Interesting. Back in 2000 I was involved with the design of a ballasted timber railroad bridge. The deck was made of longitudinal parallams stressed together with transverse prestressing rods. Luckily before the thing got built, we found some literature that said that parallams can swell transversely as much as 8%. We were able to specify certain manufacturing details and wood species to minimize swelling. We also put plate spacers between parallams to give the swelling a place to go. What a nightmare it would have been if the prestressing rods failed and went shooting down the river. At the time, I did a lot of research into stress laminated decks and did not see much about swelling.
My understanding is the deck was constructed of about 3' or 4' sections of some type of glulam lumber laying on edge. I do not know what species of wood it was. There was no room left anywhere for the wood to expand and contract. A clip is attached to a recessed bolt in the lumber. The clip goes under the top flange of the stringers to hold the deck onto the stringers. Apparently as the timber expanded the clips just slid along the top flange, they weren't enough to restrain the expansion forces.
The bridge is over an interstate, so I imagine there couldn't be any openings in the deck to protect traffic below. I'm not sure why no one thought about the expansion and contraction of the wood.
