Elsevier

Journal of Computational Physics

Volume 326, 1 December 2016, Pages 312-333
Journal of Computational Physics

A multi-dimensional finite volume cell-centered direct ALE solver for hydrodynamics

https://doi.org/10.1016/j.jcp.2016.08.050Get rights and content

Abstract

In this paper we describe a second order multi-dimensional scheme, belonging to the class of direct Arbitrary Lagrangian–Eulerian (ALE) methods, for the solution of non-linear hyperbolic systems of conservation law. The scheme is constructed upon a cell-centered explicit Lagrangian solver completed with an edge-based upwinded formulation of the numerical fluxes, computed from the MUSCL-Hancock method, to obtain a full ALE formulation. Numerical fluxes depend on nodal grid velocities which are either set or computed to avoid most of the mesh problems typically encountered in purely Lagrangian simulations. In order to assess the robustness of the scheme, most results proposed in this paper have been obtained by computing the grid velocities as a fraction of the Lagrangian nodal velocities, the ratio being set before running the test case. The last part of the paper describes preliminary results about the triple point test case run in the ALE framework by computing the grid velocities with the fully adaptive Large Eddy Limitation (L.E.L.) method proposed in [1]. Such a method automatically computes the grid velocities at each node defining the mesh from the local characteristics of the flow. We eventually discuss the advantages and the drawback of the coupling.

Introduction

Numerical simulations of fluid flow have been extensively performed on a large range of hydrodynamic phenomena over the last 60 years. The governing equations can be implemented using the finite volume approach in a given coordinate system. Several schemes have been developed and these can be divided into three categories: the first is based on a Lagrangian description of the matter for which the computational mesh follows the flow [2], [3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13]. Mass conservation is naturally ensured in this approach, which makes Lagrangian schemes the best candidate for preserving and following interfaces in multi-material simulations. However, when the flow is subject to sharp local velocity gradients or local vorticity, mesh distortion can occur and a remeshing/remapping step is inevitable. Unfortunately, this procedure is usually expensive and the remap can be inaccurate and introduce excessive diffusion. The second category adopts an Eulerian point of view for which the computational mesh is fixed, leading to the appearance of convective terms in the governing equations, even in the mass conservation equation [14], [15], [16], [17], [18], [19]. Difficulties arise in multi-material simulations when following material interfaces. Capturing or reconstructing the interfaces can be very complex, time consuming and inaccurate (for example, the order of Young's VOF method is less than one [20]), although significant improvements have been made recently [21], [22], [23].

In the 1970s, the combination of the best properties of the Lagrangian and Eulerian methods led to the creation of a third category, known as arbitrary Lagrangian–Eulerian (ALE) methods [24]. The main feature of ALE is to decorrelate the flow motion and the mesh motion to avoid any mesh tangling which may appear in purely Lagrangian Simulations. ALE methods may be divided into two classes, indirect [25], [26], [27], [28], [29], [30] and direct [31], [32], [33], [34]. Indirect methods are implemented in three steps: (1) a Lagrangian stage in which the solution and the computational mesh are updated; (2) a rezoning stage that regularizes the tangled mesh; (3) a remapping stage in which the Lagrangian solution is transferred to the rezoned mesh. For indirect ALE the decorrelation between mesh and flow is only performed occasionally when the Lagrangian stage becomes critical and might result in simulation failure. In this paper, only direct ALE methods are considered. In these methods, the hydrodynamic equations are solved in an arbitrarily moving coordinate frame, introducing advective terms in the governing equations that depend on the grid velocity. The result is a continuous rezoning capability, with no extra time cost. Overall, ALE methods can manage greater distortions than purely Lagrangian methods but with a better accuracy than purely Eulerian approaches.

In this paper we describe a second order multi-dimensional scheme, belonging to the class of direct Arbitrary Lagrangian–Eulerian (ALE) methods, named DALEC3 since it is a Direct ALE Cell-Centered Code. It is used to solve the well-known Euler equations with the Geometrical Conservation Law for the simulation of compressible hydrodynamic problems. Recall the integral form of these equations in a frame moving at the velocity w (see e.g. [34]):ddtV(t)dVS(t)wνdS=0,ddtV(t)ρdV+S(t)ρμνdS=0,ddtV(t)ρvdV+S(t)(ρvμν+pν)dS=0,ddtV(t)ρEdV+S(t)(ρEμν+pvν)dS=0,ddt(x)=w, where the volume integral is computed over an arbitrary volume V(t) moving at velocity w. The boundary of V(t) is denoted by S(t), and ν(t) denotes the unitary normal on S(t) pointing outwards. Properties characterizing the fluid are the density ρ, velocity v, total energy E and pressure p. The relative velocity will be denoted μ=vw. System (1)–(5) is closed with an equation of state in order to compute the pressure.

The first part of the paper describes the discrete formulation of the system (1)–(5). It is constructed upon a cell-centered Lagrangian solver completed with edge-based upwinded advective fluxes to obtain a full ALE formulation. We fill focus on the two cell-centered schemes GLACE and EUCCLHYD which have demonstrated their properties in references [6], [8], [9], [10]. The treatment of the boundary conditions is presented, as well as the CFL condition used to compute the time step and the second-order extension of the scheme. We also derive the entropy balance and show that our scheme satisfies the 2nd law of thermodynamics.

The second part of the paper presents several numerical results that assess the robustness and the accuracy of the scheme. To do so, most of the results have been obtained by computing the grid velocities as a fraction of the nodal Lagrangian velocities. However, our strategy is to couple the scheme with a method that adaptively computes the grid velocities. Such strategy is a common purpose for the development of direct ALE schemes, and a well-known method (coming from the indirect ALE methods community) is to rely on a rezoning procedure that keeps an acceptable geometrical mesh quality [32], [35], [36]. Recently, a promising technique, the Large Eddy Limitation method, has been developed by Costes and Ghidaglia [1]. The main feature of this technique is to express, from the Hodge decomposition, the nodal Lagrangian velocities as the sum of an irrotational part and a divergence-free part, and to limit each component regarding the local characteristic of the flow to obtain the grid velocities. Our purpose is to couple such method with the DALEC3 scheme. The last part of the paper then proposes a brief theoretical description of the L.E.L. method and preliminary results obtained for the single material triple point problem. Conclusions and perspectives end the paper.

Section snippets

Notation

The computational mesh is assumed to be composed of polygons. We denote the set of nodes, cells and edges defining the mesh respectively N, C and E. In the rest of the paper, specific indices {j,r,k} will be used respectively to denote cells, nodes and edges.

In Fig. 1, we represent a 2D polygonal cell ‘j’ whose area will be denoted Vj. The cell is defined by the set of nodes N(j) and the set of edges E(j). The length of the edge ‘k’ is denoted lk, while νj,k is the discrete outward pointing

Results

We propose in this section various 2D results to show the robustness of the scheme. Our methodology of validation acts in two phases. First, the grid velocity is computed as a fraction of the Lagrangian velocity to assess the robustness of the ALE scheme. Second, the grid velocity is given by a genuine rezone strategy [1] in order to prove that our approach can be coupled to any artificial mesh motion.

Conclusions and perspectives

In this paper, we have presented in detail a multi-dimensional Direct ALE scheme for the numerical solution to the compressible Euler equations. The scheme has been developed from a Godunov-like cell-centered Lagrangian scheme, like GLACE or EUCCLHYD, which has been supplemented with upwinded advective fluxes. The upwinded values are chosen from the sign of the edge relative velocity μν. The second-order extension is obtained by computing the numerical fluxes from the MUSCL-Hancock method.

Acknowledgements

This work has been supported by the CMLA located in Cachan and the French Alternative Energies and Atomic Energy Commission, DIF. Authors would like to thank Christophe Fochesato, Antoine Llor, Thibaud Vazquez-Gonzalez and Raphaël Loubère for their valuable comments and suggestions to improve the manuscript.

References (47)

  • C.W. Hirt et al.

    An arbitrary Lagrangian–Eulerian computing method for all flow speeds

    J. Comput. Phys.

    (1974)
  • M. Kenamond et al.

    Compatible total energy conserving and symmetry preserving arbitrary Lagrangian–Eulerian hydrodynamics in 2D rz – cylindrical coordinates

    J. Comput. Phys.

    (2014)
  • M. Kucharik et al.

    Conservative multi-material remap for staggered multi-material arbitrary Lagrangian–Eulerian methods

    J. Comput. Phys.

    (2014)
  • R. Loubère et al.

    ReALE: a reconnection-based arbitrary-Lagrangian–Eulerian method

    J. Comput. Phys.

    (2010)
  • R. Liska et al.

    Optimization-based synchronized flux-corrected conservative interpolation (remapping) of mass and momentum for arbitrary Lagrangian–Eulerian methods

    J. Comput. Phys.

    (2010)
  • R. Loubère et al.

    A subcell remapping method on staggered polygonal grids for arbitrary-Lagrangian–Eulerian methods

    J. Comput. Phys.

    (2005)
  • A. Lopez Ortega et al.

    A geometrically-conservative, synchronized, flux-corrected remap for arbitrary Lagrangian–Eulerian computations with nodal finite elements

    J. Comput. Phys.

    (2011)
  • W. Boscheri et al.

    A direct arbitrary-Lagrangian–Eulerian ADER-WENO finite volume scheme on unstructured tetrahedral meshes for conservative and non-conservative hyperbolic systems in 3D

    J. Comput. Phys.

    (2014)
  • W. Boscheri et al.

    Direct arbitrary-Lagrangian–Eulerian ADER-MOOD finite volume schemes for multidimensional hyperbolic conservation laws

    J. Comput. Phys.

    (2015)
  • R. Loubère et al.

    ReALE: a reconnection-based arbitrary-Lagrangian–Eulerian method

    J. Comput. Phys.

    (2010)
  • J. Donea et al.

    An arbitrary Lagrangian–Eulerian finite element methods for transient dynamic fluid structure interactions

    Comput. Methods Appl. Mech. Eng.

    (1982)
  • Alan M. Winslow

    Numerical solution of the quasilinear Poisson equation in a nonuniform triangle mesh

    J. Comput. Phys.

    (August 1997)
  • A.J. Barlow

    A high order cell centred dual grid Lagrangian Godunov scheme

    Comput. Fluids

    (2013)
  • Cited by (9)

    • An immersed MMALE material point method for FSI problems with structure fracturing

      2022, Computer Methods in Applied Mechanics and Engineering
      Citation Excerpt :

      The MMALE is successful in simulation of multi-material compressible flow due to its high accuracy and precisely tracing of the material interface. The MMALE also adopts a hybrid Lagrangian and Eulerian description of indirect ALE [53] which consists of three phases, a Lagrangian phase [54,55] in which the physical variables and mesh are updated, a rezoning phase for defining a new mesh with better quality, and a remapping phase [52,56] wherein the physical variables are conservatively interpolated from the old Lagrangian mesh onto the newly rezoned one. It is inspiring that the MPM and MMALE are analogical, where both of them consist of Lagrangian momentum updating and Eulerian remeshing.

    • The δ-ALE-SPH model: An arbitrary Lagrangian-Eulerian framework for the δ-SPH model with particle shifting technique

      2021, Computers and Fluids
      Citation Excerpt :

      Specifically, the ALE approach allows one to rewrite the compressible Navier-Stokes equations with an arbitrary advection velocity and to recover the Eulerian or the Lagrangian formulations when this velocity is set equal to zero or to the actual fluid velocity respectively. From a wider perspective, the above ALE framework can be classified as a “direct” ALE approach, as for example in Clair et al. [63]. This is opposed to “indirect” approaches which are characterized by a remeshing stage after the advection of the calculation point (see e.g. Galera et al. [58]) and are related to remeshed or regularised SPH schemes as in Chaniotis et al. [59], Børve et al. [60].

    • Multi-moment finite volume method for incompressible flows on unstructured moving grids and its application to fluid-rigid body interactions

      2019, Computers and Structures
      Citation Excerpt :

      ALE formulation in finite element framework was developed by Hughes et al. [13–15]. Besides, ALE method was also actively employed in the finite volume framework [16–20] for both compressible flows and incompressible flows. In the practical applications, unstructured grids are always demanded for complex geometry involved simulations especially for FSI problems.

    View all citing articles on Scopus
    View full text