We were building a fusion reactor prototype, and as far as I am aware the NEC code doesn't really have a section for that. I'm sure that much of what we did was probably against best practices.
We had a number of high voltage power supplies running in one room of our lab, and on days when the humidity in the air was low, the ionization in the air would charge up all the metal surfaces in the room. So, even if something was not connected to a high voltage, you would get a nasty shock when you touched it. All these devices were connected to the normal, "third prong" wall socket ground, but apparently the route through the building to ground was too resistive to handle the large amount of ionization in the air.
When working in a room with 180,000 volt power supplies, random shocks rapidly become more than annoying.
Spence, our electrical engineer, had the idea that the quickest way to fix the problem was to run a grounding rod into the soil right outside the window. We pounded a rod into the ground with a sledgehammer as far as we could, which was about 5 feet, I guess. Then we ran a 1/4" copper cable from the rod, through the window to connect to the main reactor chamber.
This worked great! At least for a couple of months. Our lab was located in a swamp, and the soil was basically a few meters of dried leaves. You could actually feel the soil move up and down when you walked on it. When August came around, it had not rained for quite a while, and the water level in the soil had dropped below the level of the grounding rod. The dried leaves in the soil insulated it quite well. We started getting shocks again, and it took us many hours of troubleshooting before we realized that the ground was not actually ground anymore.
We went out and pounded the rod a few feet deeper, which was easier now that the soil was dry. This reached the water table, and fixed the problem.