The data you put into a SAT solver has substantial effect on it's performance in practice. Real world inputs tend to be much nicer than pathological ones, so I don't think there's much practical application, but I also haven't read the paper yet.