By Rick Matus, Vice President, Sales & Marketing
You read in John Chawner's article how he and John Steinbrenner made the decision to go out on their own and start Pointwise, but where do I fit in? I worked with both of them in the computational fluid dynamics (CFD) group at General Dynamics back in the 1980s. Not long after they left to join MDA Engineering, I left to join Fluent, Inc., in Lebanon, New Hampshire, where I was product manager for their first unstructured CFD code, RAMPANT, and later was manager of their automotive-aerospace-turbomachinery industry-oriented team.
In January 1995, I ran into John Chawner at the AIAA Aerospace Sciences meeting in Reno. He excitedly (for John) told me about the company he and John Steinbrenner had just started focused around Gridgen. As a past Gridgen user, I knew they had a good product and wished them well. I did not think anymore about it until a few months later when John phoned and asked if I would be interested in joining them to help out with marketing and sales. It took a lot of soul searching to decide that it was a good idea to leave an established company to join two guys just getting going, but in September 1995 I joined Pointwise. (Incidentally, that is when Pointwise, Inc., had to move out of John Chawner's spare bedroom and into a real office.)
I met much of the same skepticism John and John ran into about going into the mesh generation business. My Fluent colleagues seemed amazed anyone could eke out a living just working on gridding.
At the time Gridgen was being developed, 3D graphics workstations running versions of the UNIX operating system were just entering the market, and Silicon Graphics was leading the way. It was natural that Gridgen was originally developed on Silicon Graphics workstations using their proprietary graphics language, IRIS GL. This worked well and was used for all graphics functions in Gridgen, including drawing the buttons and text for the graphical user interface.
However, by the time Pointwise, Inc., was started, other computer companies were catching up to Silicon Graphics in terms of 3D graphics performance, and we needed to make Gridgen available on these other platforms.
Our first foray into other hardware was the IBM RS/6000 workstation, which had IRIS GL emulation. This was a good first step, and the IRIS GL emulation made the port a reasonable task, but we still needed a path to other workstations. Luckily for us, Silicon Graphics was seeing the same trends we were seeing and decided to turn their IRIS GL application programming interface into an open standard, OpenGL. Very quickly other workstation vendors adopted OpenGL, which gave us an easier (but still not trivial) path to other hardware platforms.
First, there are enough differences between IRIS GL and OpenGL that it was a significant effort to move Gridgen to OpenGL. Second, in the early days there were some differences between the OpenGL functions supported by different hardware vendors, which caused many problems in porting to different platforms. Third, there were the typical problems encountered with porting to different hardware and operating system because each workstation vendor had its own proprietary microprocessors and version of UNIX. Eventually Gridgen ran on 3D UNIX workstations from SGI, IBM, HP, DEC, and Sun.
If that were not enough, in the late 1990s it became apparent that Microsoft Windows workstations were catching up to UNIX workstations in processing and graphics speeds and significantly undercutting them in cost. Obviously, we needed a Windows version of Gridgen. Lacking significant in-house Windows development experience, we decide to hire an outside contractor, Pat Baker, to port Gridgen to Windows. Gridgen for Windows was released in November 1998. It was a very successful port for two reasons: the number of people using Gridgen on Windows grew rapidly and Pat later became manager of the Product Development Team at Pointwise.
Linux and Mac OS X versions of Gridgen followed in 2001 and 2004, respectively.
In the beginning, Gridgen only made multiple-block, structured grids. That was the state of the art in CFD when Gridgen development began in the 1980s, but unstructured grid-based CFD solvers were rapidly gaining popularity in the 1990s because of their speed and geometric flexibility. It was clear Gridgen needed to be extended to handle unstructured and hybrid grids.
Gridgen support for triangular and tetrahedral unstructured grids was released in July 1998. Users could select whether to make hexahedral or isotropic tetrahedral grids on a block-by-block basis to build up hybrid grids. In January 2000, we added the ability to extrude triangular surface grids into prism layers to make building high-quality boundary layer grids easier.
We continued to make incremental improvements to Gridgen's structured and unstructured mesh generation techniques over the next several years by adding features like the ability to elliptically smooth the interfaces between structured blocks, extrude prism and hexahedral layers together, and re-extrude a boundary layer grid with different grid control parameters. Gridgen's ability to handle more complex geometries was also increased by adding native CAD readers and solid modeling capability.
The next major step forward in meshing technology was the introduction of the T-Rex (anisotropic tetrahedral extrusion technique) in March 2007. T-Rex extrudes stretched, right-angle tetrahedra in a fully unstructured fashion so each layer is not required to have the same number of elements. This allows T-Rex to stop locally when it detects a cell quality problem, whether it is caused by colliding layers or some other geometric feature, but to keep advancing in other regions. When the extrusion is complete, the tetrahedra can be combined into prisms to reduce cell count and improve boundary layer cell quality.
Gridgen has been around for 30 years, and there are still people using it. (You old timers know who you are. Have you tried Pointwise?) We will save the story of our current meshing flagship, Pointwise, for a future issue.
You probably will not be surprised that the original versions of Gridgen were only available in the U.S., given that it originated in the U.S. aerospace industry with early funding from the U.S. Air Force and NASA. The technology transfer agreement between NASA and Pointwise that gave us rights to Gridgen spelled out very clearly how we had to modify the code to make it exportable. That was one of the earlier tasks worked on at Pointwise, Inc., and it resulted in the first internationally available version of Gridgen in January 1996.
Since then, we have set up a network of international distributors to support our software in other countries. In countries where we do not have a distributor, we provide customer support directly from our headquarters office in Fort Worth, Texas. As of this writing, we have 12 distributors worldwide. Our oldest distributor, VINAS Co., Ltd. in Japan, has been with us since October 1996.
In the early days, Gridgen was distributed by shipping each customer a user manual and a magnetic tape cartridge. Our early manuals were in three-ring binders with 8.5-inch-by-11-inch inserts. The magnetic tapes came in different sizes, were formatted differently for different operating systems, and took quite a few minutes to write. Back then, shipping a new software release involved a lot of manual labor: find the correct size and format tape for each customer, then write a copy of the correct hardware version of the software onto the tape, package the tape with the user manual and ship it away.
Things improved somewhat when we shifted to CDs and preprinted, compact manuals for software distribution. The same format CD worked for every platform and there was enough room on the CD to fit a version of the software for every supported hardware platform. However, the CD and user manual still had to packaged by hand and physically mailed. For major releases of Gridgen, even the developers and support engineers were conscripted to help with packaging.
Of course, now all our software distribution is done through downloads. This has distinct advantages for customers and is easier for us, too. We can make more frequent updates to the software and documentation, and you can download the new updates immediately.
In the next issue of The Connector, John Steinbrenner will write about technical aspects of the software. We hope you will join us when we celebrate our first 20 years in conjunction with our Pointwise User Group Meeting this October in Anaheim, California. We will outline our plans for the future and host a party so you can help us celebrate this milestone. The user group meeting will include information about our latest releases and demonstrations of what other Pointwise users are doing with the software.