FEA Information Engineering Journal - April 2012

The present work will introduce some of the resent developments in the Incompressible CFD (ICFD) solver currently under development in LS-DYNA. The main feature of this solver is its ability to couple with any solid model to perform Fluid-Structure interaction (FSI) analysis. Highly non-linear behavior is supported by using automatic re-meshing strategies to maintain element quality within acceptable limits. In this work we will introduce the additional features for conjugate heat transfer, turbulence model, biphasic flow, some new feature in terms of mesh generation like boundary layer meshing and MPP. Introduction Incompressible flows cover a vast number of engineering problems ranging from car aerodynamics to arterial flows and parachute simulation. As a rule of thumb a flow may be considered incompressible if the Mach number presented in any part of the domain is not larger than 0.3. In LS-DYNA there are other two options to do CFD depending upon the kind of problem. The CESE solver is highly accurate CFD solver for compressible fluids. The ALE solver has support for both compressible and incompressible and it is a good option for highly transient problems. Due to the requirement of industrial applications a number of new features have been added to the ICFD solver prior to the release version. They will be briefly described bellow. Turbulence Models The majority of the problems that involve real life applications fall in the category of high Reynolds number problems, where turbulent effects play an important role. Since full resolution of the problem is not possible due to resource limitations robust turbulent models are critical to provide realistic results. In ICFD turbulence is and will continue being a work in progress due to the continue evolution of the field. At the moment three classical approaches have been incorporated namely K-e model, Smagorinsky LES model and a variational multiscale model which is still part of research work. The user will be able to modify some parameters of the models from the input deck to adjust it to some particular problem. Fluid / FSI 11 International LS-DYNA Users Conference 6-2 Figure 1: the image shows a flow past a flat plate in an angle on the left, simulating the wind effect on a solar panel. The image on the right is an FSI problem of a opening valve and the flow is shown and maximum aperture when the flow rate peaks. In both cases the turbulence was modeled using a LES approach. Conjugate Heat Transfer Problems involving heat conduction in solid material have been part of LS-DYNA for a long time. Now we can couple the solid material to the fluid in an implicit way to solve conjugate heat transfer problems. This kind of coupling will also allow to solve thermo mechanical problems, providing a two-way coupling between the fluid and the structure to solve fluid-structure interaction problems. Figure 2: in this problem a cooling serpentine is used to cool down a solid part (not shown in the picture). On the left image the streamlines of the flow are shown after the flow has reached steady state. The image on the right shows the flow temperature profile at steady state. Multi-Phase Flow One of the new additions to the solver is the possibility to approximate two phase flow while preserving a sharp interface even for under resolved problems. This approach will preserve subgrid features of the interface in areas where the finite element mesh is under-resolved and restore this features to the solver later in areas of the mesh with better resolution. 11 International LS-DYNA Users Conference Fluid / FSI 6-3 Figure 3: the image series show a bubble dropping into a lighter fluid. The mesh on the surface gives an idea of the mesh resolution used. In despite of the coarse mesh the interface is well captured. Boundary Layer Meshing In terms of volume meshing the solver supports automatic mesh generation which is perform internally at run time. The mesh may also be modified by an error estimator to do adaptive mesh refinement. Recently a boundary layer mesher was added to the solver to further resolve the areas of the domain close to the walls to provide a better approximation of the shear stresses. In problems involving drag calculation boundary layer meshing is a requirement. This kind of problems involve aerodynamics of bluff bodies like the one shown in the picture bellow. Figure 4: this is a classical problem to benchmark drag around bluff bodies. It is called the Ahmed body problem. On the left a detail of the mesh is show. Image A shows the full domain with mesh refinement in the wake of the body. B and C are a closer look at the mesh around the body and C shows the boundary layer mesh next to the body wall. Fluid / FSI 11 International LS-DYNA Users Conference 6-4 Parallel Computing To improve the productivity of the solver and to satisfy the demand for high performance computing all the features of the solver have been implemented in parallel. The parallel CFD solver can be coupled to the parallel solid mechanics solver to do parallel FSI. In the same way the coupling may be done in parallel with the thermal solver to do conjugate heat transfer. 11 International LS-DYNA Users Conference Fluid / FSI 6-19 How to Use the New CESE Compressible Fluid Solver in LS-DYNA Zeng-Chan Zhang Livermore Software Technology Corporation 7374 Las Positas Road Livermore, CA 94551 Abstract This new solver is based on the conservation element and solution element (CESE) method . The CESE method is a novel numerical method for solving conservation laws, and it has many nontraditional features, such as: spacetime conservation; high accuracy (2 order for both flow variables and their spatial derivatives); novel shockcapturing strategy; both strong shocks and small disturbances can be handled very well simultaneously, etc. Because of these advantages, this CESE solver is a good choice for high-speed compressible flows with complex shocks and acoustic (noise) problems (near field).This new solver is based on the conservation element and solution element (CESE) method . The CESE method is a novel numerical method for solving conservation laws, and it has many nontraditional features, such as: spacetime conservation; high accuracy (2 order for both flow variables and their spatial derivatives); novel shockcapturing strategy; both strong shocks and small disturbances can be handled very well simultaneously, etc. Because of these advantages, this CESE solver is a good choice for high-speed compressible flows with complex shocks and acoustic (noise) problems (near field). The solver has also been used to solve fluid/structure interaction (FSI) problems. For these problems, the fluid solver is based in an Eulerian frame while the structure solver is a Lagrangian frame. Their meshes are independent of each other, and the structural boundaries (fluid-structure interfaces) are tracked by the fluid solver automatically. The fluid solver gets the displacements and velocity of the interfaces from the structural solver and feeds back the fluid pressures (forces). Current status Currently, both serial & MPP solvers are available for this compressible fluid & FSI solver (in the ls980 β-version). The fluid mesh can be made up of hexahedra, wedges, tetrahedra, or a mixture of these elements, while the structural mesh can be made up of shells (thin) or solid volume elements for CESE FSI problems. Input deck setup: In this presentation, we will talk about how to use this new solver, this will include: • How to use LS-prepost to create a CESE input deck & display the final results • How to setup an input deck, including ⎯ setting some control parameters in the CESE method ⎯ initial flow field setup ⎯ boundary condition (BC) choice at each boundary • some example input decks In addition, a couple of new features will be introduced, followed by some remarks. The limitations of this solver will be pointed out too. Some examples With the new release of the ls980g beta version, we will provide ten examples (fluid & FSI). In each example, we will have problem description, input deck, numerical results and comparisons with analytical or experimental results (where available). Here we show two of them. Fluid / FSI 11 International LS-DYNA Users Conference 6-20 Fig.1 Sod’s 1-D shock tube problem: the comparison of numerical results with the analytical solutions for pressure, density and velocity at t=0.2.