A Godunov-type tensor artificial viscosity for staggered Lagrangian hydrodynamics
Introduction
Lagrangian hydrodynamics, in which the computational grid moves simultaneously with the fluid, is widely used for multi-material flow simulations. Its property of resolving sharp contact discontinuities has a broad range of applications in fields such as astrophysics and inertial confinement fusion.
Early investigations [1] led to the development of staggered grid hydrodynamics (SGH), which employs a staggered placement of variables. In SGH, the momentum equation is solved around a nodal control volume, while the energy equation is solved around a zonal control volume. A classical SGH scheme [2] uses the support operator method [3], which discretizes the governing equations in a compatible way. The advantage of this scheme is that it exactly ensures the conservation of momentum and total energy. In the presence of shock waves in the SGH, an artificial viscosity is needed to eliminate the possible numerical oscillations. For 1D calculations, Von Neumann and Richtmyer [1] introduced a viscosity with a quadratic term that is added to the pressure in the momentum and energy equations. Later, Landshoff [4] proposed an additional linear term that improved Von Neumann's method by ensuring fewer oscillations behind the shock. Another well-known artificial viscosity proposed by Kurapatenko [5] uniformly matches the quadratic term and the linear term. This artificial viscosity is a good complement of the formulation proposed by Landshoff. For 2D and 3D calculations, many artificial viscosities have been established on the basis of the 1D artificial viscosity with minor modifications. Advanced multi-dimensional artificial viscosities can be edge based [6], tensorial [7], [8], [9], or Laplacian-based [10].
A different approach designed by Godunov [11] employs a collocated placement of all variables in one control volume. This Godunov-type method is known as cell-centered hydrodynamics (CCH) in the Lagrangian framework. Classically, such schemes overcome the discontinuities at cell edges using an approximate Riemann solver. A remarkable study using CCH was conducted by Despres [12], who constructed a node solver based on a Riemann invariant relation that satisfies the time-averaged equations for momentum conservation, total energy conservation, geometric conservation law (GCL), and entropy inequality. Subsequently, Maire [13], [14] established a sophisticated node solver that exactly recovers the 1D acoustic solver by allocating two pressures on each subcell. In [15], Burton considers the ancillary equation of rotational equilibrium to establish a new node solver with increased mesh robustness. For more complex discontinuous conditions, Jia [16] improved Maire's work by recovering the numerical flux in the Harten–Lax–van Leer contact (HLLC) scheme using the finite element framework. Based on the above node solvers, important progress in CCH has been made in many areas (e.g., 2D/3D numerical schemes, planar/cylindrical coordinates, finite volume/element frameworks) [17], [18], [19], [20], [21], [22].
According to previous investigations, the artificial viscosity and approximate Riemann solution are two different shock-capturing mechanisms. However, their gaps have been bridged by the introduction of Godunov-type artificial viscosities. Related works can be traced back to the application of Godunov's artificial viscosity scheme in the early 1990s. Christensen [23] noticed that the artificial viscosity in SGH can be obtained by the HLL approximate Riemann solver under certain conditions. Luttwak et al. [24], [25] proposed to use a uniaxial tensor pseudo-viscosity based on Riemann problem in staggered meshes. Their method is known as the Staggered Mesh Godunov scheme. Later, a compatible Lagrangian discretization and associated Riemann solver-based artificial viscosity were developed in both the 2D [26], [27], [28] and 3D [29] cases. As such artificial viscosity specifies the jump condition explicitly, high-order accuracy in space can be extended through Taylor expansions of the velocity field. Detailed information can be found in the implementation of a second-order least-squares approach with a frame-invariant limiter [14]. A further potential benefit of such viscosity is that, as long as the connection between the viscosity and the Riemann solution is determined, different types of Riemann solvers (e.g., HLL, HLLC, HLLE) can capture the shock in high resolution. However, a disadvantage of the Godunov-type artificial viscosity is that little attention has been paid to the nature of its tensor. As a result, inaccuracies arise in the viscosity discretization, and these might amplify the hourglass instabilities with spurious grid distortion [8].
In this paper, the general form of the artificial viscosity is derived in the full tensor space using a discrete gradient operator method. Moreover, the relationship between the artificial viscosity and the Generalized Riemann Invariant relation is presented, allowing us to determine the viscosity coefficients. Detailed analysis shows that the developed tensor artificial viscosity is superior to the Godunov-type scalar viscosity [27] in that it is applicable for an arbitrary distribution of the velocity gradient. A series of numerical tests containing strong discontinuities demonstrates the robustness of the developed viscosity.
The remainder of this paper is organized as follows. Section 2 briefly introduces the SGH discretizations that satisfy the total energy conservation condition. In Section 3, the theory and derivation of the Godunov-type artificial viscosity are described in detail. The symmetry of the developed viscosity is proved in Section 4. In Section 5, typical numerical examples are used to verify the performance of the Godunov-type artificial viscosity. Finally, Section 6 presents the conclusions to this study.
Section snippets
Governing equations
In a 2D Lagrangian frame, the gas dynamics equations are where ρ is the density, p is the pressure, u is the fluid velocity, and E is the total energy. Eq. ((1a), (1b), (1c)a) refers to the GCL, which requires the rate of change of a Lagrangian volume to be consistent with the node motion. Eqs. ((1a), (1b), (1c)b) and ((1a), (1b), (1c)c) are the momentum and total energy conservation equations, respectively.
For a smooth flow, the total energy
Godunov-type artificial viscosity
Closely following Maire et al. [27], one can apply the Generalized Riemann Invariant relation in each subcell. This allows us to define an edge-based Godunov-type artificial viscosity. Through the surface integration of the artificial viscosity, a subcell-based matrix is specified. In addition, the above process involves a new degree of freedom due to the unknown cell center velocity . Using the Galilean invariance, can then be determined iteratively. In this section, we will mostly
Symmetry property of the tensor artificial viscosity
In this section, the developed tensor artificial viscosity is shown to maintain 1D cylindrical symmetry on an equi-angular initial grid.
Cylindrical symmetry of the tensor artificial viscosity should consider variables including the momentum, internal energy, and density. For an equi-angular grid, the variable values should be identical along the angular directions, and the momentum should be equal along the radial direction.
Theorem For an equi-angular initial grid, the tensor artificial viscosity
Numerical examples
A series of numerical simulations were conducted with the ideal gas EOS. The tensor artificial viscosity was applied to Sod's problem, the Taylor–Green vortex, the Rayleigh–Taylor instability, and the Sedov and Saltzman problems to test its numerical accuracy. The Noh and triple-point problems were also examined to demonstrate the anti-hourglass feature of the tensor artificial viscosity.
In the numerical tests, the scalar artificial viscosity and the tensor artificial viscosity are both
Concluding remarks
This paper has described a Godunov-type tensor artificial viscosity that is motivated by the scalar form [13]. The proposed method introduces a subcell-based non-negative semidefinite matrix invoking Galilean invariance and entropy inequality. With the construction of the Generalized Riemann Invariant relation in the tensor space, as well as consistency with the GCL, the artificial viscosity was designed in the form of the differential operator μG, where μ is a scalar coefficient. The surface
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References (39)
- et al.
The construction of compatible hydrodynamics algorithms utilizing conservation of total energy
J. Comput. Phys.
(1998) - et al.
Formulations of artificial viscosity for multi-dimensional shock wave computations
J. Comput. Phys.
(1998) - et al.
A tensor artificial viscosity using a mimetic finite difference algorithm
J. Comput. Phys.
(2001) - et al.
A tensor artificial viscosity using a finite element approach
J. Comput. Phys.
(2009) - et al.
A framework for developing a mimetic tensor artificial viscosity for Lagrangian hydrocodes on arbitrary polygonal meshes
J. Comput. Phys.
(2010) - et al.
A symmetry preserving dissipative artificial viscosity in an r–z staggered Lagrangian discretization
J. Comput. Phys.
(2014) A high-order cell-centered Lagrangian scheme for compressible fluid flows in two-dimensional cylindrical geometry
J. Comput. Phys.
(2009)- et al.
A cell-centered Lagrangian Godunov-like method for solid dynamics
Comput. Fluids
(2013) - et al.
A new high-order discontinuous Galerkin spectral finite element method for Lagrangian gas dynamics in two-dimensions
J. Comput. Phys.
(2011) - et al.
A 3D GCL compatible cell-centered Lagrangian scheme for solving gas dynamics equations
J. Comput. Phys.
(2016)
A high-order one-step sub-cell force-based discretization for cell-centered Lagrangian hydrodynamics on polygonal grids
Comput. Fluids
A cell-centered Lagrangian scheme with the preservation of symmetry and conservation properties for compressible fluid flows in two-dimensional cylindrical geometry
J. Comput. Phys.
Reducing spurious mesh motion in Lagrangian finite volume and discontinuous Galerkin hydrodynamic methods
J. Comput. Phys.
A Lagrangian staggered grid Godunov-like approach for hydrodynamics
J. Comput. Phys.
Arbitrary Lagrangian–Eulerian methods for modeling high-speed compressible multimaterial flows
J. Comput. Phys.
A general, non-iterative Riemann solver for Godunov's method
J. Comput. Phys.
Elimination of artificial grid distortion and hourglass-type motions by means of Lagrangian subzonal masses and pressures
J. Comput. Phys.
A survey of several finite difference methods for systems of nonlinear hyperbolic conservation laws
J. Comput. Phys.
Vorticity errors in multidimensional Lagrangian codes
J. Comput. Phys.
Cited by (4)
A parameter-free staggered-grid Lagrangian scheme for two-dimensional compressible flow problems
2024, Journal of Computational PhysicsStudy on a matter flux method for staggered essentially Lagrangian hydrodynamics on triangular grids
2023, International Journal for Numerical Methods in FluidsSymmetry Preservation by a Compatible Staggered Lagrangian Scheme Using the Control-Volume Discretization Method in r–z Coordinate
2023, Communications in Computational Physics