Accuracy of predicting strip defects in fillet welds on type-B sleeves using S₁ curve

Objective Defects in fillet welds connecting type-B sleeves, particularly strip defects such as incomplete penetration and cracks within horizontal weld junction zones, present significant engineering challenges in phased array ultrasonic testing (PAUT). To effectively detect these defects, it is crucial to address practical issues in fillet weld testing, including the optimization of testing techniques and the improvement of both defect detection probabilities and accuracy in length measurements.

Methods A finite element model for fillet welds on B-type sleeves was developed, incorporating artificial notches to simulate strip defects in the welds. This model was used to acquire acoustic scattering signals from notches at various orientations and to compute scattering coefficient matrices. The results enabled the derivation of an S₁ curve, which reflects the relationship between scattered energy and notch orientations. Additionally, the influence of wedge angles and pipeline thickness on accuracy in notch length measurements was analyzed using the S₁ curve. Total focus simulation imaging and experiments were conducted to validate the accuracy prediction based on the S₁ curve.

Results The total focus method exhibited a 100％ detection probability of notches at all orientations within the horizontal weld junction zone. The combined application of a linear array probe and a 35° wedge yielded the optimal length measurements for notches ranging from 120° to 150°, achieving accuracy greater than 70％, with total focus images closely resembling the actual notch morphology. Conversely, for notches beyond this angle range, only 1 or 2 end corners were detected. The combined use of 30°, 35°, and 40° wedges in fillet weld testing expanded the range of notch angles suitable for testing by about 20°. Compared with the measurement results obtained via total focus imaging, the prediction results for accuracy in notch length measurements based on the S₁ curve showed errors not exceeding 14％.

Conclusion The S₁ curve approach provides a means to optimize defect detection parameters for fillet welds connecting B-type sleeves in practical engineering applications. Suitable wedge angles can be selected based on the rough orientations of defects, effectively improving detection accuracy.