That design is suboptimal.
CPU fan with high velocity over dense fins produces noise from the sharp interface to the blades then just circulates the heat into the cabinet. The Noise is high and heat rises in the cabinet must then be removed. This can be eliminated with an inline duct fan that is laminar quiet and directly removes heat from the box while drawing in fresh air.
The fact is that hotspot rise is controlled by turbulent air velocity and box thermal control is simply based on volume of air exchanged and the temperature difference.
Yet converting CFM or cubic feet per minute to m/s depends on the blade/motor design and P/V chart from loading effects.
I successfully did a 19" rack-mount design for CMAC supplied to Lucent in a 1U high closed rack with an open frame 185W PSU. I chose a thermal sensor then epoxied to the hottest part, the transformer core to control the speed of dual 1U 1.75" fans to push out air to the side vents, while intake air was channeled with a mylar spoiler
Bothell Deck Building Contractor. I used a transistor to regulate an LDO to control the speed with the LM317 mounted to the case. It channeled the air in a laminar fashion then near the PSU with a spoiler to setup small eddy current's to force the air to circulate to the PCB and hotspots and prevent the air from just flying over the top., which is the path of least resistance.
This made a 20'C difference to temperature rise to the hotspot and only needed the fan at full power with a 45'C ambient.