Why is the "UCI hour record" using the normal drop position roughly 6 km/h slower than the "Superman" position ? Is the Obree-position really so much faster ? How much faster will I ride if I use the aero position instead of the normal drop position ? What happens if I use disc wheels? On this page you can calculate the effect.

First some basics:(we need some math to understand the effect...)
The total resistance opposing the motion of a cyclist on a flat terrain is the sum of the aerodynamic drag of air  F_Air and the rolling resistance  F_Roll.
The wind resistance F_Air (in Newton) depends on the aerodynamic drag coefficient cw, the projected frontal area A of bike and rider (in square meters), the air density r (about 1.225 kg/m³ at sea level for 15° C), and the square of the air speed v (in m/sec).
The air density is important if one rides at different altitudes, for a detailed view of this topic look at my hour record at altitude page.
Since it is difficult to quantify the area A precisely, one associates cw and A  to the effective frontal area cwA.
F_Air = 0.5 ×cw×A×r×v² =  0.5 ×cwA×r×v²
The power needed to overcome the air resistance is F_Air multiplied with the velocity v:
P_Air = F_Air×v = 0.5 ×cwA×r×v³
Typical values for cwA are around 0.25 m² (see table below). Example: For a cwA value of 0.25 m² you need 92 Watts to overcome the air resistance at 30 km/h, that's quite easy... But if you double the speed to 60 km/h, you would need a factor of 2³ = 8 more, that's 736 Watts!

The rolling resistance  F_Roll depends on on the rolling coefficient cr, the mass m of the rider in kg, and the acceleration due to gravity g (9.81 m/s²). It is independent of the velocity v.
F_Roll = cr×m×g
Values for cr are around 0.003-0.006 for racing bikes on a track, giving 3-5 Newton as rolling resistance for a 70 kg rider plus 10 kg bicycle.
Of course the power needed to overcome the rolling resistance depends on the velocity v:
P_Roll = F_Roll×v= cr×m×g×v
Putting in some values you'll see that at low speed the rolling resistance is not negligible compared to the aerodynamic drag. But when you go faster, the rolling resistance gets less and less important compared to the aerodynamic drag.

The total power needed is then:   P_tot = P_Air + P_Roll

Though the efficiency Eff of the bicycle is quite high (around 95-98%), but not equal to 100%  due to losses in the drivetrain and hubs, the power output the rider has to deliver is:  P_Rider = P_tot / Eff

Using those equations we can caclulate how much power we need for a given input. Most important at high speeds is the value for cwA.

In the literature, magazines or the internet one will find cwA values for upright position of about 0.37 m²,  for a dropped position about 0.28 m², for a time trial position about 0.22 m², and for the Obree position about 0.18 m²
A specific bike setup with standard 36-spoke wheels might have 0.27 m² while the same setup with disc wheels will have 0.22 m².
For more details see the tables below.

Now let's calculate the power! (Efficiency of the chain = 98%) (Click on button "Power")
You can calculate the power for two different setups (change cwA, cr, or the mass of the rider plus bike). If the values on the right side are smaller than on the left side, the power will be smaller. If you click on the button "Calculate Improvement Table", a window will open which shows how much faster the rider will be. Clicking "Plot sec per km" will create a plot showing how many seconds per kilometer one gains for a given speed, clicking "Plot km/h per km" will do a plot showing the gain in km/h.

Values for cwA (in m²)

Table from F. Malizia & B. Blocken: Cyclist aerodynamics through time: Better, faster, stronger
Drag area results from Windtunnel (WT) tests, CFD simulations and field tests for different cyclist positions:
upright position (UP), brake-hoods position (BHP), dropped position (DP), time trial position (TT), Obree position (OP) and superman position (SP).

^a cwA values are calculated assuming an air density of 1.225 kg/m³.
^b Blockage ratio larger than 10%.
^c Results of the study divided in two rows because the large drag area differences between the two riders tested.
^d CFD simulations including only the cyclist body. The bicycle was not included.
^e Field tests using PTV measurements in an indoor environment.
^f Field tests using PTV measurements in an outdoor environment.
^g Static simulations performed at different leg positions.

Bike with aerobar setup. Tested were different wheels on a track using a SRM powermeter at 45 km/h. cwA is calculated with F_Roll= 4,5 Newton and 98% Efficiency.