How GRIDGEN Got Its Groove

Conference proceedings. Technical papers. Software user manuals. Trade magazines. (Anyone interested in back issues of SGI’s Iris Universe?) Old versions of our own training notes and documentation.

I was cleaning off a bookshelf to make room for even more documents when I found a bit of history. (OK, using the word “history” might sound a bit overblown. Instead let’s call it ephemera.) Technical Proposal for Viscid/Inviscid Flow Zonal Methods, dated 21 April 1987, was the proposal we submitted in response to an Air Force solicitation for a multi-block, structured grid generator. As luck would have it, the proposal was selected for funding and we began 3 years of work that eventually produced GRIDGEN Version 6.

Viscid/Inviscid Flow Zonal Methods from 1987

By “we” I mean myself, John Steinbrenner, and Chris Fouts with liberal assistance and editing by George Howell. At the time we all worked together at General Dynamics, now Lockheed Martin Aeronautics in Fort Worth. John Steinbrenner led this effort, having completed his M.S. three years earlier with a thesis on adaptive grids. John had also spent the previous three years in the nascent CFD group developing gridding capability from scratch (i.e. GRIDGEN versions 1-5). As you know, John is a co-founder of Pointwise and the man behind most of its “science,” including T-Rex.

Chris Fouts also had an M.S. in aerospace engineering, but his strength was programming – expressing engineering and gridding ideas in code. He has almost an intuitive feel for how engineering, programming, and computers interact and is able get the best out of all of them. To this day he is perhaps the best software developer with whom I’ve ever worked and we’re privileged to have him with us at Pointwise. (He also is responsible for the term “connector,” something he’ll likely regret for the rest of his career.)

Back in 1987 I had only recently transferred to the CFD group having started my professional career in Propulsion Analysis. But while working on my M.S. I had the good fortune to take a grid generation course from Rich Hindman. He gave grid generation such a positive appeal that I knew that’s what I wanted to work on. (I know, the appeal of grid generation should be obvious to everyone.) When it became clear that we were writing the proposal I ingratiated myself onto the team.

The reason I’m writing this post is the image below, a fold-out chart that was inserted in the back of the proposal that readers could refer to while poring over the detailed text. (Sorry about the image quality. Click to see the full size version.)

GRIDGEN V6 was proposed as three separate programs – GRIDBLOCK, GRIDGEN2D, and GRIDGEN3D. (Click for full-size version.)

What you all know as Pointwise (or Gridgen) originally consisted of three separate programs: GRIDBLOCK for creating topology, GRIDGEN2D for surface gridding, and GRIDGEN3D for volume gridding. The proposal is chock full of vintage factoids and here are a few that I found notable.

  • The first code developed was the surface gridder, GRIDGEN2D. The early production versions of GRIDGEN2D ran on a VAX and displayed on a VT-240. However, GRIDGEN2D was the first program ported to the first Silicon Graphics workstation acquired by General Dynamics (a Model 2400 if memory serves me correctly). The SGI gave us a huge productivity boost because of its interactivity and from then on GRIDGEN was available exclusively on SGIs.
  • The proposal indicated a potential 6x improvement in gridding time for a 500,000 point grid for the F-16. (That grid had taken 6 person-months of effort back in 1986 and it was the basis for the first 3D CFD simulation of a complete fighter aircraft with operating inlet and nozzle.)
  • A 6x improvement sounds good, but to give you an idea of the state of the art at that time, a two-block grid (each block 61x41x21) for a blunt cone was said to take six hours but could be reduced to only 2 hours with GRIDGEN V6. (You could probably make that grid now in 2 minutes or 2 seconds with a script.)
  • One of the best aspects of the proposal was combining two “pre-preprocessing” codes into one. GRIDBLOCK was a sketchpad for creating block topology in 3D and GRIDBOUND was a unique interactive program for looking at the blocks in computational space (i.e. parallelepipeds) and defining how they all connect. GRIDBOUND was absorbed into GRIDBLOCK much to everyone’s benefit.
  • There was a fifth tool in the GRIDGEN system that was, for better or worse, given secondary status – GRIDVUE for examining the volume grids generated by GRIDGEN3D.
  • If you are keeping score, Steinbrenner worked primarily on GRIDGEN2D, Fouts on GRIDBLOCK, and I did GRIDGEN3D.

Those are a few of the highlights of how GRIDGEN Version 6 came into being. We were fortunate to be able to extend that into the (mysteriously unreleased) GRIDGEN Version 7, Version 8, and Version 9 followed by the founding of Pointwise and 17+ years later we have Pointwise (the software) Version 17.

If you want to learn more about how far we’ve come since GRIDGEN V6, click the link below to see what Pointwise V17 is all about.

This entry was posted in People & Places, Software and tagged , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s