At my first engineering job, there was a phrase I came to hate: “Let’s sharpen our pencils.” It was intended to mean “if we redo these calculations more precisely, the numbers we get may be closer to our goal.” In practice, it meant a lot of work for me (since our work was done using pencils and calculators) that was unlikely to reach the desired goal, but that’s not why I was so unhappy with it. The problem is that phrase was based on deliberately conflating precision and direction.
It’s probably easiest if I use an example. Let’s say I’m looking at the possibility of adding solar panels to a 1920s building where I know the roof was designed for 40 pounds per square foot. (That was the standard snow load in NYC codes at that time.) A fast and easy way to feel confident that I’m not overloading the roof is to make sure that the total load I expect on the roof now is 40 psf or less. Current-code snow load starts around 21 or 22 psf in New York and goes up depending on various conditions. If the roof has parapets and a fair amount of bulkheads, for example, the drifts will affect most of the roof area. This can be calculated but I’d estimate it for a lot of old buildings around 30 psf. Interestingly enough, 30 psf was the snow load in the NYC code from 1968 to 2008, a period when drift effects were not generally considered. There are various types of solar panel, but most seem to weigh 5 to 8 psf; to be conservative I’ll use 8 psf. Unless I’m mistaken, 30 plus 8 is less than 40, so it seems like we’re okay. But much of the time the weight of roofing material on an old roof has increased over the years, and if there are multiple layers of old tar paper and gravel present, that could easily add another 5 to 10 psf. Suddenly we’re put to 48 psf and I have to deliver bad news. What to do? At that old job, the answer was “sharpen your pencils.”
To be more precise I can take the time to calculate drift at all the various obstructions and add those results together, rather than estimating drifted snow as 30 psf. I could get the exact weight of the solar panels, and we could try to measure the thickness of the roofing membrane to get a better estimate of its weight. The latter two ideas might actually work in reducing the total load, since I took the high end of the range for my earlier estimates. The first is more difficult to predict, since it depends on how much the drift from a parapet overlaps the drift from the stair bulkhead, and so on. It might be less than 30 psf total, it might be more. So doing the extra work will give me a more precise answer but not necessarily, in terms of the goal, a better answer. And note that I was trying to be conscientious and I used conservative estimates for the panels and the roofing. If I had chosen middle-of-the-range values, my total estimate would be 30 psf (snow estimate) plus 6.5 psf (middle panel estimate) plus 7.5 psf (middle roofing estimate) for a total of 44 psf. That’s still too high for the goal, but now all three factors that go into the total may go up or may go down when I switch from estimating to more precise calcs.
In other words, if you tried to be realistic with your first-pass rough numbers, redoing the calcs more precisely is about as likely to make things worse as to make them better. If you were very conservative with the first pass, more precision will likely make things better, but not necessarily more so than simply estimating realistically rather than conservatively.
In short, “sharpening your pencil” is not how you move towards a goal.
