There are other alternatives too, such as allowing the tubing to move sideways a bit in a long curve. That would change the overall shape of the trajectory but the air cushion they talk about could take care of that.
If they do end up using expansion joints they could actually use the fact that some air might leak in to purposefully increase local air density to use that as an impromptu air bearing.
Anyway, there is plenty of challenging engineering in hyper loop, but the least interesting part is how to make a long metal tube, either in one piece or out of segments.
And of course even more challenging than the engineering will be the economics and the paper tigers.
I hadn't considered this strategy! Using the curves as a geometric asset instead of a liability makes a lot of sense.
What do you mean by the 'paper tigers'?
Right-of-way issues, liability. America loves its paperwork and setting up a new very large chunk of unproven infrastructure is going to be a total headache from a paper point of view.
Hyperloop has all the drawbacks of high speed rail and has unproven tech on top of that. It'll be a legal feeding frenzy.
If you care to, check out the conversation that sillysaurus3 and I had down-thread. I would love to have someone experienced in MechE glance at my napkin math!
TL;DR it looks like simply pushing all the slack to the end stations is actually feasible, and it's easier to deal with there since the cars aren't moving very fast.