ITC Level II common mistakes
ITC Level II · Understand the traps and how to avoid them.
Adjusting emissivity up or down until the displayed temperature matches what the candidate expects the surface to be, then reporting that number as the measured temperature.
Candidates have seen a 'known' answer (e.g. a fuse should be near ambient) and use emissivity as a tuning knob instead of a physical property; the camera's live display makes this feel like calibration rather than fabrication.
What to do instead
Set emissivity from the material/surface condition (or from a taped/ painted reference of known emissivity applied to the target), measure or estimate reflected apparent temperature independently, and report the value with its uncertainty rather than forcing agreement with an expectation. If surface condition is unknown, change it (tape, paint) instead of changing emissivity.
Applying steady-state reasoning and radiation formulae to a target that is still warming up or cooling down, and inserting Celsius values into Stefan-Boltzmann or radiation-exchange calculations.
Field measurements are usually taken as soon as the camera is pointed, and °C is the convenient display unit; candidates forget that thermal time constants can be minutes and that radiation scales with absolute temperature to the fourth power.
What to do instead
State and verify the thermal steady state (repeat the measurement after an interval, or check the time series) before applying steady-state criteria, and always convert to kelvin before any radiation calculation. Sanity-check the magnitude of the radiative term against convection and conduction for the geometry.
Leaving the camera on auto-adjust or on the widest available range for every shot, so a 3-5 °C temperature difference is displayed in a handful of counts or the target clips at the top of the scale.
Auto mode always produces a visually pleasing image, and a wide span feels 'safer' because nothing is missing; candidates do not check the level/span readout or the saturation indicator.
What to do instead
Set level and span manually around the temperatures of interest, keep the target inside range without clipping, and confirm the span is narrow enough that the expected delta-T occupies a useful fraction of the scale. Check the over-range/saturation indication before capturing.
Using the spotmeter or a small measurement box on a component that subtends only a pixel or two, and treating the resulting value as the temperature of that component.
The spot marker on screen looks like it measures exactly what is inside it, so candidates ignore detector geometry, lens focal length and distance and never compute the spot-size ratio.
What to do instead
Compute the measurement field size from the detector resolution, lens and distance (or use the camera's built-in spot-size calculator), and get close enough or use a telephoto lens so the target fills at least a 3x3 pixel area; otherwise report the reading as an area average, not a component temperature.
Shooting with the reflected apparent temperature left at the default and ignoring distance/atmospheric effects when working outdoors, near furnaces, or through sight glasses.
Reflected temperature is invisible in the image and defaults look plausible; low-emissivity targets such as shiny metal or sky-facing surfaces return a strong reflected component that silently offsets the reading.
What to do instead
Determine reflected apparent temperature with the reflector (crumpled foil) method or a known reference at the same geometry, enter it, and correct for distance, humidity and transmission when the path is long or obstructed. Record these settings in the report so the reading is reproducible.
Capturing before the camera has thermally stabilized, without critical focus, and without any independent check that the reading is plausible.
Modern cameras produce an image within seconds, so warm-up drift and slight defocus are invisible on the LCD, and small targets are especially sensitive to focus because blur spreads energy.
What to do instead
Allow the stated warm-up time, focus critically on the target plane (use the sharpest edge, not the display centre), and cross-check at least one reading per survey with a contact probe or a reference emitter before trusting absolute values.
Comparing two thermal images captured with different level/span, palettes or distance, and reading the palette colour as if it were an absolute temperature (e.g. 'red is hot, so this is worse').
Palettes are display mappings, not data; the same colour means different temperatures on different images, and rainbow palettes exaggerate small differences at the edges of the scale.
What to do instead
Compare isotherms and numeric spot/line profiles, not colours, and re-display both images with identical level/span and palette before comparison. Quote measured values and deltas from the same analysis settings rather than describing colours.
Assigning a severity to an electrical or mechanical finding from one thermal snapshot without recording load, without comparing like components, and without a prior baseline for the same asset.
Temperature rise is only meaningful relative to load and to a reference; a warm connection at high current may be normal, and a bearing that is 15 °C above ambient may be its normal running condition.
What to do instead
Record load current (or mechanical duty/speed) at the time of the survey, compare phase-to-phase and to similar components under similar load, apply the applicable delta-T or absolute-temperature criteria, and use the asset's own baseline or a repeat measurement to confirm a trend before classifying severity.
Writing a report that gives temperatures and a severity label but omits emissivity, reflected apparent temperature, distance, ambient conditions, load and the standard or criterion the severity was based on.
The measurements are treated as the deliverable, and severity categories are applied from memory or from a summary table without citing the source or the assumptions behind the thresholds.
What to do instead
Use a fixed report template that captures every measurement parameter alongside each image, cite the specific standard or procedure and the criterion used (e.g. delta-T over ambient or over a reference), state measurement uncertainty and limitations, and have a second person check that each severity call matches the cited criterion.
Running surveys as one-off image captures: no fixed routes, no consistent camera settings between visits, and no stored baseline images, so repeat surveys cannot be compared or trended.
Thermography is perceived as an inspection activity rather than a program, and images are saved with inconsistent naming, geometry and settings, which makes later comparison unreliable.
What to do instead
Define routes with fixed survey points, standardize camera settings and geometry per point, store baseline and subsequent images with a consistent naming convention, and trend delta-T over time. Document the program's written procedure, acceptance criteria and technician qualification requirements so results are repeatable across personnel.