How to practice user centered design with no / limited access to users?

We have an interesting project which is a prototype for a hypothetical domain specific userbase which does not currently exist.

I'd like to use good design practices and wish to adapt our existing user driven approaches to the absence of actual users.

A few things seems like actual users could be replaced with suitably researched / estimated persona's but certain other techniques like validation of design decisions and gathering metrics seem near impossible.

Here's the best way I can describe the situation:

We have a customer who wants a system to repair flying cars. Flying car's don't exist (quite yet) but if we split the problem in two, we have cars and flying.

• Car repair is well understood and users can be located.
• Flying is well understood (but perhaps slightly less so being a newer / more complex technology) but users can also be located.

The combination of flying cars doesn't quite exist and neither do the people who might repair them (in that role). The users who might be the repair people could be mechanics, or aerospace engineers, or neither, or both.

We could create a persona to represent a mechanic, and this would be a fairly sensible approach, it's likely they would be involved in repairing flying cars.

However the process to repair a flying car may be significantly different from their current role, we could model the goal of replacing the tires, or changing the oil from the mechanic's perspective, but we can't be sure those roles would actually exist in the flying car.

Does anyone have any suggestions for this kind of situation?