Convergent Science

No User Grid Generation

The fundamental principle behind the uniqueness of CONVERGE™ is that the user does not directly supply a grid as an input. Instead, the user supplies a triangulated surface and a series of guidelines from which CONVERGE™ creates the grid at run-time. For simulations with moving boundaries or changing embedding, the grid is recreated at each time step.

There are several ways the user controls the nature of the grid during the simulation. These are:

  • Base dx, dy, and dz. The user specifies the cell size of the base grid. All refinement that takes place in the grid using any of the following methods will be powers of two refinement of the base grid.
  • Fixed Embedding. The user can specify where and when refinement is desired by using fixed embedding. The refinement regions can be specified to exist at certain locations in the grid, or along specified boundaries. This can be useful for simulations where some part of the domain requires additional resolution, e.g. flow through valves or near nozzles.
  • Adaptive Mesh Refinement (AMR). The code can also adjust the mesh during the simulation to better resolve flow variables of interest. The user can specify which variables are critical (e.g. velocity for a flow case, temperature for a combustion case) and CONVERGE™ will modify the grid to enhance accuracy.
  • Grid Scaling. The user can also change the base grid size during the simulation. This can be helpful in reducing run-times if part of the simulation is not as critical and can be run coarser.

Each of these four methods for controlling the nature of the grid during the simulation are handled through simple input files and are not part of a user grid generation procedure. The setup time in preparing a simulation involves preparing the surface so that it is clean, closed, and has the boundaries properly marked.

CONVERGE™ represents the surface as a series of connected triangles. The preprocessor can read in an STL file, a triangulated surface file format output from any CAD package. The preprocessor has some functionality to clean up and repair surface files that may not be properly closed or singly defined. Once the surface is repaired, if necessary, the boundaries are marked and the surface is ready to run.

The image below shows the surface of a small-bore Diesel engine geometry that has two helical intake ports.

lighted-1

 

The following image is for the same geometry zoomed in to show the triangulation at the helical portion of the intake ports. The exhaust is represented as stub ports in the foreground of the image.

 

surface-1

 

 

By cutting away the surface to look inside, an intake valve at maximum lift can be seen.

 

valve-1

 

 The boundaries must be uniquely identified with an identifier that corresponds with an entry in the boundary input file. The following image shows how boundaries have been separated so they can be assigned different boundary conditions. CONVERGE™ allows local embedding on individual boundaries, thus making it useful to separate out boundaries that may actually have the same boundary condition. The image below has the valve defined as two different boundary types, colored blue and yellow. The reason for doing this is so that additional embedding can be specified on the yellow boundary when the valve is at low lift to help resolve the flow in the valve-seat area.

boundaries-1

 

 Once the surface is cleaned and the boundaries are marked, the surface is ready to go!

 

grid-sandia2

 

Share This Page