Pro/User 1999 Conference

Finally, the last report on this conference. Still plenty of room for comments and other reports from anyone else attending, still seems we spend more time on pro-user talking about details, like the conference date, than about the conference itself.

I liked the 1997 Orlando newsletter best myself, daily reports, instead of reports spread out over 2 weeks. Might be a chance to return to that mode for the Orlando 2000 conference next year. Rick Snider has asked for comments about a Web page to store reports and pictures and comments. Although email still has some unique ability to communicate, especially if right on the spot, daily, that's a good goal.

If you want to contribute to organizing the program for next year's conference, let Paul Wilson know (wilsonp@adc.com). Paul is the Pro/User Director of Events and Presentations, so he's the person who has most of the responsibilty for arranging the presentations at the conference. Apparently this year PTC volunteered more presentations than originally scheduled, which might be why PTC had more overall time in the presentations. Especially since in past years conference attendees in the surveys have rated PTC presentations most useful. Something to keep in mind: as Pro/E has passed its 10th anniversary, it's more of a mature software product, that seems to happen to most CAD products at about the 10th anniversary. So, there may now be just somewhat less new user experience to discuss at Pro/E conferences, since with a mature product users have worked out a lot of the issues. There's nothing like a brand new CAD product to stimulate user discussions and conferences, since the questions how to use it are all brand new too.

Peter Nurkse

peter.nurkse at sun.com

Topics in this issue:

• Pro/Intralink in Large Installations
• Dell's Exploitation of Pro/Engineer
• Maintaining Part-to-Part Feature Alignment in Assemblies
• Pro/Intralink with Windchill
• Data Exchange
• Assembly Maze and Amazing Assemblies
• Waxahachie (travel supplement)

Pro/Intralink in Large Installations

Victor Gordes, PTC's manager of Pro/PDM and Intralink marketing, spoke to a full room of over 200 people, about 1/6 of the total users attending the conference---substantial interest in this subject.

To cut to the chase, Victor said there were now Intralink installations of up to 150 users, at from 2 to 4 sites. That's the most so far, that's the largest Intralink installs so far. Victor didn't mention the size of installations converting to Intralink from Pro/PDM, but probably they are smaller---biggest conversion so far from Pro/PDM to Intralink mentioned on datamgt@prouser was 6GB.

Although bandwidth is how most people might measure a network connection, Victor singled out latency as equally important. Latency is how long it takes a message to get from a client to the server and back. A network connection can have high bandwidth, big capacity, but still low latency, if there are lots of routers and other devices slowing down traffic along the way.

If bandwidth is adequate, latency can determine how interactive a network application like Intralink seems to the user. One measure of latency is the 'ping' command, which returns a time in milliseconds, the average round trip time for a series of packet from one machine to another. Victor said that with Intralink latency is probably painful above 200 milliseconds, which is certainly true. For a very intensive interactive application, like Pro/E itself, latency probably starts getting painful at about 10 milliseconds.

Intralink 2.0 will let you fetch a configuration as stored, or the latest, or a custom configuration that you define at a particular release level. PTC realizes there's room for improvement, and is working "as hard as they can" to improve speed, reliability, and performance. Metadata (information about the data itself, such as attributes) slows down performance enough that it's best placed on a separate server of its own. And similarly, a user displaying attributes will have better performance if you only select attributes you will modify.

Dell's Exploitation of Pro/Engineer

Ed Stevenson, a dynamic guy himself, gave a dynamic talk about Pro/E implementation at Dell, a dynamic company. Dell is a company which ships about $55 million worth of computers a day, or about $18 billion a year, and where the typical product design time is 6 to 8 weeks. Pro/E data doubles every 8 months, and average return on investment is 180% annually (yes, 180%, that's the average ROI, so your project had better aim for that at least).

As a fast growing company, Dell was finding itself limited by shifting plans for Pro/E, a reactive rather than proactive mode, overextended Pro/E support staff, and little higher level visibility for Pro/E at Dell. It could take days to organize models for release (and these were "hack and whack" models), and sometimes they couldn't even shut a server down because they didn't know exactly what was on it. Ed had to get special permission to share this candid view with us, but it certainly added conviction and realism to his talk.

Here are some of the ways they made progress from that point:

• abandoned a separate CAD network and now run on the Dell company network
• moved from Unix for Pro/E to Windows and now NT, company standard
• Pro/E staff report directly to the IT manager
• CAD development lab with Unix, NT, and W95 and W98 systems, running on optical fiber, Ethernet, and ISDN networks
• rely now on corporate resources for printing and servers, no separate CAD printers or servers

Results are:

• Pro/E will run now on any system at Dell
• any system can be configured with Pro/E in 2 hours, ready to go
• $1.4 million freed up from existing budgets
• tooling engineers can now take Pro/E out into the field to suppliers, on laptops
• ROI on the Pro/E project was just 41%, small by Dell standards, but major improvements.

Maintaining Part-to-Part Feature Alignment in Assemblies

Brian Adkins got the raw material for this subject with an Internet survey of methods in use for maintaining part-to-part feature alignment in assemblies. Brian summarized the results with the following chart, which appeared on the screen during his presentation, but which was entirely blank in the hardcopy of the proceedings. Here it is:

(H = High, M = Medium, and L = Low, relative to the other methods)

This example should show the major benefits of a chart, particularly when comparing several different possible procedures (which seems to be a common need with any CAD package). Here you can see that no one method is the complete answer, and you might in fact choose any one of these methods depending on your particular situation. Charts could be a particular benefit of user conferences, PTC may not have the same overall experience with all possible methods for any one problem.

Training overhead, for example, is a big consideration: there's no benefit if a Pro/E wiz recommends a High training overhead procedure, if most of the people doing the work don't understand it. Interesting too that training overhead can be High, but setup time can still be Low, for the same method.

If you weren't at the conference, but are interested in this subject, you can email Brian and ask him for the complete presentation, in either PowerPoint or PDF format. Lots of information, including complete examples of the different methods. That's how I got a copy of the missing chart. Brian's email address is: badkins@cgn.net.

Here are brief notes on each method, assuming the classic example of two holes in different parts which need to match. The hole which determines locations is the driver hole, and the other hole is the driven hole.

• Simple Holes: both holes created separately at part level, parts are completely standalone
• Simple Relations: simple holes created as before, but then relations written at assembly level to maintain alignment
• Layouts: simple holes created as before, but then create layout with dimensions as parameters, declare layout to both parts, then write relations at part level using layout parameters
• Coaxial Assembly Holes: assembly holes drive hole locations, but result is two parents, hard to make parts stand-alone
• Assembly Cuts: create driver holes at part level, then part cuts for driven holes at assembly level, then unalign edges and dimension driven holes relative to that part
• Top-Down Skeleton: create skeleton at assembly method, then add datum planes and axes to skeleton, pass axes to parts via copy-geom
• Bottom-Up Skeleton: create driver holes at part level, copy geometry feature (axes) to skeleton, create driven holes at assembly level
• Up & Down Skeleton: create driver holes at part level as before, copy axes to skeleton, then copy them to both parts, and create coaxial holes in each part

If you want to know more, definitely email Brian for a copy of the complete presentation. Makes a useful summary for a group meeting or discussion on the subject.

Pro/Intralink with Windchill

Fredrik Hansson from PTC gave this presentation. The general theme was already expressed by Victor Gordes in the Intralink Large Installations talk: "There are no Pro/Intralink users, only Pro/E users". Meaning, Pro/Intralink is only now intended to be used by Pro/E users, and other people interested in the data will probably be using Windchill---the non-Pro/E users, usually the vast majority in any company.

That's a good example of the difference between Intralink and Windchill, which has been changing: Intralink is now seen by PTC as a product used just by Pro/E users, while Windchill is used by everyone, Pro/E users and everyone else interested in the data.

One example of the effect of this attitude on Intralink is the case of workflow. Pro/PDM has a simple workflow mechanism, but Intralink was missing it from the beginning. The original PTC plan was to add workflow to Intralink 18 months after the initial release of Intralink. Then that plan was modified to import workflow into Intralink from Optegra (that was the plan a year ago, at the Anaheim meeting). Now this year the plan is apparently to put workflow into Windchill, and not into Intralink---although Intralink 2.0 may gain some Release Management functions.

Both Intralink and Optegra (the Computervision equivalent of Intralink) are now seen as examples of gateways to Windchill. A gateway is always the master source for the particular data that it stores (Pro/E data for Intralink, or CADDS data for Optegra). But the gateway publishes that data to Windchill, and Windchill stores a copy of the data, Windchill replicates the data. Once published in Windchill, the data is then available to everyone, all the people who don't use the CAD tool itself. PTC has plans to establish more gateways for other CAD data management systems, besides Intralink and Optegra---that may be a challenge, even more of a challenge than exchanging CAD data with IGES or STEP.

Right now what gets published to Windchill is part/assembly/drawing information: 2D and 3D graphics files, and also the BOM (product structure). The 2D and 3D graphics files can be marked up and redlined independently of the originals, which stay in Pro/E or in Optegra. And Windchill can support different BOM structures for the same product: for example, in addition to the original engineering BOM, also a manufacturing BOM, or a sales BOM, each a different view of the same product.

Gateways can publish information to Windchill manually, or automatically, by a polling process. For example, you can specify a period for updates, or you can specify a step in the process that triggers an update to Windchill (new release level, change in a file, etc.).

Remarkably enough, the Windchill browser doesn't recognize Pro/E files: if you download a Pro/E file, you have to tell Windchill that this object is a Pro/E file, Windchill doesn't know that. Perhaps this feature is a remnant of Windchill's origins with Computervision, independent of PTC. But if Windchill can establish an even playing field, and show no favorites among different CAD applications, then that could help increase Windchill's credibility as a generic data management tool.

Windchill does depend on the same Graphics Server that Product View uses, and that server runs only on NT. Probably a remnant of the Graphics Server's origins at Object Logic (later acquired by Division, itself in turn bought up by PTC). As typical with most visualization vendors, Object Logic was probably just targeting the Windows and NT market, they weren't trying to develop visualization tools for the Unix market---because the Unix market is mostly people who have the CAD tools on their desktop already.

But now PTC has a core product, this Graphics Server, used by both Product View and by Windchill, which runs only on NT. The Graphics Server creates images in Windchill, and provides access to specific versions of the original data on demand.

Fredrik said that 98288 is the golden version for Intralink now, the version to use, benefitting from a major effort to increase speed and reliability (speed increases of up to 10x on some operations, according to user tests).

Data Exchange

Asa Trainer from PTC gave one of his usual knowledgeable presentations on data exchange. He gave an outline of steps for successful data exchange, which goes far beyond the data exchange step itself, what most people concentrate on. Here are those steps:

• Design Standards: if you don't have these in place, your data exchange problems will increase.
• Model Quality: procedures and checks on model quality, while parts and assemblies are being constructed---finding problems during or after the data exchange may just be too late.
• Data Presentation: at this point, you're preparing the data for the data exchange, with a view to the recipient and what you want to give them and what they want to receive.
• Data Exchange: the data exchange process itself, what most people see to the exclusion of all the other steps.
• Data Validation: recipient checks they did get what they were supposed to get.
• Data Repair: especially with solids and surfaces, tolerances or other problems may mean the recipient has to do extra work to make the data useable in their system.
• Application: using the data in the receiving system.

The STEP application protocols supported by PTC are just the same as Asa reported at the Orlando conference two years ago:

• AP203 (developed principally in US, for parts)
  • conformance class #2: wireframe and surfaces
  • conformance class #4: shells
  • conformance class #6: boundary representation solids and assemblies
• AP214 (automotive, comes out of the European auto industry)
  • conformance class #1: components
  • conformance class #2: assemblies
  • conformance class #3: product presentation (layers, also now colors, but not groups)
• AP202 (associative draughting, the British spelling, the British still influence the language of international standards)
  • conformance class #3: wirefame
  • conformance class #5: surfaces

Why not more application protocols in two years? Well, AP214 isn't yet an international standard, for example, although it's close to that point. Development of real international standards goes slowly. Associative draughting (however you spell it) is a real feat, tranferring not just a drawing but also the relationships between the elements of the drawing and 3D models.

Asa narrated an inspiring live real time demo of the Associative Topology Bus (ATB), the method PTC is using to transfer associative geometry between different applications. With the ATB, the source data can only be modified by the source application, but it can be used freely in other applications, which can then update when the source changes. Being live and real time, the demo had some pauses, but that's the real world.

These were the components of the demo:

• an automobile firewall from Computervision CADDS5
• an ignition coil from Pro/Desktop, inserted into a Pro/E bracket
• wiring routed from the ignition coil to a relay from CATIA, using both Pro/E and CADDS5 geometry references

Once everything was assembled in Pro/E, then Computervision was brought up to modify a hole in the original firewall through which the wiring passed (a pleasant sight for the CV users in the audience), and then back in Pro/E an update incorporated the new hole position and size into the Pro/E assembly, and then modified the wiring path accordingly.
And everything worked.

Future plans for data exchange at PTC include:

• heterogenous assemblies (many different CAD applications, using Division) more 2D data from CADAM (Pro/CDM) to fix 3D models
• capture solid model accuracy in the import phase
• create a file for the validation process
• include configuration management data (who modified it, when, what authorization, STEP stuff)
• more Topology Bus work

Assembly Maze and Amazing Assemblies

By popular demand, Ted Bradshaw from John Deere repeated this presentation twice---and still didn't have time to cover it all. However the key problem he was addressing is pretty familiar: how to limit unwanted external references within an assembly. For example, you construct one part using features from another part, in the assembly, and suddenly you fin d you the first part now references the whole tractor, or the whole airplane, or the whole locomotive, or some other mammoth assembly.

The solution Ted proposed to this old problem seems pretty new, using IGES as a filter, like this:

• create a temporary part
• assemble the temporary part to the part you want to modify
• now copy the surface geometry you want to reference into the temporary part
• export IGES from the temporary part
• now import that IGES file into the part being modified
• delete the temporary part (that deletes the external reference back to the assembly, the reference is gone)
• now go ahead and do your modification using that imported IGES feature

You probably could use STEP instead of IGES as the filter, it might even work better overall, since PTC has put more effort into STEP recently than into IGES. But the basic principle is still to use data transfer into a different data format as a filter to exclude any trace of an external reference. If you're ever imported an IGES or STEP file you know how it comes in as a single feature, with all associativity and references lost. That's how you can filter out external references, working just within one assembly. It's as if you step outside of Pro/E for a moment, into IGES or STEP, just in order to come back into Pro/E clean of external references.

For managing layers within an assembly, Ted recommended carefully stuffing your assembly with every single layer used in any component in that assembly, so that you can then control all layers in any component right from the top assembly level. To get all those layers into the assembly itself, use start parts and start assemblies, and default layers. Uniform identical layers make layer control easy, and it doesn't matter at all if most layers defined at the top assembly level are empty in the assembly itself, they're still useful for controlling layers in components lower down.

Waxahachie (travel supplement)

Perhaps the Pro/User conference will return to Dallas, after all surely the rest of the country (and the rest of the world) must acknowledge that the Texas regional user groups are the most formidably organized and efficient Pro/E user groups anywhere (this written by an admiring West coast resident).

So, if you come back to Dallas, where can you go to see a bit of Texas outside of Dallas/Ft. Worth? I suggest Waxahachie, a small town 30 miles south of Dallas, which happens to have a good 20% of all the National Historical Buildings in all of Texas.

But it isn't a cute carefully preserved antique town, such as you might find elsewhere. Instead it's a living working town, which happens to incorporate dozens of historical buildings, both home and business, in daily life. As proof that Waxahachie isn't cute, there are a few buildings here and there that are simply falling down, you don't see that in the cute antique towns. Only one gift shop on the main square, and all the inevitable antique shops relegated to a side street.

The biggest treasure of Waxahachie is right in the main square, in fact it pretty much fills up the main square. It's the Ellis County courthouse, and it's built in the very best Romanesque style. Romanesque was an architectural style which was all the rage in England and France over 800 years ago (before the Gothic fad took hold). Romanesque buildings have columns and heads carved in stone and porches and arches and carved flowers and towers and many kinds of stone textures, and the Ellis County courthouse has all of that in abundance.

The general effect of the Ellis County courthouse is of a rather short and stocky Romanesque cathedral from 1100 A.D. in Europe, converted into an office building or courthouse, and then dropped right into the middle of a small town in central Texas. Pretty remarkable, perhaps even, you might say, a miracle. Worth a trip from Dallas to see it.