TY - JOUR
T1 - Three-dimensional adaptive grid-embedding euler technique
AU - Davis, Roger L.
AU - Dannenhoffer, John F.
N1 - Funding Information:
This work was supported by the United Technologies Cor-poration under the Corporate Sponsored Research Program. The authors wish to thank Dan Dorney from the United Technologies Research Center for his work in grid generation and Bob Haimes from the Massachusetts Institute of Technology for supplying the VISUAL3 code that was used in the visualization of the numerical results.
PY - 1994/6
Y1 - 1994/6
N2 - A new three-dimensional adaptive-grid Euler procedure is presented that automatically detects high-gradient regions in the flow and locally subdivides the computational grid in these regions to provide a uniform, high level of accuracy over the entire domain. A tunable, semistructured data system is utilized that provides global, topological unstructured-grid flexibility along with the efficiency of a local, structured-grid system. In addition, this data structure allows for the flow solution algorithm to be executed on a wide variety of parallel/vector computing platforms. An explicit, time-marching, control volume procedure is used to integrate the Euler equations to steady state. In addition, a multiple-grid procedure is used throughout the embedded-grid regions as well as on subgrids coarser than the initial grid to accelerate convergence and properly propagate disturbance waves through refined-grid regions. Upon convergence, high flow gradient regions, where it is assumed that large truncation errors in the solution exist, are detected using a combination of directional refinement vectors that have large components in areas of these gradients. The local computational grid is directionally subdivided in these regions and the flow solution is reinitiated. Overall convergence occurs when a prespecified level of accuracy is reached. Solutions are presented that demonstrate the efficiency and accuracy of the present procedure.
AB - A new three-dimensional adaptive-grid Euler procedure is presented that automatically detects high-gradient regions in the flow and locally subdivides the computational grid in these regions to provide a uniform, high level of accuracy over the entire domain. A tunable, semistructured data system is utilized that provides global, topological unstructured-grid flexibility along with the efficiency of a local, structured-grid system. In addition, this data structure allows for the flow solution algorithm to be executed on a wide variety of parallel/vector computing platforms. An explicit, time-marching, control volume procedure is used to integrate the Euler equations to steady state. In addition, a multiple-grid procedure is used throughout the embedded-grid regions as well as on subgrids coarser than the initial grid to accelerate convergence and properly propagate disturbance waves through refined-grid regions. Upon convergence, high flow gradient regions, where it is assumed that large truncation errors in the solution exist, are detected using a combination of directional refinement vectors that have large components in areas of these gradients. The local computational grid is directionally subdivided in these regions and the flow solution is reinitiated. Overall convergence occurs when a prespecified level of accuracy is reached. Solutions are presented that demonstrate the efficiency and accuracy of the present procedure.
UR - http://www.scopus.com/inward/record.url?scp=0028450394&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0028450394&partnerID=8YFLogxK
U2 - 10.2514/3.12116
DO - 10.2514/3.12116
M3 - Article
AN - SCOPUS:0028450394
SN - 0001-1452
VL - 32
SP - 1167
EP - 1174
JO - AIAA journal
JF - AIAA journal
IS - 6
ER -