Quantitative Characterization of Thermographic Sequence Data

Interpretation of image data sequences from active thermography systems has typically been based on analysis of temporal variations in the contrast of each pixel, relative to a real (or synthesized) defect-free reference point on the sample. Various schemes for quantitative analysis of contrast data have been developed, in which times for events such as maximum contrast or the maximum slope of the ascending contrast are identified and empirically correlated to physical characteristics such as depth or thermal diffusivity. The Thermographic Signal Reconstruction (TSR) method offers an effective alternative to contrast-based methods for quantitative measurement, as it significantly extends limits of detectability, and enables automated pass/fail processing. In the TSR method, a noise reduced replica of the time history of each pixel is expressed in series form, so that an entire data set, normally comprising several hundred frames, comprises only the coefficients of the series expansion for each pixel. The resulting series offers a significant improvement is signal to noise performance. However, the series is readily differentiated so that features may be detected based on the detection of inflection points or other characteristics of the pixel time history. The TSR process can be applied to an entire time sequence in order to create a single “fingerprint” image that facilitates comparison of samples to a defect free reference. Introduction: The use of Pulsed Thermography for Nondestructive Inspection (NDI) has increased dramatically in the past few years. Applications range from maintenance of in-service aircraft to process control for the manufacture of large aerospace structures (1,2). For many applications, Pulsed Thermography compares favourably to conventional inspection technologies in terms of its sensitivity, speed, curvature tolerance and non-contact nature. Modern systems are capable of subsurface defect detection and materials characterization in both polymer and ceramic matrix composites, offering reliable measurement of sample thickness, defect depth and thermal diffusivity. Despite the apparent benefits of Thermographic NDI, the technology was not widely accepted until recently. Early attempts at implementation were only capable of identifying gross defects that were detectable using simple coin tap or visual inspection methods. The most frequently cited shortcoming of the technology was its subjective nature, i.e. interpretation of results depended heavily on the skill and experience of the inspector. The situation has improved dramatically with new approaches to signal processing, excitation and modeling, as well as improvements in IR camera technology and computer processing and data communication speed. In the most widely used implementation of Pulsed Thermography, the surface of a sample is heated with a brief, spatially uniform pulse of light, and an IR camera interfaced to a PC is used to monitor and analyze the time dependent response of the sample surface temperature to the thermal impulse. In areas of the sample surface directly above a buried defect, the transient flow of heat from the surface into the sample bulk is partially obstructed by the flaw, thus causing a transient, local temperature increase. For a semi-infinite, defect free sample, the time-dependent surface temperature response to an instantaneous heat pulse is given by t c Q T t T Surf Surf κρ = − ) 0 ( ) ( , (1 where Q is the input energy per unit area, κ the thermal conductivity, ρ the density and c the specific heat of the sample (3). Plotting the natural logarithm of both sides of Eq. 1 reveals a slope = -0.5 Back wall Deeper delamination Shallow delamination