Fatigue limit estimates based on stair-case tests

摘要

The Bayesian approach for analysing staircase test result is introduced and compared to the traditional Dixon-Mood method. The Bayesian approach takes advantage of the engineer's prior knowledge about the product in question and combines this knowledge with the experimental results. By simulations we then can get some answers to the following interesting questions: How much prior knowledge is needed to get better estimates by this advanced method than by the simple traditional method? How much is gained by increasing the number of specimens in the test? How much prior knowledge is needed to completely avoid expensive tests? A great number of Monte Carlo simulations of stair-case tests are also performed in order to derive a useful expression for correction of variance estimates from Dixon-Mood analysis of tests.