It’s never sat right with me since my first Statics-related course. When learning about reaction forces at a support, one type of support is a roller? Is this what allows an expansion joint to open, basically? hard to picture a bridge, for one, with one end floating free along a roller. Always bugged me…
Can some Civil Eng confirm if this is the correct mental picture?
The use of a theoretical roller in statics class is useful to illustrate that a member/end of the bridge is not rigidly connected to the bridge abutment or support. (IIRC, and it has been a long time since classes, it is impossible to solve a statics problem for a totally rigid bridge.) In the real world rollers are not used. Instead one side is not pinned/bolted/welded. The bridge beam is allowed to sit on a flat support on the abutment and move freely (fractions of an inch) as temperature expands and contracts the bridge materials. That’s why you are having a hard time picturing it. A roller holding a bridge would be doomed to fail – too much weight to support. The function can be adequately handled by flat plates resting on each other and being allowed to slide.
Caveat – I am a Civil Engineer. However, I do not design bridges but have designed road and trail approaches and worked with bridge designers. This very simplistic answer should work until a real bridge designer comes along. 
Hi again. I checked with the structures/bridge engineer at work. I was mostly right. In the classroom, assuming a frictionless bearing (roller) on one end allows analysis using statics. If a structure is pinned at both ends, it is “statically indeterminate.” It can still be analyzed using moment distribution which is learned in advanced classes and is done IRL. There are computer programs that do this very well.
When analyzing older bridges he calculates loads both ways and there is usually not much difference in the results. The reason a bridge may go from free floating on one end to being considered a solid support is that dirt, rocks and rust can get into the joint and make it so the two plates don’t slide across each other properly.
OK, I am a bridge engineer (or, actually, engineer-in-training; I don’t have my PE yet). In, Texas, we use steel reinforced bearing pads for the beam bearing. The bemas are designed as simple spans. As a rule of thumb, we place an expansion joint at least every 300 feet or so, which is usually two or three spans. The concrete slab atop the spans forms a continuous beam unit with an expansion joint at each end. The expansion joints are usually reinforced with steel to prevent the slab edges from crumbling. They’re easy to spot on a bridge; when you hear the “kachunk” as you enter a bridge, that’s your tires hitting the edges of the expansion joint. They are absolutely vital, I’ve seen pictures of a bridge where the contractor let concrete get down in the expansion joints and there’s major destructive cracking.