Charpy V notch tests – Risks associated with testing with 3 samples

Abstract

To estimate the risk of brittle fracture of a component in service, Charpy tests are generally carried out on 3 samples at the same temperature. Work done over many years on welded joint fatigue using probability and statistics has shown that to have a good assessment of minimum strength, it is clearly necessary to have a larger number of test results. In order to analyse the risks associated with this practice of 3 samples tested at the same temperature, a series of 17 Charpy impact test results of samples of the same steel plate was chosen. From this series, by translation, 2 series were generated, one with a test of 3 specimens showing acceptable steel and one with a test of 3 specimens showing unacceptable steel. The 3-test method was applied to these series leading to a possible simulation of 12 tests each. It was found that the 3-test practice leads to a risk of error of 30 to 40%. It is also found that the determination of the transition curve of a DH36 steel finally shows that its temperature for a CV of 27 J is of the order of −50 °C whereas the regulations require only −20 °C. In conclusion, a possible improvement of the test procedure is indicated, which is the subject of a study within the framework of the “CCRS rupture and fatigue” working groups of the Société des Ingénieurs Soudeurs (SIS).