Development of an Expert System to Characterize Weld Defects Identified by Ultrasonic Testing

Abstract

Welded structures are examined nondestructively, particularly for critical applications where weld failure can be catastrophic, such as in pressure vessels, load-bearing structural members, and power plants. Ultrasonic Testing (UT) is used in the examination of welds in thinner and thicker gauge materials where the size and location of the flaws are important to detect and interpret. Despite the advantages of the ultrasonic technique, the classification of defects based on ultrasonic signals is still frequently questioned, since the analysis and the identification of defect types depend exclusively on the experience and knowledge of the operator. The problem becomes more acute when high inspection rates, high probability of detection, and low number of false results are required. Thus, the correct classification of the type of flaw present in the material reduces measurement errors, increasing the confidence in the test and consequently the safety of the welded structure during service. In the present study, a new algorithm that allows for the detection and measurement of the length and type of weld defects is proposed. The system is based on a coupled dynamic and static patterns in an A-Scan and uses the defects cited in DIN EN 1713 standard as reference for evaluation. The proposed expert system has been evaluated and validated by examining several specimens containing various types of natural (non-artificial) defects identified in the mentioned standard. The results indicate that, the proposed algorithm has a clear potential in automatic defect detection and presents many advantages to the manual method for defect detection and characterization.