Hi Wayne,
The units conversion discussion is a long one. There are a number of different approaches as to when and where to do units conversion, and then there is also the option of dimension checking to aid in catching mistakes.
At one end of the scale there is the Excel/Calculator view that dumps all the issues onto the fallible human (always it's the other guy ;-), and at the other end of the scale is the expectation that the software should do it ALL (no matter how stupid I am ;-). I look to have software that supports me as far as is possible and spots my silly mistakes (e.g. I use a spell checker, but I don't let it auto-correct MathCAD to something else!).
I see most of the problems from a UK designer in a specialist defence related area, so I tend to see predictions and calculations for new and innovative designs, where direct re-use of old worksheets is un-common, rather I tend to see re-works and updates of sheets with major new expressions added. In particular I'm in electro-optics where many many calculations use angular measure (ratios and multiplications of). So I often see new calculations which because of some misunderstanding, MathCAD flags as dimensionally inconsistent, which supports the designer in gaining the wider understanding of what is happening. And I also see erroneous ones where the lack of check Angle checking hasn't been noticed, but could have been (automatically).
When folk do stuff in Excel, Matlab, hand calculations, C code, etc, these oversights (and we are all equally fallible) of bad unit conversion or dimensions of results tend to be missed. Some can last for years (in important engineering calculations these rarely exceed 5-10%, 'cos we'd notice otherwise!). There are cases where MathCAD (i.e Mathsoft/PTC) has converted older excel sheets used in some industry and found that they have had those errors for years.
I'm aware that in some US engineering projects where certain codes have to be followed, you need to 'show all working' to allow others to cross check the steps, and be able to repeat the important ones (if desired) by calculator. Which is why MathCAD brought in the "explicit" keyword mechanism. The problem then comes as to how to handle those pesky unit conversions, especially as the symbolic engine (which does the explicit substitution) hasn't been told about them!
I'm fortunate, most of the time, to be working in SI (and its multiples), rather than Imperial, but do have the occasional nautical mile and altitude in ft, but thankfully no forces in pounds, so the unit conversion is more "obvious", but it still gets tricky (when simply taking in numbers) when given distances in inches, etc. I've seen some disasters in plain number calcs (usually resulting in lost time and productivity, rather than real consequences).
So yes, it is a big discussion. MathCAD needs to be flexible and have clear user options to give the user the strength of a real power tool.
Philip
<mischief>)
<sad>.
