Will also compare shrouded, unshrouded, suck (pull), and blow (push) air-flows.
Was just messing with the equipment a bit last night, putting a Yate-Loon on a PA120.1 radiator with a shroud, in suck or blow arrangement. I was able to position the anemometer on the other side of the radiator to get an idea of the evenness of the air-flow through the radiator. Used the tachometer to measure the fan speeds, and the noise meter the fan noise levels. I haven't calibrated the fan noise levels yet to any base reference point, so I won't report the absolute values I got, just the relative values.
Free-air: +0dBA noise, 1480rpm speed
Shroud Suck: -2.7dBA noise (i.e. quieter), 1350rpm speed
Shroud Blow: -3.0dBA noise (i.e. quieter), 1460rpm speed
I found the noise level differences, and the fan speed differences to be quite intriguing. If the fan is spinning faster, it's pushing more air-flow in the blow-mode.
However, using the anemometer on the other side found that with the shroud, in blow mode, the air coming out the edges of the radiator had about an 80% higher velocity than the air coming out the dead-center. In suck mode, the overall air-velocity appeared lower (will need to measure properly later on full-testbed), but it was more even, with only about a 25% variation between the edges (higher) and the dead-center (lower).
So yeah - this throws a cloud over the general suck/blow debate. The fan appears to like to blow much better than to suck, but the air-flow is less even, even though it's of a higher velocity. The fan is also quieter in blow-mode.
It would appear (without further testing) that putting a fan into blow-mode on a radiator with a really deep shroud (60cm) to straighten out the air-flow would be the best way to go, but that's not terribly practical.
It seems to be a scenario of while blow creates more air-flow, the unevenness of the flow through the core hurts performance. Performance does not scale linearly with air-flow, and so it's always better to have even air-flow over the entire core area, rather than some areas receiving twice the air-velocity of other areas.