A reliability based optimal design(RBOD) is cosidered more rational and ideal but needs more information on distribution of random variables, as compared to a safety factor based optimal design. In this Thesis, an approximate method of correlating a safety factor to a desired level of reliability is suggested. It is obtained from a relation between the advanced first order second moment method(AFOSM) and a partial factor of safety and then defining the central factor of safety.
Through several numerical examples, the results of the two approach are compared with each other along with deterministic optimal design results. Although it is difficult to quantify the correlation, the formulation by the simpler approach with the safety factor obtained seems meaningful when coefficients of variations are small. There seems, however, not much consistency in terms of conservativeness, and this requires further research before any practical use is possible.
