Steel framing in buildings began in the 1880s, extending technology already in use for wrought-iron framing. That included hot-driven rivets for most connections, and bolts for lightly-loaded connections that would only have load applied in one direction, like beam-to-girder connections. Rivets shrank as they cooled, creating tension in the rivet shaft that clamped the pieces of metal (“plies”) together, while ordinary bolts did not have that tension. Welding in a recognizably modern form existed in the 1910s, but wasn’t widely used in buildings until the 1960s. High-strength bolts, which can be tightened until they have tension in the shaft, also became common starting in the 60s. High-strength bolts more or less replaced rivets on a one-for-one basis, keeping the same connection forms.
Those connection forms – used first with wrought iron and rivets, and then with steel and rivets, and now with steel and high-strength bolts, all have one thing in common: the connected plies are always parallel planes. You can only use steel fasteners to clamp together two parallel planes. In order to deal with three-dimensional reality, you need to create brackets using angles and other three-dimensional shapes. If you bolt a horizontal-plane leg of an angle to a beam flange, you can bolt the other leg, which will be in a vertical plane, to a column. By using multiple angles, you can turn any corner, and create any connection you need…at the cost of the connections becoming large brackets with many connector pieces and many rivets or bolts.
Welding changed this. Ordinary welds – the fillet welds between two pieces of steel that are the easiest welds to make – can join two plies at right angles to one another. The use of welding has eliminated those big brackets because we now can put the connector pieces in the orientation we want. Because welding is more expensive and slower than bolting, the most common way to make connections between two steel members is to weld connector pieces to one member in the shop, where the cost differential is low, and then bolt the connector pieces to the other member in the field, where the cost differential is high. So much of connection design consists of figuring out what connectors are needed to transfer load, and how to attach them.
The picture above is from an ordinary – i.e., unphotogenic – project. It’s one corner of a steel frame built to hold a new water tank, for a sprinkler system, above a mid-rise building. The frame is basically a table, with four legs (columns) supporting four edge beams, supporting some infill to hold the actual tank. The frame was designed, by Michael Lo and Gabi Pardo, without bracing, so the connections have to be able to resist bending moments from wind load on the tank, and for general lateral stability. The geometry is not symmetrical, so those moments are not the same in the east-west and north-south directions. The end result: the vertical reaction of each beam is taken by a single-plate connection, with a vertical plate welded to the column and bolted to the beam. The moments are carried by horizontal plates at the beam flanges, also welded to the column and bolted to the beam. The beam on the left has smaller moments, so fewer bolts are needed.
This is is bog-standard stuff for structural engineers, to the point where we sometimes forget that it can look odd to other people. Why bolts here and welds there? Why different size plates? Why do the beams and columns have that shape? The answers are a combination of looking for structural efficiency, looking for cost efficiency, and practical geometry.