Abstract:The ability to detect and classify damages in complex materials and structures is an important problem from both safety and economical perspectives. This paper develops a novel approach based on geostatistical analysis for the corrosion prediction. Our approach is based on using geostatistics for modeling oil tank thickness from nondestructive data. First of all the semivariogram was obtained from the scatter points set. Then computing the ordinary kriging predictions at unmonitored points. Finally, simulating the thickness along the surface of the tank.