As I understand, two fans with the same diameter and with the same rotor speed(usually measured in RPM) can have different air pressure(usually measured in mmH2O) and air speed(usually measured in CFM) because of fan blades design. Higher air flow speed should provide worse penetration(for example, penetration to radiator with thin fin gaps) abilities. However, it is counterintuitive to imagine that air molecules with higher speed have smaller penetration ability than the ones with lower speed. Why is that so? Or is it because of complexity of air flow dynamics and air flow with slower speed has indeed for some (weird) reason better penetration abilities? At least I know that this is not because of Bernoulli's principle because this should apply only for tubes.
3 Answers
The most likely explanation is that at lower speeds you get a more laminar flow. To quote the wiki article:
At low velocities, the fluid tends to flow without lateral mixing, and adjacent layers slide past one another like playing cards. There are no cross-currents perpendicular to the direction of flow, nor eddies or swirls of fluids.[2] In laminar flow, the motion of the particles of the fluid is very orderly with all particles moving in straight lines parallel to the pipe walls.[3] Laminar flow is a flow regime characterized by high momentum diffusion and low momentum convection.
Background
The issue of higher speeds having smaller penetration results from the ram pressure, which scales as: $$ P = \frac{1}{2} \rho \ v^{2} $$ where $\rho$ is the mass density and $v$ is the flow speed. Even if you have thin fins in a radiator, they can start to "look like" sails to the incident flow if the flow moves fast enough causing the drag to increase.
You can see a similar effect if you watch pine trees under different wind conditions compared to deciduous trees (i.e., leafy trees). At low wind speeds, many types of pine trees will show little movement because the air can easily pass between the needles. At higher speeds, suddenly they begin to sway/move just like the deciduous trees. I noticed this when I saw that a small pine tree had been uprooted during a tornado during graduate school. I was initially puzzled how the wind had enough force to actually accomplish such a feat (since it was a weak, F0, tornado).
I then recalled that ram pressure is $\propto \ v^{2}$ and that the Reynolds number scales as: $$ Re = \frac{ v \ L }{ \nu } $$ where $L$ is linear the scale size of the obstacle and $\nu$ is the kinematic viscosity. The easy way to think about $Re$ is that for large values (i.e., $> 10^{3}$) the flow can become rather turbulent downstream of the obstacle.
So for your radiator fins, after a certain inflow speed the air flow over the fins will start to become unstable at the immediate downstream edge. This will start to cause a kind of "pile up" that will disrupt the incident flow and impede the penetration depth of the air.
Answer
So the answer is that the lower penetration depth of air flow for higher inflow speeds is a combination of the effects of ram pressure an increasing Reynolds number.
However, it is counterintuitive to imagine that air molecules with higher speed have smaller penetration ability than the ones with lower speed. Why is that so?
The velocity has an influence, but it is almost purely an issue of the shape of object which it needs to "penetrate". This influence is not even linear, but can even decrease when the velocity increases. It's all well educated in this old video; Fluid Dynamic of Drag, (Part I) At 8 min it's well shown.
Or is it because of complexity of air flow dynamics and air flow with slower speed has indeed for some (weird) reason better penetration abilities?
The decreased penetration ability comes mainly from so called "separation of the boudary layer", which shortly said simply causes a reduction for the open area of flow.
At least I know that this is not because of Bernoulli's principle because this should apply only for tubes.
Bernoulli apply everywhere.
ANSWER: Higher air speed $V$ always provides more "penetration" as this penetration is simply flow amount $Q$ through Area $A$, and $V/A=Q$. This "penetration"/flow might be really unefficient, the efficiency can be seen from the rise in the pressure. And basically the pressure generally rises along the velocity. But there is an optimum point. In water flow this is defined as a Froude number. At certain airspeed the area might be blocked because of separation, and if there is possibility to bypass flow, then the "penetration" is reduced.
It's not reasonable to answer more broadly to a question in this level; here's some material to study further;
