CRACK DETECTION BY DIFFERENTIAL LASER THERMOGRAPHY

Abstract

Crack formation can be detrimental to the integrity of structures. Cracks can be difficult to detect, especially sub-surface cracks. This thesis creates a framework using laser point thermography, curve fitting, and multivariable regression to determine the location and characteristics of vertical cracks. It utilizes finite element modeling to create a two-dimensional cross section of steel and model fifty-two surface and internal cracks. It then fits a curve to the modified data for each crack. Multivariable regression determines a matrix that relates the curve fitting coefficients to the crack characteristics. This matrix is used to determine the characteristics of four cracks given the thermal data. The result is a very high accuracy in determining the location of a crack and less accuracy in determining the length and depth of the crack. This thesis shows potential for continued work using this framework for crack detection after experimentation and extruding it to three-dimensions.