一种改善几何非线性边值问题求解效率的组合法

AN INNOVATIVE GROUPING METHOD FOR IMPROVING THE SOLVING EFFICIENCY OF GEOMETRICAL NONLINEAR BOUNDARY VALUE PROBLEMS

  • 摘要: 基于更新拉格朗日理论,推导了梁单元几何大变形控制方程,并建立了相应边值问题的有限元求解方法,通过已有文献数据验证了求解方法的正确性。为了提升现有几何非线性问题求解效率,对几何非线性边值问题按极限点类型进行分类和效率影响因素分析,通过改进已有增量迭代法预测与校正策略,发展了一种可改善几何非线性边值问题求解效率的组合法,并采用Lee Frame等三个常用算例验证了组合法的有效性和量化分析了组合法的加速效果。结果表明:对于十万自由度梁结构模型,无极限点类型问题求解耗时可缩短至原算法的1/3,结果最大相对误差不超过1%;多极限点类型问题求解耗时可缩短至原算法的2/3,且结果最大相对误差小于1%。建立的几何非线性组合求解法可拓展应用于连续体几何非线性、材料非线性等边值问题求解。

     

    Abstract: The finite element method for solving geometrical nonlinear boundary value problems (BVPs) of beam structures was first established upon the basis of Lagrange theory updated, and its correctness was verified here after by using the open literature data. To improve the computational efficiency of the commonly applied incremental iteration methods, the geometrical nonlinear BVPs were classified into two categories, i.e., BVPs with no-limit point and those with multi-limit point, according to the types of limit points, and the influence factors of the algorithm efficiency were analyzed simultaneously. An innovative combination method was developed by improving the prediction and correction strategy of the incremental iteration method, and its acceleration performance was quantitatively analyzed using three commonly-adopted examples, such as Lee Frame and so forth. Computational results indicate that the solution time of geometrical nonlinear BVPs of beam structures with no limit-point and 100,000 degrees of freedoms can be reduced to only 1/3 of the original algorithm, while the maximum relative error is less than 1%. The solution time of geometrical nonlinear BVPs involving multi limit-points can also be shortened to 2/3 of the original algorithm, and the maximum relative error is less than 1% as well. In future works, the combination methodology developed will be expected to be extended to geometrical nonlinear BVPs of continuum structures.

     

/

返回文章
返回