李鸿晶, 杨筱朋, 梅雨辰. 一种多自由度阻尼体系动力问题的高阶分析方法[J]. 工程力学, 2021, 38(1): 15-26. DOI: 10.6052/j.issn.1000-4750.2020.03.0162
引用本文: 李鸿晶, 杨筱朋, 梅雨辰. 一种多自由度阻尼体系动力问题的高阶分析方法[J]. 工程力学, 2021, 38(1): 15-26. DOI: 10.6052/j.issn.1000-4750.2020.03.0162
LI Hong-jing, YANG Xiao-peng, MEI Yu-chen. A HIGH-ORDER PROCEDURE FOR DYNAMICS OF MULTI-DEGREE-OF-FREEDOM DAMPING SYSTEMS[J]. Engineering Mechanics, 2021, 38(1): 15-26. DOI: 10.6052/j.issn.1000-4750.2020.03.0162
Citation: LI Hong-jing, YANG Xiao-peng, MEI Yu-chen. A HIGH-ORDER PROCEDURE FOR DYNAMICS OF MULTI-DEGREE-OF-FREEDOM DAMPING SYSTEMS[J]. Engineering Mechanics, 2021, 38(1): 15-26. DOI: 10.6052/j.issn.1000-4750.2020.03.0162

一种多自由度阻尼体系动力问题的高阶分析方法

A HIGH-ORDER PROCEDURE FOR DYNAMICS OF MULTI-DEGREE-OF-FREEDOM DAMPING SYSTEMS

  • 摘要: 传统动力时程直接积分法多采用低阶数值格式,需要选择非常小的时间步距才能获得满足精度要求的动力分析结果。该文将结构动力时程分析的积分求微法推广至多自由度情形,发展了一种具有较高计算效率的多自由度阻尼体系的动力时程高阶分析方法。将相邻的ρ个时步组成一个待求解时段,基于多自由度体系动力响应积分解,以精细积分法结合秦九韶算法计算各时间节点的矩阵指数。逆用微分求积原理发展一种针对含有矩阵指数卷积的高精度数值方法,逐时段求解得到体系各时刻的动力响应。该文方法为动力时程高阶分析方法,全部分析过程仅表现为一系列矩阵乘法及其递推计算,无需求解方程及额外插值,一步计算便可同时获得时段内ρ个时步的全部动态响应(实际动力分析时可取ρ=10~15),具备高效显式算法的特点。由于不必直接计算动力响应积分解中的定积分项,因而避免了对动力方程非齐次项进行特殊处理时所面临的困难。数值试验进一步表明,该方法能很快收敛到精确解,具有较高的计算精度,且数值稳定性好,在较大的时间步距下依然能得到较精确可靠的结果。

     

    Abstract: The lower order schemes are usually adopted in the traditional time-history integration methods, in which time steps must be selected small enough to meet requirements in the accuracy of computational results. We extend the integral differentiation procedure for dynamics of structures to the multi-degree-of-freedom (MDOF) systems. A high-order dynamic method is developed for the MDOF damping systems without more computational efforts. The duration is divided into a few time intervals consisting of ρ equidistant time steps, and the matrix exponential at each time node over the time interval in consideration is obtained by combining precise time-step integration method (PTIM) with the Qin Jiu-shao algorithm, based upon the dynamic solution of Duhamel’s integral for MDOF system. The differential quadrature (DQ) rule is employed in the inverse way to obtain the responses from the matrix exponential solution with convolution form interval by interval. According to this high-order dynamic procedure, only a series of matrix multiplications and their recursions are required in the whole analysis process, without solving the equation and performing extra interpolation. The dynamic responses at ρ discrete time instants can be acquired simultaneously (generally ρ=10~15 may be chosen for dynamic analysis in practice), and it indicates the characteristics of efficient and explicit algorithm. Since it is unnecessary to directly calculate the definite integral term in the integral solution of the dynamic response, the difficulty in special processing of the inhomogeneous term of the dynamic equation would be avoided. Numerical examples further show that this approach owns better numerical stability, the result can quickly converge to an accurate solution, and high calculation accuracy can be still achieved even in the case of large time steps.

     

/

返回文章
返回