Modeling and identification of uniform corrosion damage on a thin plate using a Bayesian framework

Corrosion remains one of the primary mechanisms of structural degradation facing the civil and commercial infrastructure. In this study we consider the problem of estimating the degree of corrosion in a plate structure using only the structure׳s vibrational response to impact excitation. Specifically, a thin aluminum plate was twice placed in an accelerated corrosion environment resulting in varying levels of corrosion damage. At each damage level, the impulse response at three different locations was simultaneously measured with resistive strain gauges. From this data, a Ritz model with Timoshenko beam basis functions was used in a Bayesian framework to estimate the average plate thickness. The approach is first illustrated on a numerical model of the plate, and then applied to the experimental structural response. While the approach is able to accurately estimate the thickness of the “healthy” plate, the quality of the estimates degrades as the corrosion begins to alter the thickness in localized regions of the plate. This is reflected in the thickness estimates which show increasing variance with the degree of corrosion.