Pro/User 2000 Conference
Orlando, Florida
June 20, 2000

This newsletter brought to your desktop by Sun Microsystems and Dell Computer

Peter Nurkse
Sun Microsystems
Peter.nurkse@sun.com

Subjects for this newsletter:

PTC overview

Dick Harrison this year shared his overview slot with Jon Stevenson, VP of the newly formed MDA (MCAD) business unit, but promised to come back tomorrow for the management panel session for more detailed questions and answers.

Dick said that PTC has over the last 2 years been making the transition from a one product company, based on Pro/E, to a two product company, based on Pro/E and PDM (product data management). He said that during this time PTC management did "take our eye off the MCAD ball a little bit", while building up the new PDM business based on Windchill. He further said that this situation was "unfair" to some PTC executives, like Jim Baum, because they were being asked to allocate resources between both Pro/E and PDM product lines. Dick said that he and others in the management team were guilty.

Dick said that the transition in product lines is now complete, after these two years, and that the company should now be focused equally on both product lines, MCAD and PDM. He expects that PTC should now be able to eliminate the self-imposed stress and pressure of the transition, and introduce new applications calmly and comfortably---which should be more comfortable for PTC customers too. He said that PTC actions wouldn't be dictated by the stock price and Wall Street, and the audience reacted with widespread applause.

PTC now has 3 product divisions:
  • Pro/E and MDA (Jim Stevenson), MCAD business, probably 95% of current revenues
  • Windchill (Jim Baum), the Windchill product, the "Christmas present" (Dick's description at the '98 conference) acquired with Computervision
  • NetMarkets (James Heppelman), a "lighter weight" Windchill, for Internet based exchanges and markets, both outside and inside firewalls
Each of these VPs has the entire development (R&D), marketing, and service organizations for their respective products. Sales is separate, and sales is now two different divisions: Primary Account sales and resellers (Paul Cunningham), and Major Account sales (Trent Brown). So, your sales rep will still sell the entire product line.

MCAD overview

Jon Stevenson spoke next, the new Pro/E and MDA VP. That's his title, but like Dick before him Jon said 'MCAD' instead of saying 'MDA'. So, should be OK for users too to say the good old MCAD instead of the less familiar MDA (mechanical design automation).

Jon said that his MCAD business unit has 1900 people, including 870 in programming (probably that includes about 250 or 300 programmers at the old Computervision development center in Pune, India). His total R&D budget is over $90 million.

Jon emphasized several times that Pro/E is fundamentally "a feature based solid modeling system", and that the competition is still trying to catch up. He pointed out that Euclid and Applicon Bravo and Sherpa have all been acquired and then discontinued recently, but that PTC has continued to support all their acquisitions.

PTC will be opening an Online Store on the Web next month (July), to make it easy for anyone to evaluate PTC products. Ease of use is going to be further developed (certainly seems that ease of use is a constant struggle, it just never is perfected). And another major initiative is 'Multi-CAD Openness', recognizing requirements among all customers (regardless of size) for heterogenous data formats and associativity. Last major innovation is Webcentric collaboration (with Windchill).

Jon listed the following six specific new capabilities for Pro/E in the coming year:
  • Large Assembly Modeling (seems another constant struggle, like ease of use). PTC is working with Hyundai shipyards in Korea on handling large amounts of data and (equally important) the associated design processes). Hyundai is an old Computervision customer, and Computervision was the most popular CAD package among shipbuilders worldwide, so this shipbuilding collaboration might be another benefit of the CV purchase.
  • Complex Routed Systems. By the end of this year (2000i3), there should be a 2D routing and schematic package based on Medusa, and a data driven driver for routed systems. Dick didn't mention Medusa as a Christmas present in the CV purchase, but perhaps Medusa has shown more capabilities and more dedicated users than PTC expected at the time of the CV purchase. At BMW in Germany, Medusa has been described as 'the Queen of CAD systems' (that's dedication, even enthusiasm).
  • Surface Modeling. ID and Styling (the CDRS package, also known as Pro/Designer) should be integrated within Pro/E this year (2000i3, again), with Class A automotive surfacing (ICEM Surf) later.
  • Behavioral Modeling. Continuation of development on this front.
  • Manufacturing. Expert Machinist has been out for about a year, will now be followed by an Expert Toolmaker package, and a tools database for managing tool inventories.
  • Simulation. Pro/Mechanica is going to include radiation analysis next.

Large Assemblies

Leo Greene from PTC gave a presentation on large assemblies. But some Computervision users at the conference remarked that his example, a race car transmission, was far smaller than the assemblies which they were designing in production mode on CV ten years ago, with ten years old hardware. Perhaps if PTC can incorporate some of the large assembly expertise from CV, then Dick Harrison might be able to find still another Christmas present in that CV purchase. Large assemblies haven't just been a constant theme at Pro/E user conferences for the last 10 years, they've also been a major reason why the largest automotive and aerospace companies have chosen other CAD systems over Pro/E in competitive benchmarks.

Leo said for example that "to put everything on the screen is very limited value", and that you usually want to step back, for less detail, and "you may never need to bring up the whole assembly". Seems that didn't much impress some users of other CAD packages, who're used to bringing up much larger assemblies than the transmission in any level of detail they want. PTC's competitors are said to lick their chops when they see a benchmark with large assemblies, they figure they'll win there. That type of benchmark test probably excludes companies like Caterpillar or Deere, who started with Pro/E on smaller parts and assemblies long ago, and have followed a long route since then to get to larger assemblies.

Couple of years ago, PTC was said in the press to be working on a new CAD package, called Newton (apparently named after the Boston suburb, near Waltham). Could be by now Newton is no longer, or is folded into Pro/E. But a new package would make some sense, to break away from Pro/E and try a new approach. And a new approach might be most needed, and give the most benefit to PTC, in large assemblies. This new approach could be something that's developed from the ground up for large assembly work. Could be it sits on top of Pro/E, Pro/E still needed for the piece parts and smaller subassemblies.

Enough of this speculating. Back to Leo's presentation. But it sure seems that large assemblies are a constant source of interest to Pro/E users, to PTC (remember Dick's mention of a collaboration with Hyundai), and also PTC's competitors.

Leo had good specific tips:
  • raise no. of user colors to 32 at least
  • in the View Presentation window, set 6 rotations per second and enable LOD (level of detail). Enabling LOD means you'll see smaller components disappear as you rotate the assembly, to reappear when you slow rotation or stop. 6 rotations per sec means you shouldn't lose track of the geometry while you rotate
  • turn off the silhouette edges
  • do HLR for all views in one drawing sheet all at once, if you don't use HLR or silhouette edges you'll get faster display (old CV users may recognize this tip, it probably is a tip for any CAD user working with a large assembly)
  • make basic reps of parts, eliminating rounds and other smaller features. Leo said you could reduce part size by up to 99% this way. In case you're skeptical, think of a flat plate with various holes and cutouts and other internal features. Just represent it as a simple flat plate, six blank flat surfaces, and you'll probably match a 99% reduction
  • erase all views, when you aren't using them (another old CV tip), regen them only when you want to regen that specific view
He mentioned new features on 2000i2:
  • config.pro 'open_simplified_rep_by_default yes' to remind you to chose a simplified rep, instead of accepting the master rep by mistake
  • there's a graphical interface for defining simplified reps by rule, helps you keep track of AND and OR statements
  • can replace a component by an unrelated nonassociative copy---just as long as it has the featues you need to place it
  • simplified reps of drawings themselves, include view type and location, saved with the last sheet and pan/zoom information
  • on-demand simplified reps lets Pro/E figure out what you want to work with at any one time, automatically, but you need to remember to save any rep you make this way

Shrinkwrap on 2000i2:

Shrinkwrap was the biggest theme in Leo's talk. Perhaps this might be how some of those other CAD competitors got their advantage with large assemblies, they just aren't as ambitious as PTC, didn't try to include so much information in a large assembly. Shrinkwrap is an easy to use way to gut your assembly of as much information as you don't need to work in your area.

  • shrinkwrap can be surfaces or faceted solids or merged solids (the most information, capable of HLR)
  • shrinkwrap on 2000i2 should be associative, but only for the surfaces method (if you wait for a faceted solid or merged solid shrinkwrap to be associative, you'll wait a long time)
  • shrinkwrap entities can be local (only update within current design) or external (created in a separate .prt file, and updates whenever that shrinkwrap and the assembly are in session together)

Large Assemblies and Drawings

More on this perpetual large assembly theme from Ted Rzasa, of Hamilton Sundtrand Space Systems, but this time a user point of view. Like all user points of view, it depends on the company and their business. HSSS produces one of a kind space hardware, that's already a difference from many companies. And Ted uses parts that he designed on the drawing board 25 years ago now today, there's another possible difference.

A good part of Ted's valuable contribution was a table of Options for Faster Models. Here are some conclusions from that table:

  • store work locally. You may retrieve files from a server, but store them in some local directory on your own hard disk, and work on them there. When you're ready to submit back to the server, then do the submit. Pro/E is faster when it works on local files. Only drawback of this approach is that often (usually) files stored on a local disk are not backed up (but servers themselves are usually only backed up once a day, so if you submit files to server once a day you could be covered)
  • use 2D for certain parts (you probably won't hear this recommendation from PTC). So some parts are illustrated only on a drawing, not on the 3D model, although they exist as a 3D .prt with datums just in the assembly just for a BOM
  • erase views whenever not needed, erase views constantly, as default, while drawing is in process (same old generic CAD user advice as Leo gave)
  • use accelerator files and simplified reps

At Hamilton, they have a refreshingly simple approach to layers: they view layering as more work with minimal gain, those words. So they use just exactly four layers, and that's that.

These are config.pro options they use specifically for large assemblies:
  • retain_display_memory yes (object display is kept in memory)
  • save_display yes (and object display also saved on a save)
  • display_in_adding_view minimal_wireframe
  • allow_refs_to_geometry_reps_in_dwg
  • auto_regen_views no (the old CV feature too)
  • display wireframe (big divide here, some others swear surfaces are faster)
  • display_trimetric no (new views don't show at all until they are oriented)

Hamilton does generate fully constrained assemblies for release. But seems unlikely that PTC and Hyundai will manage to generate a fully constrained ship, with millions of components. Probably will be limits to fully constrained assemblies. Even among users comparing assemblies with the same number of components, whether fully constrained will be productive or not may depend on a lot of other factors---the business, the customers, the role of drawings, etc.

Ted shared with us this valuable information on retrieval times for a test assembly:

Master rep. 4:00 mins (data on local disk, 7 to 9 mins if working on a server)
Same with accelerator files 1:35 mins
Geom rep (no accelerator files, but apparently consumes time looking for them) 5:31 mins
Geom rep (with accelerator files) 0:32 mins

So the geometry rep without accelerator files is dead last, when working on the local disk, but same rep with accelerator files is about 1100% faster, and first in speed.

And here are retrieval times for an assembly drawing:
Master rep 5:02 mins (local disk, 8 to 10 mins if working on a server)
Same with accelerator files 2:36 mins
Master rep with saved display 4:08 mins
Geom rep (again no accelerator files, but again apparently looking for them) 9:24 mins
Geom rep with saved display 9:10 mins
Master rep with accelerator files and saved display 1:43 mins
Geom rep with accelerator files and saved display 0:56 mins

Well, that's going to have to be it for today. In a few minutes I'll be meeting with Ed to send this newsletter out. Perpetual problem of these user group meetings, there's so much to report. Hope this newsletter helps give you some of the flavor and the contents of what's happening here. Writing on a laptop in the hotel lobby has probably helped me keep the flavor, better than writing in some empty room or suite.

Couple of today's topics I'll cover when I have time (this could be a longer series of newsletters, but then they might be easier to read, less of an information dump):

  • Productivity tools
  • Parametric models from imported geometry