Essentially, the need for cooling is a product of the inherent inefficiencies of the internal combustion engine. Even an advanced modern F1 engine is relatively inefficient when it comes to converting the power available from the fuel/air mixture into power at the rear wheels. This is measured in terms of 'thermal efficiency', and is typically in the region of 30%: that's to say, if a typical F1 engine produces slightly under 650 KW (approx. 850 bhp) on the dyno, something like 1500KW (or potentially 2000 bhp) of the energy is lost.
So where does it go? A small percentage is turned into the distinctive sound of an F1 car. The vast majority, though, must be dissipated as heat from a number of areas: for example, the oil dissipates around 120KW and the water system 160KW. The inefficiencies of the gearbox will mean it has to dissipate around 15KW, while the hydraulics represent a further 3 KW. However, as much as 34% of the remainder is lost through the exhausts as heat, while up to 15% of the available energy can be accounted for in unburnt fuel.
Moreover, this energy wasteage provides significant challenges when it comes to controlling temperatures. While the heat exchangers on a racing car are extremely efficient, their ability to cool the engine is a function of the 'air-side capacity' - essentially, how big a mass of air you can make flow through the radiator for a given area. This depends, of course, on generating high air velocities in the radiator intake ducts: however, typically, air velocity in the radiator ducts will only be 10-15% of the car's velocity, so even if the car is travelling at 300 kph, the air in the ducts is probably only at 30-35 kph.
Furthermore, temperatures in the oil and water systems vary according to different criteria: water temperature is a function of the average power used around the circuit, while oil temperature is approximately a function of power and also average engine speed around the lap.
Given how complicated cooling management is, you need a good reason to tackle so many contradictory problems, and that reason is aerodynamics. Essentially, we must find the correct balance between cooling and aero performance because the more air we channel through the radiators, the less efficient the overall aerodynamics become. In fact, changing between minimum and maximum cooling can reduce downforce by as much as 5%, which translates to a lap-time deficit of around 0.4s on an average circuit.
Airflow is controlled by different configurations of radiator outlet, and the R24 has 13 different possible configurations to cope with all manner of conditions. The configuration used at a particular circuit is defined according to the ambient temperatures, 'circuit factors' such as how much full throttle is used, and the temperature limits we can run the engine at.
Typically, we will run oil temperatures of over 100°C, while pressurising the water system at up to 3.75 bar allows the boiling point to be delayed until around 120°C: running these higher temperatures means we require less airflow through the radiators, thus improving aerodynamic performance. As ever, though, these choices carry a penalty: each extra 5°C of water temperature we run, allowing the radiator outlets to be smaller, robs the engine of over 1 bhp.
However, the importance of aerodynamics in modern F1 means we continue devoting significant resources and wind tunnel time to cooling. This is nowhere better illustrated than by the fact that the penalty in terms of aero efficiency we must accept for a 10°C drop in car temperatures, is 80% smaller than it was just four years ago.