[CCoE Notice] Master's Thesis Defense

[Khator, Suresh]

NEW INELASTIC DISPLACEMENT RATIO OF SDOF STRUCTURES USING EXPERT SYSTEMS

Selma N. Ozkul
Tuesday, 11/29/2011 at 10:30 am, CEE Conference Room N-137

Committee Members: Ashraf Ayoub - Associate Professor, Civil and Environmental Engineering, Cumaraswamy Vipulanandan - Professor, Civil and Environmental Engineering, Bora Gencturk- Asst. Professor, Civil and Environmental Engineering, Abdusselam Altunkaynak - Associate Professor, Istanbul Technical University - Faculty of Civil Engineering

The existing approximate methods for estimating the inelastic displacement ratios of SDOF structures subjected to seismic excitation are built upon some assumptions, which ignore the effect of uncertainties on the concerning phenomenon.  Assumptions are generally good in a way to understand the problems at a simpler level.  However, vital imprecise parts of a fact may be lost with such simplifications.  Therefore, uncertainty techniques should be preferred while modeling such phenomenon that inherits impreciseness.  This research presents a new method predicting the inelastic displacement ratio of moderately degrading SDOF RC structures subjected to earthquake loading using fuzzy logic approaches and expert systems.
A well defined degrading model was used to conduct the dynamic analyses.  300 earthquake motions recorded on firm sites, including recent ones from Japan and New Zealand, with magnitudes greater than 5 and PGA values greater than 0.08g were selected.  These earthquake records were applied on five RC columns that were chosen among 255 tested columns based on their beam-column element parameters reported by the Pacific Earthquake Engineering Research Center (PEER).  A total of 96,000 dynamic analyses were conducted.  The results from these dynamic analyses were used to develop the fuzzy inelastic displacement ratio model inheriting uncertainties in terms of strength reduction factor (R) and period of vibration (T).  The performance evaluation of the new fuzzy logic method and four existing approximate methods were investigated using different input data sets.  As a result, more accurate results were predicted using the new method.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://Bug.EGR.UH.EDU/pipermail/engi-dist/attachments/20111122/1dc4f838/attachment.html