Model Updating Of Offshore Platform Structures Based On Modal Parameter Identification

Offshore platform structures always serve in the harsh marine environment. Theirsafety and durability are extremely important. Once the incident occured, not only causedenormous economic losses and casualties, but also caused environmental pollution,ecological damage and other adverse consequences. To protect the safety and to preventthe occurrence of major accidents, the safety assessment is necessary for the offshoreplatform structures during the service.An assessment model reflecting the actual structure is the basics for the safetyassessment of the offshore platform structure. But disagreements often exist between themodel and actual structure. Model updating is a common method to improve it. However,the traditional model updating methods more or less have some defects, which limit theirapplication in the large offshore platform structures. In contrast, Cross Model Cross Mode(CMCM) method is more promising to applicate in it because of its advantages.Because of the complexity of offshore platform structures, there are manyuncertainties in modeling. However, measured data is very limited and containing noise,which limits the practical application of the model updating methods. Problems to besolved contains:1) the selection of updating parameters,2) ill-conditioned problems ofmodel updating,3) model updating of boundaries, etc. While CMCM method is alsofaced with the ablve problems in practical application.Considering the characters of offshore platform structures, the difficult problems ofCMCM method faced in the practical applicating are researched based on those existingin the literature. Major contend in this thesis are summarized as follows:1. The selection of updating parameters is researched from the angle of sensitivityanalysis. A multi-steps model updating method based on sensitivity grading is proposedin this thesis. In the method, parameters can be properly selected by sensitivity analysisand grading, which can also improve the ill-condition of model updating equations. Basedon the sensitivity grading results, the selected paramenters can be updated step by step,which can improve the solving accuracy of the low sensitivity parameters. A numericalexample of a jacket structure verified the feasibility of the method.2. Model updating equations of large structures always presents ill-condition. Smallerrors in the measured data will cause oscillation of the solutions. In this thesis, thesolution of ill-conditioned problems with noisy data is researched. Regularization methods commonly used in mathematics are introduced to CMCM method. Theapplicability is verified through a numerical example, and then the noise robustness ofCMCM method is discussed based on the application conditions. Results indicate thatproper regularization can solve the ill-conditioned model updating problems, but need tojudge the solvability conditions: discrete Picard conditions. In addition, for the sameupdating object, less CMCM equations will have better robustness.3. This thesis extends the CMCM method to update the boundary conditions ofconstraint and piles. Because of the lack of modal shapes near the boundary, onlyfrequencies are used in the updating procedure, which can avoid the influence of shapeerrors of the upper nodes. And also, multi-steps mehod is used to update the wholestructre (including the boundary and the superstructure), which can avoid theill-conditioned problem. Numerical and experimental examples verified the feasibility ofthe method.4. Experimental study of a steel jacket scale model is conducted. Model updatingresults verified the practicability of the methods proposed in the thesis.To some extent, this thesis solved the difficult problems of model updating faced inthe practical applicating and advanced the CMCM method form the theory level to theenginering applicating level. The multi-steps updating method based on sensitivitygrading presented in this thesis is also applicable to other method and has greatersignificance.