Air Cooled VS. Water Cooled

[rv6ejguy]

Hoerner's book outlines the differences in air and water cooled installations and as has been previously discussed, there are many factors involved in the entire process right from the heat sink process. What was not obvious to me was that the velocity of air past the typical air cooled engine fin was twice as high as on a typical rad setup and the heat transfer coefficient was substantially less per unit fin area so heat dissipation rates were little different despite the higher delta T. Radiators however show very high drag at high speeds unless flow is throttled. Air cooled cylinders show less drag penalty at high speed by comparison and were clearly superior with both in a fixed geometry outlet configuration.

His analysis stated that both air and liquid cooled installations could theoretically produce negative cooling drag (net thrust) if properly designed. The opposed engine offers lower frontal area than a radial but the cooling air path essentially turns through 180 degrees compared to straight through on radial and liquid cooled installations. Any time you turn air at high velocities, you have a pressure or momentum loss. If the cooling air exit velocity is less than the free stream velocity, you have net drag.

Hoerner explained in detail on how important throttling the cooling air (cowl flaps) was to reduce drag at high speed on both air and liquid cooling installations and showed that is was impossible to realize net thrust on either setup without them if cooling was to be adequate in climb also.

It's clear from many flying liquid cooled aircraft that ground cooling is worse with rad inlets close to the prop shank and best when the rads are wetted from air coming from the outer half of the prop disc. In flight, most forward facing components see about the same pressure so inlets close to the spinner work well in that respect.

Interestingly with improperly designed inlets and exits, inlet spillage at high speeds was quite common. This could be compared to an overflowing funnel where filling rate exceeds draining rate. This creates substantial drag.

Several examples of different rad placements and duct shapes show very large variations in drag despite only subtle differences.

[ergie63]

rv6ejguy:

I was wondering whether or not the fins in your radiator are louvered. In automotive apps they are used to promote turbulence to enhance heat transfer. Is this also true in aircraft apps? Turbulence, it seems, is anathema to aircraft design.

[rv6ejguy]

Yes, the fins are louvered on my heat exchangers. The flow within all radiator systems is clearly within the turbulent range anyway so this is not a factor in itself but flow bench studies show that louvered fins do create about 20% more pressure drop than unlouvered fins.

It should be noted that the rads on WW2 aircraft where much of the data originates were generally not louvered and were of thicker copper construction in most cases. While copper has higher thermal conductivity than aluminum, the solder joints interfered considerably with heat transfer. The modern Visteon type furnace brazed aluminum cores with louvered fins are at least 25% more efficient per unit face area than WW2 era copper rads. This would result in less overall drag and potentially higher Delta T on the inlet vs. outlet air compared to 60 year old designs.

It is interesting to compare percentages of glycol and water vs. heat transfer coefficient, boiling points and Delta T. While high percentages of glycol result in higher boiling points, glycol has a much lower rate of heat transfer than water. A low percentage of glycol to prevent freezing and minimize corrosion and higher system pressures actually gives the best of all worlds- best cooling, high boiling point and lowest drag. As I mentioned previously, the P51 pressurizes the system to 50psi and uses a 50/50 mix. Pre-war, many systems used 100% glycol and lower system pressures.

As hp development increased on the British Merlins, glycol percentages were reduced to 30% as the existing radiator designs were not capable of rejecting the increased heat without redesign. With the Merlin 61 and Griffon engines, larger rads had to be designed eventually despite this as hp was doubled finally in the 130 series Merlins and Griffons in 1944-1945.

[robpar]

Originally Posted by rv6ejguy

As I mentioned previously, the P51 pressurizes the system to 50psi and uses a 50/50 mix. Pre-war, many systems used 100% glycol and lower system pressures.

Ross:

I have had dealings with the "Boilers Branch" if you even think of running high pressure cooling systems (over 15 psi) do not let them know about it or the plane will be so heavy you couldn't get it off the ground with a 1000 HP. Just kidding but there was an boiler inspector here that had bug in his bonnet about Cappuccino machines, if you can believe it. He wanted them certified! Calmer heads prevailed.

Higher boiling points would be very helpful. I have heard that the F1/IRL people are looking into additives that raise the boiling point but do not lower the Cp value of water. Raising the engine temperature would reduce the area of radiator required for a given amount of heat to reject, assuming constant Cp.

During a course on heat transfer I took many years ago the material the heat exchanger was made of did not effect the amount of heat that was transferred as long as the material was thin. There was a Length/Thinckness ratio that specified what was too thick. But in our designs we were using guage material and it made very little difference whether we used copper, aluminum or SS. The maximum difference if I remember was about 3%. Scaling on the surfaces of the heat exchanger made a much bigger difference than the material, so materials were chosen for compatibility with the fluid.

An interesting thread. The choice of air or liquid cooling is more complicated than it would at first appear.

Bob Parry

[rv6ejguy]

Yes, thinner is better from a thermal gradient point of view and that is one more reason why modern aluminum cores are more efficient than thicker brass or copper cores.

Some of the heat exchangers used for water and oil radiators on British and American engines were of the round honeycomb design and rather deep. These had relatively low pressure loss through the core but were not as efficient thermally as a traditional tube and fin type core. The very innovative Westland Whirlwind used two round honeycomb water radiators and one round oil cooler buried in each wing root. The inlets were located in the wing leading edges and air flowed straight though the spar which was a trusswork of streamlined tubing. Air exited out the rear spar through a long hinged door on the top wing surface just ahead of the flap nose.

An air cooled engine fin is very thick in comparison with a radiator tube and fin structure. What it loses in efficiency, it may make up with somewhat higher Delta T.

If you look at almost any WW2 design both air and liquid cooled, you will see cowl or rad flaps. The designers clearly understood the drag reduction possible in high speed flight using these. I submit that most of the liquid cooled experimental aircraft flying today suffer a much higher drag penalty than need be due to less than careful rad layout, poor inlet and exit ducting and a lack of rad flaps to recover much of the lost momentum lost to the rad core. Certainly my 6A is really bad at present.

[Rotary10-RV]

Ross,
I'm sure you understand the duct shape and length are very crucial to good water-cooled applications. The P-51 was one of the first aircraft that had proper diverging converging ducts. And that was only on the later versions. The use of a proper diverging duct slows the flow with less drag losses, which provides a proper speed of flow through the heat exchanger and the converging (exit) duct also speeds the flow with minimal loss. With proper duct work drag loss can be minimised.
The biggest problem with the Egg Subaru is the minimal distance allowed by the cowl to the front of the radiator face. There is little if any divergence in the duct unless the owner makes one on their own. The second sin radiator wise is the rads themselves are backed up close to the engine cylinder heads. A good system provides clear area for the radiators to exhaust into. A good rule of thumb is 3-4" of clear space before tapering the exit duct work. Obviously the Subaru FWF from Egg and others have ignored the exit side clearance rule due to the convenience of mounting the radiators near the cowl inlet holes. K&W (Kuchemann and Weber) found the exit to be equally important to the entrance to efficient cooling and low drag. Most of the inside cowl radiator mountings ignore this.
Bill Jepson

[rv6ejguy]

Absolutely and this is why I mentioned a few of the people like Finley and Clarke who have actually done things correctly and have had excellent results and even good ground cooling to boot.

This is why I suggest that people don't copy what I've done on my RV6A as there are no exit ducts. What I have works well for cooling but also must be creating high drag.

The data has existed for a very long time on how to do it right but this research has been largely ignored in recent times for ease of installation.