Ok, so I had 4 or 5 strands to my molecular biology PhD and I worked on all of these simultaneously. The idea was that if 1 or more of these turned out to be a dead end, I would still have 3 separate date chapters. It worked out that 3 of them got decent results (so I got 3 papers) and 2 of them only had basic info that went into thesis chapters with other data.
Out of these 5 strands, 4 were distinct ie using totally different methods and looking at different areas of biology and 1 used the same methods as another but with a different purpose. 2 out of these 5 took up most of my time, and the others were more long term experiments that ran in the background most of the time.
These projects were all related because they were all working with the same organism.
I think this was a good approach to take, because I too have seen students that basically have to start again with a different project in their third year because their previous projects weren't going to generate any data.