本文選題：結(jié)構(gòu)動力分析 + 非線性�。� 參考：《中國地震局工程力學(xué)研究所》2012年博士論文

【摘要】：隨著科學(xué)技術(shù)的快速進(jìn)步和經(jīng)濟(jì)條件的迅速改善，世界各地對大型結(jié)構(gòu)的需求越來越多。對這些大型結(jié)構(gòu)進(jìn)行地震反應(yīng)分析時(shí)，存在兩個科學(xué)技術(shù)上的挑戰(zhàn)：（1）結(jié)構(gòu)有限元模型的單元數(shù)和自由度數(shù)巨大；（2）該類結(jié)構(gòu)受高階振型影響較大，需要建立高階的求解方程。并行計(jì)算技術(shù)的出現(xiàn)，為順利的求解該類型結(jié)構(gòu)，確保計(jì)算精度，提供了很好的方法和手段。本研究利用并行計(jì)算技術(shù)，研究了大型結(jié)構(gòu)的地震反應(yīng)分析方法。研究內(nèi)容主要包括以下幾個方面： 1．深入研究了區(qū)域剖分方法的國內(nèi)外研究現(xiàn)狀，對部分方法編制了C/C++程序并進(jìn)行區(qū)域剖分效果的對比，分析了各個方法的優(yōu)缺點(diǎn)及適用范圍。 2．對各種區(qū)域分解算法進(jìn)行了深入的研究，分析總結(jié)了各種常用算法的優(yōu)缺點(diǎn)。 3．專門對子結(jié)構(gòu)分析方法進(jìn)行了研究，編制了相應(yīng)的C/C++程序，并利用ANSYS作前處理器，得到相應(yīng)的結(jié)構(gòu)剛度矩陣和荷載矩陣后，對二個單元和四個單元的簡單懸臂梁進(jìn)行了試驗(yàn)性并行計(jì)算算法的研究，此方法的詳細(xì)研究為后續(xù)方法的研究奠定了基礎(chǔ)。 4．詳細(xì)研究了BDD（Balancing Domain Decomposition）方法，BDD方法是一種子結(jié)構(gòu)方法，其主要是把結(jié)構(gòu)分成若干個子區(qū)域，每個子區(qū)域分成區(qū)域與區(qū)域間的邊界節(jié)點(diǎn)和剩余的內(nèi)部節(jié)點(diǎn)，每個子區(qū)域首先進(jìn)行自由度靜力凝聚，形成界面節(jié)點(diǎn)自由度方程。然后應(yīng)用BDD預(yù)處理子的預(yù)處理共軛梯度法求解各個界面方程，在求解過程中要交換界面之間的信息，最終得到各個節(jié)點(diǎn)的位移和應(yīng)力情況。 5．基于區(qū)域分解算法對結(jié)構(gòu)的靜力和動力問題進(jìn)行了并行計(jì)算方法研究，在進(jìn)行大型結(jié)構(gòu)的動力并行計(jì)算分析時(shí)，利用Newmark-β法對時(shí)間離散化積分。對于動力分析，基于ADVENTURE編制了并行計(jì)算程序，實(shí)例證明，編譯的程序可以用于大型結(jié)構(gòu)的動力并行計(jì)算分析。 6．基于區(qū)域分解算法對靜力非線性和動力非線性有限元進(jìn)行了并行計(jì)算方法的研究，在非線性方程求解中，在每個荷增量步內(nèi)采用了牛頓-拉夫遜迭代法，，在牛頓-拉夫遜的每個迭代步內(nèi)使用了共軛梯度法進(jìn)行相應(yīng)的迭代計(jì)算，對于動力非線性分析，也基于ADVENTURE編制了并行計(jì)算程序，實(shí)例證明，編譯的程序可以用來求解大型結(jié)構(gòu)的動力非線性反應(yīng)。 7．應(yīng)用上述方法對意大利的萬神廟進(jìn)行了三維地震反應(yīng)分析，該結(jié)構(gòu)總共劃分為1,329,027個四結(jié)點(diǎn)的四面體單元，結(jié)果表明，當(dāng)采用96個計(jì)算節(jié)點(diǎn)時(shí)，只需要短短的十一分鐘左右的計(jì)算時(shí)間。
[Abstract]:With the rapid progress of science and technology and the rapid improvement of economic conditions, there is more and more demand for large structures all over the world. In the seismic response analysis of these large structures, there are two scientific and technical challenges: 1) the finite element number and the free degree of the finite element model of these large structures are very large. The emergence of parallel computing technology provides a good method and means to solve this type of structure smoothly and ensure the calculation accuracy. In this paper, parallel computing technique is used to study the seismic response analysis of large structures. The research mainly includes the following aspects: 1. In this paper, the current research situation of the regional division method is deeply studied, and the C / C program of some methods is compiled and compared with the results of the regional partition method. The advantages and disadvantages of each method and its application range are analyzed. 2. Various domain decomposition algorithms are studied, and their advantages and disadvantages are analyzed and summarized. 3. The substructure analysis method is studied, and the corresponding C / C program is compiled, and the corresponding stiffness matrix and load matrix are obtained by using ANSYS as the preprocessor. The experimental parallel computing algorithm for simple cantilever beams with two elements and four elements is studied in this paper. The detailed study of this method lays a foundation for the further study of the method. 4. In this paper, the BDD(Balancing Domain Decompositionmethod is studied in detail. It is a substructure method, which divides the structure into several sub-regions, each subregion is divided into boundary nodes and residual internal nodes. In each sub-region, the degree of freedom (DOF) is firstly condensed to form the interface node degree of freedom (DOF) equation. Then the preprocessing conjugate gradient method of BDD preprocessor is used to solve each interface equation. The information between interfaces is exchanged and the displacement and stress of each node are obtained. 5. Based on domain decomposition algorithm, the parallel computation method for static and dynamic problems of structures is studied. Newmark- 尾 method is used to discretize time integration in dynamic parallel computation and analysis of large structures. For dynamic analysis, a parallel computing program based on ADVENTURE is developed. The example shows that the compiled program can be used for dynamic parallel computation and analysis of large structures. 6. Based on the domain decomposition algorithm, the static nonlinear finite element method and the dynamic nonlinear finite element method are studied. In solving the nonlinear equations, Newton-Raphson iteration method is used in each incremental step. In each iteration step of Newton-Raphson, the conjugate gradient method is used to carry out the corresponding iterative calculation. For dynamic nonlinear analysis, a parallel calculation program based on ADVENTURE is also developed, which is proved by an example. The compiled program can be used to solve the dynamic nonlinear response of large structures. 7. Three-dimensional seismic response analysis of Panshenmiao in Italy is carried out by using the above method. The structure is divided into 1329027 tetrahedral elements with four nodes. The results show that 96 nodes are used. It only takes about eleven minutes to calculate.
【學(xué)位授予單位】：中國地震局工程力學(xué)研究所
【學(xué)位級別】：博士
【學(xué)位授予年份】：2012
【分類號】：TP338.6;TU311.3

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 胡寧,張汝清;一種改進(jìn)的有限元子結(jié)構(gòu)并行算法[J];重慶交通學(xué)院學(xué)報(bào);1990年04期

2 郭迅;;地震工程試驗(yàn)聯(lián)網(wǎng)最新進(jìn)展[J];地震工程與工程振動;2006年04期

3 田石柱;蔡新江;;遠(yuǎn)程協(xié)同結(jié)構(gòu)試驗(yàn)方法研究與發(fā)展[J];地震工程與工程振動;2006年05期

4 鐘萬勰;;結(jié)構(gòu)動力方程的精細(xì)時(shí)程積分法[J];大連理工大學(xué)學(xué)報(bào);1994年02期

5 胡寧,張汝清;一種多波前并行處理的有限元算法[J];重慶大學(xué)學(xué)報(bào)(自然科學(xué)版);1991年04期

6 龔強(qiáng);;地震工程遠(yuǎn)程協(xié)同試驗(yàn)支持系統(tǒng)——NEES Grid CN的研究[J];信息技術(shù);2006年03期

7 李強(qiáng)，鄒經(jīng)湘;獨(dú)立子結(jié)構(gòu)并行計(jì)算方法[J];哈爾濱工業(yè)大學(xué)學(xué)報(bào);1996年02期

8 余天堂,姜弘道;一種求解結(jié)構(gòu)動力響應(yīng)的并行解法[J];河海大學(xué)學(xué)報(bào)(自然科學(xué)版);1999年03期

9 茹忠亮;葉亞齊;趙洪波;;區(qū)域分解并行有限元計(jì)算方法研究[J];河南理工大學(xué)學(xué)報(bào)(自然科學(xué)版);2007年02期

10 李麗君,金先龍,李淵印,李根國;有限元軟件結(jié)構(gòu)分析模塊的并行開發(fā)及應(yīng)用[J];上海交通大學(xué)學(xué)報(bào);2004年08期

本文編號：1987047