Ph.D. Tezi Görüntüleme

Student: Vedat TOĞAN
Supervisor: Prof. Dr. Ayşe DALOĞLU
Department: İnşaat Mühendisliği
Institution: Graduate School of Natural and Applied Sciences
University: Karadeniz Technical University, Turkey
Level: Ph.D.
Acceptance Date: 25/9/2009
Number of Pages: 115
Registration Number: Di722

      With the improving technology today, there are many algorithms available to obtain the optimum solution among the designs that satisfy some predefined conditions for a certain problem. The common property of these algorithms is that the parameters of the optimization such as material, load, geometry etc., are considered as deterministic. However, kinds of randomness exist for most of the parameters used in engineering problems. That is, considering these parameters as random variables is more truthful.

The aim of this study is to obtain the minimum weight of the offshore structures by performing their optimizations under the randomness that is unavoidable for the parameters used in the optimization. A computer program is coded in FORTRAN for the purpose. Program includes the sequential quadratic programming and differential evolution as the optimization methods in addition to the first order reliability and inverse reliability methods. Three different types of offshore structures are optimized considering the component reliability under self-weight, maximum wave height and the total mass at the top. The uncertainties related with both the material and the loading are taken into account in the optimization process.

      It is concluded that when the parameters defined in the optimization are considered as random the weight of offshore structures are calculated heavier than those obtained with deterministic parameters. The performance measure approach based on the inverse reliability method is robust compare to the reliability index approach based on the first order reliability method in terms of both the convergence rate and the computation time. Although it needs more computation time differential evolution produces nearly the same results with the sequential quadratic programming and it can be employed.

Key Words: Random Variables, Reliability, Reliability Based Design Optimization, Offshore Structures, Differential Evolution, Sensitivity