April 3rd, 2006 by Bob Sheaves
by RW (Bob) Sheaves, CEO, catNET Incorporated
My business (https://www.catnetsolutions.com) performs business process planning for the automotive industry, as a whole. Because of our associations, I see a lot of people in this industry, in general, concentrating on tools and their price, and NOT concentrating on the job at hand- meaning “What are you going to DO with these tools?”
In a real example, a particular customer asked us to provide an evaluation of various CAD packages based on cost. (We do not sell any software or hardware, but provide planning, setup, integration, support, and training.) This customer was looking solely at price, not how his company used the tools:
MAKING DRAWINGS:
The first question needed to be asked was simple-”How are you using the information from the engineer’s mind?” Paper drawings are a proven waste of time, effort and resources in any manufacturing business. For example, look at a single cast lower control arm, such as used in the Dodge Durango in the MY2000 vehicles (as well at its stablemate, the Dodge Dakota).
http://www.consumeraffairs.com/automotive/dodge_dur_steer.html directly addresses customer complaints on this vehicles steering and suspension. No verification of individual claims has been made by this writer.
This control arm has no moving parts, in and of itself, and incorporates a compression style ball joint (this balljoint design was found to be faulty by NHTSA investigators and subject to recall.) at the wheel end, with 2 rubber isolators providing geometric rotation around a common axis to allow jounce and rebound travel. Made from cast ductile iron, this part only contains 5 machined areas:
1. Forward bushing bore and spot faced length control
2. Rearward bushing bore and spot faced length control
3. Ball joint mounting surface (for mechanical hot riveting)
4. Sway bar end link mounting surface
5. Sway bar end link threaded hole
To completely design this part and detail all the necessary forms and shapes needed for tooling and fixturing, took a total of 59 drawings of multiple views and GD&T (Geometric Dimensioning and Tolerancing). The time expended on this was a total of approximately 500 man hours of labor, including the checking of the drawings for completeness, fit, and function. This time does NOT include the separate engineering procedures needed to ensure a quality part that performs to design and program intent. The cost, at an industry average wage, including overhead, of $75 per hour means that $37,500 was spent on a design that was destined to failure and cost the company millions in recall costs.
A modern 3 dimensional modeling tool, such as CATIA or BRL-CAD, generates a part in that same manner as you would manufacture it, the model has mass, dynamic properties, and tooling information, all bound into the definition of the part, subsystem, or assembly. There is simply no reason to make drawings as the cost of that paper will ALWAYS outweigh the costs of going from model, with simultaneous FEA analysis, simultaneous dynamic analysis, and manufacturability analysis, to manufactured part, when all the costs are added (Root Cause Method.
PARTS DEFINITION:
A “part” can be defined as a piece of material, manufactured to provide functionality of some sort. This part has a mass (or “weight” in common terms), a moment of inertia (dynamic properties when it moves in its function), material properties (what it is made from and how it forms in various manufacturing methods), thermal characteristics (heat transfer), and other esoteric properties the average person does not care about, but is of prime interest to various engineers, based on the use of the part.
Accurate modeling of all these parameters are necessary if you are to make efficient reuse of all the information in the overall business process. Dr. Deming teaches us that “re-” anything is a loss of efficiency…re-work, re-pair, re-build, etc. To be able to capture all this specified information is paramount for the initial choice of a CAD/CAM system. The model represents what is called “Design Intent” and is the “stake in the ground” against which all SPC, GD&T (or FD&T if you wish), and other processes utilized in manufacturing are measured.
Without the most accurate model possible, you fall subject to the “law of unintended consequences” and end up, in an extreme instance, of killing people because you just didn’t know something.
Boeing is very aware of this fact, IBM the same, DCX the same, Ford the same, etc…..
What is the common denominator among the companies listed? CATIA or Computer Assisted Three dimensional Interactive Application. In CATIA v4 or v5 (unix or PC based) you have a single location of a part in the process, which all information is centered around, through ENOVIA (functionally, a relational database, when combined with it’s component pieces making a PLM system) enabling all parts to only be designed one time, and most importantly, in one place. This alone was worth (in 1989 dollars) over $2.5 million to a particular automotive company in saved revenues. This also allowed a reduction of time, indirectly, as CATIA allowed only 1 version of a design to be used - all could see the part in the virtual space the vehicle is built in, but only ONE could modify that assigned part. All those using that particular part in their “space claim” were immediately updated with the latest changed. By using solid modeling technology, this company was also able to automatically check for assembly, function, and fit without human intervention and “pissing contests”. If the part did not fit or function within the assembly, it was flagged for reprocessing, along with the interfering ancillary components.
IS THE PART MADE CORRECTLY OR NOT?
Data translations between systems, through either direct translators, or intermediate file format conversions (such as IGES and STEP 203/213) are fought with alligators in the path of the unwary. Class “A” surface definition is lost during translations, competing CAD programs do not utilize the same math in computing elements of the designs (Bezier curves vs. NURBS ), system accuracy is not the same between the systems (19 digit accuracy, versus 7 digit accuracy), etc. 19 place accuracy sounds like it is meaningless, since for all practical purposes, you are defining a length less than the difference between colors in the light wave spectrum.
This simplistic approach has one VERY fatal error built into it….that is, the purpose of the design intent - a STAKE IN THE GROUND - does NOT exist.
If the “stake” is constantly moving, there is no way to determine what is correct , except by judgment A very poor and expensive learning process, as Chrysler found out.
This means the CAD system MUST integrate directly between and “outside” analysis programs (such as ADAMS and DADS for dynamics analysis, NASTRAN for FEA, LS-DYNA for crash simulation, etc.) Failure to accommodate this interoperability brands such companies as “poseurs” and will not be taken seriously by those that will control the funding for research and manufacture.
WHAT ALL THIS MEANS?
In summation, it is my opinion that the choice of a CAD system is way too early at this time, since the parameters of the basic design have neither been decided nor published on record here. Choosing what are the important parts of the design, how it is to function, and what the general approach to take is FAR more important to the success to the project, at this point in time. Also, the fact that such a closed source system as CATIA exists, is NOT a recommendation, but rather, by concentrating on setting the design operational parameters, an intelligent choice may be made by comparing OPEN SOURCE software, such as BRL-CAD, to the CLOSED SOURCE software, such as CATIA.
Best regards,
Bob Sheaves
CEO
catNET Incorporated
