How to Read a Temperature Profile Along a Run

Why this matters

One temperature at the end of a long run tells you that something is wrong and nothing about where. So the tech guesses, and the guess is usually the most visible thing on the run: the equipment at one end, or the insulation they can see. Meanwhile the actual fault is often three feet of pipe at a tee nobody looked at, quietly dumping a large share of the loop's heat into a place it does not belong.

A temperature profile fixes that. Take the same reading at several points along the run, and each segment's change becomes attributable to what physically happens inside that segment. The trick is that a profile is only readable if every point on it was taken under the same conditions, which is where most attempts fall apart.

The steps below are ranked by how big a hole skipping each one leaves, not by the order you perform them. Execution order is given at the end of the list.

Before you walk the run

Each hazard on this job has a different control, and the one that gets skipped is the airborne one.

• Pipe insulation in an older building is presumed to contain asbestos until it is sampled. Thermal system insulation is exactly the material that historically carried it, and lifting it to reach a pipe surface releases fibers you will not see or smell. Do not cut, tear, or peel suspect insulation. Read on an already-exposed section, at a fitting where the jacket is open, or have it sampled and abated first. The general industry duty sits at 29 CFR 1910.1001 and the duty covering maintenance and renovation work that disturbs it sits at 29 CFR 1926.1101, and a field-service shop can fall under either depending on the job.
• Hot water burns fast. At around 140 F a serious scald takes only seconds of contact, and the time to injury lengthens sharply as the temperature falls toward 120 F. Read pipe surface with a strap-on or infrared instrument rather than opening a port, and do not loosen a fitting on a live line to insert a probe.
• Attics and crawlspaces. Step only on framing or laid planking; ceiling material will not hold you. Attic temperatures in summer drive heat illness quickly, so work in short intervals with water on hand and a second person aware you are up there.
• Metering a pump or a control to confirm the system state: electrical work is governed by 29 CFR 1910.333(b)(2), not by the mechanical lockout standard. Prove your meter live, prove the circuit dead, prove the meter again, the sequence in NFPA 70E-2021, 120.5. If you are isolating the pump itself rather than metering it, lock and tag it under 29 CFR 1910.147 for the mechanical and stored-energy hazard.

The seven steps, ranked by what skipping each one costs

1. Map the run into segments and write down what each segment is supposed to do to the temperature. Skipping this costs you the entire exercise, which is why it sits first. Without a written expectation per segment you have a column of numbers and no basis for calling any of them wrong. A segment is defined by its endpoints, its length, what it passes through, and whether anything ties into it. Record all four before you take a reading.

2. Fix one instrument and one method, and prove it against a known point. This is the most expensive step you can still get a number without, because it produces a confidently wrong answer rather than no answer. A profile is built from differences, and differences are where instrument offset does the most damage: two probes an honest 1.5 F apart turn a 2 F segment loss into anything from 0.5 F to 3.5 F. Use the same instrument for every point, apply it the same way every time (same contact pressure, same surface preparation, same dwell to a stable reading), and check it once against a stable known reference such as an ice bath before you start. An offset common to every point cancels out of every difference. An offset that varies point to point does not.

3. Take the entire set inside one steady operating window. Skipping this makes the profile fictional rather than merely uncertain. If the equipment cycles, a draw occurs, or a zone valve opens partway through your walk, the later points describe a different system than the earlier ones and the segment between them shows a change that never happened. Establish steady state first, note the start and end time of the window, and abandon the set if the state changed inside it.

4. Compare each segment against its own expectation, not against its neighbours. Skipping this reliably names the wrong culprit, and it is the single most common way a correct set of readings produces a wrong repair. The largest drop on a run is very often the longest segment, or the one passing through the coldest space, doing exactly what it is supposed to do. Normalize before you compare: express each segment's change per unit of length, then divide by that segment's own temperature difference to its surroundings, because loss to the space scales with both.

5. Measure flow and convert each segment's change into duty. Skipping this leaves you with a shape but no magnitude, so you can rank segments against each other and cannot tell whether the whole finding is worth a repair. For water, Btu/h equals 500 x gpm x the temperature change in F, where the 500 covers about 8.33 lb per gallon, 60 minutes per hour, and a specific heat near 1.0; use the fluid's own properties for a glycol mix. Measure flow at the equipment; a flow taken from a drawing propagates its own error into every segment equally and silently.

6. Record the surrounding air temperature and the pipe surface condition at each point, alongside the fluid reading. Skipping this leaves you knowing where without knowing why, which turns into a return visit. Loss to a 50 F attic and loss to a 70 F corridor are not comparable numbers, and a drop that survives normalization for both length and ambient is telling you the heat is not leaving through the surface at all.

7. Re-read the first and last points at the end of the walk. Skipping this costs you only the ability to detect drift, which is the smallest hole on this list but a real one. If the boundary points have moved since you started, the system did not hold still and step 3 was not actually satisfied.

Execution order: set up 1, 2 and 3, then walk the run once taking the fluid reading and the ambient together at each point, then close with 7 before you pack up. Steps 4 and 5 happen at the truck with the sheet in front of you.

The worked run

Domestic hot water recirculation loop in a small two-storey commercial building. Complaint: long wait for hot water at the far fixture, worse than it used to be. No draw during the survey, recirculation pump running continuously, measured recirculation flow 1.0 gpm. All readings on pipe surface under lifted, already-open insulation with the same infrared instrument, inside one 15 minute window.

Segment drops: 2, 3, 8, 13, 4, 1. They sum to 31 F, which matches 138 minus 107, so nothing was mis-transcribed.

Total loop loss: 500 x 1.0 x 31 = 15,500 Btu/h.

Raw reading of that column names the wrong thing twice. S4 has the biggest drop at 13 F, and S3 is second at 8 F, and the attic segment is the one everybody wants to blame because it is long and it is cold up there.

Normalize per step 4. Each segment gets its drop expressed per 100 ft, then divided by that segment's own mean fluid temperature above its surrounding air, expressed in hundreds of degrees.

Five of the six segments land between 14.3 and 17.8, averaging about 15.9. The attic run, the biggest genuine surface loss on the job, sits at 16.9, barely above that cluster, which says it is losing heat in proportion to its length and its exposure and is not a fault. S4 sits at 893, about 56 times the cluster average.

Nothing three feet long loses that through its surface. A segment that fails normalization by that margin is not losing heat to the room; it is receiving colder water. Converted to duty, S4 accounts for 500 x 1.0 x 13 = 6,500 Btu/h, which is 42 percent of the entire loop's loss inside 3 feet of pipe.

The tee at P3 to P4 serves a branch with a tempering valve. A failed check on that branch was letting cooler water cross into the loop. That is a mixing loss, and mixing losses are precisely what step 6 exists to expose, because they are invisible to any analysis that assumes all heat leaves through the pipe wall.

Verify by predicting, then re-measuring

Do not verify by asking whether it feels better. Predict a number first, then go get it.

Isolating the suspect branch should do two things. The 13 F drop across S4 should collapse to whatever three feet of pipe genuinely loses, which the cluster index puts under half a degree. And every downstream segment should then run hotter, so its own loss should rise in proportion to its new temperature above ambient: S5 was losing 4 F at 40 F above ambient and would now sit around 51 F above ambient, so expect roughly 5 F, and S6 correspondingly about 1.3 F. That predicts a return temperature near 118 F against the 107 F recorded.

Measured after isolating the branch, at the same 1.0 gpm re-verified at the pump: P4 at 124.7 F, and 118 F at the heater return. Total loop loss 500 x 1.0 x (138 - 118) = about 9,850 Btu/h, a reduction of roughly 36 percent against the 15,500 measured before, both figures computed at the same measured flow so the comparison is like for like.

A prediction that lands is worth far more on the invoice than a repair that helped. It tells the customer, and your own file, that the mechanism was understood rather than stumbled into.

Where a profile misleads

Three conditions break the assumption that a segment's change describes that segment.

Flow that is not the same in every segment. The method above assumes one stream passing through every point in series. On a run with active takeoffs, or a two-pipe system with intermediate returns, the flow past P4 is not the flow past P1, and duty per segment computed on a single flow figure is wrong by exactly the ratio of the two. Confirm the run is series before trusting duty numbers, or measure flow in each leg.

A recent draw or a recent cycle. Pipe and its insulation store heat, and a run that has just delivered a large draw is still equalizing for minutes afterward. The profile taken during that window shows a slope that is the pipe re-warming rather than the pipe losing. Wait until two consecutive boundary readings hold before you start.

A reading taken on a fitting, a valve body, or a hanger. These are thermal shortcuts and thermal masses. A valve body reads lower than the water inside it because the body itself is losing heat around its whole exposed surface, and a reading taken on one and compared against a reading taken on straight pipe manufactures a drop that is not in the fluid. Keep every reading on straight pipe, at least a few diameters clear of fittings and supports, and note the exception on the sheet if you cannot.

References

• 29 CFR 1910.1001 (asbestos, general industry) and 29 CFR 1926.1101 (asbestos in construction, covering maintenance and renovation that disturbs thermal system insulation)
• 29 CFR 1910.333(b)(2) for electrical work and 29 CFR 1910.147 for mechanical isolation and stored energy; NFPA 70E-2021, 120.5 for live-dead-live proving
• ASHRAE Handbook, Fundamentals volume, for pipe heat loss and insulation thickness relationships
• Insulation manufacturer published heat-loss tables for the specific pipe size, insulation type and thickness on the job
• See related: The Temperature You Measure and the Temperature That Matters; What a Heat Exchanger Is Actually Doing