I believe I covered the shroud business in an earlier chapter. The best improvement I got was 0.5c. But sometimes a shroud worsened the cooling. I think it has to do with the spread of air from the push fan moving faster when unobstructed. The frame of the shroud may produce islands or a ring of obstruction, and like the bow of a ship cause a reduction of airflow speed in front of the obstruction. So I quit using a shroud.
One thing that came out very clearly was when I put my Mechatronics on push, there was a lot of what I call side spill. I could feel lots of airflow on both sides. Then I put one of the Deltas on pull, and there was a lot less side spill. It makes sense of course, but having such a dramatic demonstration was . . . dramatic.
Noctua's own fans: Actually I think the TY-140 is a better choice for center fan. Ironically, this may be due to its having wide intake and exhaust flares -- when the 160mm side is installed to make the fan wide rather than tall I got better results.
Overall, I think the D14 benefits when the edges of its fins get airflow. When I wrapped up the sides the cooling got worse.
I think between Chapters 4, 5 and 6 you can see the tradeoffs between quiet and cooling performance. Chunky is right when he points to the heatsink being optimized for quiet.
As for fans, a pair of PFB1212UHE's would get used once or twice on the D14 for testing the extreme datapoint, and then a few times on the upcoming Mugen 2 and the revisit to the Megahalems.
Upcoming for the D14 is the PWM testing. I was really hoping I would have some Noctua PWM fans by now.