I'd recommend 140mm fans for this case due to the third 120mm being cut off from the psu shroud. You'd get less airflow to the motherboard area than 2x 140s because of that. In more open cases, I would recommend 3x 120s because they do flow more than 2 140s as well as having better air pressure due to the extra fan. While thermalbench shows the SW3 120 and 140 performing roughly the same when multiplied x3 and x2 respectively, that is airflow through a radiator. Restriction through say a dust filter and maybe a covered front panel will paint a different story.
Unfortunately, I can't find any websites out there to back up my claims. Cooling technique does have open airflow measurements of the
ML120 and
ML140, but it is that - open airflow - and we need some restriction in front of the fan for a scenario that will more accurately reflect performance in a computer case.
But since that's all we have right now other than Thermalbench's ML120/140 and SW3 120/140 radiator airflow tests (has to be these fans in particular since the 140mm version is a scaled up 120mm, so it's as apples to apples as we can get), let's compare results from those links up top.
~13dba normalized
ML120
50.4cfm
0.889mmh20
ML140
49.8cfm
0.699mmh20
~26dba normalized
ML120
89.1cfm
2.921mmh20
ML140
89.8cfm
2.413mmh20
You multiply the 120 results by 3 and 140 by 2 and you start to see why I think that 3x 120s are superior. While a third fan will introduce more noise, with a performance gap that large between the two, you can lower fan speeds slightly and still come out in the positive.
So, we have two extremes. One shows 3x120s performing roughly equally to 2x 140s on radiators, and one showing 3x 120s performing much better than 2x140s in open air. The TB test can give us an idea of a worst case scenario where the intake area is severely obstructed, so this could apply to your cases with dust filters as well as a covered front panel with little ventilation on the sides. The coolingtechnique results can give us a good idea on how these fans would flow in a less restrictive and open case.
Are both extremes accurate enough for what we're trying to gauge? No, but at least we have some boundaries to work with. At the worst, the 3x 120s can perform the same if not a little worse due to the extra noise of a third fan compared to 2x 140s. At best, the 120s will perform much better even when accommodating for the extra noise of a third fan. But...we still have a lot of more variables that need to be accounted for even still. Air cooling? Water cooling? Graphics card(s) and cooler type? Case design? It really is a tricky question to answer.
You can look at it from a reliability perspective as well. Two is less than three, so one less point of failure. It is also cheaper. Oh, and there is one more thing to consider when it comes to cases with a closed off front panel with side ventilation. There is an efficiency curve for that front panel, and once you pass that threshold, you will have diminishing returns on airflow performance. For instance, the Fractal R6's front panel efficiency may peak at say 100cfm. If you go beyond that, the airflow to noise scaling will be a lot worse with a lot more noise/rpm generated to get more air through those vents. Kinda like with processors. Some hit a voltage wall, and in order to get higher clockspeeds, you'll have to put in a lot more voltage than usual to get there. Ie. 5ghz at 1.35v, but 5.2ghz might need 1.45v when you only needed 1.32v at 4.9ghz. That efficiency curve can potentially put 3x 120s in a bad light because even though they're capable of more airflow, they are limited by the front panel's airflow efficiency/potential.
