3.2 Heat Transfer Modes and Conduction Variables
Key Takeaways
- The three modes are conduction (Fourier: q = kA·ΔT/L), convection (Newton: q = hA·ΔT), and radiation (Stefan-Boltzmann: q = εσA·T⁴) — only radiation is what the camera actually detects
- Thermal conductivity spans four orders of magnitude from copper at about 400 W/(m·K) to fibreglass insulation at about 0.04, which is why insulation defects and metal hot spots look completely different
- Convection coefficients for natural air convection run roughly 2–25 W/(m²·K) and forced air 25–250, so wind can multiply surface heat loss by an order of magnitude
- Steady state means temperatures have stopped changing; transient conditions after load steps, startup, or weather changes invalidate comparison against any ΔT criteria table
- Heat always flows from hotter to colder, so the hottest point in a thermal pattern is normally at or nearest the source — a pattern whose peak is displaced points to conduction paths or airflow, not to the fault location
The ITC Level II outline lists nine separate heat-transfer objectives — more than any other topic in the syllabus. That weighting is deliberate. Nearly every misdiagnosis in the field comes from attributing a thermal pattern to the wrong transfer mechanism.
The Three Modes
| Mode | Governing relationship | Requires | What it looks like |
|---|---|---|---|
| Conduction | Fourier: q = k·A·ΔT / L | Physical contact, a medium | Smooth gradients along solid paths; hot spot smears toward good conductors |
| Convection | Newton: q = h·A·(T_s − T_∞) | A moving fluid (air, oil, water) | Plumes rising above hot components; washed-out surfaces in wind |
| Radiation | Stefan-Boltzmann: q = ε·σ·A·T⁴ | Nothing — works in vacuum | Reflections, sky effects, and the signal your camera actually measures |
where σ = 5.67 × 10⁻⁸ W/(m²·K⁴).
The point that matters most: the imager detects radiation only. Conduction and convection are inferred from the pattern radiation produces. Every diagnostic statement about a conduction path or an air leak is an interpretation, not a measurement.
Direction of Heat Flow
The second law fixes the direction: heat moves spontaneously from higher to lower temperature, never the reverse without work input. Three field consequences:
- The source is the hottest point. In a bolted joint the peak temperature sits at the resistance, and temperature falls with distance along the conductor. If the hottest point on a lug is offset from the bolt, look for a different resistance — a strand break, a crimp, or a corroded washer.
- A "cold" anomaly is still a heat-flow story. A cool spot on an operating machine can mean blocked flow, a closed valve, a plugged tube, or refrigerant where it should not be.
- Load paths are heat paths. Current follows the metal; so does conducted heat. A hot spot appearing several centimetres from the actual defect usually means the defect is buried where you cannot see it.
Conduction Variables
Fourier's law identifies exactly what you can change: conductivity (k), area (A), temperature difference (ΔT), and path length (L).
| Material | k, W/(m·K) | Note |
|---|---|---|
| Copper | ~400 | Spreads heat so effectively that a hot spot may appear diffuse |
| Aluminium | ~237 | Common bus material; similar spreading behaviour |
| Carbon steel | ~50 | Roughly one-eighth of copper |
| Stainless steel | ~16 | Poor conductor for a metal — enclosure panels lag badly |
| Glass | ~1.0 | Windows are conduction bridges in envelopes |
| Water | ~0.6 | Fifteen times better than air — wet insulation loses R-value |
| Wood | ~0.12 | The classic framing thermal bridge |
| Fibreglass batt | ~0.04 | Insulation works by trapping air |
| Still air | ~0.026 | The actual insulator in every insulation product |
Two Level II inferences follow directly:
- Wet insulation fails because water conducts. Replacing trapped air (k ≈ 0.026) with water (k ≈ 0.6) raises conductivity by roughly twenty times, which is why moisture shows as a thermal defect in Chapter 11 long before it shows as a leak.
- Metals hide small faults. High conductivity spreads a localised defect over a large area, lowering the peak ΔT. A modest ΔT on heavy copper can represent more dissipated power than the same ΔT on stainless.
Thermal bridging
Where a high-k element penetrates a low-k assembly — a steel stud through insulation, a concrete slab edge, a metal window frame — heat short-circuits the insulation. The signature is a linear, regularly spaced pattern that follows construction geometry. Regularity is the diagnostic: framing repeats on a module; moisture and air leakage do not.
Convection and the Effect of Wind
q = h·A·(T_s − T_∞), where h is the convection coefficient — not a material property but a function of fluid, geometry, and velocity.
| Condition | Typical h, W/(m²·K) |
|---|---|
| Natural convection, air | ~2–25 |
| Forced convection, air | ~25–250 |
| Natural convection, water | ~50–1000 |
Wind is the single largest environmental threat to outdoor electrical thermography. Moving air raises h by an order of magnitude, stripping heat from a hot connection and collapsing the measured ΔT. A genuine Priority 1 finding at still air can read as a mild anomaly at 8 m/s.
Practice rules that follow:
- Infraspection's standard requires inspections to be performed when wind conditions are favourable, and requires wind speed and direction to be recorded with every exception where significant.
- Do not invent a correction factor. Correction tables circulate widely and disagree with one another; if your program uses one, it must be documented in the written procedure and cited in the report.
- Where wind cannot be avoided, treat findings as potentially understated and re-inspect. A large ΔT measured in wind is real and probably worse; a small ΔT measured in wind proves nothing.
Indoors, convection produces plumes — warm air rising above a heat source, warming surfaces above it. A warm band above a breaker may be convective carry-over rather than a second defect.
Steady State versus Transient
| Steady state | Transient | |
|---|---|---|
| Definition | Temperatures no longer changing with time | Temperatures still rising or falling |
| When | Stable load, stable ambient, sufficient elapsed time | After load steps, startup, shutdown, weather change, sunrise/sunset |
| Criteria valid? | Yes | No — criteria tables assume equilibrium |
Infraspection's standard makes this the end user's duty: the qualified assistant must "allow sufficient time for recently-energized equipment to produce stable thermal patterns." The heavier the assembly, the longer that takes. A small fuse may stabilise in minutes; a large oil-filled transformer takes hours.
A transient survey is not worthless — it is just not gradeable. Document that conditions were transient and re-inspect.
Reading the Pattern: A Diagnostic Sequence
- Where is the peak? At a connection, a whole component, or spread across a region?
- How does it decay? Sharp local peak suggests a resistive point defect; broad smooth gradient suggests conduction from a buried source or bulk overload.
- Is it geometric? Regular, repeating, straight lines suggest construction (thermal bridging), not a fault.
- Does it move with air? Plumes and streaks that follow airflow are convective.
- Does it change when you change viewing angle? If so, it is a reflection, not a thermal feature — the radiation mode fooling you.
- Is it cooler than surroundings and amorphous? Consider evaporation or flow blockage.
Worked Examples
Example 1 — Conduction versus contact resistance. A 400 A aluminium bus run shows a 6 C° rise over 1.5 m with no localised peak. That is bulk resistive heating consistent with loading, not a joint defect. Compare against load and against the sister phase before reporting.
Example 2 — Wind suppression. An outdoor cutout reads 12 C° above its sister phase at 9 m/s wind. Forced convection at that speed can easily halve the apparent ΔT. Treat 12 C° as a floor, not a value, and re-inspect in calm conditions.
Example 3 — Transient trap. A chiller compressor started 6 minutes earlier shows discharge line at 38 °C. The system has not reached steady state; comparison to a baseline recorded at 40 minutes of run time is invalid.
Example 4 — Bridging versus moisture. An interior wall shows cool vertical stripes at 400 mm centres. The regular spacing identifies framing thermal bridges, not wet insulation, which would appear amorphous and irregular.
Quick Answer: Conduction follows q = kA·ΔT/L, convection q = hA·ΔT, radiation q = εσA·T⁴ — and only radiation reaches the camera. Conductivity runs from copper (~400) to still air (~0.026 W/(m·K)); water at ~0.6 is why wet insulation fails. Forced-air convection coefficients (~25–250) dwarf natural convection (~2–25), so wind suppresses outdoor ΔT and small readings in wind prove nothing. Criteria tables assume steady state; transient surveys must be documented and repeated.
Which heat transfer mode does an infrared imager actually detect?
An outdoor bus connection is measured at 12 C° above its sister phase while wind is blowing at roughly 9 m/s. How should the Level II thermographer treat the result?
An interior wall survey shows cool vertical stripes at regular 400 mm spacing during the heating season. What is the most likely cause?
Why can a ΔT criteria table not be applied to equipment energised six minutes earlier?