Once in a while, I read something on someone else’s blog that makes me say out loud “Yes! That’s it!” Fortunately, these days I’m mostly alone in a room, so I don’t have to explain to anyone why I’m talking to a screen. This most recently happened when reading “Why Structural Engineers Use Excel” by Maryanne Clare. She’s a structural engineer by former training and occupation now (as far as I know from reading her public comments) moved into software development. In other words, she’s well-qualified to discuss the dismal state of structural engineering calculations.
For the non-engineers reading this, a side note: engineering, as performed on a daily basis, is less math intensive than you probably think. The page above, from a 1908 textbook, shows a method of solving statically-determinate trusses. What might jump out at you is that the math is simple algebra. I don’t think my calculations have included anything more advanced than solving a quadratic equation or some trig in the last ten years, which means I’ve been using math I learned before tenth grade. Of course, I sometimes use FEM programs that contain a great deal of linear algebra in their code and I use (all the time) beam formulas based on differential equations but in both of those examples I’m not doing the hard work. I’m using someone else’s hard work. The beam formulas, for example, were solved in the nineteenth century, and there’s no reason to solve them again when we can simply use them. I remember being taught the derivation of one of the simpler cases in college but I couldn’t recreate that derivation now under any circumstances. The point of structural engineering practice is structures, not math. Math is a tool to be used in the simplest form that gets to the result.
If math is just a tool, then there’s nothing wrong with using a spreadsheet, everybody’s at-hand math tool, right? Wrong. Ms. Clare hits the high points on why engineers shouldn’t use excel, as does one of her sources, “Excel is Evil” by Colin Caprani. Her point about opacity is the one that bothers me the most, with the lack of proper unit handling being the next worst. Unit checks are one the simplest and most effective ways to see if your results make sense. (Unit checks are particularly useful in the imperial unit system, where missing inch-to-foot conversions can leave mistaken factors of 12 floating around.) We use three of the same alternates she mentions – Mathcad, Tedds, and Blockpad – because the commercial software market has limited options.
I can’t help but think that Excel remains so popular because it is heavily used by the business executives who make decisions about software development. It works for them, so it must be the right tool for everyone, right?