Yep, the thing I've found is that the idea and design stage is quite easy compared to the often unanticipated problems and complications that arise in actually constructing a working model.Lots of things to consider and the bubbler is actually almost as complex as the reactor itself.
Regarding freezing, I should imagine that dry cells are much more prone to problems due to their design. I used ordinary car anti-freeze in a wet cell I employed for a time, with seemingly no ill-effects.
I have to consider this with my new designs, but due to the fact that I will be employing cavitation and plasma discharges, I'm reluctant to use additives which under these conditions may just react unpredictably.