Uncertainty in Design

What they don't tell you in engineering school, as you are quietly studying how to do boxed problems with every even answer out of the back of the book, is that most problems are ill defined or not defined at all.  Some problems that you think are insignificant are, in fact, big and drive the design, and others that you think are huge; not so much.  And that's just in identifying the problem.  That says nothing about defining it, and it says nothing again about actually going about solving it.

Mechanical systems, I think, are more prone to this than problems in other engineering fields.  With electronics, the main design problem that is ill defined is the issue of "noise" or "EMI".  These are generally dealt with by guess-and-check; make the design using the best "design practices" that you can come up with, and check to see if it works.  It's an ill defined problem, but the problem is not a design feature.  In other words, it does not drive the design so much as constrain it.  The only design issue is to mitigate the problem; there is no optimization; the less of it the better.

Mechanical systems, and thermal systems in particular, have all this design uncertainty as a driving force in the design.  It is an optimization problem.  You cannot just make fans bigger and bigger and the enclosure bigger and bigger to make everything cooler.  Even if you could specify huge fans, there remains the problem of getting the airflow to where you need it.  it's just not that obvious.

So, how to deal with this engineering uncertainty?  That's the 64 million dollar question they don't really teach you in school.

There is guess and check.  Effective but expensive.  There is software.  There is back of the envelope correlations or sophisticated design tools.

Right now I am stuck on a design.  I don't know if it will work worth beans.  There are lots of components and lots of integration and it is all very confusing.  At some point I will just have to make the jump.  See if it works.