I've always been interested in the "Meredith Effect" as theorized back in 1936 by F.W. Meredith. He surmised that with a well shaped duct enclosing the radiator, it was possible to recapture the cooling air momentum lost passing through the rad after adding the heat and constricting the exit. These thoughts were applied in the famous rad setup in the P51.
I've been flight testing my 6A with the new ventral radiator scoop for a couple months. It's been fitted with 2 Magnehelic pressure gauges, one for rad inlet pressure and one for outlet pressure, an exit air temperature probe and a pitot tube to measure exit velocity.
It has always been the position of the air cooled advocates that those installations would always suffer less cooling drag simply because they were working with heads at about 400F vs. the 200F we have available in liquid cooled aircraft engines. In thermodynamics, Delta T or the difference in temperature between the object and the cooling medium means a great deal. All things being equal (which they are not in this case) you can sink off more heat or energy when the Delta is higher.
I always found this logic simplistic and flawed. While the the total energy we need to dissipate should be about the same to cool a 200hp Lycoming and a 200hp Subaru, the point of heat exchange and the duct geometry are vastly different.
In a Lycoming setup, we have a very short distance to slow down the air from the cowl inlet to the cooling fins and the air must turn first 90 degrees to go down through the head and cylinder fins, then must turn another 90 degrees to head towards the exit. We also have the steel barrels which are not nearly so efficient as aluminum at dissipating heat. Then we have inefficient changes in volume as the air passes through the system- large at the inlet, small through the fins, very large exiting into the cowling and then hitting many obstructions including the flat firewall on its way to the actual smaller outlet.
On the ventral rad setup, we have several times the length to slow down the air efficiently and gradually and recover the maximum pressure at the rad face. We don't change the air path 180 degrees like the Lycoming. Whenever you turn air suddenly, you have a loss of velocity unless it is done very gradually. We call this momentum loss. Whenever we take air on board the aircraft internally, unless we can return it to the same velocity, we create drag. By slowing down the air at the rad face, we reduce drag through the core. Then we reduce the cross sectional area of the duct gradually to accelerate the air again before it leaves the exit. The radiator core has many times the surface to volume ratio of an air cooled engines coarse fins and divides each hot part (rad tubes) into very thin sheets. This means that it is very efficient at transferring heat to the atmosphere.
I've been following Dan Horton's threads and test results on his IO-390 RV8 cooling experiments with great interest so I'll draw on some of the numbers he posted on VAF. This is not meant to degrade the fine work and testing Dan has done, as we can all appreciate that. I am just comparing the science of the two setups and I want to say, that these are just 2 points, who is to say either one is fully optimized?
So what do the numbers show?
Pressure recovery. The ventral rad duct appears more efficient. We recover 84% of the theoretical pressure available compared to 77% for the Lycoming, both with exit doors closed.
CHT / coolant to exit air Delta. These are hard to compare apples to apples as in the Lycoming setup, cooling air from the inlets also goes through the oil cooler and over the exhaust system before exiting. The exhaust likely contributes somewhat to the temperature rise of the cooling air since it is around 1400F and there is quite a bit of surface area there. If we compare at face value, we see the Lycoming setup is in the low to mid 50% range, depending on speed and CHT (I guessed a bit here). The radiator setup peaked at 92%. So despite the much lower Delta of coolant vs. cylinder heads, the radiator is far more efficient per unit temperature at transferring energy to the atmosphere.
Momentum recovery. I was not able to compare Dan's values here (maybe he can fill this part in). In the rad setup, I measured a 97% momentum recovery at the exit with the door mostly closed with a coolant temperature of 176F. This suggests that by raising the coolant temp and the Delta by 30F, we could probably reach velocity momentum parity inlet vs. outlet and possibly even add velocity (net thrust).
Now the remaining question is how much drag, internal and external does the ventral scoop present vs. the clean contours of the air cooled setup. (airplanes are all full of compromises). I can't answer at this point. I did my best to place the scoop in the frontal "shadow" of the existing lower cowl exit so the frontal area gain is minimal but I've also added some more wetted area.
Finally we have a couple more points to compare, more as food for thought:
My setup has excess inlet area still for the oil and intercoolers but my friend, Russell Sherwood has a well optimized (SARL winning) Glasair with a 230hp Subaru six. He has about 48 in2 of total inlet area (induction, oil, gearbox and rad vs. 56in2 for Dan's clean RV8. This suggests that if the same exit sizes are used, the liquid cooled setup is taking less mass flow on board to do the job with roughly the same hp levels. Russell's exit area in normal cruise with oil and coolant combined is around 27 to 30in2, Dan said his was around 30in2 if I got that correct.
Dan, please step in here to correct anything I posted incorrectly and your observations as well.
I welcome any other views on this fascinating topic.
Here are a couple videos doing some tuft testing: http://www.youtube.com/watch?v=QT8njoirTkU
http://www.youtube.com/watch?v=gPpWHvr1kJw
I've been flight testing my 6A with the new ventral radiator scoop for a couple months. It's been fitted with 2 Magnehelic pressure gauges, one for rad inlet pressure and one for outlet pressure, an exit air temperature probe and a pitot tube to measure exit velocity.
It has always been the position of the air cooled advocates that those installations would always suffer less cooling drag simply because they were working with heads at about 400F vs. the 200F we have available in liquid cooled aircraft engines. In thermodynamics, Delta T or the difference in temperature between the object and the cooling medium means a great deal. All things being equal (which they are not in this case) you can sink off more heat or energy when the Delta is higher.
I always found this logic simplistic and flawed. While the the total energy we need to dissipate should be about the same to cool a 200hp Lycoming and a 200hp Subaru, the point of heat exchange and the duct geometry are vastly different.
In a Lycoming setup, we have a very short distance to slow down the air from the cowl inlet to the cooling fins and the air must turn first 90 degrees to go down through the head and cylinder fins, then must turn another 90 degrees to head towards the exit. We also have the steel barrels which are not nearly so efficient as aluminum at dissipating heat. Then we have inefficient changes in volume as the air passes through the system- large at the inlet, small through the fins, very large exiting into the cowling and then hitting many obstructions including the flat firewall on its way to the actual smaller outlet.
On the ventral rad setup, we have several times the length to slow down the air efficiently and gradually and recover the maximum pressure at the rad face. We don't change the air path 180 degrees like the Lycoming. Whenever you turn air suddenly, you have a loss of velocity unless it is done very gradually. We call this momentum loss. Whenever we take air on board the aircraft internally, unless we can return it to the same velocity, we create drag. By slowing down the air at the rad face, we reduce drag through the core. Then we reduce the cross sectional area of the duct gradually to accelerate the air again before it leaves the exit. The radiator core has many times the surface to volume ratio of an air cooled engines coarse fins and divides each hot part (rad tubes) into very thin sheets. This means that it is very efficient at transferring heat to the atmosphere.
I've been following Dan Horton's threads and test results on his IO-390 RV8 cooling experiments with great interest so I'll draw on some of the numbers he posted on VAF. This is not meant to degrade the fine work and testing Dan has done, as we can all appreciate that. I am just comparing the science of the two setups and I want to say, that these are just 2 points, who is to say either one is fully optimized?
So what do the numbers show?
Pressure recovery. The ventral rad duct appears more efficient. We recover 84% of the theoretical pressure available compared to 77% for the Lycoming, both with exit doors closed.
CHT / coolant to exit air Delta. These are hard to compare apples to apples as in the Lycoming setup, cooling air from the inlets also goes through the oil cooler and over the exhaust system before exiting. The exhaust likely contributes somewhat to the temperature rise of the cooling air since it is around 1400F and there is quite a bit of surface area there. If we compare at face value, we see the Lycoming setup is in the low to mid 50% range, depending on speed and CHT (I guessed a bit here). The radiator setup peaked at 92%. So despite the much lower Delta of coolant vs. cylinder heads, the radiator is far more efficient per unit temperature at transferring energy to the atmosphere.
Momentum recovery. I was not able to compare Dan's values here (maybe he can fill this part in). In the rad setup, I measured a 97% momentum recovery at the exit with the door mostly closed with a coolant temperature of 176F. This suggests that by raising the coolant temp and the Delta by 30F, we could probably reach velocity momentum parity inlet vs. outlet and possibly even add velocity (net thrust).
Now the remaining question is how much drag, internal and external does the ventral scoop present vs. the clean contours of the air cooled setup. (airplanes are all full of compromises). I can't answer at this point. I did my best to place the scoop in the frontal "shadow" of the existing lower cowl exit so the frontal area gain is minimal but I've also added some more wetted area.
Finally we have a couple more points to compare, more as food for thought:
My setup has excess inlet area still for the oil and intercoolers but my friend, Russell Sherwood has a well optimized (SARL winning) Glasair with a 230hp Subaru six. He has about 48 in2 of total inlet area (induction, oil, gearbox and rad vs. 56in2 for Dan's clean RV8. This suggests that if the same exit sizes are used, the liquid cooled setup is taking less mass flow on board to do the job with roughly the same hp levels. Russell's exit area in normal cruise with oil and coolant combined is around 27 to 30in2, Dan said his was around 30in2 if I got that correct.
Dan, please step in here to correct anything I posted incorrectly and your observations as well.
I welcome any other views on this fascinating topic.
Here are a couple videos doing some tuft testing: http://www.youtube.com/watch?v=QT8njoirTkU
http://www.youtube.com/watch?v=gPpWHvr1kJw
Last edited: