Almost everything done with the construction of large underground caverns is from experience. I was involved with the field I now call Rock Dynamics, and it's time for some injection of physics and the Scientific Method.
This picture represents the basic large configuration of an underground powerhouse. There are limits to the size of the cavern, and limits imposed by the quality of the rock. In general, you need very good rock to construct large openings.
Experience only gets you to where you've been before. The Scientific Method can advance things. I want the caverns to go into tectonic areas. This is where the real benefit can happen. A nice active mountain range provides lots of elevation for hydro power. Right now, we are at a point where we can do large caverns in good rock, in active mountain building.
That requires knowing your basic physics. The societal myth is that you can't build in an earthquake zone, but the SM should address that. On preliminary evidence, caverns cannot be bothered by seismic waves, but don't do well if a rupture goes right through it.
The testable hypothesis is such: It is impossible for a cavern to be damaged by a seismic wave, if it is a hard inclusion in a softer rock. This is like a steel tube embedded in rubber, being hit by a rubber hammer.
If such a cavern would be built, I would embed instruments into everything. The cavern would be exposed to 10 minor events before a major. So much fun watching the seismic wave zoom right through the complex with the stress never going above ambient seismic noise.
Faults would be a different matter, but could be isolated. There is a mine in Canada with a greasy fault going through it, and you can watch it move. This is actually a good stress reliever and can be engineered.
