Skip links

One-Sided Accuracy

I was going to wrap up this week’s discussion of loading with my fifth rant about accuracy (see here) but I have a better idea. Engineering as a whole (including subfields other than structure) goes from extraordinary precision (the circuits on integrated chips; the paths plotted for interplanetary rockets) up to building-construction tolerances of “close enough.” Structure is at the imprecise end of the spectrum, with allowable tolerances measured in quarters and halves of inches, but we get things pretty close. After all, if the structure were badly out of alignment, the architectural elements that require accuracy, like glass curtain walls, wouldn’t fit because they only have a limited ability to be adapted to imprecise contextual geometry.

Despite inaccuracy issues and construction tolerances, we have a pretty good idea of the strength of our structures. The simplifying assumptions we make in modeling are generally not so far from reality. What we don’t know and generally cannot know, is the actual loading on them. Every structural analysis has two sides: capacity and demand. I’ve discussed this week how we don’t really know the demand for occupancy loading very well even though we have a reasonable idea of the maximum loading for a given occupancy. We have a reasonably good idea of the maximum expected wind loads across the US, statistically speaking, but of course at any given moment the actual wind load is likely to be much, much lower. (Wind pressure varies by the square of the wind speed, so if the expected 100-year-recurrent wind has a speed of 100 miles per hour and that’s the basis for the code wind load, a hurricane with 80 mph winds only exerts 64 percent of the design pressure.) We have some idea about seismic loading, but the fact that we revise the loading after every major quake suggests we’re still learning that.

So we know capacity pretty accurately but we don’t know demand very well because we’re still learning the natural forces (and they change over time) and we can’t know demand very well because anything other than a defined statical maximum is a random variable on any given day. An engineer working in design alternates between the two sides of the equation, one accurate and one not, multiple times per day, and sees nothing out of the ordinary about it. It’s when we interact with people not in the AEC community that this issue become difficult to explain. In lawsuits, for example, engineers are often asked “Is such and such true to a reasonable engineering certainty?” The problem with the question is not that engineers can’t answer it, it’s that our answer means something different to us than it does to non-engineers unless we explain our logic, and that explanation may be unwelcome for reasons that have nothing to do with engineering analysis.

Above, the timber framing of the Schwab mansion on Riverside Drive, under construction. The house was replaced by an apartment house long ago, or, in engineering terms, we went from a structure mostly designed by rule of thumb to one with defined loading and capacity.