【作者】 雷国东；

【导师】 柳贡民；

【作者基本信息】 哈尔滨工程大学 ， 轮机工程， 2008， 博士

【摘要】 非结构网格有限体积法适合应用于复杂几何结构的流场计算,而SIMPLE系列算法适合应用于工业上广泛存在的低速流、不可压流计算,欧拉-拉格朗日方法适合应用于含稀疏颗粒相的流场计算。本文正是以这些理论为基础展开工作,包括CGNS网格前后处理,压力修正算法的实现和拓展,化学组分源项模型,颗粒相与气相之间的相互作用等。分述如下:本文提出了两种将CGNS非结构网格点数据结构转化为非结构网格面数据结构的快速算法,以作为本文操作非结构网格的基础:内部网格面的整数hash表算法和内部双面壁边界网格面的实数hash表算法。由于CGNS非结构网格存储的是网格结点数据,例如网格结点坐标、每一个网格由哪些结点构成等数据信息;而有限体积法离散过程中,需要用到大量的面积分,所以更多的需要网格面数据结构。内部网格面的构造根据网格面两侧的网格在这个公共网格面上的结点序号相同的特点设计算法,双面壁网格面则根据重合在一起的两个边界网格面中心距离接近于零,或者说小于某一个特征小量,比如该网格面面积的算术平方根来设计算法。两种hash表算法具有快速,内存消耗量小于后续有限体积求解器内存峰值的特点,使前处理算法不会成为整个计算的瓶颈。在压力修正方程的构造过程中,考虑温度修正量对密度修正量的影响,通过静焓方程构造温度修正量与压力修正量的关系,而这种做法的实际效果相当于增强了压力修正量方程线性代数方程组的主对角线。对于可压缩流计算,在流量计算时,在线性插值和迎风格式的基础上,网格面上的密度值采用低阶和高阶的混合格式,动量方程中的压力项处理为滞后的压力梯度项,即线性插值计算方法,和网格面上的压力积分之和,网格面上的压力值则可以采用低阶和高阶混合格式。这些做法兼顾了分辨率和计算的稳定性。在喷雾燃烧数值模拟中考虑了湍流反应组分源项模型如EBU模型、ED模型、Arrinius/EBU/ED模型和Presumed PDF模型。对于颗粒相和气相之间的相互作用,本文一是将颗粒相对气相的反作用阻力中的部分阻力系数剥离,由拉格朗日系统转入欧拉系统,加入到气相动量方程线性代数方程组的主对角线上,这一系数为正实数,增强了主对角线的系数,也增强了含颗粒流的气相迭代计算稳定性;二是在采用时间推进方法计算稳态流场过程中,对耦合的颗粒相与连续相之间计算采用显式和隐式的计算方法进行了研究,这样做的目的在于相对传统的显式时间计算方式,可以采用一个较大的时间步长值进行迭代,从而减少时间推进时的计算量,使得喷雾燃烧数值模拟的计算稳定性得以增强同时减少计算量。在课题组研制的自主知识产权流体力学软件开发过程中,本文作者承担了部分功能模块的研发工作,取得了上述研究成果。并就一些算例与商用软件和经典计算结果予以对比,验证了本文算法的正确性。更多还原

【Abstract】 The unstructured FVM is fit for the complex flow field calculation, and the SIMPLE series algorithms are fit for the low march number, or incompressible flow field calculation, Euler-Largrange method is fit for the sparse spray flow field calculation. Based on these above, the preprocessing and postprocessing with CGNS system, the implementation and modification of SIMPLE series algorithms, the models of the species source term, the ineraction between the particle phase and the gas phase, and so on are carried out. The details are below:Two hash methods forming the face structure from the node structure of the unstructured CGNS mesh are the bases to operate the unstructured mesh in this paper: the integer number hash method for the inner faces formation and the real number hash method for the two-side wall boundary faces formation. Because the node data is stored by the unstructured CGNS mesh, for example, the node coordinates, the node index list of every grid. The face intigral is very usual in Finite Volume method, so the shifting from node to face is needed. The same node index list in the interface of two neighbour cells is employed to formate the inner face, and the distance between the centers of two faces in a two-side wall boundary is approximate to zero, or else less than a specified real number for example the square root of the area of the faces, and this is the key point to formate faces in this kind of boundarys. The two method make the faces formation is very quick, and the computer memory consumed is less than the FVM solver, so the CGNS preprocessing is not the bottle neck of the whole computation.In the formation of the pressure correction equation, the influence of the temperature correction to the density correction is considered, the relationship between the temperature correction and the pressure correction is based on the static enthalpy equation. This method enhanced the primary diagonal coefficient of the pressure correction equation. To the compressible flow calculation, some hybrid methods of low order and high order blending interpolation are employed, such as the hybrid method to get the density of the face mass flux, the hybrid method for the pressure term in the momentum equations, the transient term calculation. These methods compromise btween the resolution and stability.The turbulent combustion species models such as EBU, Eddy Dissipation, Arrinius/EBU/ED, presumed PDF are employed to simulate the spray combustion. In this paper, the momentunm interaction between particle phase and gas phase is modeled by splitting the drag force coefficients, shifting to the Euler system from the Largrange system, adding these coeffients to the diagonal coefficients of the gas phase momentum equation linear algebraic matrix, and the added coefficients are positive, so the diagonal coefficients are enhanced and the calculation stability is enhanced. With the time marching method calculating the steady fields, the explicit and implicit methods of the coupling particulate flow are researched in this paper. With this method a larger time step can be used to reduce the calculation of the time marching progress.In the in-house CFD software developing of our team, some tasks are finished by the writer of this paper. Some calculation cases compared to the commercial software prove the correctness of the methods in this paper.更多还原