Simulation: Fact or Fantasy?

Joel, Thanks for your commentary and your compliments. I am the author of that paper, so please allow me to clarify. (The paper is available here, registration required: http://www2.spaceclaim.com/LearnMoreNow/WhitePapers/CAE-White-Paper-Down...)

My writing treats "simulation" as an equivalent term for "computer-aided-engineering" (CAE). Although that simplification holds for the context of the paper, it is not entirely true. CAE is a subset of simulation. Instead of saying, "engineers can quickly understand the design approach that will best meet performance objectives...," I could have said "...are likely to best meet performance objectives...." Your feedback is appropriate.

However, the spirit of the paper is not limited to computer-based simulation. For example, we learn dimensional analysis so we can build and test scale models that simulate how full-scale models will perform. We create mockups so we can test usability before committing to a product concept. From aircraft to toys, early-phase optimization is critical to product success. But the idea of simulation goes even further than that. For many products, the cost of modeling a detailed design in CAD exceeds the cost of making prototypes for destructive testing. In these cases, CAE may only offer incremental benefits over physical testing. Even physical testing of prototypes, however, cane be considered a form of simulation. Prototypes can be machined or 3D-printed, giving them different physical properties than cast or molded prodcution parts. And even production parts vary. In a sense, every test is a form of simulation.

In summary, test early, test often, and test any way you can. Test with CAE, test in the lab, and test in the field. Use whatever tools you have at your disposal to make sure your design ideas are going to work. Once you're confident in the basic idea, optimize the model. Then, detailed design can commence with a well-specified and highly-engineered concept.

-Blake

Joel, I've heard a couple great comments about the value of mistakes in engineering. The first was John Muskivitch who said mistakes are how we learn our limits. He'd rather drive across a bridge designed by someone who'd made a few mistakes and learned from them than a bridge designed by someone who'd never made a mistake because this person's luck may run out at the wrong time.

The other comment on a similar vein was at a PDMA conference by a professor from Stanford (can't recall his name) that said if you aren't making mistakes, you aren't trying to innovate. The true measure of an engineering organization isn't how many mistakes made but how quickly the mistakes are identified and the lessons put into practice.

A Gedankenexperiment--a thought experiment. Imagine you had the perfect simulation tool, that modeled the physics of your problem 'perfectly'. Set up your problem, the iteration algorithm is optimized, you arrive at the perfect solution. Do you think you might unintentionally miss any of the thousands of accidental discoveries that are due to nonoptimal solutions to your particular problem, but instead lead to an optimal solution to a problem you weren't even trying to solve? I believe you would. For instance, say you had the perfect chemical reaction simulator. You're trying to discover the chemical formula of a new refrigerant. You plug your initial conditions, boundary conditions, etc. into the perfect simulator, and crunch away. You found a new refrigerant, which MAY OR MAY NOT be better than current refrigerant. What you did NOT discover, if you were Roy Plunkett of Kinetic Chemicals, and you had this magical tool in 1938, is Teflon, which, like many chemicals, has bad and good qualities, but in the main, is one of the 20th centuries greatest discoveries in chemistry.