Structural redesign with forced response constraints by large admissible perturbations.

LargE Admissible Perturbations (LEAP) is a general methodology which solves sizing redesign problems of complex structures without trial and error or repetitive finite element analyses. The corresponding code RESTRUCT (REdesign of STRUCTures) produces an optimal redesign of minimum structural change or minimum weight by changing structural properties. In previous work, algorithms with modal dynamic constraints and static displacement constraints have been implemented in the code and tested. The objective of this research is to include forced vibration constraints in the redesign process. Since non-structural terms, i.e. damping, dead weight, and fluid added mass, are present in most practical problems, they must be included in the redesign formulation, which leads to theoretical and numerical challenges. The existing LEAP algorithm, referred to as the Incremental Method, is modified in order to deal with non-structural terms. Also, a new algorithm, referred to as the Direct Method, is developed and implemented in this research. The Direct Method offers significant advantages in terms of computational effort and practicality. The selection of mode shapes, which is shown to have a significant impact on the efficiency and accuracy, is extensively studied in order to provide valuable guidelines to the designer. The problem of integrated redesign including modal dynamic, static, and forced vibration constraints at multiple frequencies of excitation, and multiple locations is studied in the context of large complex structures. Four structural applications are used in this work: a cantilevered beam and a cantilevered plate used to assess and validate the methodology, a complex three dimensional offshore tower, and an idealized ship hull structure.PhDApplied SciencesMechanical engineeringOcean engineeringUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/127244/2/3029303.pd