Reliability-based experiment planning with linear regression models

In many applications experimental data are fitted to linear regression models. In fatigue analyses of offshore steel platforms fitted linear models are used both in the SN-approach and in the fracture mechanics approach. In this paper the problem of how to make optimal planning of experiments to be used in the linear regression models is considered. Optimal planning for a certain class of problems can be performed using preposterior analyses from Bayesian decision theory. It is shown how cost optimal reliability-based experiment plans can be obtained using First Order Reliability Methods. An illustrative example is shown.