TY - GEN
T1 - Automated creation of 3-D overset grids directly from solid models
AU - Dannenhoffer, John F.
AU - Haimesy, Robert
PY - 2011
Y1 - 2011
N2 - One of the challenges when generating overset grids is the decomposition of a configuration into the components around which the various grids will be generated. This step frequently requires that the analyst view the configuration and graphically select the surfaces that will form the boundaries of each grid. Fortunately, many configurations are now generated through solid modelling systems, in which a configuration is generated by combining various primitives into the final configuration. This paper describes a technique for automatically generating overset grids directly from the solid modeling primitives. The basic idea here is to generate a structured grid for each primitive, which is fairly easy since most solid modellers use simple shapes for their primitives. Once all the grids have been generated, then the three Boolean operators - union, intersect, and difference - are used in the same way as they were used in generating the solid model. The only information needed here is the feature tree that is already stored in the CAD model. Four main challenges emerge from such an approach. First, conventional overset methods only "union" grids; in the current approach, the hole-cutting and interpolation schemes must be modified to allow "intersection" and "difference" operators as well. Second, the selections of a grid hierarchy must be automatic, in the sense that the system must automatically decide which other grid to use in the interpolation operation; here the scheme uses a combination of grid density and feature tree information to make the choices. Third, many of the grids that were created end up not participating in any useful computations; techniques for pruning these grids are also included. And finally, in order to adequately capture the intersections of the various primitives, it has been found to be necessary to automatically generate collar grids; the latter is done with a special hyperbolic grid generator. The new technique is applied to a complex 3-D configuration that contains over 50 features. A flow-solver-ready set of 229 overset grids with over nine million grid points is generated automatically in less than seven minutes on a laptop computer. Through a set of configurations that are generated by defeaturing the original in various ways, it is shown that the technique scales approximately linearly with the number of features.
AB - One of the challenges when generating overset grids is the decomposition of a configuration into the components around which the various grids will be generated. This step frequently requires that the analyst view the configuration and graphically select the surfaces that will form the boundaries of each grid. Fortunately, many configurations are now generated through solid modelling systems, in which a configuration is generated by combining various primitives into the final configuration. This paper describes a technique for automatically generating overset grids directly from the solid modeling primitives. The basic idea here is to generate a structured grid for each primitive, which is fairly easy since most solid modellers use simple shapes for their primitives. Once all the grids have been generated, then the three Boolean operators - union, intersect, and difference - are used in the same way as they were used in generating the solid model. The only information needed here is the feature tree that is already stored in the CAD model. Four main challenges emerge from such an approach. First, conventional overset methods only "union" grids; in the current approach, the hole-cutting and interpolation schemes must be modified to allow "intersection" and "difference" operators as well. Second, the selections of a grid hierarchy must be automatic, in the sense that the system must automatically decide which other grid to use in the interpolation operation; here the scheme uses a combination of grid density and feature tree information to make the choices. Third, many of the grids that were created end up not participating in any useful computations; techniques for pruning these grids are also included. And finally, in order to adequately capture the intersections of the various primitives, it has been found to be necessary to automatically generate collar grids; the latter is done with a special hyperbolic grid generator. The new technique is applied to a complex 3-D configuration that contains over 50 features. A flow-solver-ready set of 229 overset grids with over nine million grid points is generated automatically in less than seven minutes on a laptop computer. Through a set of configurations that are generated by defeaturing the original in various ways, it is shown that the technique scales approximately linearly with the number of features.
UR - http://www.scopus.com/inward/record.url?scp=85088755871&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85088755871&partnerID=8YFLogxK
U2 - 10.2514/6.2011-3540
DO - 10.2514/6.2011-3540
M3 - Conference contribution
AN - SCOPUS:85088755871
SN - 9781624101489
T3 - 20th AIAA Computational Fluid Dynamics Conference 2011
BT - 20th AIAA Computational Fluid Dynamics Conference 2011
PB - American Institute of Aeronautics and Astronautics Inc.
T2 - 20th AIAA Computational Fluid Dynamics Conference 2011
Y2 - 27 June 2011 through 30 June 2011
ER -