Incorporation of Mixed Finite Element Methods in Compositional Simulation for Reduction of Numerical Dispersion

Incorporation of Mixed Finite Element Methods in Compositional Simulation for Reduction of Numerical Dispersion
复制标题

DOI:
10.2118/12267-ms
复制
发表时间:
1983
期刊:
--
影响因子:
--
通讯作者:
R. Ewing;R. F. Heinemann
R. Ewing;R. F. Heinemann
中科院分区:
其他
文献类型:
--
作者:
R. Ewing;R. F. Heinemann

文献摘要

被引文献

相似文献

以往的研究表明,标准的有限差分技术造成数值色散和网格方向的问题时,用于模拟提高采收率的过程与不利的流度比。在组分模拟中,数值色散可以扩散尖锐的流体界面,从而产生流体组分的错误预测和混相锋前进速度的相应误差。数值色散也会影响单相和两相流区域边界的计算位置。不准确的流体速度和次优使用上游加权的运输条款联合收割机结合起来,造成许多方面的数值分散和网格定位问题。一个混合有限元方法已被开发,以获得更准确的近似流体速度。在该方法中,达西速度被认为是与总流体压力一起作为主要变量。虽然有限元技术被用来计算更准确的流体速度,这些速度,然后被纳入一个更标准的有限差分法的模拟过程的大部分。本文介绍了使用混合方法在二维有限差分成分模拟器,以减少数值色散所造成的问题。与标准的有限差分模拟器进行了比较,涉及不混溶位移更多»和多个接触的可压缩性现象的问题。«少
Previous studies have shown that standard finite difference techniques cause numerical dispersion and grid orientation problems when used to simulate enhanced recovery processes with adverse mobility ratios. In compositional simulation, numerical dispersion can diffuse sharp fluid interfaces yielding erroneous predictions of fluid compositions and corresponding errors in the velocities of the miscible frontal advance. Numerical dispersion can also effect the computed locations of the boundaries of the regions of single-phase and two-phase flow. Inaccurate fluid velocities and suboptimal use of upstream weighting of transport terms combine to cause many aspects of the numerical dispersion and grid orientation problems. A mixed finite element method has been developed to obtain more accurate approximations to the fluid velocities. In this method the Darcy velocities are considered as primary variables together with the total fluid pressure. Although finite element techniques are used to compute the more accurate fluid velocities, these velocities are then incorporated into a more standard finite difference method for the bulk of the simulation process. This paper presents the use of mixed methods in a two-dimensional finite difference compositional simulator to reduce problems caused by numerical dispersion. Comparisons are made with a standard finite difference simulator on problems involving immiscible displacementmore » and multiple contact miscibility phenomena.« less