|
I'm not sure about the initial mistake either. But then again, this was a lesson in one of my structural classes, so you don't forget about that sort of issue. Basically, I'm surprised that no one did the basic worst case sigma=MY/I for the side and quartering loads. The old grizzled supervising engineer generally will run through a bunch of ballpark estimates to see if the designer has actually checked all the worst case combinations. I don't really buy the article's factor of safety discussion -- the factor of safety is for uncertainty, not blunders. I _might_ save you from a blunder, but that's not what it's for. It's for variation in material dimension, quality, and loading variation. As for the bolts vs welding, There's definitely scope for changes at the shop drawing stage. When an engineering firm designs a building, they'll generally design a set of typical connections, but when the shop goes to build them, they have to take this column:floor beam connection and actually turn it into "cut this cope, drill here, ..." and they'll generally do the simplest, cheapest thing they can. The engineers have to review the shop drawings, and if they don't have a good set of calcs for what the loads are in that particular bit, then they aren't going to necessarily see that this substitution of bolts won't work. (Note -- bolts can be just as good, and potentially better than welds, depending on what you're doing, how you're erecting, and the space you have available. Full Penetration welds are a pain, but they're the easiest way to specify a full strength join. But doing them on site, in the air, when you're putting the thing together is not something that you'd generally want to spec.) There have been a significant number of historical disasters connected to shop drawing changes that the engineers didn't pick up. (e.g., the Hyatt Regency walkway collapse). Also remember, this was done in the 70's, so they wouldn't have had Finite element models of the building where it's super easy to just run a couple more loads. Everything there would have been indeterminate elastic analysis, so it's not quick to run an accurate calculation if you're trying to find beam sizes. |