Uniformly robust preconditioners for incompressible MHD system☆
Introduction
We consider the following incompressible dimensional MHD equations under the influence of the body force and applied currents: for with and a fixed . Here, is a connected, bounded domain which is either convex or has a boundary , denotes the velocity field, is the pressure, is the magnetic field, is a known body force, and is a known applied current with and , where denotes the outer unit normal of . The physical parameters (fluid Reynolds number, we assume ), (magnetic Reynolds number) and the coupling coefficient are given by where is the characteristic velocity, the characteristic length, the kinematic viscosity, the magnetic permeability, the electric conductivity, the characteristic magnetic field, the fluid density. The system is considered in conjunction with the following boundary and initial conditions with , .
The MHD equations model the dynamic behaviors of an electrically conducting fluid under the effect of an imposed magnetic field. Its applications involve in geophysics, astrophysics, fusion reactor blankets and confinement for controlled thermonuclear fusion, see [1], [2], [3]. The governing equations behind MHD couple the Navier–Stokes equations for hydrodynamics and Maxwell’s equations for electromagnetism. The two types equations are coupled by the Lorentz force, which governs the effect of a magnetic field on fluid flow, and the appearance of the fluid velocity in Ohm’s law, which accounts for the influence of hydrodynamics on the electric current. Concerning the corresponding extensive theoretical modeling/numerical analysis of the MHD system, we refer to [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21] and the references therein.
In recent years, there have been many numerical investigations for the MHD system and other corresponding works such as [22], [23], [24], [25], [26], [27]. For example, in [8], [28], optimal error estimates for semi-implicit scheme of the MHD system were proved, and numerical performances were also presented. In [29], [30], convergence analyses of Crank–Nicolson extrapolation scheme were performed. In [31], some coupled and decoupled schemes for approximating weak solution and regular strong solution were proposed and analyzed. In [32], stability and convergence of a BFF2 linear scheme were investigated. In [9], two partitioned schemes for a reduced MHD system at small magnetic Reynolds number were proposed and analyzed. In [33], under low regular assumption of the solutions and domain geometries, an optimal error estimate for the semi-implicit scheme was established. In [34], [35], [36], [37], [38], stability and convergence of decoupled schemes for the MHD problems were discussed. In [39], convergence analysis was carried out for the MHD model in a nonconvex, nonsmooth and multiconnected domain. In [40], the author developed a parallel, scalable multigrid-based and physics-based preconditioning strategy for the compressible MHD equations. Numerical tests demonstrate the excellent scalability properties of the solver in terms of grid refinement. In [41], the authors employed Newton–Krylov methods that are preconditioned by fully-coupled algebraic multilevel preconditioners for the MHD problem. They verified the preconditioner is robust and scalable. In [5], [42], the authors studied the block preconditioners by approximate block factorization approach for the MHD system, and demonstrated the robustness of these preconditioners with respect to mesh refinement. In [43], the authors designed block LU preconditioners for the thermally coupled MHD equations with good properties with respect to the mesh size and the Hartmann number. In [16], the authors developed block preconditioners for a new MHD model that preserves divergence free of magnetic field. They proved the preconditioners are robust with respect to most physical and discretization parameters. In [44], the authors proposed a Newton–Krylov solver for stationary MHD equations. They demonstrated the robustness of the proposed solver with respect to mesh refinement.
Seen from the above literatures, in the numerical studies for the MHD problems, there mainly includes two respects: (i) proposing efficient schemes and carrying out numerical analysis and (ii) designing robust solvers for solving the linear systems. In the most existing literatures, people often focus on one respect. In this paper, we attempt to connect the two respects. We propose the first order and second order time-marching schemes for the MHD equations and provide their optimal error estimates. Then, we design block diagonal preconditioners for the schemes using the operator preconditioning method [45], [46] and prove their robustness.
In most of MHD solvers literatures, such as [5], [40], [41], [42], [43], [44], people often focus on the robustness of the preconditioners with respect to mesh refinement, and pay a little attention to the robustness on the physical parameters. In [16], the authors designed preconditioners that are robust with respect to most physical parameters. They demonstrated the condition number of the preconditioned system depends on the coupling coefficient and the numerical solution of magnetic field . In this paper, we aim to study uniformly robust preconditioners for the MHD problem. In order to avoid such a condition like [16], we subtly introduce a symmetric term in magnetic equations, and define an appropriate block diagonal preconditioner for the augmented system. We rigorously prove the condition number of the preconditioned operator is uniformly bounded by a constant which only depends on the domain . Thus, the proposed preconditioner is robust with respect to mesh size , time step size , and physical parameters . We also present the and optimal error estimates of the velocity field and magnetic field , and the optimal error estimates of the pressure for the time-marching scheme. Finally, numerical experiments, including convergence tests and physical benchmark problems, are implemented to test the accuracy of our schemes and validate uniform robustness of the proposed preconditioner.
The rest of paper is organized as follows. In Section 2, we propose the first order time-marching scheme and provide its optimal error estimates. In Section 3, we design a block diagonal preconditioner for the scheme and prove its uniform robustness. In Section 4, we extend the discussions to the second order schemes. In Section 5, ample numerical experiments are presented to test our theoretical results. Finally, some concluding remarks are given in Section 6.
Section snippets
First order scheme and error estimates
In this section, we propose the first order time-marching scheme for problem (1.1) and demonstrate error estimates. We first define some notations of function spaces and norms. For two vector functions , we denote the inner product as and norm . For function setting of this MHD model, we also define several standard Sobolev spaces: and use for the norm in . In and
Robust preconditioner
In this section, we study the robust preconditioner for the scheme (2.3). We first present spatial discretization for (2.3) based on finite element method. For , , we define the conforming finite element spaces which consist of continuous piecewise polynomials of degree and , respectively. Moreover, we assume and satisfy the inf–sup condition: where the constant only depends on the domain . A well known
Second order schemes and robust preconditioners
In this section, we generalize our discussions to second order time-marching schemes. With the notations the two second order schemes based on BDF2 and Crank–Nicolson methods are presented as follows.
Second order scheme I(BDF2): Find satisfying
Numerical examples
In this section, we present accuracy tests and benchmark problems to test the accuracy of the schemes and the robustness of the proposed preconditioner (3.13). Here, we verify the robustness of the preconditioner (3.13) for the first order scheme (3.5). We expect the preconditioner (4.3) for the second order schemes has the similar performance. For spatial discretization, we use conforming finite elements to discretize velocity, pressure and magnetic field for island coalescence
Conclusion remarks
In this paper, we propose the first order and second order time-marching schemes and devise robust preconditioners for solving the MHD problem. We provide the optimal error estimates for the proposed time-marching scheme. Furthermore, we design a block diagonal preconditioner for the proposed scheme using the operator preconditioning framework, and rigorously prove that the condition number of preconditioned system is bounded by a constant that only depends on domain . Numerical experiments
References (51)
- et al.
Global small solutions of 2-d incompressible MHD system
J. Differential Equations
(2015) - et al.
Robust preconditioners for incompressible MHD models
J. Comput. Phys.
(2016) - et al.
Central discontinuous galerkin methods for ideal MHD equations with the exactly divergence-free magnetic field
J. Comput. Phys.
(2011) - et al.
Locally divergence-free central discontinuous Galerkin methods for ideal MHD equations
J. Comput. Sci.
(2013) - et al.
Efficient, non-iterative, and second-order accurate numerical algorithms for the anisotropic Allen–Cahn Equation withprecise nonlocal mass conservation
J. Comput. Appl. Math.
(2020) - et al.
Efficient and energy stable method for the Cahn–Hilliard phase-field model for diblock copolymers
Appl. Numer. Math.
(2020) - et al.
Efficient numerical scheme for a dendritic solidification phase field model with melt convection
J. Comput. Phys.
(2019) - et al.
Fast, provably unconditionally energy stable, and second-order accurate algorithms for the anisotropic Cahn–Hilliard model
Comput. Methods Appl. Mech. Engrg.
(2019) - et al.
A novel decoupled and stable scheme for an anisotropic phase-field dendritic crystal growth model
Appl. Math. Lett.
(2019) - et al.
Optimal convergence analysis of crank-nicolson extrapolation scheme for the three-dimensional incompressible magnetohydrodynamics
Comput. Math. Appl.
(2018)
A semi-implicit energy conserving finite element method for the dynamical incompressible magnetohydrodynamics equations
Comput. Methods Appl. Mech. Engrg.
Decoupled schemes for unsteady mhd equations ii: Finite element spatial discretization and numerical implementation
Comput. Math. Appl.
Convergence analysis of an unconditionally energy stable projection scheme for magneto-hydrodynamic equations
Appl. Numer. Math.
Fully decoupled, linear and unconditionally energy stable time discretization scheme for solving the magneto-hydrodynamic equations
J. Comput. Appl. Math.
Towards a scalable fully-implicit fully-coupled resistive MHD formulation with stabilized fe methods
J. Comput. Phys.
Block recursive LU preconditioners for the thermally coupled incompressible inductionless MHD problem
J. Comput. Phys.
A robust solver for the finite element approximation of stationary incompressible MHD equations in 3D
J. Comput. Phys.
Operator preconditioning
Comput. Math. Appl.
Magnetohydrodynamics, Vol. 3
Advanced Magnetohydrodynamics: With Applications To Laboratory and Astrophysical Plasmas
Advances in Solar System Magnetohydrodynamics
Some mathematical questions related to the MHD equations
Comm. Pure Appl. Math.
A block preconditioner for an exact penalty formulation for stationary MHD
SIAM J. Sci. Comput.
Global small solutions to an MHD-type system: The three-dimensional case
Comm. Pure Appl. Math.
Unconditional convergence of the Euler semi-implicit scheme for the three-dimensional incompressible MHD equations
IMA J. Numer. Anal.
Cited by (9)
Error analysis of vector penalty-projection method with second order accuracy for incompressible magnetohydrodynamic system
2024, Computers and Mathematics with ApplicationsA coupled nonlinear finite element scheme for anisotropic diffusion equation with nonlinear capacity term
2024, Journal of Computational and Applied MathematicsStability and convergence of spatial discrete stabilized finite volume method for the unsteady incompressible magnetohydrodynamics equations
2022, Applied Numerical MathematicsA fully decoupled linearized finite element method with second-order temporal accuracy and unconditional energy stability for incompressible MHD equations
2022, Journal of Computational PhysicsCitation Excerpt :The combined magnetic field produced by the two magnetic islands produces Lorentz forces, which pull the islands together over time. Concerning the physical background of this problem and the extensive numerical simulations for it, we refer to [44–49]. The Kelvin-Helmholtz (K-H) instability in sheared flow configurations is an efficient mechanism that cause fluid mixing, momentum and energy transfer, and turbulence development.
Optimal error estimate of the penalty method for the 2D/3D time-dependent MHD equations
2023, Numerical Algorithms
- ☆
This work is partially supported by National Natural Science Foundation of China under Grant Numbers 11601468 and 11771375, and Shandong Province Natural Science Foundation, China (ZR2018MA008).