共查询到20条相似文献，搜索用时 31 毫秒
1.
Iterative solvers preconditioned with algebraic multigrid have been devised as an optimal technology to speed up the response of large sparse linear systems. In this work, this technique was implemented in the framework of the dual delineation approach. This involves a single groundwater flow linear solution and a pure advective transport solution with different right-hand sides. The new solver was compared with other preconditioned iterative methods, the MODFLOW's GMG solver, and direct sparse solvers. Test problems include two- and three-dimensional benchmarks spanning homogeneous and highly heterogeneous and anisotropic formations. For the groundwater flow problems, using the algebraic multigrid preconditioning speeds up the numerical solution by one to two orders of magnitude. The algebraic multigrid preconditioner efficiency was preserved for the three dimensional heterogeneous and anisotropic problem unlike for the MODFLOW's GMG solver. Contrarily, a sparse direct solver was the most efficient for the pure advective transport processes such as the forward travel time simulations. Hence, the best sparse solver for the more general advection-dispersion transport equation is likely to be Péclet number dependent. When equipped with the best solvers, processing multimillion grid blocks by the dual delineation approach is a matter of seconds. This paves the way for its routine application to large geological models. The paper gives practical hints on the strategies and conditions under which algebraic multigrid preconditioning would remain competitive for the class of nonlinear and/or transient problems. 相似文献
2.
A Parallel PCG Solver for MODFLOW 总被引：2，自引：0，他引：2
In order to simulate large-scale ground water flow problems more efficiently with MODFLOW, the OpenMP programming paradigm was used to parallelize the preconditioned conjugate-gradient (PCG) solver with in this study. Incremental parallelization, the significant advantage supported by OpenMP on a shared-memory computer, made the solver transit to a parallel program smoothly one block of code at a time. The parallel PCG solver, suitable for both MODFLOW-2000 and MODFLOW-2005, is verified using an 8-processor computer. Both the impact of compilers and different model domain sizes were considered in the numerical experiments. Based on the timing results, execution times using the parallel PCG solver are typically about 1.40 to 5.31 times faster than those using the serial one. In addition, the simulation results are the exact same as the original PCG solver, because the majority of serial codes were not changed. It is worth noting that this parallelizing approach reduces cost in terms of software maintenance because only a single source PCG solver code needs to be maintained in the MODFLOW source tree. 相似文献
3.
To evaluate the use of general‐purpose graphics processing units (GPGPUs) to improve the performance of MODFLOW, an unstructured preconditioned conjugate gradient (UPCG) solver has been developed. The UPCG solver uses a compressed sparse row storage scheme and includes Jacobi, zero fill‐in incomplete, and modified‐incomplete lower‐upper (LU) factorization, and generalized least‐squares polynomial preconditioners. The UPCG solver also includes options for sequential and parallel solution on the central processing unit (CPU) using OpenMP. For simulations utilizing the GPGPU, all basic linear algebra operations are performed on the GPGPU; memory copies between the central processing unit CPU and GPCPU occur prior to the first iteration of the UPCG solver and after satisfying head and flow criteria or exceeding a maximum number of iterations. The efficiency of the UPCG solver for GPGPU and CPU solutions is benchmarked using simulations of a synthetic, heterogeneous unconfined aquifer with tens of thousands to millions of active grid cells. Testing indicates GPGPU speedups on the order of 2 to 8, relative to the standard MODFLOW preconditioned conjugate gradient (PCG) solver, can be achieved when (1) memory copies between the CPU and GPGPU are optimized, (2) the percentage of time performing memory copies between the CPU and GPGPU is small relative to the calculation time, (3) high‐performance GPGPU cards are utilized, and (4) CPU‐GPGPU combinations are used to execute sequential operations that are difficult to parallelize. Furthermore, UPCG solver testing indicates GPGPU speedups exceed parallel CPU speedups achieved using OpenMP on multicore CPUs for preconditioners that can be easily parallelized. 相似文献
4.
Due to complex dynamics inherent in the physical models, numerical formulation of subsurface and overland flow coupling can be challenging to solve. ParFlow is a subsurface flow code that utilizes a structured grid discretization in order to benefit from fast and efficient structured solvers. Implicit coupling between subsurface and overland flow modes in ParFlow is obtained by prescribing an overland boundary condition at the top surface of the computational domain. This form of implicit coupling leads to the activation and deactivation of the overland boundary condition during simulations where ponding or drying events occur. This results in a discontinuity in the discrete system that can be challenging to resolve. Furthermore, the coupling relies on unstructured connectivities between the subsurface and surface components of the discrete system, which makes it challenging to use structured solvers to effectively capture the dynamics of the coupled flow. We present a formulation of the discretized algebraic system that enables the use of an analytic form of the Jacobian for the Newton–Krylov solver, while preserving the structured properties of the discretization. An effective multigrid preconditioner is extracted from the analytic Jacobian and used to precondition the Jacobian linear system solver. We compare the performance of the new solver against one that uses a finite difference approximation to the Jacobian within the Newton–Krylov approach, previously used in the literature. Numerical results explores the effectiveness of using the analytic Jacobian for the Newton–Krylov solver, and highlights the performance of the new preconditioner and its cost. The results indicate that the new solver is robust and generally outperforms the solver that is based on the finite difference approximation to the Jacobian, for problems where the overland boundary condition is activated and deactivated during the simulation. A parallel weak scaling study highlights the efficiency of the new solver. 相似文献
5.
Modeling large multicomponent reactive transport systems in porous media is particularly challenging when the governing partial differential algebraic equations (PDAEs) are highly nonlinear and tightly coupled due to complex nonlinear reactions and strong solution-media interactions. Here we present a preconditioned Jacobian-Free Newton-Krylov (JFNK) solution approach to solve the governing PDAEs in a fully coupled and fully implicit manner. A well-known advantage of the JFNK method is that it does not require explicitly computing and storing the Jacobian matrix during Newton nonlinear iterations. Our approach further enhances the JFNK method by utilizing physics-based, block preconditioning and a multigrid algorithm for efficient inversion of the preconditioner. This preconditioning strategy accounts for self- and optionally, cross-coupling between primary variables using diagonal and off-diagonal blocks of an approximate Jacobian, respectively. Numerical results are presented demonstrating the efficiency and massive scalability of the solution strategy for reactive transport problems involving strong solution-mineral interactions and fast kinetics. We found that the physics-based, block preconditioner significantly decreases the number of linear iterations, directly reducing computational cost; and the strongly scalable algebraic multigrid algorithm for approximate inversion of the preconditioner leads to excellent parallel scaling performance. 相似文献
6.
多重网格方法在求解由偏微分方程的边值问题离散所得线性系统时,具有非常高的计算效率.但常用的几何多重网格法在处理带跃变系数的偏微分方程时存在一定缺陷,限制了其应用.本文应用代数多重网格（AMG）方法求解三维直流电阻率法正演模拟形成的有限差分线性方程组,通过求解二次场的方法消除了总场中由点电源导致的奇异性,从而获得快速、精确的三维电阻率数值模拟.对两个存在大的电性差异的模型进行了模拟计算,以验证代数多重网格法的收敛效率.计算结果表明,与不完全Cholesky共轭梯度（ICCG）方法相比,代数多重网格方法具有更高的计算效率及稳定性.而且,随着三维网格节点数的增加,代数多重网格方法计算的高效性更加明显. 相似文献
7.
A multigrid solver for 3D electromagnetic diffusion 总被引：2，自引：0，他引：2
W.A. Mulder 《Geophysical Prospecting》2006,54(5):633-649
The performance of a multigrid solver for the time‐harmonic electromagnetic problem in geophysical settings is investigated. The frequencies are sufficiently small for waves travelling at the speed of light to be negligible, so that a diffusive problem remains. The discretization of the governing equations is obtained by the finite‐integration technique, which can be viewed as a finite‐volume generalization of Yee's staggered grid scheme. The resulting set of discrete equations is solved by a multigrid method. The convergence rate of the multigrid method decreased when the grid was stretched. The slower convergence rate of the multigrid method can be compensated by using bicgstab2 , a conjugate‐gradient‐type method for non‐symmetric problems. In that case, the multigrid solver acts as a preconditioner. However, whereas the multigrid method provides excellent convergence with constant grid spacings, it performs less than satisfactorily when substantial grid stretching is used. 相似文献
8.
Upscaling pore-scale processes into macroscopic quantities such as hydrodynamic dispersion is still not a straightforward matter for porous media with complex pore space geometries. Recently it has become possible to obtain very realistic 3D geometries for the pore system of real rocks using either numerical reconstruction or micro-CT measurements. In this work, we present a finite element–finite volume simulation method for modeling single-phase fluid flow and solute transport in experimentally obtained 3D pore geometries. Algebraic multigrid techniques and parallelization allow us to solve the Stokes and advection–diffusion equations on large meshes with several millions of elements. We apply this method in a proof-of-concept study of a digitized Fontainebleau sandstone sample. We use the calculated velocity to simulate pore-scale solute transport and diffusion. From this, we are able to calculate the a priori emergent macroscopic hydrodynamic dispersion coefficient of the porous medium for a given molecular diffusion Dm of the solute species. By performing this calculation at a range of flow rates, we can correctly predict all of the observed flow regimes from diffusion dominated to convection dominated. 相似文献
9.
When managing large-scale ground water contamination problems, it is often necessary to model flow and transport using finely discretized domains--for instance (1) to simulate flow and transport near a contamination source area or in the area where a remediation technology is being implemented; (2) to account for small-scale heterogeneities; (3) to represent ground water-surface water interactions; or (4) some combination of these scenarios. A model with a large domain and fine-grid resolution will need extensive computing resources. In this work, a domain decomposition-based assembly model implemented in a parallel computing environment is developed, which will allow efficient simulation of large-scale ground water flow and transport problems using domain-wide grid refinement. The method employs common ground water flow (MODFLOW) and transport (RT3D) simulators, enabling the solution of almost all commonly encountered ground water flow and transport problems. The basic approach partitions a large model domain into any number of subdomains. Parallel processors are used to solve the model equations within each subdomain. Schwarz iteration is applied to match the flow solution at the subdomain boundaries. For the transport model, an extended numerical array is implemented to permit the exchange of dispersive and advective flux information across subdomain boundaries. The model is verified using a conventional single-domain model. Model simulations demonstrate that the proposed model operated in a parallel computing environment can result in considerable savings in computer run times (between 50% and 80%) compared with conventional modeling approaches and may be used to simulate grid discretizations that were formerly intractable. 相似文献
10.
Karst aquifers represent dual flow systems consisting of a highly conductive conduit system embedded in a less permeable rock matrix. Hybrid models iteratively coupling both flow systems generally consume much time, especially because of the nonlinearity of turbulent conduit flow. To reduce calculation times compared to those of existing approaches, a new iterative equation solver for the conduit system is developed based on an approximated Newton–Raphson expression and a Gauß–Seidel or successive over‐relaxation scheme with a single iteration step at the innermost level. It is implemented and tested in the research code CAVE but should be easily adaptable to similar models such as the Conduit Flow Process for MODFLOW‐2005. It substantially reduces the computational effort as demonstrated by steady‐state benchmark scenarios as well as by transient karst genesis simulations. Water balance errors are found to be acceptable in most of the test cases. However, the performance and accuracy may deteriorate under unfavorable conditions such as sudden, strong changes of the flow field at some stages of the karst genesis simulations. 相似文献
11.
12.
This pore-scale modeling study in saturated porous media shows that compound-specific effects are important not only at steady-state and for the lateral displacement of solutes with different diffusivities but also for transient transport and solute breakthrough. We performed flow and transport simulations in two-dimensional pore-scale domains with different arrangement of the solid grains leading to distinct characteristics of flow variability and connectivity, representing mildly and highly heterogeneous porous media, respectively. The results obtained for a range of average velocities representative of groundwater flow (0.1–10 m/day), show significant effects of aqueous diffusion on solute breakthrough curves. However, the magnitude of such effects can be masked by the flux-averaging approach used to measure solute breakthrough and can hinder the correct interpretation of the true dilution of different solutes. We propose, as a metric of mixing, a transient flux-related dilution index that allows quantifying the evolution of solute dilution at a given position along the main flow direction. For the different solute transport scenarios we obtained dilution breakthrough curves that complement and add important information to traditional solute breakthrough curves. Such dilution breakthrough curves allow capturing the compound-specific mixing of the different solutes and provide useful insights on the interplay between advective and diffusive processes, mass transfer limitations, and incomplete mixing in the heterogeneous pore-scale domains. The quantification of dilution for conservative solutes is in good agreement with the outcomes of mixing-controlled reactive transport simulations, in which the mass and concentration breakthrough curves of the product of an instantaneous transformation of two initially segregated reactants were used as measures of reactive mixing. 相似文献
13.
A terrain-following grid formulation (TFG) is presented for simulation of coupled variably-saturated subsurface and surface water flow. The TFG is introduced into the integrated hydrologic model, ParFlow, which uses an implicit, Newton Krylov solution technique. The analytical Jacobian is also formulated and presented and both the diagonal and non-symmetric terms are used to precondition the Krylov linear system. The new formulation is verified against an orthogonal stencil and is shown to provide increased accuracy at lower lateral spatial discretization for hillslope simulations. Using TFG, efficient scaling to a large number of processors (16,384) and a large domain size (8.1 Billion unknowns) is shown. This demonstrates the applicability of this formulation to high-resolution, large-spatial extent hydrology applications where topographic effects are important. Furthermore, cases where the analytical Jacobian is used for the Newton iteration and as a non-symmetric preconditioner for the linear system are shown to have faster computation times and better scaling. This demonstrates the importance of solver efficiency in parallel scaling through the use of an appropriate preconditioner. 相似文献
14.
The χMD matrix solver package is incorporated into USGS groundwater modeling software, such as MODFLOW-NWT, MODFLOW-USG, and MT3D. The solver is used to solve matrices assembled through numerical discretization of the groundwater flow equation, and solute transport equations. χMD has demonstrated its higher robustness, faster execution speed, and more efficient memory usage compared to the existing solvers for many types of groundwater flow problems. χMD uses preconditioned iterative Krylov-subspace methods and consists of preconditioning and acceleration modules. Because the solver package uses a variety of preconditioning features including level-based incomplete lower-upper (ILU) factorization method with a drop tolerance scheme, users must choose optimal preconditioning parameters to improve execution speed and robustness. In order to examine how the preconditioning parameters, ILU factorization level, and drop tolerance values affect the overall performance of the matrix solver, we evaluated five different groundwater model applications using MODFLOW-USG that include different numerical complexities. For those five cases, the number of discretization nodes varied from 10,000 cells to 730,300 cells. From the analysis, we found that the preconditioning parameters greatly affect execution times and memory usage of the preconditioning and acceleration procedures. In addition, a combination of the ILU level between five to seven and the drop tolerance value between 10−2 and 10−3 usually resulted in shorter overall execution time. Our study suggests that the users can elicit higher performance and robustness of the χMD matrix solver using this combination of the parameters and enhance computational efficiency of solving groundwater and solute transport problems. 相似文献
15.
Eric D. Morway Richard G. Niswonger Christian D. Langevin Ryan T. Bailey Richard W. Healy 《Ground water》2013,51(2):237-251
The MT3DMS groundwater solute transport model was modified to simulate solute transport in the unsaturated zone by incorporating the unsaturated‐zone flow (UZF1) package developed for MODFLOW. The modified MT3DMS code uses a volume‐averaged approach in which Lagrangian‐based UZF1 fluid fluxes and storage changes are mapped onto a fixed grid. Referred to as UZF‐MT3DMS, the linked model was tested against published benchmarks solved analytically as well as against other published codes, most frequently the U.S. Geological Survey's Variably‐Saturated Two‐Dimensional Flow and Transport Model. Results from a suite of test cases demonstrate that the modified code accurately simulates solute advection, dispersion, and reaction in the unsaturated zone. Two‐ and three‐dimensional simulations also were investigated to ensure unsaturated‐saturated zone interaction was simulated correctly. Because the UZF1 solution is analytical, large‐scale flow and transport investigations can be performed free from the computational and data burdens required by numerical solutions to Richards' equation. Results demonstrate that significant simulation runtime savings can be achieved with UZF‐MT3DMS, an important development when hundreds or thousands of model runs are required during parameter estimation and uncertainty analysis. Three‐dimensional variably saturated flow and transport simulations revealed UZF‐MT3DMS to have runtimes that are less than one tenth of the time required by models that rely on Richards' equation. Given its accuracy and efficiency, and the wide‐spread use of both MODFLOW and MT3DMS, the added capability of unsaturated‐zone transport in this familiar modeling framework stands to benefit a broad user‐ship. 相似文献
16.
A significant body of current research is aimed at developing methods for numerical simulation of flow and transport in porous media that explicitly resolve complex pore and solid geometries, and at utilizing such models to study the relationships between fundamental pore-scale processes and macroscopic manifestations at larger (i.e., Darcy) scales. A number of different numerical methods for pore-scale simulation have been developed, and have been extensively tested and validated for simplified geometries. However, validation of pore-scale simulations of fluid velocity for complex, three-dimensional (3D) pore geometries that are representative of natural porous media is challenging due to our limited ability to measure pore-scale velocity in such systems. Recent advances in magnetic resonance imaging (MRI) offer the opportunity to measure not only the pore geometry, but also local fluid velocities under steady-state flow conditions in 3D and with high spatial resolution. In this paper, we present a 3D velocity field measured at sub-pore resolution (tens of micrometers) over a centimeter-scale 3D domain using MRI methods. We have utilized the measured pore geometry to perform 3D simulations of Navier–Stokes flow over the same domain using direct numerical simulation techniques. We present a comparison of the numerical simulation results with the measured velocity field. It is shown that the numerical results match the observed velocity patterns well overall except for a variance and small systematic scaling which can be attributed to the known experimental uncertainty in the MRI measurements. The comparisons presented here provide strong validation of the pore-scale simulation methods and new insights for interpretation of uncertainty in MRI measurements of pore-scale velocity. This study also provides a potential benchmark for future comparison of other pore-scale simulation methods. © 2012 Elsevier Science. All rights reserved. 相似文献
17.
有限差分法求解Helmholtz方程,依赖于两点:1差分格式的构造;2高效的求解算法.本文采用平均导数法离散Helmholtz方程.该差分格式有三点好处:1能适用于横纵不等间距采样;2在完全匹配层区域(PML),差分方程与微分方程逐点相容;3能将一个波长内的采样点数减少至少于4.求解离散的Helmholtz方程的算法一般分为直接法和迭代算法.直接法由于内存需求太大而无法适用于大规模问题;基于Krylov子空间的迭代方法结合多重网格预条件算法是一种快速高效求解方法,然而对于横纵不等间距采样(在多重网格中称为各向异性问题),经典的多重网格方法失效.本文分析了经典多重网格的三个重要组成部分:完全加权限制算子,点松弛技术以及双线性延拓算子,进而采用了半粗化技术代替全粗化技术,线松弛技术代替点松弛技术以及依赖差分算子的延拓算子代替双线性延拓算子,使得各向异性问题变得收敛;而且对于非均匀介质中-低频率的迭代问题,我们获得了较为满意的收敛速度. 相似文献
18.
三维反演解释是电磁法勘探发展的重要趋势,而如何提高三维反演的可靠性、稳定性和计算效率是算法开发者们目前的研究重点.本文实现了一种频率域可控源电磁(CSEM)三维反演算法.其中正演基于拟态有限体积法离散化,利用直接矩阵分解技术来求解大型线性系统方程,不仅准确、稳定,而且特别有利于含有大量发射场源位置的CSEM勘探情况;对目标函数的最优化采用高斯牛顿法(GN),具有近似二次的收敛性;使用预条件共轭梯度法(PCG)求解每次GN迭代所得到的法方程,避免了显式求解和存储灵敏度矩阵,减小了计算量.以上这些方法的结合应用,使得本文的三维反演算法准确、稳定且高效.通过陆地和海洋CSEM勘探场景中的典型理论模型的反演测试,验证了本文算法的有效性. 相似文献
19.
本研究针对大地电磁测深法有限元数值模拟中,迭代法求解线性方程组效率较低的问题,利用亥姆霍兹分解原理,将电场矢量双旋度方程的预条件问题转化为基于矢量位的泊松问题和基于标量位的拉普拉斯问题,并在四面体非结构化棱边元离散的情况下,借助节点元辅助网格离散上述预条件问题,进一步利用代数多重网格方法（AMG）实施求解,最终实现预条件算法.利用经典的COMMEMI理论模型进行试算并与前人的积分方程解进行对比,验证了本文数值模拟程序与预条件方法的正确性和可靠性.此外,利用不同自由度规模的实验模型对这一预条件算法的效率进行了测试.结果表明,这一算法可以有效地提升大地电磁测深法棱边有限元数值模拟迭代法的收敛性,计算效率较通用的不完全LU分解预条件算法明显更高;在较大自由度网格（>1000万）数值模拟计算中,其算法效率及内存占用相对直接解法有较大优势,也使小型工作站上利用较大自由度的有限元网格进行大地电磁测深数值模拟计算成为可能. 相似文献
20.
Seokkoo KangAnne Lightbody Craig HillFotis Sotiropoulos 《Advances in water resources》2011,34(1):98-113
We develop an efficient and versatile numerical model for carrying out high-resolution simulations of turbulent flows in natural meandering streams with arbitrarily complex bathymetry. The numerical model solves the 3D, unsteady, incompressible Navier-Stokes and continuity equations in generalized curvilinear coordinates. The method can handle the arbitrary geometrical complexity of natural streams using the sharp-interface curvilinear immersed boundary (CURVIB) method of Ge and Sotiropoulos (2007) [1]. The governing equations are discretized with three-point, central, second-order accurate finite-difference formulas and integrated in time using an efficient, second-order accurate fractional step method. To enable efficient simulations on grids with tens of millions of grid nodes in long and shallow domains typical of natural streams, the algebraic multigrid (AMG) method is used to solve the Poisson equation for the pressure coupled with a matrix-free Krylov solver for the momentum equations. Depending on the desired level of resolution and available computational resources, the numerical model can either simulate, via direct numerical simulation (DNS), large-eddy simulation (LES), or unsteady Reynolds-averaged Navier-Stokes (URANS) modeling. The potential of the model as a powerful tool for simulating energetic coherent structures in turbulent flows in natural river reaches is demonstrated by applying it to carry out LES and URANS in a 50-m long natural meandering stream at resolution sufficiently fine to capture vortex shedding from centimeter-scale roughness elements on the bed. The accuracy of the simulations is demonstrated by comparisons with experimental data and the relative performance of the LES and URANS models is also discussed. 相似文献