The meeting was called to order at 2:00 Eastern time. There were 14 attendees, listed below.
The minutes of the 24 August meeting were reviewed and approved as posted on the web site.
Chris Rumsey said the draft version of the User's Guide is almost ready. This includes some sample code, which he will also release. The software will be released both with the User's Guide, and under the Utilities web site. Comments are solicited when the Draft User's Guide is released.
David Edwards, reporting for Steve Legensky, stated that the paper
abstract for a CGNS applications paper should be ready by 31
October. A survey to collect information for the paper was sent to
CGNS users with responses due by 7 November. This data will be used
for the paper, and responses by all parties is requested. We
responded to a request from the AIAA Aerospace America year in
review - Interactive Computer Graphics
for a contribution to an article.
Steve Feldman suggested contacting the 'owner' of CFD-Online to see
if we could or should add a CGNS forum to that popular site. There
being general support for the idea, Steve will contact them.
A discussion of the need for summary information
followed. After some discussion it became clear there is a
need for summary information that describes the content of a
particular file. A consensus seemed to form around the idea
of a utility that would scan the file for its contents, and
another utility that would use this information to classify
the information. Given the classification, applications could
make assumptions about the file and avoid messy checks for
data availability. There is a need for a classification system
(e.g. solution with conservation variables, non-dimensional
structured grid, heirarchical structures with at most 2
children,?). There were no volunteers to develop this system,
but it was agreed that we should continue to pursue the idea,
the classification system and the utilities.
The January meeting is tenatively scheduled for Monday night at Reno.
|Chris Rumsey||NASA Langley|
|Ed Ascoli (for Armen Darian)||Boeing|
|Charlie Towne||NASA Glenn|
|Diane Poirier||ICEM CFD Engineering|
|Francis Enomoto||NASA Ames|
|David Edwards||Intelligent Light|
|Doug McCarthy||Boeing Commercial|
|Todd Michal||Boeing Phantom Works|
Outstanding developments in Y2K included: collaborative systems for better utilization of graphics hardware, co-processing systems for visualizing large-scale simulations, an emerging ISO standard for CFD data exchange, and commercialization of automatic feature extraction tools.
Many aerospace companies and government laboratories have established Visualization Lab's that contain specialized SGI graphics workstations to generate hi-resolution computer graphics images. These labs have been poorly utilized in the past because engineers often had to travel great distances to use the graphics hardware. This situation is rapidly changing as vendors adapt their software to utilize the Internet and corporate intranets. For example, EnLiten is a new product from Computational Engineering International (CEI) that allows high-end visualizations to be easily and cost-effectively shared throughout an enterprise. EnLiten supports 3D stereo viewing and a full range of computing and display systems, from notebook computers to workstations to semi-immersive environments such as the NCSA ImmersaDesk. It plugs into well-known browsers and viewers such as MS Internet Explorer, MS PowerPoint, and Netscape.
Parallel computing is rapidly emerging as the method of choice for visualizing large-scale CFD solutions. Intelligent Light has built upon MIT's pV3 client-server infrastructure with the popular Fieldview software. The result is a fully supported, robust commercial "graphics client" for the pV3 system. pV3 is used throughout the world for visualization co-processing, that is, concurrent simulation and interactive visualization. Intelligent Light's implementation supports systems ranging from laptop Windows PC's through UNIX clusters and supercomputers and enables enterprise-wide visualization.
Every CFD code that enters the marketplace generates datasets with a slightly different format than their competitor. Products that utilize these datasets, i.e., pre- and post-processing software, have to support approximately 50 different file formats at the present time. The lack of standardization forces vendors to spend an inordinate amount of time implementing and maintaining file exchange software. To combat this software crisis, a consortium of CFD users and vendors have adopted a formal standard for exchanging CFD data called the CFD General Notation System (CGNS). The CGNS system facilitates the exchange of data between sites and applications, and stabilizes the archiving of aerodynamic data. An Application Program Interface, implemented by ICEM CFD Engineering, is platform independent and simplifies the installation of CGNS in C, C++, and Fortran applications. CGNS was conceived back in 1994 as a means to promote "plug-and-play" CFD via a file specification standard and supporting code that could be distributed freely. In 2000, the CGNS Steering Committee affiliated with AIAA as a subcommittee, made progress in proposing CGNS as an ISO standard, and added capabilities to the standard to handle increasingly specialized and complex meshes. Currently, there are over 20 CFD applications that support or have translators for CGNS. Further information about this effort can be found at http://www.cgns.org/.
Automatic feature extraction tools have transitioned from research
institutions to commercial vendors of post-processing software in the
past six months. The open-source FX (Feature eXtraction) library
released by MIT in June 2000, has been adopted by two leading suppliers
of CFD visualization software; Intelligent Light (Fieldview) and CEI
(Ensight). Feature extraction tools automatically deduce the location,
shape, and strength of specific features without human intervention, and
do so in substantially less time than their human counterpart. This is a
different paradigm than typical graphical analysis methods where the
engineer has to do the work, interpreting the data from imagery and
using their insight or experience to infer and find specific patterns.
Feature extraction algorithms are programmed with domain-specific
knowledge, so they do not require a-priori knowledge of places to look
for interesting behavior. The latest feature extraction tools, developed
at MIT and NASA Ames, automatically calculate vortex cores, separation
and re-attachment lines, and shock surfaces. These tools will be
available in Fieldview 7 and Ensight Gold in the final quarter of Y2K.
Figure 1 [sh_side35.jpg] (to be supplied by CEI)
Figure 2 [f16.jpg] An F16 solution from the USAF Cobalt group. Vortex cores are colored by strength and were used to guide the creation of streamlines. (Image courtesy of Intelligent Light)
Dr. David N. Kenwright
Principal Research Scientist
Department of Aeronautics and Astronautics
Massachusetts Institute of Technology
Bldg 37-462, 77 Vassar Street
Cambridge, MA 02139
Tel. (617) 253-3505
Fax. (617) 258-5143