Uncertainty Quantification In Radiative Power Sensors Using Monte Carlo And Latin Hypercube Sampling

This study introduces an experimental rig developed around a radiative heat flux sensor, designed to estimate the power emitted by a combustion flame over time. The system is built around a radiative heat flux sensor composed of a thermally responsive metallic disk, directly exposed to the flame’s radiation and instrumented with a surface-mounted thermocouple. The temporal evolution of the disk’s temperature, recorded during exposure, allows for the indirect estimation of key thermophysical properties of the source, such as radiative power density and its variation over time. This approach aims to provide a fast, minimally intrusive, and repeatable method for quantifying heat release in harsh thermal environments. To assess the reliability of the measurement chain and understand how input uncertainties propagate through the system, a probabilistic analysis is carried out using both Monte Carlo and Latin Hypercube Sampling (LHS) methods. These techniques were applied to account for variability in key parameters, such as disk’s thermal properties and the boundary conditions of the system. The comparison between Monte Carlo and LHS highlights not only the sensitivity of the system to each input, but also the advantages of stratified sampling in terms of computational efficiency and convergence behaviour. The integration of this experimental rig with a robust uncertainty quantification approach results in a practical and scalable framework for the thermal characterization of radiative sources. This can be particularly valuable in aerospace applications involving solid propulsion systems [1], where flame behaviour and thermal loads play a central role in both performance and structural design considerations.