Comparison of numerical methods for simulating strongly nonlinear and heterogeneous reactive transport problems—the MoMaS benchmark case
Comparison of numerical methods for simulating strongly nonlinear and heterogeneous reactive transport problems—the MoMaS benchmark case
复制标题
模拟强非线性和异质反应输运问题的数值方法比较——MoMaS 基准案例
DOI：
--
复制
发表时间：
2010
期刊：
影响因子：
--
通讯作者：
K. MacQuarrie
中科院分区：
文献类型：
--
作者：
J. Carrayrou;J. Hoffmann;P. Knabner;S. Kräutle;C. Dieuleveult;J. Erhel;J. Lee;V. Lagneau;K. Mayer;K. MacQuarrie
虽然多组分反应输运模型在各个地球科学领域得到了广泛的应用，但它仍然存在着重大的数学和计算挑战。有必要解决和比较复杂的基准问题的解决方案，使用各种代码，因为这种相互比较可以揭示有前途的数值解方法，并增加在反应运输代码的应用的信心。在这篇文章中，针对MoMaS基准的所谓简单测试案例的1D和2D子问题，比较了五种当前反应式传输代码的结果和性能（Carrayrou等人，Comput Geosci，2009，本期）。这个基准测试提出了一个简单的虚拟反应性运输问题，突出了真实的反应性运输问题中遇到的主要数值困难。作为一个组，代码包括迭代和非迭代算子分裂和全球隐式解决方案的方法。1D容易平流和1D容易扩散的情况下，使用所有的代码解决，并在一般情况下，有一个很好的协议，解决方案的差异仅限于快速浓度变化的地区。计算需求通常与各种解决方案的预期一致。解决2D问题的三个代码给出的解决方案之间的差异更重要。2D问题所需的非常高的计算工作量说明了并行计算的重要性。基准测试的最重要成果是，所有代码都能够为非常复杂和计算困难的问题产生可比较的结果。
Although multicomponent reactive transport modeling is gaining wider application in various geoscience fields, it continues to present significant mathematical and computational challenges. There is a need to solve and compare the solutions to complex benchmark problems, using a variety of codes, because such intercomparisons can reveal promising numerical solution approaches and increase confidence in the application of reactive transport codes. In this contribution, the results and performance of five current reactive transport codes are compared for the 1D and 2D subproblems of the so-called easy test case of the MoMaS benchmark (Carrayrou et al., Comput Geosci, 2009, this issue). This benchmark presents a simple fictitious reactive transport problem that highlights the main numerical difficulties encountered in real reactive transport problems. As a group, the codes include iterative and noniterative operator splitting and global implicit solution approaches. The 1D easy advective and 1D easy diffusive scenarios were solved using all codes, and, in general, there was a good agreement, with solution discrepancies limited to regions with rapid concentration changes. Computational demands were typically consistent with what was expected for the various solution approaches. The differences between solutions given by the three codes solving the 2D problem are more important. The very high computing effort required by the 2D problem illustrates the importance of parallel computations. The most important outcome of the benchmark exercise is that all codes are able to generate comparable results for problems of significant complexity and computational difficulty.