Computational Reduction of Specimen Noise to Enable Improved Thermography Characterization of Flaws in Graphite Polymer Composites

Flaw detection and characterization with thermographic techniques in graphite polymer composites are often limited by localized variations in the thermographic response. Variations in properties such as acceptable porosity, fiber volume content and surface polymer thickness result in variations in the thermal response that in general cause significant variations in the initial thermal response. These result in a "noise" floor that increases the difficulty of detecting and characterizing deeper flaws. A method is presented for computationally removing a significant amount of the "noise" from near surface porosity by diffusing the early time response, then subtracting it from subsequent responses. Simulations of the thermal response of a composite are utilized in defining the limitations of the technique. This method for reducing the data is shown to give considerable improvement characterizing both the size and depth of damage. Examples are shown for data acquired on specimens with fabricated delaminations and impact damage.