This is a Cooperative Agreement. The purpose of this project is to develop a modern reliability-based set of procedures for the evaluation, design and re-assessment of offshore structures. The dual coupled bases are: (1) state-of-art non-linear structural analysis, and (2) explicit structural reliability. The objective is to exploit the best tools available to the industry to insure safe yet economical facilities.
A large suite of real records (from one bin, i.e., very nearly the same magnitude and distance causative event) was run through a large number of different single and multiple degree of freedom models of frames, and found the median and standard deviation of response to these real (untouched) records. The records were 'compatibilized' (using two different professional, state of art/practice compatiblizing programs) to have virtually the same response spectrum: namely the median response spectrum of all these records. Then these 'artificial' records were run through all the same models and processed in the same way.
At interesting levels of ductility, (degree of nonlinearity) the artificial records were good in that they produced much less dispersion (a factor of about 4 less - implying 'cost'; savings of a factor of 16 for comparable accuracy in prediction of median response). But, they were unconservatively biased: the median predicted was about 20 percent too low, at ductilities of 4 or more for virtually every model looked at.