This paper presents a methodology to quantify systematic error bounds in THEODOR-based heat flux inversion due to tile property uncertainty, using a first-order perturbation analysis, interval analysis, Bayesian inference, and validation with synthetic data and high-fidelity simulations.
Key findings
Develops a first-order perturbation analysis framework for uncertainty propagation.
Deterministic error bounds are derived using interval analysis for bounded parameter uncertainties.
A Bayesian inference approach is formulated incorporating prior distributions on tile properties.
Validation strategy includes synthetic data benchmarks and high-fidelity COMSOL Multiphysics simulations.
Limitations & open questions
The methodology's applicability is limited to scenarios where tile thermophysical properties are uncertain.
Assumptions made in the Bayesian framework may not hold in all experimental conditions.