I don't like this idea. It seems like it would place too much importance on a random event. If you're lucky, you get a big advantage; if you're unlucky, you get screwed. In a strategy game, a risky move should be one that exposes you to potential punishment from other players' actions, not from a roll of the dice.
OTC does have random events, like Marsquakes and duststorms. But they mostly are things that players have to react to, not things that they can prepare for in hopes of getting lucky. That's the right approach. This volcano idea is different. There's no reacting to changing circumstances; instead, you just build geothermals and then either win if you get lucky and lose if you get unlucky. Or, you don't build geothermals, and lose if the guy who did build them gets lucky and win if he gets unlucky.
Here's an alternate idea: Geothermals get damaged over time due to seismic / volcanic activity, either at a constant rate (possibly randomized per map) or in a series of small, intermittent, frequent bursts. The damage manifests as requiring water as an input to the plant's operation. It starts off small, requiring 0.1 or 0.05 water to make 1 power, but it builds up over time. The plant can be repaired at any time, just like a destroyed facility can be, and for the same cost in resources and downtime. Depending on the how damaged the plant is, and depending on the relative prices of power and water, the plant might become unprofitable to operate. Players with geothermals would have to react to the intermittent accumulation of damage by monitoring the damage, monitoring the relative prices of power and water, deciding if and when to shut down the plant, and deciding if and when to begin repairs.
