News Item: Two U.S. Postage Stamps, issued in 2006, commemorate X-Planes, a series of experimental vehicles involving NASA, the U.S. military, and aerospace companies. The computer-generated images shown on the stamps were produced from results using the CFL3D code (computed by the NASA Langley Geometry Laboratory).
CONTENTS
These pages provide the documentation for the newest version of CFL3D, Version 6.5. CFL3D is a long-standing Navier-Stokes CFD code developed at NASA Langley Research Center for solving 2-D or 3-D flows on structured grids. The code has been in existence since the late 1980's, is very reliable and stable, and has many capabilities and options. We invite you to start off by perusing the Test Cases and Image Gallery, by clicking on the buttons to the left.
Version 6.5 (released on May 29, 2007) represents only a minor upgrade over Version 6.4 (released on October 18, 2005). Most of the changes reflect minor bug fixes or new capabilities, particularly with respect to several new keyword input capabilities for input/output control and aeroelasticity. See New Features for details on these. Furthermore, details on all of the differences and updates made to the code since 2000 can be found in Updates.
There is no formal User's manual available for Version 6, but there is a manual for Version 5 and virtually all of this documentation is also applicable to Version 6. There are many options in V6 that were not available in V5, and a few that are handled slightly differently. Usage of these new options is described in New Features. You may download a pdf file of the Version 5 User's Manual (either by chapter or in total) by clicking on the "Version 5 Manual" selection in the index to the left.
MAJOR ENHANCEMENTS IN VERSION 6
GENERAL FEATURES OF THE CFL3D FLOW SOLVER
Version 6 is currently available for general distribution to a "U.S. person" within the United States. A "U.S. person" is a citizen of United States, a lawful permanent resident alien of the U.S., or someone in the U.S. as a protected political asylee or under amnesty. The word "person" includes U.S. organizations and entities, such as companies or universities - see 22 CFR §120.15 for the full legal definition.
Before requesting the code, you should consider the following;
If you would like a copy of the code, please
follow the request procedure below:
Note: If you are already an approved user of the Version 6 code, and would like to update to the latest point release, only step 1 above is necessary.
Conditions of use:
CFL3Dv6 can be run in either the standard (sequential) mode or in distributed/ parallel mode. Whereas in the sequential mode all blocks at all multigrid levels reside on one processor, in the distributed/parallel mode each processor will have one or more grid blocks residing on it. The coarser levels required for multigrid reside on the same processor as the finer parent block. The computations for a particular grid block are not currently split among multiple processors. As a result, single-block grids cannot be run in parallel. However, a Block Splitter tool is provided to facilitate conversion of an existing grid and input file to run in a parallel environment.
The parallel code utilizes the Message Passing Interface (MPI). The required MPI libraries may be "native" ones that exist on parallel processors such as the Origin 2000, the SP2 or the T3E, or, for workstation clusters, the portable MPI implementation, MPICH.
The code displays reasonably good scalability with increasing processor count for load balanced cases. Scaling tests have been conducted on SGI Origin 2000, CRAY T3E, and "Beowulf" Pentium systems.
Initial development of CFL3dv6 was done under contract to Computer Sciences Corporation, with Jerry Mall carrying out the initial port of Version 5.0 to MPI. The basic structure of the code was modeled after the earlier parallel version, CFL3Dv4.1hp. After the code was passed back to NASA, additional modifications were made to implement the full range of features available in the standard release of Version 5.0, as well as to add the additional features that are described within this document.
Along the way P. Sundaram and collegues at Boeing Phantom Works-Long Beach made significant contributions by aggressive application of the parallel code. James Hager of BPW-LB provided a number of helpful suggestions for improvement. Thomas Faulkner of MRJ Technology Solutions has done extensive benchmarking of the earlier CFL3DV4.1hp on the NAS Origin 2000 and NAS Whitney cluster and provided valuable feedback on code performance.
GRID GENERATORS AND POST-PROCESSORS
The following links are to companies or organizations that provide grid generation and/or postprocessing visualization software, neither of which are included with CFL3D. This list does not constitute an endorsement. Also, it is not an exhaustive list; there are likely other sources. For input, CFL3D can work with grids in any one of 3 formats: CGNS, PLOT3D, or CFL3D. For output, the code can create results in two widely-used formats: CGNS and/or PLOT3D. (These links will open new windows.)
(These links will open new windows.)
General issues & Parallel/MPI:
Dr. Robert T. Biedron
phone: (757) 864-2156
FAX: (757) 864-8816
e-mail: r.t.biedron@larc.nasa.gov
General issues & Turbulence models:
Dr. Christopher L. Rumsey
phone: (757) 864-2165
FAX: (757) 864-8816
e-mail: c.l.rumsey@larc.nasa.gov
Aeroelastics/Mesh deformation:
Dr. Robert E. Bartels
phone: (757) 864-2813
FAX: (757) 864-8678
e-mail: r.e.bartels@larc.nasa.gov
Page Curator and NASA Responsible Official: Christopher L. Rumsey
Last Updated: May 29, 2007
