Article for 3/98 issue of Cadence Magazine
(uncut version for the Internet)
Copyright (c) Michelle Pillers 1998 - All rights reserved



MCAD Renaissance of the 90's
A Report on the State of the Art

by Michelle Pillers, P.E.

Have you been following everything that has been going on in the Mechanical Computer Aided Design (MCAD) industry? Can you believe it? MCAD firms are being acquired left and right. Sworn enemies are forging partnerships. Who is going to win the 3D modeling kernel wars, ACIS or Parasolid? And what is the difference between them? Everything is moving so fast. It is really hard to keep up.

These issues are of special importance to someone such as me. I run a commercial product design engineering firm near Seattle called Evolution Design Engineering, LLC (EDE). It is important that we know what the industry is doing. We need to be prepared to ensure that we can interface with our broad client base and subcontractors. We need to be ready to offer product design/development services to new clientele without a "hiccup" in the process. For us, being on time, on budget, and able to shorten the development process is everything.

The good news is that amid all this chaos, engineers are reaping the benefits big time! Midrange MCAD modeling system developers are vying for the attention of the estimated 800,000 mechanical engineering desktops in the United States, not to mention the rest of the world. The result is that 3D MCAD solid modeling systems are becoming so powerful and easy, that even engineer's kids are using them to design science projects at home.


A Little History

Before continuing, it might be interesting to reflect on where we have been, as a base for where we are going.

In the 60's

I am not a historian, but I did a little digging. Here's what I discovered. Conceived during 1957, the first CAD system was actually developed during the 1960's. Dr. Hanratty is widely known as "the Father of CADD/CAM" for his pioneering contributions to the field of computer aided design and manufacturing. While at General Motors, Dr. Hanratty was co-designer of DAC (Design Automated by Computer), the first production interactive graphics manufacturing system. In 1971, Dr. Hanratty founded Manufacturing and Consulting Services, Inc. (MCS), who's products include Anvil-Express. He is still President of MCS. In their early years, MCS supplied code to companies like McDonnell Douglas (Unigraphics), Computervision (CADDS), AUTOTROL (AD380), and Control Data (CD-2000). Industry analysts have estimated that 70% of all 3D MCAD/CAM systems available today can trace their roots back to MCS's original code.

Another notable contributor to the computer graphics industry during this period was Drs. David Evans and Ivan Sutherland, who founded Evans and Sutherland in 1968. However, rumor has it that Computervision sold the first commercial CAD package to Xerox in 1969.

In the 70's

The 1970's focused on automating 2D drafting. The key here was placing lines and circles on a computer screen and automating the process by using a macro/programming interface. During this period, CAD operators not only had to be good drafters, but they also had to be good programmers.

Notable applications came from United Computing, Intergraph, and IBM. Most of these applications were single-precision, but of course, that was due to the limited power of the computing hardware of the time. Although Unigraphics was a 3D CAD/CAM system from it's earliest version, they did not release a double precision version until 1979.

The 1980's

A number of early Computer Aided Manufacturing (CAM) systems were introduced in the early 1980's to partially automate the manufacturing process by creating digital control tapes to drive machine tools, and CAE tools were introduced for very complex design analysis. In 1981, Unigraphics introduced the first solid modeling system, Uni-Solids. It was based on PADL-2, and was sold as a stand-alone product to Unigraphics.

Commencing in the mid-1980's, major technological advances, including the introduction of feature-based parametric modeling systems, allowed CAD/CAM/CAE software products to become a more integral part of the product design process.

One of the most exciting things that happened during this era was the advent of the personal computer and the beginning of Autodesk. John Walker, genius programmer and visionary founded Autodesk in 1982. He and his small band of original thinkers set off to develop five different desktop automation applications. They did this with the notion that one of the applications would take off. They would then band together behind the winner and develop it further. That product turned out to be AutoCAD, a clunky little 2D graphics program

that allowed graphics vendors to show off their high-end video hardware. The net result was a small army of hardware salesman demonstrating AutoCAD, that Autodesk didn't need to pay. These demonstrators were at every computer graphics trade show! By 1987 Autodesk had shipped 100,000 copies of AutoCAD.

In the mean time, CADKEY joined the market, but focused on 3D. In fact I remember that initially they even had a special 3D input device similar to coordinate measuring instruments available today. Working in 3D was a very exciting concept to me at the time and still is today. But 3D was still very hard to work with on a PC. It was not until a later release that CADKEY was able to become a serious player as a 3D wireframe layout tool and for drafting. Regardless, AutoCAD managed to lead the market with a more familiar 2D paradigm, but still managed to add 3D wireframe modeling to AutoCAD in Release 10.

The drive toward 3D solid modeling got a real shot in the arm in 1988 with the first copies of Pro/ENGINEER. Pro/ENGINEER turned out to be an extremely robust modeler that worked as advertised. In 1988 Unigraphics acquired Shape Data Ltd. (developers of Romulus, Romulus-D, and Parasolid) and began commercializing the Parasolid kernel as a stand-alone product. Parasolid is an extremely powerful B-rep solid modeling kernel that allows the joining of boundary represented surfaces together as a solid.

After six banner years as Autodesk's CEO, John Walker decided to step down and pass the day to day management onto Alvar Green. He then rejoined his programming buddies in developing new products for Autodesk in 1989.

The early 1990's

No doubt the 90's have brought us unbelievable strides in MCAD modeling capabilities. In 1990, Spatial Technologies brought us ACIS, a commercial solid modeling engine that provides a common database part that had the ability to be accessed & utilized in multiple environments, i.e., part model, analysis model, assembly model, and manufacturing model.

Being flush with cash, Autodesk began purchasing companies to gain access to technologies. Notable acquisitions like Micro Engineering Solutions in 1992 helped bring us AutoSurf, and later Woodburne in 1993 brought us Autodesk's first parametric modeler AutoCAD Designer.

With Release 13, ACIS became the underlying 3D technology upon which AutoCAD and Mechanical Desktop would be built. In 1994 Autodesk released AutoCAD 13 and shipped their 1,000,000th copy!

During the last 2 1/2 years

These last few years have seen fierce competition for MCAD market share. Pro/ENGINEER continues to make headway into the high-end market and threatens the current power base of old timer enterprise systems like CATIA, CADAM, Unigraphics and Intergraph. This push has made them the 10th largest independent software company in the world. As a point of reference, Autodesk ranks as the 5th largest PC software company in the world.

See Timeline

As of this year, Autodesk has shipped over 1.5 million copies of AutoCAD. Regardless, Autodesk faces serious competition from a number of very formidable MCAD products. As of this writing, many mid-range products like Solidworks98, SolidEdge 4.0, and Anvil Express have many capabilities that we have yet to see in Mechanical Desktop 2.0. These are capabilities like lofting, variable sweeps along a 3D spline, sheetmetal part bending, dynamic kinematic motion, animation of assemblies, and so on.

A big advantage for the newer products is that they do not need to work with legacy data. This is something that Autodesk has to deal with. It is important that Mechanical Desktop work seamlessly with old AutoCAD data or Autodesk could end up with about 2 million upset customers. This leaves the competition to focus on adding newer features.

However, many of these newer and smaller firms are being gobbled up by larger concerns. Solidworks merged into Dassault Systèmes last June, EDS/Unigraphics has acquired control of SolidEdge, and PTC bought financially troubled Computervision (CV) the same day CV rolled out DesignWave.


State of the Art

In spite of the mergers, it appears that the key technologies are surviving and mid-range MCAD systems are succeeding. All systems are migrating toward Windows/NT and Windows/98. ACIS and Parasolid are at the core of nearly every system. Hardware prices keep plummeting. Not to mention that powerful, robust MCAD software like PT/Modeler and CADKEY97 are available for under $3,000. Communication between computer

systems is becoming easier. Internet, Intranet, and web development has opened data sharing between MCAD systems at remote sites and vendors like never before. With web specific formats like VRML, DWF, and Internet browser plug-in add-ons for viewing native part and assembly files, users can communicate on the telephone while revolving around a part and viewing it simultaneously using Microsoft NetMeeting.

MCAD vendors realize that the competition is fierce, and that their products must be extremely easy to use, do more than ever before, and be affordable. MCAD systems are finally reaching a point that they provide the designer with no limits.

Also, programming techniques are improving for the developers and MCAD vendors are including cool features into their products that they find in competitor's applications in just a matter of months. Consequently, the average mid-range MCAD product can do Lofting and allows the operator to work in real time on a shaded perspective image of their model. Most applications include data translation utilities like IGES, STEP, STL, ACIS, Parasolid, DXF, and DWG as standard. Ease of use has improved dramatically with the introduction of the Feature/Assembly tree, introduced over 3 years ago by in TriSpectives 1.0 and then exploited by SolidWorks95. Now nearly all MCAD systems have this interface.

There have also been dramatic improvements in visualization and analytical inquiry. The MCAD operator can now see the consequences of his or her next move in real-time with the use of feedback like AutoSnap in Mechanical Desktop and dynamic highlighting in SolidWorks. Nearly every dialog box comes with a preview button to see what will happen next. The operator spends far less time correcting mistakes and performing UNDO's. Instead, MCAD operators spend more time designing and doing analysis.

Mid-range MCAD products like new Solidworks98, SolidEdge 4.0, Microstation, and Anvil Express include powerful sheetmetal modules in their base product. Interestingly, high-end MCAD systems like Pro/ENGINEER and SolidDesigner do not. Mechanical Desktop 2.0, Microstation, and Anvil Express also include powerful NURB based surface creation tools that can apply freeform surfaces to solids.

Nearly all MCAD packages include Application Programming Interfaces (API). These interfaces allow seamless integration of powerful add-on analysis packages for doing finite element analysis (FEA), interactive product simulation (DADS and Working Model 3D), rendering (AccuRender & PhotoWorks), add-on surfacing, CAM, and animations.

Thanks to Windows, the graphical user interfaces (GUI) between different products are not only becoming easier and more intuitive, but they are amazingly similar to each other. Use of tear-off toolbars, feature browsers, intuitive icons, common pull-down menu terminology, and graying out of toolbar buttons that won't work with the current selection, allow the user to come up to speed quickly and get productive in record time. This translates into real money - because there is less training, more billable work (important to a consulting firm like EDE), or getting a product out to the market quicker.

I see the evolution in interface technology as creating an important equalizing force in our industry, and leveling of the playing field. It is now possible to hire the best design/engineering talent available without having to worry about whether they know or don't know a particular MCAD system. The fact is that they will probably be able to determine how to use it in no time at all, even without training. 


What of the Kernel Wars?

The race for modeling engine dominance is fierce. Two core products are at the forefront of this race - Unigraphics' Parasolid and Spatial Technologies' ACIS. Parasolid has been used in high-end modelers for about ten years. Whereas Spatial, founded in 1990, has had their core-modeling engine, ACIS, in products like AutoCAD for around 4 years. The nice thing about using a core modeler is that it creates a common part file format providing a data pipeline between design, analysis, rapid prototyping, manufacturing, and other CAD systems. Vendors, such as Autodesk, SolidWorks, Intergraph and others, can focus on user interface and assembly modeling issues while handing off core solid modeling functionality, such as

sweeping, Boolean operations, knitting surfaces, complex blending, chamfering and filleting, tapering and draft, uneven scaling, offsetting of solids, thickening of surfaces, mass properties, shelling, exporting of geometry, session rollback for Undo, foreign geometry interface, graphical data for viewing, and so on,

to the core modeling engine. As you can see, it is not only important that the core modeler be feature rich, but it must also be robust and dependable.

Interestingly, I have been seeing MCAD products (like Solidworks for instance) that are licensing and integrating technologies from both, Unigraphics and Spatial. Consequently, some MCAD vendors can import and export part models in both formats.

On the other hand, concern has been expressed as to which core modeler is better, and why it might be better to go with one MCAD vendor versus another. Personally, I don't think it is as big an issue as some people would like you to believe. Case in point, Integraph's SolidEdge is currently an ACIS based product, but due to a joint venture, they are switching to Parasolid. Another case in point, Bentley has decided to move Microstation core over to Parasolid. In fact, I even think SolidWorks was once an ACIS based product early in its existence (before its first release).

The point is, if the MCAD product has the proper software architecture, it should be straightforward to swap out the modeling core. I suspect that every MCAD provider does this. This is good news for the user when considering their workstation and long term investment.


What of MCAD value? Which system is best?

No discussion of the MCAD state of the art would be complete without looking at what kind of bang you can get for your buck. Please note tClick to see Chart showing best MCAD value according to author.hat due to limited resources and my reliance on experience and anecdotal evidence, the following analysis is very unscientific. Nevertheless, I believe it to be of considerable value if viewed with a grain of salt and a wide margin of error.

In an effort to be fair and objective, I polled a number of Internet newsgroups, reviewed volumes of literature sent to me by MCAD vendors, got released and unreleased copies of MCAD software, and learned them in record time. As the former Director of Product Development at Teague, I gained experience using Pro/ENGINEER, and made a number of contacts within The Boeing Company, where I got information on CATIA. I have completed projects using SolidDesigner, have experience using Mechanical Desktop since it was simply AutoCAD Designer. And most recently (and to their credit), I did a crash course learning SolidEdge v4.0. I relied on anecdotal evidence for all other information in the chart.

I thought the chart would be a good idea considering all the motion, volatility, and options in the current market. However, I urge everyone to be his or her own judge. Try out the software before you buy, and get a money back guarantee in case you find it doesn't meet your needs. Each industry is unique and has its own set of needs. Mine are in bringing state of the art commercial products to the market better faster and cheaper. With that, let us proceed.

In an effort to be objective, the first thing I did was to get a handle on the kind of functionality I thought was important in an MCAD product. See Table 1. This table only looks at features I felt were important, but not necessarily found in all of the packages. For instance, the table ignores things like 2D drafting, translation utilities, visualization utilities, Internet connectivity, object highlighting, feature/assembly management trees, "best in class" third party software integration, assembly and constraint management.

The table instead looks at other important productivity improvement features like Windows OLE capabilities for using the product with other Windows applications, solid sweeps with guide curves, whether or not it comes with integrated sheetmetal and freeform surfacing capabilities, and so on.  

One of the most valuable features I have found in some new MCAD products is what I call "dynamic linkage motion". This feature is standard on many systems and costs extra on others. It allows one to "grab" a part in the assembly module and move other parts in the assembly via under-constrained degrees of freedom. For instance, moving a four-bar linkage back and forth in real-time by grabbing the crank. Another indispensable feature I found is being able to design parts within the context of an assembly; thereby having features on one part (like hole placement) being tied to features on another. Although all MCAD systems are moving in this direction, not all systems currently have it yet.

Table 2 takes the results of Table 1 and creates a non weighted usability percentage which is used with an expected average learning curve to generate a Productivity Quotient. This number is plotted against how much a useable system might cost to create a Mid-Range MCAD Value Chart. As mentioned earlier, use the chart with care since it was not derived from a critical survey of any kind. However, the methodology may prove useful in your own evaluation process.

As you can see by the chart, the mid-range market coupled with todays technologies has created considerable value for the design engineer. This chart shows a clustering of solid modeling products that provide exceptional value to the end user.  


In Conclusion

It is truly amazing to think about it, but we, as designers and engineers, are finally in a position to become "one" with our tools and totally focus on the design effort instead of worrying about syntax and how to do the most basic things. Because of these things, using different MCAD systems may not turn out to the big concern it used to be. A design engineer will soon be able to move from MCAD system to MCAD system and remain productive. The fierce competition, core modeling engines, advanced graphical user interfaces, Internet/Intranet connectivity, and lower prices have really put us in the middle of an MCAD Renaissance.

 


Michelle Pillers, P.E., is a licensed mechanical engineer, product designer, and a principal at Evolution Design Engineering, LLC. As an avid user of computer technology and solid modeling applications, she is often engaged in lively debates on the Internet in the Mechanical Desktop, Solidworks, and Pro/ENGINEER Internet newgroups. She can be reached by email at mpillers@edeinc.com. Be sure to visit EDE's website at www.edeinc.com.