In this study, the capability of detecting the planar (crack-like) flaws in weld metal of carbon steel welded joints was studied by using the X-ray radiography and the ultrasonic testing.

UT indications (data analysis level was set at 5% DAC) were acquired by automated UT system and evaluated for the flaw length by using the conventional amplitude-based sizing technique (10% DAC & 20% DAC) and the flaw depth (in the thickness direction) by using the tip echo technique. Defect sizes estimated by UT were compared with the true defect sizes of the fracture surface on the specimens after fatigue test.

The results showed that some of the defects detected by UT did not agreed with by RT, and that the accuracy of UT defect sizing was -1. 0 mm mean error (2. 5 mm standard deviation) for the amplitude-based method measuring the defect length and +0. 6 mm mean error (0. 9 mm standard deviation) for tip echo method measuring the defect depth.