Multi-spectral infrared thermography using finite element emissivity modelling for high-temperature measurements on metallic surfaces

Abstract

Accurate high-temperature radiation thermometry of opaque surfaces (from 800 to $1200K$) is critical for controlling industrial processes and ensuring product quality. However, its accuracy is limited by inadequate knowledge of the material’s optical properties and their spectral and thermal dependencies. Multi-spectral thermography addresses this challenge by identifying the surface temperature while considering its emissivity spectra. Here, we propose a method combining infrared radiometric measurements with deterministic inverse modelling, where emissivity is interpolated using a P1-1D finite element polynomial basis to attenuate the systematic and stochastic components of the identification error. Monte Carlo simulations and experimental validations using an induction heater and a multi-spectral infrared camera calibrated at four wavelengths demonstrate transient-state temperature uncertainties below 5%. Numerical hyperspectral assessments with 20 wavelengths further demonstrate significant bias reduction with increased spectral resolution. These findings establish an advancement towards reliable, dynamic, non-contact temperature measurement in industrial metallurgical environments.