The reason this contradicts your reading is likely because, as others have commented, it's not a real space elevator. It doesn't go to space, and it is more like a super-tall building with an interesting adaptive stability system than a tether connected to a counterweight as in traditional space elevator designs.
So this also means it may not be prior art. The innovation is located in however it is they plan to keep a 20km tall structure stable and attach an electric elevator for rockets to it.
I don't think a model is necessary for a patent application, but some check on bullshit is necessary. If you are granted the patent, and following the instructions in it does not yield a working device--or if the instructions cannot be followed, because they require an impossibility or something that does not yet exist--that should be grounds for immediate disqualification.
There has to be some way to encourage "this is how you do it" over "so I have this really cool idea that I think is worth lots of money".
Counterexample: designing TCP/IP was an expensive, R&D-intensive process, but it's pretty easy to write a TCP/IP stack given the RFCs.
The written description also has to be sufficient enough to show that you actually possess the invention when the patent is filed.
So technically, if this can't be built today because we don't know of any materials that can actually behave this way, it's not a valid patent.
I'm not totally sure what happens in cases where the invention is possible but so prohibitively expensive that it's not really possible. I'd imagine that is valid.
Traditionally, invalidating a patent based on written description or enablement issues was hard. But the Fed Circuit has made it easier in recent decades.
I'm pretty sure it wouldn't be valid, but I'm not familiar with the sort of building materials necessary to build a space elevator. My understanding is that nothing we have today could do it, but that's based on light reading.
EDIT: I'd also be willing to bet that some popscience mag or text book discloses this general idea anyway, so it'd be invalid based on prior art too.