Overset Grids in STAR-CCM+: Methodology, Applications and Future Developments Eberhard Schreck and Milovan Peri ć CD-adapco
Introduction • Overset grids: History Advantages and challenges • Overset grids in STAR-CCM+: Methodology User interface Tips and tricks • Examples of application • Future developments
Overset Grids History • Overset grids were used already 30 years ago… • The main motivation has been to use multiple regular grid blocks to handle complex geometry… • In October 2012, the 11 th Overset Grid Symposium was held in Dayton (most presentations available at Symp. Website) : The multi-block approach still dominating (tens to hundreds of grid blocks, up to 10% of grid points involved in interpolation…); Unstructured grids being used by few groups, resulting in a smaller number of grid blocks; Different numerics, grid coupling (interpolation) and hole- cutting algorithms – same problems (“orphan” nodes)…
Advantages of Overset Grids • Easier to perform and automate parametric studies: – With a single set of grids, many different configurations can be computed; – Grid quality not affected by changing position/orientation of bodies; – Boundary conditions easier to set… • Easier to handle relative motion of bodies: – Arbitrary motion can be handled; – Paths can cross; – Tangential motion at close proximity can be handled…
Challenges with Overset Grids • Complex logic and coding is required for an automatic handling of arbitrary body motion and multiple overset grids… • Situations can arise where coupling of predefined overset grids is not possible (“orphan” cells…)… • Parallelization and load balancing are challenging…
Overset Grids Method in STAR-CCM+, I Control volumes are labelled as: Active cells, or Passive cells. In active cells, regular discretized equations are solved. In passive cells, no equation is solved – they are temporarily or permanently de-activated. Active cells along interface to passive cells refer to donor cells at another grid instead of the passive neighbours on the same grid... The first layer of passive cells next to active cells are called acceptor cells...
Overset Grids Method in STAR-CCM+, II Currently, triangular (2D) or tetrahedral (3D) interpolation elements are used, with either distance-weighted or linear interpolation... Other (higher- order) interpolations will come… Background grid N 1 , N 2 , N 3 – Neighbors from the same grid; N 4 , N 5 , N 6 – Neighbors from the overlapping grid. Overset grid
Overset Grids Method in STAR-CCM+, III No explicit interpolation of solution is performed… Solution is computed on all grids simultaneously – grids are implicitly coupled through the linear equation system matrix...
Overset Grids Method in STAR-CCM+, IV Different interpolation functions can be used to express values at acceptor cells via values at donor cells (different interpolation elements)… Interpolation elements are not unique – when grids move, continuity is important… Donor cells must be active cells. The change of cell status is controlled by the solver and happens automatically. The user can visualize the cell status as a scalar field (this can help in case of problems – mostly due to inadequate grids)...
Overset Grids Method in STAR-CCM+, V Overset grids usually involve: One background mesh, adapted to environment; One or more overset grids attached to bodies, overlapping the background mesh and/or each other. Each grid represents a separate Region in STAR-CCM+ terminology... Both background and overset mesh(es) can be generated (or imported) in the usual way, region by region…
Overset Grids Method in STAR-CCM+, VI Each grid (background and overset) can move according to one of the standard motion models available in STAR- CCM+… Each grid can also deform (e.g. in a coupled fluid- structure interaction simulation) using any available morphing technique… Overset grids can fall out of solution domain (cut-out by boundary surface). Overset grids can overlap each other.
Working with Overset Grids, I • STAR-CCM+ infrastructure for interfaces has been extended – overset grids are another type of volume interface… • New intersector-module was added to STAR-CCM+ to handle: Cell status (“hole - cutting” algorithms); Searching for donors to each acceptor cell; Definition of interpolation factors, etc… • The solver is almost unaffected – almost all models can be used (coupled and segregated solver, VOF, Lagrangian and Eulerian multiphase flows etc.)…
Working with Overset Grids, II • No compromises on usability: Any grid type can be used; Most physics models can be applied; All motion models can be used; Processing pipeline (meshing, solving, analysing) is unaffected; Only two additional set-up steps: New region interface (with interface options) • New boundary condition •
Working with Overset Grids, III Set-up of overset grid computation of flow around a pitching foil in a channel: one background grid for the channel and one overset grid for the region around foil. Background region Overset region Overset interface for regions “Background” and “Over”
Working with Overset Grids, IV Front and back planes are symmetry planes. The overset region has one boundary that is fully submerged within background region... Background region Overset region Overset boundary Overset grid surface has boundary type “ OversetMesh ”
Working with Overset Grids, V “Volume Mesh Representation” includes active cells – used to plot results.. . Active cells in overset grid Active cells in background grid
Working with Overset Grids, VI Acceptor cells (value “ - 2”) Active cells (value “0”) Background Passive cells (value “1”) Checking “Overlap Cell Status” (scalar field): acceptor cells must separate active and passive cells – direct contact is not allowed...
Working with Overset Grids, VII Acceptor cells (value “ - 2”) Over Active cells (value “0”) Checking “Overlap Cell Status” (scalar field): the overset grid here contains only active and acceptor cells...
Tips and Tricks… In the overlapping zone, cells should be of comparable size in both meshes (recommendation) : Interpolation errors in the coupling equation should be of the same order as when computing convective and diffusive fluxes (interpolation over half a cell); The coarser of the two coupled meshes determines the error level… Between two body walls, at least 4 cells on both background and overset grid are needed to couple them (requirement) . The overset grid should not move more than one cell per time step in the overlapping zone (recommendation) .
Visualization - Isolines Pressure contours with lines: small imperfections (two lines visible within overlap zone) visible only at few locations – most contours are almost perfectly continuous (grid from previous slides)
Convergence of Iterations Residuals history for a laminar flow around an object… Implicit coupling of grids allows convergence to round- off level of residuals…
Overlap of Overset Regions Example of overset grids overlapping each other.
Overset + Morphing, FSI Example of combination of overset grids and morphing when simulating large deformation of structures.
Overset-Lagrangian Example of overset grids in combination with Lagrangian multiphase flow model (overset grids move and fall partly outside solution domain; particles are not affected by internal grid motion).
Examples of Application • Parametric studies (varying angle of attack) • Bodies moving relative to each other • Engineering problems that can be solved with overset grids easier than otherwise…
Application to Parametric Studies, I Flow around a body at different angles of attack A horizontal section through both grids (only active cells are shown). Total number of cells: ca. 1 million Vertical section through the two grids (only active cells are shown). Same grids and boundary conditions – many positions (easy to automate).
Application to Parametric Studies, II -30° -15° 0° 15° 30° Velocity distribution in a section parallel to bottom wall for different angles of attack
Application to Parametric Studies, III Residual history from the computation of flow around a vehicle in a wind tunnel at different angles of attack: time step 1000 s, rotation 15° per time step, standard k- ε turbulence model, under-relaxation 0.9/0.1/0.9 for velocities/pressure/turbulence, wind speed 40 m/s…
Application to Parametric Studies, IV History of computed forces from the computation of flow around a vehicle in a wind tunnel at different angles of attack (since the time step is very large, steady-state solutions are obtained).
Application to Parametric Studies, V Simulation of motion of a container ship in Stokes waves propagating from right to left: initial vessel orientation 30° (upper) and - 30° (lower) relative to the direction of wave propagation. Single set of grids, same boundary conditions, different vessel orientations – easy to automate…
Recommend
More recommend
