Research On Parallel Computing Algorithm For Long Span Self-Anchored Suspension Bridge Structure And The Behavior Of Main Beam Under Dead Load

In the modern bridge designing, more and more complex bridge structures with long span are choosen to be preferred plans, leading to an increase in computational analysis time. Computational performance in engineering analysis has been greatly improved with the development of computer hardware, but it still cannot meet the needs of designers. In This paper, Focusing on Long span self-anchored suspension bridge structure and the behavior of main beam under dead load, research is mainly carried out on the designing and optimization of bridge structure finite element program, the main methods of which are node numbering optimization algorithms and OpenMP parallel computing. Then comprehensive analyses of the state of self-anchored suspension bridge main beam under dead load are done with self-developed program. Main contents are as follows:Firstly, facing the fact that there is a long processing time in analysis with existing bridge finite element programs, methods and development status of improving computing speed of FEA program are discussed in terms of program development, and then researching methods are determined.Next, a program is developed and related theories are discussed, which mainly focuses on self-anchored bridge structural finite element program development and its theoretical bases, loads involved in the analysis program such as weight, shrinkage and creep of concrete, prestressing effect, temperature effect and sling force.Then, focusing on the calculation speed optimization of program, the research is mainly on bandwidth optimization algorithms of overall stiffness matrix and parallel programming.In bandwidth optimization algorithms research, artificial and computer automatically numbered methods are discussed for reducing bandwidth of the overall stiffness matrix. Three node numbering algorithms, AD, CM and GPS(name of algorithm) algorithm, are discussed and related programs are developed. Then five bridge structures are taken as examples to verify the correctness of the optimization programs and to do some comparison in efficiency, in which GPS algorithm is found superior to the others when used in general beam structure programs.In parallel programming research, features of the parallel programming library OpenMP, which is based on the shared-memory model, are introduced. Pointcut and methods of parallelization in Bridge finite element program are studied and put into practice. Parallelized and serial program are both used in sample structure analyzing, and computing efficiency is compared with these results.Finally, the current self-developed bridge structure finite element program is applied to analyse the behavior of main beam of the Mudanjiang Eleventh Road Birdge under dead load. The main contents are:time-dependent effects analysis, additional prestress loss analysis of main beam, analysis of influence parameters of main beam, long-term time-dependent effects analysis of lane load and temperature load.