适用于间断Galerkin方法的限制器研究

开发了一种适用于高精度间断Galerkin方法的斜率（多项式系数）限制器。与现有的斜率限制器不同,该限制器实施过程不考虑网格单元类型（三角形或四边形）,通过全微分方法构造新的多项式系数,因此,该限制器能够适用于各种类型网格——结构化网格、具有单一单元的非结构化网格和具有混合单元的非结构化网格。由于该限制器能够方便地应用于具有混合单元的非结构化网格,因此,本文使用的程序能够方便地求解具有复杂几何结构的流动问题。本文利用一些典型算例对其性能进行了验证,表明该限制器适用于不同类型的网格单元,能够在光滑解区保证高的精度,并能够在阊断区抑帛3非物理振荡。     

一个基于非均匀网格上的二阶基本振荡差分格式

在基本无振荡格式的构造中，将通常的对流通量f的逼近方式推广到对通量导数的逼近，这一构造方法可以有效地应用到非均匀或非结构网格。直接基于非均匀网格上，构造了一个二阶的基本无振荡（ENO）差分格式，该格式具有形式简单，对网格的划分灵活，对传统格式相比不增加计算量等优点，几个数值算例证明了格式的有效性。     

对流扩散方程的绝对稳定高阶中心差分格式

将作者提出的数值摄动算法改进为区分离散单元内上游和下游并分别对通量进行高精度重构的双重数值摄动算法,与原(单重)摄动算法相比,双重摄动算法既提高了格式精度又明显扩大了格式的稳定域范围.利用双重摄动算法,即分别利用上游和下游基点变量的摄动重构将高阶流体力学关系及迎风机制耦合进二阶中心格式之中,由此构建了对流扩散方程的对网格Reynolds数的任意值均稳定(绝对稳定)高精度(四阶和八阶精度)三基点中心TVD差分格式,通过解析分析以及3个算例计算证实了构建格式的优良性能;3个算例包括一维线性、非线性(Burgers方程)和二维变系数对流扩散方程.数值计算表明:构建的格式在粗网格下不振荡,构建格式在粗网格时的最大误差L_∞和均方误差L_2与二阶中心格式在细网格时的相应误差一致,对线性方程,构建格式在细网格下可达到L_2精度阶.     

非结构混合网格化学非平衡粘性绕流数值模拟

在非结构混合网格上对化学非平衡粘性绕流进行了数值模拟。控制方程为考虑了化学非平衡效应的二维Navier-Stokes方程,化学动力学模型为7组元、7反应模型。控制方程中的对流项采用VanLeer逆风分裂格式处理,并应用MUSCL方法及Minmod限制器扩展到二阶精度,粘性项用中心差分格式处理。时间推进采用显式5步龙格-库塔方法。为了适应高超声速流场计算,对VanLeer通量分裂方法进行了改进,并引入了化学反应时间步长。对RAMC-II模型的飞行试验流场进行了数值模拟,计算结果与试验测量数据符合较好,并与参考文献中的数值模拟结果吻合。     

基于非结构/混合网格的高阶精度格式研究进展

尽管以二阶精度格式为基础的计算流体力学(CFD) 方法和软件已经在航空航天飞行器设计中发挥了重要的作用, 但是由于二阶精度格式的耗散和色散较大, 对于湍流、分离等多尺度流动现象的模拟, 现有成熟的CFD 软件仍难以给出满意的结果, 为此CFD 工作者发展了众多的高阶精度计算格式. 如果以适应的计算网格来分类, 一般可以分为基于结构网格的有限差分格式、基于非结构/混合网格的有限体积法和有限元方法,以及各种类型的混合方法. 由于非结构/混合网格具有良好的几何适应性, 基于非结构/混合网格的高阶精度格式近年来备受关注. 本文综述了近年来基于非结构/混合网格的高阶精度格式研究进展, 重点介绍了空间离散方法, 主要包括k-Exact 和ENO/WENO 等有限体积方法, 间断伽辽金(DG) 有限元方法, 有限谱体积(SV) 和有限谱差分(SD) 方法, 以及近来发展的各种DG/FV 混合算法和将各种方法统一在一个框架内的CPR (correctionprocedure via reconstruction) 方法等. 随后简要介绍了高阶精度格式应用于复杂外形流动数值模拟的一些需要关注的问题, 包括曲边界的处理方法、间断侦测和限制器、各种加速收敛技术等. 在综述过程中, 介绍了各种方法的优势与不足, 其间介绍了作者发展的基于"静动态混合重构" 的DG/FV 混合算法. 最后展望了基于非结构/混合网格的高阶精度格式的未来发展趋势及应用前景.     

NND格式在非结构网格中的推广

在张涵信提出的无波动、无自由参数的差分格式（ＮＮＤ格式）的基础上，构造了适用于非结构网格的二阶精度ＮＮＤ有限体积格式，解决了现有非结构网格方法中为抑制激波附近的波动而必须引入含自由参数的人工粘性项的困难，并采用网格自适应技术以提高效率．通过对二维平板激波反射和前台阶在管道内的流动问题的计算，表明本方法可有效地用于Ｅｕｌｅｒ方程的求解．     

网格与高精度差分计算问题

研究NS方程差分求解时来流雷诺数、计算格式精度和计算网格之间的关系．给出了判定空间三个方向上的粘性贡献在给定雷诺数、格式精度和网格下是否能够正确计入的估计方法．指出在NS方程的二阶差分方法的数值模拟中,由于物面法向采用了压缩网格技术,物面附近的网格间距很小,该方向上的粘性贡献可被计入．但是如果流向和周向的网格较粗,相应的差分方程中的粘性贡献可能落入截断误差相同的量级,因此在精度上等于仍是求解略去流向和周向粘性项的薄层近似方程．指出,高阶精度的差分计算格式,可以避免对网格要求苛刻的困难．并进一步讨论了建立高阶精度格式的问题,提出了建立高阶精度格式应该满足的原则：耗散控制原则以及色散控制原则．为了避免激波附近可能出现的微小非物理振荡,建议发展混合高阶精度格式,即在激波区,采用网格自适应的NND格式,在激波以外的区域,采用按上述原则发展的高阶格式．     

一个基于非均匀网格上的二阶基本无振荡差分格式

在基本无振荡格式的构造中，将通常的对流通量$f$的逼近方式推广到对通量导数的逼近，这一构造方法可以有效地应用到非均匀或非结构网格. 直接基于非均匀网格上，构造了一个二阶的基本无振荡(ENO)差分格式. 该格式具有形式简单，对网格的划分灵活，与传统格式相比不增加计算量等优点. 几个数值算例证明了格式的有效性.     

黏弹性管道水柱分离弥合水锤有限体积法模型

基于有限体积法二阶Godunov求解格式对黏弹性输水管道中水柱分离弥合现象进行建模和模拟研究. 在传统的弹性管道模型基础上考虑管道黏弹性效应的影响. 在瞬变流控制方程中引入管道黏弹性项和动态摩阻项, 采用有限体积法进行求解, 考虑压力修正系数来模拟自由气体对计算单元的影响, 同时为避免数值模拟结果产生虚假震荡引入斜率限制器MINMOD函数; 通过虚拟单元法进行边界构建, 实现了计算区域的统一计算. 将所建模型计算结果与已有模型结果、试验结果进行对比, 并对影响模型的各参数进行敏感性分析. 结果表明, 本文模型能够准确模拟出纯水锤、水柱分离弥合水锤两种情况下的瞬态压力变化, 均能与试验数据高度吻合; 与传统的特征线方法相比, 当库朗数C_r小于1时, 有限体积法二阶Godunov格式计算结果更准确、稳定; 在压力波动的衰减过程中, 黏弹性效应相比于管道摩阻起主导作用; 与弹性管道模型相比, 考虑管道黏弹性效应后可显著提高模拟结果的准确性, 尤其是压力波峰值的相对误差明显降低.      

求解管道耦合水力瞬变模型的Godunov格式

应用特征线差分法求解耦合瞬变问题时存在难以避免的多波插值问题,会引入较大的插值误差。为了解决此问题,本文提出了一种求解管道耦合水力瞬变模型的Godunov计算格式。首先基于有限体积法对模型进行数值离散,然后采用时空均为二阶精度的三步MUSCL-Hancock方法计算单元界面上的数值通量,同时引入斜率限制器函数来抑制虚假的数值振荡。在计算边界单元时,采用Rankine-Hugoniot条件与边界条件相结合的方法建立边界方程,有效降低了计算的复杂程度。实验与仿真对比表明：本文的计算结果与实验结果吻合较好,激波捕捉准确且无虚假的数值振荡,进而证明了该方法的可行性和有效性。     

A finite volume scheme for two-phase compressible flows

In gas-particle two-phase flows, when the concentration of the disperesed phase is low, certain assumptions may be made which simplify considerably the equations one has to solve. The gas and particle flows are then linked only via the interaction terms. One may therefore uncouple the full system of equations into two subsystems: one for the gas phase, whose homogeneous part reduces to the Euler equations; and a second system for the particle motion, whose homogeneous part is a degenerate hyperbolic system. The equations governing the gas phase flow may be solved using a high-resolution scheme, while the equations describing the motion of the dispersed phase are treated by a donor-cell method using the solution of a particular Riemann problem. Coupling is then achieved via the right-hand-side terms. To illustrate the capabilities of the proposed method, results are presented for a case specially chosen from among the most difficult to handle, since it involves certain geometrical difficulties, the treatment of regions in which particles are absent and the capturing of particle fronts.     

基于OpenMP的三维显式物质点法并行化研究

基于OpenMP技术开发了三维显式物质点并行程序MPM3DMP。为了避免节点更新阶段的数据竞争,采用区域分解法将背景网格分解为均匀的子域,每个线程负责一个子域的节点变量更新,然后将更新后的节点变量装配到整体。在质点更新阶段采用了循环分解方法进行并行。针对Taylor杆碰撞的三种计算模型,在双Intel Woodcrest 4核CPU服务器下进行了测试:粗模型在4核下加速比为3.82,在8核下为6.23,中模型在4核下加速比为3.79,在8核下加速比为6.23;细模型在4核下加速比为3.75,8核下加速比为6.26。因此,本文的并行程序具有较好的并行效率和可扩展性。     

A composite grid approach for the Euler equations

A zonal grid methodology has been developed for the calculation of compressible fluid flows. The domain subdivision is based on patched grid systems composed of zones or blocks within which a distinct curvilinear grid is generated. The flow simulation is then carried out with a modified scheme based on the Euler finite volume solver of Ni. This scheme uses a distribution procedure that provides an easy and accurate way for the transfer of information from one block to another. This method results in a naturally conservative computation at the interfaces. It is analysed and developed for the treatment of embedded grids with a grid point common to more than four blocks.     

Data assimilation for non-linear tidal models

A data assimilation procedure to incorporate measurements into a non-linear tidal model by using Kalman-filtering techniques is developed. The Kalman filter is based on the two-dimensional shallow water equations. To account for the inaccuracies, these equations are embedded into a stochastic environment by introducing a coloured system noise process into the momentum equations. The continuity equation is assumed to be perfect. The deterministic part of the equations is discretized using an ADI method, the stochastic part using the Euler scheme. Assuming that the system noise is less spatially variable than the underlying water wave process, this stochastic part can be approximated on a coarser grid than the grid used to approximate the deterministic part. A Chandrasekhar-type filter algorithm is employed to obtain the constant-gain extended Kalman filter for weakly non-linear systems. The capabilities of the filter are illustrated by applying it to the assimilation of water level measurements into a tidal model of the North Sea.     

多相流局部混合型质点网格法

提出模拟多相流的局部混合型质点网格法,该方法能稳定地模拟高浓度流体颗粒两相 流. 在每一个颗粒团尺度的欧拉网格下(本文称之为欧拉微元),基于Lagrangian 追踪原理,可直接估计体积内颗粒的总量,从而准确求出欧拉微元和控制容积内颗粒的 浓度(即颗粒在容积内的体积含量). 同时,假设在新的时间步下,颗粒在欧拉微元里充分 混合,形成新颗粒团. 作者对竖直和倾斜容器中单粒径颗粒沉降和竖直容器中双粒径颗粒双 峰悬浮液沉降过程进行了计算,结果与实测数据相符.     

Numerical investigation of Richtmyer-Meshkov instability driven by cylindrical shocks

In this paper, a numerical method with high order accuracy and high resolution was developed to simulate the Richtmyer-Meshkov(RM) instability driven by cylindrical shock waves. Compressible Euler equations in cylindrical coordinate were adopted for the cylindrical geometry and a third order accurate group control scheme was adopted to discretize the equations. Moreover, an adaptive grid technique was developed to refine the grid near the moving interface to improve the resolution of numerical solutions. The results of simulation exhibited the evolution process of RM instability, and the effect of Atwood number was studied. The larger the absolute value of Atwood number, the larger the perturbation amplitude. The nonlinear effect manifests more evidently in cylindrical geometry. The shock reflected from the pole center accelerates the interface for the second time, considerably complicating the interface evolution process, and such phenomena of reshock and secondary shock were studied. The project supported by the National Natural Science Foundation of China (10176033, 10135010 and 90205025). The English text was polished by Yunming Chen.     

An unstructured/structured multi‐layer hybrid grid method and its application

A multi‐layer hybrid grid method is constructed to simulate complex flow field around 2‐D and 3‐D configuration. The method combines Cartesian grids with structured grids and triangular meshes to provide great flexibility in discretizing a domain. We generate the body‐fitted structured grids near the wall surface and the Cartesian grids for the far field. In addition, we regard the triangular meshes as an adhesive to link each grid part. Coupled with a tree data structure, the Cartesian grid is generated automatically through a cell‐cutting algorithm. The grid merging methodology is discussed, which can smooth hybrid grids and improve the quality of the grids. A cell‐centred finite volume flow solver has been developed in combination with a dual‐time stepping scheme. The flow solver supports arbitrary control volume cells. Both inviscid and viscous flows are computed by solving the Euler and Navier–Stokes equations. The above methods and algorithms have been validated on some test cases. Computed results are presented and compared with experimental data. Copyright © 2006 John Wiley & Sons, Ltd.     

A particle–gridless hybrid method for incompressible flows

A particle–gridless hybrid method for the analysis of incompressible flows is presented. The numerical scheme consists of Lagrangian and Eulerian phases as in an arbitrary Lagrangian–Eulerian (ALE) method, where a new‐time physical property at an arbitrary position is determined by introducing an artificial velocity. For the Lagrangian calculation, the moving‐particle semi‐implicit (MPS) method is used. Diffusion and pressure gradient terms of the Navier–Stokes equation are calculated using the particle interaction models of the MPS method. As an incompressible condition, divergence of velocity is used while the particle number density is kept constant in the MPS method. For the Eulerian calculation, an accurate and stable convection scheme is developed. This convection scheme is based on a flow directional local grid so that it can be applied to multi‐dimensional convection problems easily. A two‐dimensional pure convection problem is calculated and a more accurate and stable solution is obtained compared with other schemes. The particle–gridless hybrid method is applied to the analysis of sloshing problems. The amplitude and period of sloshing are predicted accurately by the present method. The range of the occurrence of self‐induced sloshing predicted by the present method shows good agreement with the experimental data. Calculations have succeeded even for the higher injection velocity range, where the grid method fails to simulate. Copyright © 1999 John Wiley & Sons, Ltd.     

Flux‐difference splitting‐based upwind compact schemes for the incompressible Navier–Stokes equations

Third‐order and fifth‐order upwind compact finite difference schemes based on flux‐difference splitting are proposed for solving the incompressible Navier–Stokes equations in conjunction with the artificial compressibility (AC) method. Since the governing equations in the AC method are hyperbolic, flux‐difference splitting (FDS) originally developed for the compressible Euler equations can be used. In the present upwind compact schemes, the split derivatives for the convective terms at grid points are linked to the differences of split fluxes between neighboring grid points, and these differences are computed by using FDS. The viscous terms are approximated with a sixth‐order central compact scheme. Comparisons with 2D benchmark solutions demonstrate that the present compact schemes are simple, efficient, and high‐order accurate. Copyright © 2008 John Wiley & Sons, Ltd.     

GASEOUS FLOW ORIGINATED BY A GUNSHOT,I.

SUMMARY

The numerical simulation of the gaseous flow originated by a gunshot both during the projectile motion within the muzzle and out of it is considered. The numerical solution of the Euler equations is obtained via a second order accurate ENO scheme. A special algorithm of cells reorganization is developed for computing the projectile motion out of the muzzle. It reveals an interesting property of the ENO schemes and Roe's approximate solution of automatically capturing discontinuities by means of the choice of the corresponding grid velocity. The algorithm developed describes sufficiently well all gasdynamics phenomena which accompany the gunshot.