CAD in creative design, problem solving

Awl --
In a pointless response to the Village Idiot (Jon Banquer, unemployed ex-thief of Qualcomm, in his absurd notion of him asking difficult Qs), I brought up the notions of CAD being more for communication than the design process itself. Bonkers of course confuses "difficult" with "rancorous".
Some refinements of the design vs. communication notion.
In "art" type design, like, say, for a new car body, certainly CAD could be useful, as a simple curvature of a line can alter the visual effect in auto design.. But this is more of an "illustrating" context than say a parts/machining context.
Ito of the actual function/design of mechanical parts, has anyone actually solved a "design problem" using CAD? And by design problem, I don't mean radius blends, geometry problems, etc, for which CAD can really shine, but rather the solution to a "how do I do this" engineering-type problem.
For me, the problem is always solved on paper, the essence of the design clearly present on a napkin. CAD, for me, is just for telling other people where the holes go, or for keeping track of the history/evolution of a design....
Now admittedly, for things like carburetors, these holes can get perty complicated, and mebbe CAD would be useful in doping a few things out in something like that, but I would think mostly it's a tool for communicating/building, not really thinking per se.
After all, some perty complicated carburetors were built before CAD, and I don't know that CAD really improved them.
But mebbe they did.
Opinions? Experiences?

Existential Angst wrote:
  Quoted Text Here  
I'm not sure what you mean by "geometry". I'm too cheap to use 3D cad (it gets in the way of me visualizing stuff in my head), but I've seen mechanical engineers use it to guarantee fit of multiple parts -- guy A designs part A, guy B designs parts B and C, then guy C puts them together into an assembly and finds that part C and part A overlap -- ah ha! -- and things get changed.
So from that perspective, and from the perspective of being able to use it to visualize how a complicated assembly is going to go together and work -- particularly when you use animation for moving mechanism -- yes, I've seen it used for design.

  Quoted Text Here  
http://en.wikipedia.org/wiki/Finite_element_method
http://www.youtube.com/watch?v=Bt6Q_pVxwl0

jsw

wrote:
  Quoted Text Here  
http://en.wikipedia.org/wiki/Finite_element_method
Funny, how calculus (the integral) is in fact the result of FEM, in the limit as x --> 0, but done analytically (power rules and all that). FEM is kinda like calculus in reverse, when analytic solutions are not possible.

http://www.youtube.com/watch?v=Bt6Q_pVxwl0

Nice. But I would distinguish CAD from this type of analysis/animation program.

Existential Angst wrote:
  Quoted Text Here  
I guess that depends on what word you assign to the 'D'. If you think that 'CAD' means "computer aided drafting", then yes, it's different. But if you think that 'CAD' means "computer aided _design_" then this sort of analysis program is nothing other than a fulfillment of at least a small part of the promise.
I will grant that -- strictly speaking -- it's still not design: analysis is an aid to telling you whether your design process is on the right track, and in the right hands it will even put up road signs telling you where to turn next. But I've yet to see a program that can come close to successfully carrying out the order "check these ten possibilities and tell me which one I want*", much less "I want to increase the speed of my wire brush assembly machine without losing product quality".
But I would make the -- entirely semantic -- claim that 'CAD' means "computer aided design", and that in a limited sense we have just that.
* With scripting you can check ten possibilities, but you still have to look at the results and figure out what they mean.

  Quoted Text Here  
One might propose that design implies 3D in most cases but that simple drafting/drawing needs only 2D.

wrote:
  Quoted Text Here  
Take that sample truss and adjust the element dimensions until all areas are the same color under load.
I've been using electronic design CAD programs for ~25 years. Simulation, analysis and rules checking are essential components of them.
jsw

  Quoted Text Here  
Why not just solve a fairly simple problem in Mechanics ?
  Quoted Text Here  

wrote:
  Quoted Text Here  
Because "redundancies" in structural mechanics render a simple looking problem not so simple, in fact unsolvable analytically (at least with traditional analytic methods), thus requiring numeric methods. Ergo the finite element analysis et al in that wiki link.

On Tue, 18 May 2010 10:12:38 -0400, "Existential Angst"
  Quoted Text Here  
The statemnt was until all areas are the same color under load" which implies similar stresses. What safety factors to use? Which failure modes to account for? As far as the specific issue goes it's still just simple Mechanics IIRC & AFAIK. Slide-rule stuff <g>.

wrote:
  Quoted Text Here  
Except it isn't. Once you introduce redundancies, it's not that Newton's laws no longer apply, it's that it's no long clear on what to apply them and when.
One of the first projects in a structural mechanics course I took was exactly this, a very simple model of two poles with a cross brace, subjected to some torsion. Done with Hollerith cards... goodgawd...
Don't recall the exact how's and why's, but the bottom line was that that simple model could not be solved analytically. Ergo the iteration/numerical analysis.
Other "simple" problems have either very complicated solutions the make numerical analysis more efficacious, or solutions with unsolvable (non-integratable) functions, like elliptic integrals. The three-body problem is one example.
Heat conduction and fluid mechanics problems other broad classes of problems that rarely have analytic solutions, even though they are glibly described by div, grad, curl and all that (the actual title of an iconic advanced calculus book by Schey).
Some people *think* their solutions are analytical, not realizing that the tables they are looking up answers in are the result of numerical analysis, tabulated visavis initial and boundary conditions. Other tables will be tabulations of analytical results, so there's no way to really tell, unless you know the "infrastructure" of the problem.
Redundancies in structural mechanics are apparently similar.
Funny, trig functions are a kind of hybrid. Even tho they are expressed in an elegant shorthand, ultimately they are ground out by N number of terms of an infinite series, terminated according to one's need for precision. Ditto e, pi, etc. So even some theoretical type stuff winds up being numerical, to some degree.
What a messy universe....

(Amazon.com product link shortened)74302771&sr=1-1

  Quoted Text Here  
(Amazon.com product link shortened)74302771&sr=1-1
Ha! I just re-read that book a few weeks ago. It's a good one.

wrote:
  Quoted Text Here  
Take that sample truss and adjust the element dimensions until all areas are the same color under load.
I've been using electronic design CAD programs for ~25 years. Simulation, analysis and rules checking are essential components of them.
===============================================I should have specified D = drawing in CAD. :) Analytic/animation type stuff certainly can be useful where D = design, but this is a bit more sophisticated than doodling on a screen. :) :)

  Quoted Text Here  
EXACTLY!!
but I've seen
  Quoted Text Here  
Yeah, that's geometry.... and dotting i's, crossing t's. Not saying CAD isn't important... it sure beats trial and error at a gear indexer/hobbing machine, eh?? But the basic ideas *generally* do not require CAD -- or so it seems to me.
  Quoted Text Here  
As per Jim's links, I would distinguish CAD from animation/analysis programs.

Existential Angst wrote:
  Quoted Text Here  
Think really big napkins. Endless, boundless napkins, if you like. Think about being able to draw a perfectly straight line, or a perfectly round circle, with just a flick of your wrist. Think about being able to undo a doodle that looks wrong; but without having to scribble over it and mess up the whole deal, or without losing time and train-of-thought when the napkin's full of ink and you need to start over. Think about being able to pick up one of your doodles, right off of the napkin, to turn if over and see if it still looks right. Think about having all your napkins saved in one drawer, so you can easily find an old one and compare it to something new you just thought of. Imagine that, almost by magic, all your doodles and scribbles are done at the same scale, or can be made to scale the same, so that any collection of doodles can be put onto the same napkin for comparison, brainstorming, or thought experiments about the project or problem.
Think about this, too: The first step to solving any problem is to state the problem accurately and effectively. I find that having the immediacy of napkin sketching combined with the precision of a CAD drawing can make the problem itself more visible, which often makes it more soluble. What's important is not to let the drawing become a source of delay and distraction that messes up your thought process.
The reason most CAD users, including me when I was new at it, have trouble "thinking" with a CAD system is that the mechanics of using the system get in the way of dreaming and imagining and squeezing thoughts out of your brain. When a thought appears in your mind, you want to CAPTURE IT, not go looking for the right command icon, then trying to make up numbers or mouse-clicked positions that you don't even have yet, and then extending and trimming and coloring and layering and more, just to sketch something rough and simple. By the time you've done all that, the fleeting thought that you were grasping for is gone.
The solution is not to limit your CAD system to "after-the-fact" refinement or presentation of a napkin-sketched idea; but to become as fluent and comfortable with it as you are with your pencil. Then you'll be BETTER able to play with ideas, and the mechanics of CAD system will be less limiting than the inaccuracy and messiness and size constraints and coffee stains on your napkin. When lines and circles and points of intersection and tangency flow from your mouse the way doodles now flow from your pencil, you'll think better, more easily, and more effectively.
To accomplish that level of comfort, you'll need two things. One, a CAD system that's easy and comfortable to use, and to get thoroughly used to. And two, lots of practice. Not necessarily structured practice; but the same kind of constant endless doodling that you now do on paper.
One of the reasons I still use AutoCAD Light '97 for much of my design work (despite JB's endless rants about how idiotic I am) is that I like it's UI, and I've spent so many zillions of hours with it that I can can capture ideas using only my fingers, and without distracting my brain. The CAD system is as natural for me as a pencil, but much more effective. I don't need to think about drawing; but only about what to draw. And when I get even the roughest sketch onto my screen, it's a better sketch, more useful, and more easily played with, than anything I could do on a napkin, notepad, or drafting board. It's also more immediate and spontaneous than what I do with more capable CAD software, which still demands my attention for its own needs. 3D shapes? Fitting things together? Test assemblies that actually look like they might work? Later. First I gotta get this thing working in my head. My old and outdated AutoCAD does that for me like no "stronger" system ever has.
I once attended a dinner party for a club that my wife belonged to. She knew all the other club members, but I knew nobody; and even my wife didn't know any of the spouses. While sitting around a table of 10 or 12 people, enjoying coffee after dinner, talk turned to something or other that some folks had questions about, and others offered to explain. Instantly, three people at the table, including me, reached into our coat pockets, pulled out our pens, and moved our coffee cups off of the paper napkins they'd been served on. My wife laughed out loud. "You can always tell the engineers in any crowd," she said. "They can't talk or think without a pen or pencil."
And she was dead right - about the people at the table, and about the general observation. Thinking - especially the kind involved in design work of any kind - necessarily involves capturing what we "see" in our minds. We need to grasp things that would otherwise slip away, store them outside our heads so our minds are free to keep running forward, and then look at our ideas as a way to understand them, manipulate them, and begin hunting for possibilities our original thoughts had only promised; but not made clear.
Getting ideas onto paper (or screen) really is a critical part of the process. Napkins have their virtues, therefore. But so did slide-rules, and for many of the very same reasons. When was the last time you used a slipstick, even for rough calculations or estimates?
Pick a CAD system you can learn to use without effort, that you can play and doodle with. Save the high-powered software for later. You'll be amazed at how many napkins will be spared, and how much more robust your thought processes can become.
KG

  Quoted Text Here  
The interesting case is someone who thinks multidimensionally and has artistic ability but no training in drafting. Isometric pencil drawing is also a learned skill, especially foreshortened curves. You see the same distraction and frustration when they try to sketch lets say a rocking chair they are making.
My favorite CAD exercise is drawing a bolt, with realistic vee threads and chamfered edges on the head. My best time is 2 minutes.
jsw

  Quoted Text Here  
Readers Digest version: Industrial Engineers usually come up with the concept design for a product. The look, shape "feel" of the part with direction from the all mighty "Marketing people". That shape = CAD data is then passed off to other specific engineers to design the internal "guts" & fine details of the parts required to still maintain the original aesthetics from that Industrial design. Having that info in CAD data IMO is very important for down stream work now days.
Most all solid modelers are "napkin sketchers" Paint type programs are ok just to convey idea's but eventually it should go to a CAD system. So why not skip that time used in Paint Brush or Paint Shop Pro & go right to CAD modeling? That way there's no misinterpretations as to the intent. You've got something more than a pretty picture. SLA's can be made , prototype parts can be machined off of that CAD data. No physical detail drawings are really required until the product is released for production. Accurate renderings can be made from any viewing angle to please those marketing people & do power point presentations for the "big guys".
High end graphic arts software like Corel Draw & Adobe are slick & able to export DXF files. But using that software efficiently is a whole other profession.
Heck we're machinists here, I'll stick with solid modelers any day of the week, unless of course you want to paste a foil beanie on picture of someone & post it on the net <g>
just my 2 cents (now worth about 1cent in this economy)

  Quoted Text Here  
You should see what anautomotive engine cover might look like as esigned/sketched bythe "stylist" in charge ... then compare with what will actually fit under the hood over the engine & it's bits <G>. Without excessivly denting the (we got it closed, eight?) hood if you rev the engine ...

  Quoted Text Here  
Jon, how can you model something if you don't first have a good idea or understanding of what it is you're trying to create?