Reliability-Based Derivation of Partial Safety Factor for Structural Integrity Assessment
When there are uncertainties in the input random variables, or scatter in the material properties, probabilistic assessment is a useful tool for decision making in the field of safety analysis. The partial safety factor (PSF) method was aimed on ensuring that the failure probability did not exceed a target value. In order to be conservative the input value for each random variable during the assessment procedure should be multiplied by the partial safety factors. So it is essentially a deterministic assessment using conservative values of the input random variables and a relatively simple and independent method of assessing failure probabilities using R6 failure assessment diagram. The application of partial safety factors is an important breakthrough of assessment in structures containing defects. In recent years, sets of PSFs for load, defect size, fracture toughness and yield stress had been given in two standards, BS7910 and API579. However, the recommended PSFs in both standards were larger than the original PSFs in PD6493 which was replaced by BS7910. It is therefore a new method of calculating PSFs should be found to prove which is more appropriate and convenient for engineering application. In the case of the partial safety factor method target reliabilities in the range from 0.001 to 0.00001 were considered and new series of PSFs were derived from the results of reliability analysis for the linear elastic fracture mode and elastic-plastic fracture mode. After comparing with the PSFs in BS7910 and API 579, it is concluded that the partial safety factors were generally conservative compared to our research work.