It would probably just cause problems.
It can work in theory beacuse the "coding" part of DNA is just really 1 of the strands, the second strand is just there to make copying and error preventing etc easier. But adding more moving parts just makes errors that much more likely.
It would be more intersting to have more base pairs then to have more helixes.
Having more then 1 coding strand would cause problems because both strands would have to code for something non-detrimental at the same time, which is unlikely to work out, or you would need to evolve 2 seperate transcode mechanisms (1 different one per strand). Having 3+ helixes also would remove the incredible space saving self folding up DNA can do, which is bad for error prevention.
edit: i did find this on wikipedia tho http://en.wikipedia.org/wiki/G-quadruplex might be a fun read.