Switchgear Survey Through an IR Window: Apparent vs. True Temperature
An electrician asks you to survey energized 480 V switchgear through installed IR-transparent windows (stated transmission 0.80, spectral range 8–14 µm). The imager is a 320×240 microbolometer, 8–14 µm, with a 24° lens and a stated accuracy of ±2 °C. Through the window at 2 m standoff, a busbar connection reads 78 °C; the same connection read 61 °C on last quarter's survey taken with the panel open (direct line of sight, e set to 0.95). Ambient is 24 °C, load is stated to be the same. The window manufacturer's data sheet gives transmission 0.80 but does not state a reflection figure. The imager's transmission setting is left at 1.00.
1. Explain the discrepancy between 78 °C and 61 °C in terms of the measurement chain.
Read the worked answer
Two effects act through the window:
- Transmission: the imager is set to τ=1.00 but the window transmits ~0.80. The instrument assumes all radiance comes from the target; in reality 20% is lost (absorbed/scattered) and some is replaced by window emission and reflection. The net effect on a target hotter than the window is to under-report target temperature - the opposite of what we see.
- Reflection/emission from the window: the window reflects the surroundings (and the imager's own optics/operator) and emits at its own temperature. If the window is near ambient (24 °C) and the reflected background is warm, this adds radiance the imager attributes to the target, biasing the reading high.
The observed +17 °C bias is larger than either effect alone would usually produce at these temperatures, which points to a reflected warm background (e.g., the imager and operator, or warm equipment behind) dominating, plus the uncorrected τ. The 61 °C direct reading is the better reference but is itself subject to e assumptions.
2. What corrections and field checks would you make before reporting a temperature, and what would you report?
Read the worked answer
Corrections and checks:
- Set τ=0.80 in the imager (and reflected temperature to a measured value, not default).
- Measure reflected apparent temperature using the reflector method at the window plane, or at minimum use a crumpled aluminum foil reference; do not leave the default 20 °C.
- Confirm the window's spectral band matches the imager (8–14 µm stated - verify; some windows are quartz or have coatings with poor LWIR transmission).
- Check for a warm reflector in the window - move the imager off-axis and see if the reading changes; a strong dependence indicates reflection.
- Compare to the prior direct survey and, if possible, take a direct reading with the panel open under controlled conditions (appropriate PPE and arc-flash boundaries).
What to report: a corrected temperature with an explicit uncertainty statement, not a bare number. Given ±2 °C instrument accuracy, uncorrected reflection, and window transmission uncertainty, the honest report is a relative/qualitative severity call (e.g., 'connection is significantly hotter than siblings, Class 2 per NETA/NFPA 70B guidance') plus a corrected estimate with stated uncertainty. Do not report 78 °C as a true temperature; the direct 61 °C is the better anchor until the window corrections are applied and verified.