ACOUSTIC EMISSION & NDT SCIENTIFIC PUBLICATIONS

 

Pressure Vessel Evaluation With Pattern Recognition Acoustic Emission Data Analysis

 

Proceedings of the 25^th European Conference on Acoustic Emission Testing - EWGAE 2002, September 11 – 13, 2002, Prague, Czech Republic, Editor: P. Mazal, ISBN 80-214-2174-6, Volume I, pp. 29-36

A. A. Anastassopoulos, A. N. Tsimogiannis, D. A. Kouroussis

 

Acoustic Emission (AE) has been successfully applied for the structural integrity assessment of metallic pressure vessels, during both in-service tests and hydraulic testing. The extensive testing of such vessels has led to the development of AE proof testing procedures, evaluation criteria and international standards. In addition to that industry proved procedures such as  MONPAC, extended the pass/fail evaluation of the codes to quantitative evaluation of fault severity and criticality and have provided the industry with a tool for 100% evaluation of the vessel, capable of giving early warning of developing defects, increasing, thus, the operational safety.

It might often be the case in such AE tests that complex AE signatures are present, i.e. multiple AE sources emitting simultaneously, such as propagating flaws, external mechanical noise (wind gusts, impacts, friction, nearby works etc.), turbulent noise from the filling point of the vessel, EMI, etc. In such cases, noise-related sources have to be effectively discriminated and filtered out because they might lead to non-relevant indications. Consequently, the analysis of the AE data is not always straightforward. The ever-increasing demand for an analysis tool for the characterization and understanding of the recorded AE sources has led to the development of methodologies for the mathematical classification of the corresponding AE data. In this context, Unsupervised Pattern Recognition (UPR) has recently been applied to AE data obtained during testing of pressure vessels such as spheres and bullets. The present paper outlines the basic features of Unsupervised Pattern Recognition for AE data and reports on some successful application examples of the technique, with the use of specialized Pattern Recognition software. It is shown that, upon proper selection of the UPR algorithms and parameters, the technique can successfully identify and separate noise-related AE (EMI, friction, mechanical impacts, hydraulic noise, leak) from legitimate AE and assist in signal characterization. Furthermore, Supervised Pattern Recognitions algorithms can be trained to automatically segregate the data of any similar AE test. It is concluded that further work will lead to the automation of the analysis procedure and to more effective and confident assessment of the structural integrity of the tested vessels.