Aerial Accuracy Logo

IEC 62446-3 Thermal Inspection Methodology

Complete guide to standards-aligned outdoor infrared thermography for photovoltaic systems. Anomaly classification, severity assessment, and professional deliverable formats.

Quick Reference: Severity Thresholds

S1 - Monitor
ΔTn < 10°C
~15% module output loss. Schedule for next maintenance window.
S2 - Plan Repair
10°C ≤ ΔTn < 20°C
~30% module output loss. Schedule repair within 30-90 days.
S3 - Dispatch Now
ΔTn ≥ 20°C or safety risk
~60% module output loss. Immediate action required.

Anomaly Classes

String Outages

Entire string appears uniformly cool compared to neighbors. Zero or near-zero output from affected string.

Causes: Blown fuse, tripped breaker, disconnected DC cable, inverter fault
Thermal signature: Uniform cool string (all modules same temp)
Loss: 100% of string output

Bypass Diode Issues

1/3-module hot bands or step-changes across substrings. Diode conducting due to cell shading or failure.

Causes: Shorted diode, partial shading, cell cracking
Thermal signature: 1/3 of module hot (bypass activated)
Loss: 33% (one diode) or 66% (two diodes) per module

Sub-Module Hotspots

Isolated hot cells, busbars, or ribbons at cell scale. Requires ≥5×5 px/cell sampling for reliable detection.

Causes: Cell micro-crack, PID, snail trails, delamination
Thermal signature: Single cell or busbar segment elevated
Loss: 15-60% depending on severity

Wiring/Polarity Heating

Hot home-runs, connectors, combiner inputs, or reversed strings creating resistive heating in DC circuits.

Causes: Loose connector, corroded terminal, reversed polarity, undersized wire
Thermal signature: Linear heat along cable runs, point heat at connectors
Loss: Variable; safety hazard priority

Temperature Normalization (ΔTn)

Raw temperature differences (ΔT) vary with irradiance levels. A 10°C difference at 500 W/m² represents a more severe defect than 10°C at 1000 W/m². Normalization removes this bias for consistent classification across different measurement conditions.

Normalization Formula

ΔTn = (Tdefect − Tref) × (1000 W/m² ÷ Gmeas)
Tdefect

Radiometric temperature of the anomaly (°C)

Tref

Reference temperature from healthy modules (°C)

Gmeas

Measured irradiance at time of capture (W/m²)

Reference Temperature Selection

Tref should be the median temperature of 6-10 adjacent healthy modules in the same row and orientation. This accounts for:

  • Local ambient temperature variations
  • Wind cooling effects across the array
  • Module age and soiling consistency
  • Tracker position and sun angle

Example Calculation

Measured values:

  • Tdefect = 68°C (hotspot on cell)
  • Tref = 52°C (median of adjacent modules)
  • Gmeas = 860 W/m² (from pyranometer)

Calculation:

ΔTn = (68 − 52) × (1000 ÷ 860) = 16 × 1.163 = 18.6°C

Classification: S2 (Plan Repair) — 10°C ≤ 18.6°C < 20°C

Measurement Requirements & False-Positive Controls

Reliable defect detection requires controlled measurement conditions. Log these parameters for every finding to ensure defensible classifications.

Irradiance (G)

Preferred:≥700 W/m²
Minimum:≥600 W/m²

Record actual value from calibrated pyranometer with sensor ID. Low irradiance reduces thermal contrast, masking defects.

Wind Speed

Recommended:≤5 m/s
Maximum:≤10 m/s

High wind convectively cools modules, reducing ΔT and masking hotspots. Record wind speed at time of capture.

View Angle

Required:≤30° from normal

Oblique angles reduce emissivity accuracy and cause reflections. Note module tilt/azimuth and sun angle for edge rows.

Image Quality

Focus: Sharp edges, no blur
No smear or ghosting artifacts
Adequate resolution (≥5×5 px/cell)

Reject frames with motion blur, focus issues, or lens condensation.

Repeatability Check

Re-observe suspected anomalies after 1-3 minutes to rule out transient effects (passing clouds, temporary shading, bird droppings). A repeatable thermal signature confirms the defect is real; transients disappear.

Capture → Evidence → Report Workflow

1Capture Set

Radiometric Thermal

R-JPEG or TIFF with full temperature matrix. EXIF/RTK GPS tags intact.

RGB Visual

Matching visible-light image for defect identification and context.

Flight Log

Per-photo GPS, yaw/pitch/roll, timestamp (UTC).

Ground Sensors

Pyranometer/irradiance + ambient temp + wind (attach raw CSV).

2Evidence Packet (Per Finding)

  • • Primary frame (radiometric) + RGB context image
  • • ΔT map with defect box and reference box annotated
  • • Computed ΔTn with Gmeas value
  • • Repeat frame if used for transient validation

3Dispatch-Ready Report Bundle

PDF Report

Human-readable with thumbnails, callouts, and findings summary.

CSV Data

Row-level records for CMMS import and analysis.

KML/KMZ

Pin-drop locations for truck tablets and field navigation.

GeoTIFF (Optional)

Georeferenced thermal orthomosaic for GIS integration.

CSV Schema (CMMS Drop-In)

Standardized CSV format for direct import into maintenance management systems. All timestamps in UTC, coordinates in WGS84 decimal degrees.

site_name, farm_id, array_id, tracker_id, row_id, string_id, module_pos,
lat, lon, elev_m, timestamp_utc,
anomaly_class, subtype, severity_bin,
T_defect_C, T_ref_C, deltaT_C, irradiance_Wm2, deltaT_normalized_C_1000,
wind_ms, ambient_C, tilt_deg, azimuth_deg, view_angle_deg,
image_thermal_path, image_rgb_path, evidence_zip, notes

Location Fields

  • module_pos: e.g., "S47-M2847" or X/Y within string
  • tracker_id: Tracker serial or logical ID
  • row_id: Row number within array block

Classification Fields

  • anomaly_class: string_outage, bypass_diode, hotspot, wiring
  • subtype: cell_hotspot, bypass_1/3, connector, home_run, string_down
  • severity_bin: S1, S2, S3

Measurement Fields

  • deltaT_normalized_C_1000: ΔTn at standard 1000 W/m²
  • irradiance_Wm2: Actual irradiance at capture
  • view_angle_deg: Camera angle from module normal

Evidence Fields

  • image_thermal_path: Path to radiometric original
  • evidence_zip: Folder with originals & charts
  • notes: Free-text repair hints

PDF Report Structure

Fast to read on-site. Designed for field crews and asset managers.

1. Summary Dashboard

  • • Total findings count by anomaly class
  • • Breakdown by severity (S1/S2/S3)
  • • Estimated MW at risk
  • • Inspection date, conditions, coverage area

2. Map Page

  • • KML snapshot with cluster pins
  • • Color-coded by severity
  • • Array block labels for navigation

3. Findings (One Per Page)

Title: Row R17 / String S47 / Module M2847
GPS: 34.523400, -101.777700
Time (UTC): 2025-03-15T14:32:00Z
ΔTn: 22.4°C (S3 - Dispatch)
G: 860 W/m² | Wind: 3.2 m/s
Thermal + RGB images with defect boxes; mini measurement table; repair hint.

KML/KMZ Styling (Crew-Friendly)

Styled for immediate field use on truck tablets and mobile devices.

S1
Small Yellow Pin
Monitor
S2
Orange Pin
Plan Repair
S3
Large Red Pin
Dispatch Now

Pop-up Fields

  • • Anomaly class and subtype
  • • ΔTn value and severity
  • • String/module IDs
  • • Thumbnail image
  • • "Open evidence folder" link

ID & Naming Conventions

Survey ID

SiteCode-YYMMDD-Run##

Example: HORNET-251223-R01

Finding ID

SiteCode-A####

Increment per finding. Mirror across PDF/CSV/KML.

Image Files

FindingID-THERM.jpg — Processed thermal
FindingID-RGB.jpg — Visual context
Keep radiometric originals (R-JPEG/TIFF) in evidence folder.

Field Math Quick Reference

Raw Temperature Difference

ΔT = T_defect − T_ref

Both values from radiometric image.

Normalized Temperature Difference

ΔTn = ΔT × (1000 / G)

G = measured irradiance in W/m².

Reference Temperature Selection

T_ref = median(T_1, T_2, ..., T_n) where n = 6-10

Select adjacent healthy modules in same row. Median reduces outlier influence.

Our Equipment & How We Meet These Requirements

For full camera specifications (NETD, GSD at altitude, calibration schedule) and a detailed mapping of how our Autel 640T platform meets each IEC 62446-3 requirement listed above, see our technical reference:

Standards-Compliant Solar Thermal Inspections

Our thermal inspection services follow IEC 62446-3 methodology with full deliverable packages: radiometric data, GPS coordinates, severity classification, and CMMS-ready exports.