For what it's worth, I often have piles of sorted documents to sort together, so I've used merge sort on RL objects fairly regularly. Binary search, too, though I use the optimization where it's not strictly binary when I have a guess about where the item should be.
I've probably used a few other algorithms, but I can't think of any off the top of my head.
This insight is courtesy of a sorting algorithm that's optimal for systems with very limited memory but multiple tape systems, if I remember correctly. Also IIRC insertion sort can beat doing the binary sort in situations of 7-20 or so? Much over a dozen, I can't use my fingers of one hand as bookmarks to do, say, three comparisons to find the right place, & I'd rather break it down into more piles.
First pass: put raports in the bins year by year maintaining sort order in the bins. Binary search used to place new report in the bin.
Second pass: put year by year sorted reports in the correct order on the shelfs.
I don't remeber the name of the algorithm used and the book from which it come was given to the library.