion of roughly ten stem diameters in a moving liquid, and more in gas because gas couples to the stem far more weakly. Treat that as a starting point and take the number from the probe manufacturer's installation instruction, which is the document that owns it. The tell that you are short: insulate the exposed stem and the reading moves. If it moves, it was conducting.

Contact. A surface probe on a pipe reads a blend of pipe metal and the air around the junction, weighted by which couples better, and bare that blend can sit well away from the pipe. Clean the surface, use a conforming tip or thermal compound, and cover the probe and a few inches of pipe either side with insulation. Then wait, because you have just changed the local thermal environment and the reading has to re-settle.

View. An infrared instrument averages everything inside its field of view. Spot size grows with distance at a fixed ratio, published as the distance-to-spot ratio, so a 12 to 1 instrument at 24 inches is reading a 2 inch circle whether your target is 2 inches or a quarter inch. Emissivity, the fraction of ideal radiation a surface emits, is a property of that specific surface and finish; take it from a published emissivity table or the instrument's own guidance and never from memory, because bare bright metal and painted metal at the same temperature can read tens of degrees apart.

The placements that come off the list, and why

This is the part worth memorizing, because it is where a placement looks reasonable and is not.

• Pipe surface at a hanger, clamp, bracket or valve body. Those are heat sinks with their own mass and their own path to the room. You will read the fitting, not the fluid. Move to straight pipe well clear of the hardware.
• Duct velocity anywhere near a fitting. A traverse plane wants straight duct on both sides of any disturbance; the figure air-balancing practice commonly works to is on the order of seven and a half equivalent diameters downstream and two and a half upstream, from the procedural standards a balancing contractor works under, which bind through the project specification rather than on their own. Where the duct does not give you that, you increase the traverse point count and record that the plane was compromised.
• Supply air within sight of the heat exchanger or the electric elements. A probe that can see a glowing or hot surface receives radiation directly and reads above the air. Move downstream past a turn, or shield the element from the radiant view.
• Room air at a probe that can see a window, a radiant panel or direct sun. Same mechanism, same fix; where the radiant environment cannot be avoided, an aspirated or shielded probe is the instrument for that job.
• A closed enclosure cover, read as if it were the inside. A warm cover tells you heat is being generated inside and tells you nothing about by what factor the inside exceeds it, because the internal air film and the cover's own conduction sit between the two. If you need the internal temperature, you need access to the inside, not a better reading of the outside.
• Anywhere you had to reach past a guard, into a live panel, or through insulation of unknown type to get to. There is nearly always a listed access point, a port, or an instrument with a longer reach.

Representativeness: when one point stands for the bulk

A single point describes the bulk only when the bulk is genuinely mixed, or when you know the profile and have chosen the point deliberately. Neither is the default.

Stratification is the normal state of most things a field tech measures. Air in a duct is faster in the middle than at the wall. Air in a room is warmer near the ceiling. Water in an unmixed tank layers by temperature. A plenum after a coil carries streaks that have not blended yet. In all of those, one point is one point, and the honest options are to traverse and average, to move to a location where mixing has occurred, or to state on the record that the value is a single point at a named spot.

That is why the traverse exists as a method. Nobody invented multi-point traverses because single readings were inaccurate; they invented them because single readings were unrepresentative, which is a different defect and is not cured by a better instrument.

Worked example: the 200 F lug that read under 100 F

An infrared survey of a distribution panel, cover off, done under the energized-work gate and PPE described above. The instrument's optic is 12 to 1. The tech stands at a comfortable arm's length, call it 24 inches, and sweeps the terminations. Every reading comes back between the low 90s and about 97 F. Nothing gets flagged.

Work the geometry. At 24 inches with a 12 to 1 optic, the spot diameter is 24 divided by 12, which is 2.0 inches. Spot area is pi over 4 times 2.0 squared, which is 3.14 square inches. A terminal lug is roughly half an inch across, so its area is pi over 4 times 0.5 squared, which is 0.196 square inches. The target is 0.196 divided by 3.14, or 6.25 percent, of what the instrument is looking at. The other 93.75 percent is cool panel interior.

Say the lug is genuinely at 200 F and the surrounding panel interior is at 90 F. Area-weighted, the instrument sees 0.0625 times 200 plus 0.9375 times 90, which is 12.5 plus 84.4, or 96.9 F. That is squarely inside the range the tech recorded and looks like a healthy panel.

Two qualifications on that figure. Emitted power rises steeply with temperature, so radiance-weighting gives the hot spot somewhat more influence than this linear temperature average does, which makes 96.9 F the pessimistic end. And bare bright lug metal emits far less than the painted interior, which pushes the apparent lug reading down again unless emissivity is set for the surface being read. Neither rescues the survey: a 200 F termination hid inside a spot that read under 100 F.

The fix, and the one that is not allowed. Get the spot onto the target. Halve the distance and the spot halves to 1.0 inch, which still leaves the lug at a quarter of the view. Getting to a 0.5 inch spot with this optic means 6 inches from an open energized panel, which puts your head and the instrument inside the arc-flash boundary, and that placement comes off the list. The compliant answers are a tighter optic used from the same standoff, an infrared viewport listed for the enclosure so the survey happens with the cover on, or a de-energized inspection with a contact probe and lockout applied under 29 CFR 1910.333(b)(2), proving dead with the live-dead-live sequence in NFPA 70E-2021, 120.5.

The failure mode. A panel gets surveyed annually, passes every year, and fails at a lug. The photographs in the file are real, the instrument was in calibration, and every reading was true about the two-inch circle it described. Nobody ever wrote down the standoff distance, which is the field that would have made the defect visible.

How to verify you got this right

• Change the placement slightly and see whether the number moves. Insulate the exposed stem, slide the surface probe six inches along the pipe, halve your infrared standoff. A number that jumps was coupling-limited, and the higher-coupling placement is the truer one.
• Compute your spot size before you accept an infrared reading, and write the standoff distance on the record next to the temperature. Without it, the reading cannot be reproduced or defended.
• Say what the reading represents, in words, before you write it down. "Surface of the suction line, 18 inches from the compressor, insulated over the probe" is a measurement. "Suction line temp" is a rumour.
• Ask whether the point is mixed. If you cannot argue that it is, the record says single point at a named location, and any average you report is a traverse or it is nothing.

References

• 29 CFR 1910.333(a)(1) and 1910.335(a) - the energized-work gate and the electrical protective equipment required for an infrared or contact survey with covers removed
• 29 CFR 1910.333(b)(2) and NFPA 70E-2021, 120.5 - electrical lockout and the live-dead-live proving sequence for the de-energized alternative, binding through the employer's electrical safety program
• 29 CFR 1910.147 - mechanical isolation before a probe enters a duct, plenum or guard
• 29 CFR 1910.134 and 29 CFR 1910.1001 - respiratory protection program requirements and asbestos rules where reaching pipe metal means disturbing thermal insulation
• See related: How a Reading Changes With Where You Take It; The Measurement That Changes What It Measures; Airflow Measurement and Balancing Reference