Slashdot Mirror
Getting Human Hands Back Into Digital Design
Hugh Pickens writes "Using computers to model the physical world has become increasingly common as products as diverse as cars and planes, pharmaceuticals and cellphones are almost entirely conceived, specified, and designed on a computer screen. Typically, only when these creations are nearly ready for mass manufacturing are prototypes made. But the NYTimes is running an interesting essay highlighting a little-noticed movement in the world of professional design and engineering: a renewed appreciation for manual labor, or innovating with the aid of human hands. 'A lot of people get lost in the world of computer simulation,' says Bill Burnett, executive director of the product design program at Stanford. 'You can't simulate everything.' Fifty years ago, tinkering with gadgets was routine for people drawn to engineering and invention, and making refinements with your own hands means 'you have to be extremely self-critical,' says Richard Sennett, whose book The Craftsman examines the importance of skilled manual labor. Even in highly abstract fields, like the design of next-generation electronic circuits, some people believe that hands-on experiences can enhance creativity. 'You need your hands to verify experimentally a technology that doesn't exist,' says Mario Paniccia, director of Intel's photonics technology lab."
Warning: TFA requires registration, or, bugmenot.
Give a man a fire and he's warm for a day. Light a man on fire and he's warm for the rest of his life. - Terry Pratchett
A computer model is far more useful than a piece of hardware on your desk. It does more and costs less.
One important aspect is measurements. There is no easy way to do any meaningful measurements in a microwave circuit unless it is specifically designed for that (and for nothing else.) However a CST or Ansoft model allows you to measure the field, or the current, or whatever else you want in any point of the model (and of the space around it, if you build an antenna, for example.) These measurements will be totally non-invasive, as opposed to a real-world probe that you would have to use. Some RF designs require hundreds of iterations before you achieve the desired compromise between all your design goals. Doing this in a computer will take a month. Doing this in metal will take 10 years.
Another advantage is in parametric design. Usually models are not hardcoded, but defined with a set of parameters (Excel for Autodesk Inventor, built-in spreadsheets for SolidWorks, etc.) You can manipulate these parameters and [almost] instantly see their effect. To do this in a real-world hardware you'd need weeks and thousands of dollars.
Per my current practice, the model is built only as a working prototype, when the design has been done and validated on the computer. This model can be also used as a sales demo, but the main purpose of building it is to verify the calculations, and the quality of the overall design (such as "can it be assembled?")
They sort of do it with cars but only for the sake of styling.
The number of prototypes for a new car has gone down significantly, but they're still used at all stages of the design process. Noise/vibration/harshness tests are one big area where models aren't accurate enough yet. Endurance tests are also still done.
The aviation industry is ahead in this regard because it's insanely expensive to build a prototype of an airliner.