Session: Poster & Student Poster Session

Paper Number: 171332

171332 - Differential Evolution-Based Estimation of Material Thermal Diffusivity From Pulsed Infrared Thermography 

Abstract: 

Aging metallic supporting structures of a nuclear reactor suffer from degradation mechanisms caused by high-temperature and high-radiation environments, such as creep and embrittlement. Nondestructive evaluation (NDE) enables assessment of structural materials without altering their properties. We investigate the strategy for estimation of material thermal diffusivity from nondestructive one-sided pulsed infrared thermography (PIT) measurements. Diffusivity is a function of material thermal conductivity, density, and heat capacity. PIT is a one-sided, non-contact NDE method that involves measuring the transient surface temperature response to a rapid thermal flash. As heat diffuses into the material bulk, the surface temperature decays toward a steady-state value. Surface temperature transients can be measured with a fast frame infrared camera. We hypothesize that local changes in thermal diffusivity could be associated with material degradation, which can be detected by analyzing temperature transients recorded from pulsed infrared thermography (PIT) measurements. There exists an analytic solution in the form of an infinite sum of power series for surface temperature transient in a plate. The analytic solution for transient temperature depends on thermal diffusivity and material thickness. By fitting this solution to experimental PIT data for samples with known thickness, we estimate the thermal diffusivity. We utilize a differential evolution (DE) optimization method to solve this inverse problem. DE is a continuous, single-objective optimization algorithm that iteratively minimizes fitting error. Thermal diffusivity can be estimated for every pixel in the PIT image, and analyzing the statistical distribution of thermal diffusivity can reveal information about material state. Preliminary study validates algorithm performance for computer-simulated PIT data and experimental measurements with metallic and ceramic specimens. Observed variations in thermal diffusivity indicate underlying material degradation, supporting the use of PIT combined with differential evolution as a promising approach for structural health monitoring in nuclear reactors.

Presenting Author: Hannah Havel Argonne National Laboratory and Northern Illinois University

Presenting Author Biography: Third-year undergraduate student at Northern Illinois University studying computer science and mathematics, with research interests in scientific computing.

Authors:

Hannah Havel Argonne National Laboratory and Northern Illinois University
Alex Heifetz Argonne National Laboratory