Computer Problems
I’m doing some heat transfer homework and, to check my assumptions, I just decided to plot the temperature distribution. I did this by solving the big complicated pde numerically using MAPLE and having it spit out a nice 3d plot. This took less time to type out and execute than for me to compose the problem by hand (take the problem and reduce it to something tractable with a big morass of assumptions) and I haven’t even plugged numbers in yet.
I wonder how long it will be before engineering courses no longer bother teaching these approximation methods.
For context: there are lots of graphical methods that engineers would have historically used which are largely not taught anymore, we hear about them but generally use computers instead.
Furthermore for any real problem the pdes are so complex that people universally use a numerical method. With the ubiquity of powerful computing (I have a matlab clone on my phone, and it is actually quite useful, not to mention things like wolfram alpha and the cloud in general) does it really make sense to learn the old fashioned way? Maybe we would be better served by spending more time on numerical methods.
Also: I’ve lamented before about how few people (even my peers in engineering) don’t really know how to make a computer solve their problems. Especially if there isn’t a specially made piece of software for it. While intro programming and numerical methods are two required courses, we rarely take what we learned in these and apply them in other classes. I think we might be better served as engineers if basic scripting and solving problems computationally was a part of every class. But then again, once you know how to code you see every problem as a computer problem.
Posted on November 1, 2011, in chemical engineering and tagged computer problems. Bookmark the permalink. 3 Comments.
I feel that the graphical methods give a good snapshot of the system and allow for quick estimation. They are very approximate, but they are an excellent tool to introduce concepts to undergrads.
“While intro programming and numerical methods are two required courses, we rarely take what we learned in these and apply them in other classes.”
I took these required courses but rarely had a chance to use them. I only have a basic understanding of coding (so maybe I missed the opportunities), but there were very few problems that I ran across that needed numerical techniques. Those that did could frequently be solved using Excel. Now that I’m working on more real life problems, I’m using commercial software (Aspen) that requires minimal coding knowledge. So, more coding knowledge would be nice but of limited use. As the chemical engineering curricula is so packed already, there isn’t the pressing need to justify more time to teach programming as well.
I was having a chat recently with other people from the dark side: those of us who learned to code when we were kids, and this came up. We turn every problem into a programming problem. I think it might be a case of that old adage about having a hammer and seeing everything as nails. While being able to quickly whip up some code is not the only tool in my toolbox it has been so useful in so many wide and disparate areas my first instinct when presented with a problem is to see how I could solve it with code.
I know people who treat excel the same way I treat perl and python, and I’ve seen some pretty amazing excel worksheets: One guy I know did a simple finite difference solution to a heat transfer problem entirely in excel with macros.
So yeah, practically speaking I should probably focus more on learning how to use HYSYS or some other simulation package than reinventing the wheel everytime.
Pingback: Solving an absorption tower problem using numpy and scipy « the factory