Differentiation of leakage and corrosion signals in acoustic emission testing of aboveground storage tank floors with artificial neural networks

The acoustic emission method is a well-recognized procedure of floor inspection for aboveground storage tanks as well as a preventive tool for the improvement of maintenance service time. The high importance of this method is better realized when it is used in sensitive structures while they are in operation. A laboratory-based smaller-scale tank was made and all operating conditions were simulated. This provided a databank for recognition of acoustic waves. More experiments were conducted to simulate corrosion in a tank. Data were then collected from acoustic emission measurements for several aboveground storage tanks in a refinery. The data gathered from laboratory tests and the real data were combined in order to differentiate between leakage and corrosion signals. An artificial neural network (ANN) system was used to categorize the tested aboveground storage tanks.