Air cooled inline engines - a missed opportunity?

[ThomasP]

I was going to bring up the Gipsy Twelve air-cooled V12 used on the DH Albatross.

As far as I can tell there were no significant problems keeping it cool when running to its designed power, so obviously the DH engineers knew what they were doing. BUT IIRC, unlike the Ranger which was designed to be as compact as possible to fit in different small(ish) airframes, the Gipsy Twelve was specifically designed for use with the large Albatross airframe and associated large wing/nacelle volume for cooling arrangements. Size (length in particular?) was not so important and the beautifully streamlined nacelles and the cooling ducts were designed around the engine from the start.

I do not know what the duty cycles were for one engine compared to the other, but I suspect that the cycle range and rate were lower for the Gipsy Twelve than for the Ranger.

 

[z42]

In the 20s/30s/40s, designers were learning a lot about airflow management/cooling. For the air cooled engine, "just hanging jugs out in the airflow like cauliflower leaves" no longer worked. Air, at the higher speeds of the newer airplanes, went around the engine rather than across the cooling fins. Also, if the air is moving too fast, it doesn't absorb as much heat as it could.

So, you need to get the air to enter, slow it down to the optimum speed, baffle the head/cylinder to ensure air goes over the cooling fins, cast/machine the optimum cooling fins, then extract the heated air. Note: optimum air volume changes depending on speed/power i.e. climb/cruise/combat/descent/etc.

Yes, so similar to how these liquid cooling radiators were designed, with a small opening and then a diffuser duct to reduce the velocity before going through the radiator matrix (or in this case, through the fins on the cylinders). And finally a contracting duct to increase the velocity before exhasting the air.

de Havilland Albatross with de Havilland Gipsy Kings is how you do it. We note how small the intake ducts are on the wing leading edge and how small/simple the exhaust duct is.

The challenge is the V-12 gets really long - the Gipsy King 82" long (4.646" pistons) vs Merlin 69" (5.4" pistons). The Pennine (5.4" pistons, but X-24) is 106" long! And an engine that long doesn't "swap" into an existing airframe - you need an engine swap ala reverse of P36 to P40/Fw.190A to Fw.190D, where you can demonstrate the superiority of your engine e.g.
Eliminating liquid cooling by using R-R Exe in a Barracuda. Poppet valves on the Exe probably desired given some 20/20 hindsight. Note: the air intake on Battle with Exe installed is far from optimal.

Good points.

 

[z42]

A problem with V-12s is that you are trying to turn the air twice through around 90 degrees.

Not sure that's much of a problem, the system is driven mostly by the pressure differentials, the momentum and kinetic energy of the air is small in comparison.

See e.g. turbomachinery with centrifugal compressors, or for that matter many turboprops which have reverse airflow (air inlet at the back, exhaust at the front)

It has very little to do with the temperature of the air at rear cylinders, Well it kind of does but every cylinder is supposed to be getting fresh air and not air that already gone by another cylinder.

True. In that sense it seems the cooling problem is more straightforward than e.g. those 4 row radials.

Also note that the air passages on the DH Albatross are for just over 500hp engines. You need a air passages about 3 times larger for 1500hp engines.

Thanks to the magic of the quadratic, the ducting diameter only needs to be a factor of sqrt(3) bigger.

 

[don4331]

True. In that sense it seems the cooling problem is more straightforward than e.g. those 4 row radials.

Thanks to the magic of the quadratic, the ducting diameter only needs to be a factor of sqrt(3) bigger.

Cooling problem is very similar to 4 row radial - you need to take just the right amount off each row, so from front to back all get the correct amount.

Ranger (Fairchild) probably as much as anybody did in studying air cooling. Like the size and spacing of the fins and how far away the air could be and still do any cooling.
They also spent a lot of time and money just trying to get that 520hp engine to cool in aircraft.

A problem with V-12s is that you are trying to turn the air twice through around 90 degrees.

You can play games with reverse cooling at lower speeds. When you try for 350-400mph changing the direction of the air introduces losses.

I would bring up the sport airplane racers - they are turning the air significantly twice to run through the fins of the flat 6 engines they are running - yet they are able to cool the engines successfully. And they are making >900hp to run 400+ mph, so there was still something Fairchild could have learned. p.s. I am aware they run water misting systems to assist with cooling, but they are making well over 500hp before they have to turn it on.

 

[z42]

Cooling problem is very similar to 4 row radial - you need to take just the right amount off each row, so from front to back all get the correct amount.

I meant that in an inline you have this plenum from where the air flows transversely over the cylinder banks, thus you don't have the issue of some cylinders obstructing others. Of course the plenum has to be correctly shaped to ensure even flow over each cylinder.

In contrast it seems all the 4 row radials had significant challenges in cooling the back rows, despite, I'm sure, extensive design work wrt airflow and baffling.

 

[ThomasP]

My understanding is that even the post-war R-4360 had significant cooling issues. I am not aware if they were ever satisfactorily resolved.

 

[wuzak]

The Pennine (5.4" pistons, but X-24) is 106" long!

The Chrysler IV-2220 was 122 inches long with 5.8 inch bores (x 8) and liquid cooling.

 

[Shortround6]

Not sure that's much of a problem, the system is driven mostly by the pressure differentials, the momentum and kinetic energy of the air is small in comparison.

The pressure differentials are powered by drag. In some form or other. RR figured out that the Air intake momentum for the Merlin XX varied from 32.8lbs to 14.1lbs between 15,000 and 35,000ft on a Hurricane II. It may not be large bit does exist.
The more twists and turns the air has to make means more turbulence.

I would bring up the sport airplane racers - they are turning the air significantly twice to run through the fins of the flat 6 engines they are running - yet they are able to cool the engines successfully. And they are making >900hp to run 400+ mph, so there was still something Fairchild could have learned. p.s. I am aware they run water misting systems to assist with cooling, but they are making well over 500hp before they have to turn it on.

We have several things going on here. One is that nobody wants to try to make a brand new engine so they are sort of stuck with the flat six. the two changes of direction date back to the introduction off the flat six. If fact it goes back to inline fours, like Tiger Moths. Air enters one side, goes down the side of the engine, goes through the cylinders (and baffles) exits the engine on the other side travels to the rear of cowling and either exits on one side or is allowed to cross behind the engine and exit from both sides.

Air cooled engines had the same problems as liquid cooled engines do. Getting the air flow right for 400mph at a given altitude is not that hard. Getting the airflow right at sea level and at 21,000ft (air has 1/2 the weight of sea level) is harder. Getting the airflow right for 400mph and 900hp is a lot easier than getting the airflow right at 200mph (max climb) and 900hp.
Now try maneuvering, hard turn. Plane is trying to make max power at low speed with the plane flying at a high angle of attack (8-10 degrees ).
Yes it can be done with adjustable flaps/doors/louvers.

V-12s have got some problems for air cooled aircraft engines.
As cylinders get bigger they are harder to cool. The cylinder surface (wall area, head area, piston top) does not go up at the same proportion that the volume goes up. You are burning more fuel per unit of surface area at the basic level. You also have problems on large diameter cylinders with the pistons because the main source of heat transfer for the piston crown is through the cylinder walls and the distance from the center of the piston to the cylinder wall gave problems. One reason that P & W never went over a 146mm X 152mm cylinder on any mass produced engine after the early 1930s. Other people did, but even the Bristol Centaurus used 146mm X 178mm Cylinders.

You need more space between the cylinders for an air cooled engine than for a liquid cooled engine. Which means a longer/heavier crankcase and crankshaft.

You also need to figure out the valve arrangements. Leaving Bristol out of it just about everybody used pushrods for radials, flat engines and for low powered 4-6 cylinder in-lines.
The HIgher powered versions and the V-12s which were basically two sixs on a common crankshaft often tried to use overhead cam/s either single or double.
Problem here became the cam boxes/covers took up space/volume that you needed to cool the heads of high powered engines.
Ranger engine

Note the size of the cam boxes. Also note the the designer has a choice with the air flow. The Ranger routed the air over the exhaust side of the cylinder head and over the intake side of the cylinder. Radials didn't have that choice. Both valves on each cylinder got fresh air. The rear cylinders all got fresh air.
Each cylinder on the Ranger was suppose to get fresh air.
Just about all air cooled V-12s used 2 valves per cylinder. Not sure what the valve angles were. But again, more than 2 valves are fighting for cooling space. The Wright and P & W engines used 2 valves but they used wide valve spacing and hemi heads.

Once you start going over 87 octane fuel the heat problems get worse. The higher octane/PN fuel allows you to cram more fuel into the cylinder for each stroke creating more heat.
The basic solution for air cooled engines in more fin area. Next solution was changing the fin material/construction. Liquid cooled engines had quick fix, they could increase the flow of the coolant, but not too much or it wouldn't pick up enough heat. They could design new blocks/heads with larger passages. The Air cooled engines could design more/deeper fins from new castings/forgings.

You could try X-24s to keep the size of the inline Air cooled engine cylinders smaller for better cooling ( Napier Dagger and Isotta Zeta) but they were expensive

Isotta Fraschini Zeta - Wikipedia

en.wikipedia.org

24 cylinders are rarely as cheap as 12.

 

[manta22]

Cooling problem is very similar to 4 row radial - you need to take just the right amount off each row, so from front to back all get the correct amount.


I would bring up the sport airplane racers - they are turning the air significantly twice to run through the fins of the flat 6 engines they are running - yet they are able to cool the engines successfully. And they are making >900hp to run 400+ mph, so there was still something Fairchild could have learned. p.s. I am aware they run water misting systems to assist with cooling, but they are making well over 500hp before they have to turn it on.

"I would bring up the sport airplane racers -..." that is a whole different situation. Lots of power delivered for a short period of time vs needing a reasonable amount of power delivered for hours at a time with a long time between overhauls. Drag racers build engines that deliver insane power but those are expected to run full throttle for less than 10 seconds at a time. Other applications such as off-shore boat racing, Bonneville, or NASCAR need to be run full-throttle for minutes to hours. On the other end of the spectrum, passenger cars & trucks are designed to run for 100,000+ miles and deliver adequate power reliability. Aircraft applications are similar.

 

[Spindash64]

I honestly don't see the point of an air cooled inline anymore than a liquid cooled radial: while there are other advantages and disadvantages to a given arrangement, the primary rational for using an engine with inline arrangement IS that you can use liquid cooling for a smaller total frontal area, even if the total displacement is lower. While NACA cowlings would make the total difference in drag relatively small, to the point that the higher displacement, higher power radials could generally hit the same speeds as comparable inlines, it's still important to remember that these planes are often using more propulsive horsepower to hit the same speed

Also, while I do not know anywhere NEAR enough about engines to really argue this effectively, it looks to me as though liquid cooling also tends to allow higher manifold pressure limits for a given effective octane rating than with air cooled engines: the -1 Corsairs needed water injection to hit 60" of Manifold pressure, while the P-51D was cleared for 61" as Military power, not even as WER. Of course, the R-2800 is still far more powerful, boasting 70% greater displacement, along with being able to individually run the 2 blowers rather than together, but it still demonstrates that liquid cooling offers significant advantages for maximum safe power. Without essentially sticking your engine in the refrigerator, it's much more dangerous to try and let things run that hot, because it won't be able to lose enough of that heat between combustion strokes

 

[manta22]

I honestly don't see the point of an air cooled inline anymore than a liquid cooled radial: while there are other advantages and disadvantages to a given arrangement, the primary rational for using an engine with inline arrangement IS that you can use liquid cooling for a smaller total frontal area, even if the total displacement is lower. While NACA cowlings would make the total difference in drag relatively small, to the point that the higher displacement, higher power radials could generally hit the same speeds as comparable inlines, it's still important to remember that these planes are often using more propulsive horsepower to hit the same speed

Also, while I do not know anywhere NEAR enough about engines to really argue this effectively, it looks to me as though liquid cooling also tends to allow higher manifold pressure limits for a given effective octane rating than with air cooled engines: the -1 Corsairs needed water injection to hit 60" of Manifold pressure, while the P-51D was cleared for 61" as Military power, not even as WER. Of course, the R-2800 is still far more powerful, boasting 70% greater displacement, along with being able to individually run the 2 blowers rather than together, but it still demonstrates that liquid cooling offers significant advantages for maximum safe power. Without essentially sticking your engine in the refrigerator, it's much more dangerous to try and let things run that hot, because it won't be able to lose enough of that heat between combustion strokes

A critical element in any engine is its cylinder head temperature. On many instrument panels there is a thermocouple gauge to monitor the head temperature. It is far more difficult to keep an air-cooled cylinder head cool.

 

[Shortround6]

Air cooled engines as a general rule, and we are talking real general since they went from under 40hp fours to over 2000hp 18s, were operating close to the "edge" than liquid cooled engines. And be very careful as to which air cooled radials you pick examples from. The R-2800 "C" series engines only made WW II by under 12 months and most production was in the spring of 1945, The legend has it that the only things the "C" shared with the "B" engines in the P-47Ds and F6Fs was the bore and stroke and the starter dog. Everything else was changed. There was enough extra finning on the engine to allow it to make 2000hp with 10% less airflow entering the Nacelle.
Every version of the major US and British radials were very substantial redesigns that changed the cooling ablities.
Hence the need for water injection for any large increase in power. An additional cooling system. The R-2800 B was not using over 11-12lbs of boost without it and was using a large amount of raw fuel as coolant even at that level.

The problems with an air cooled V-12 are many.
You are limited to the size cylinders you can use, even 12 R-2800 cylinders only gives you 1866 cu in. OR you can use bigger cylinders but lower the rpm. Heat is = to the amount of fuel burned per minute but reducing power by using lower rpm goes against the trying to make more power.
Going to V-12s gives up the rather neat and short intake system of the radials for the longer more convoluted intake system of a V-12 with more mixture distribution problems.
Power is restricted not by the engine in general but by the worst cylinder (hottest) in the whole engine. Fuel injection could help but that adds cost/complexity.
Getting a rigid structure for the air cooled V-12 is harder. The cylinder bores need to be spaced further apart and you can't brace the cylinders against each other like a liquid cooled engine. And a cam box with cam shaft/s and rocker arms do not provide the rigidity of complete 6 cylinder head.

A radial may have it's own issues but the big V-12 has some weight/size issues that exceed those of the radial.

 

[yulzari]

Napier went away from air cooling after the Rapier H16 and Dagger H24 for liquid cooling of the H24 Sabre and the Dagger gained a reputation for poor cooling. However an ex wartime de Havilland chap told me that they had looked at the Dagger installation of Napiers and concluded that de Havilland would have done a better job. Essentially Napier concentrated on stuffing more air in whilst de Havilland looked more at getting the air out. In a purpose built installation he said they found that the Dagger could be cooled with no problems at all stages of a flight and could cope with even more power/heat.

One should note that the biggest de Havilland air cooled engine (Gipsy Twelve) used a reverse flow with intakes in the wing leading edge and an adjustable exit beneath the engine. IIRC similar arrangements were used successfully for the post war paired Bristol Centaurus 5,300bhp air cooled radials in the Bristol Brabazon, albeit with far more complexity.

Bristol Brabazon - Wikipedia

en.wikipedia.org

de Havilland Gipsy Twelve - Wikipedia

en.wikipedia.org

Napier Dagger - Wikipedia

en.wikipedia.org

Napier Rapier - Wikipedia

en.wikipedia.org

 

[Shortround6]

Napier went away from air cooling after the Rapier H16 and Dagger H24 for liquid cooling of the H24 Sabre and the Dagger gained a reputation for poor cooling. However an ex wartime de Havilland chap told me that they had looked at the Dagger installation of Napiers and concluded that de Havilland would have done a better job. Essentially Napier concentrated on stuffing more air in whilst de Havilland looked more at getting the air out. In a purpose built installation he said they found that the Dagger could be cooled with no problems at all stages of a flight and could cope with even more power/heat.

One should note that the biggest de Havilland air cooled engine (Gipsy Twelve) used a reverse flow with intakes in the wing leading edge and an adjustable exit beneath the engine. IIRC similar arrangements were used successfully for the post war paired Bristol Centaurus 5,300bhp air cooled radials in the Bristol Brabazon, albeit with far more complexity.

A lot was learned about air cooling in the very late 30s and during WW II. de Havilland probably could have done a better job on the Dagger.
The question is how far they could have gotten.
The Dagger had a few problems of it's own.
Yes it was about 300lbs lighter than a Merlin once you throw in the liquid cooling parts and coolant.
It was also about 160-200hp less powerful even on 87 octane fuel.

It was probably more expensive to build. And it was harder to scale up. it was only 16.8 liters and when you try to scale it up you loose some of the things that helped it make the power it did.
Like the high rpm, 4200rpm.
Like the efficiency of the small cylinders.
And the cooling. If you go from 97mm bore cylinders to 127mm bore cylinders you increase the wall circumference by about 31% but you increase the volume by 71% so if you burn the same amount of fuel per unit of displacement you have to figure out how to get rid of the heat. Decreasing the rpm (longer stroke) decreases the heat load while decreasing the power.
Compared to a radial the Dagger (and the other inline/v-12 air cooled engines) used over head cams and rockers in boxes which blocked some of the airflow to the cylinder heads.

 

[yulzari]

A lot was learned about air cooling in the very late 30s and during WW II. de Havilland probably could have done a better job on the Dagger.
The question is how far they could have gotten.
The Dagger had a few problems of it's own.
Yes it was about 300lbs lighter than a Merlin once you throw in the liquid cooling parts and coolant.
It was also about 160-200hp less powerful even on 87 octane fuel.

It was probably more expensive to build. And it was harder to scale up. it was only 16.8 liters and when you try to scale it up you loose some of the things that helped it make the power it did.
Like the high rpm, 4200rpm.
Like the efficiency of the small cylinders.
And the cooling. If you go from 97mm bore cylinders to 127mm bore cylinders you increase the wall circumference by about 31% but you increase the volume by 71% so if you burn the same amount of fuel per unit of displacement you have to figure out how to get rid of the heat. Decreasing the rpm (longer stroke) decreases the heat load while decreasing the power.
Compared to a radial the Dagger (and the other inline/v-12 air cooled engines) used over head cams and rockers in boxes which blocked some of the airflow to the cylinder heads.

The raison d'etre of the Dagger was to run an engine twice as fast rather than make it twice as big. And with 24 cylinders they could be quite efficient at the task; if you can lose the heat efficiently. Improved cooling could lose that heat with no extra weight so, ceteris paribus, the Dagger, with 17 litres could equate with a Merlin at 27 litres. As to whether it could go on to cope with later Merlin pressures is another matter on which I am not qualified to comment.

Twin Daggers on a Westland Whirlwind anybody? - let the Whirlwind fanboys commence……..

 

[swampyankee]

The UK (Dehaviland, Cirrus), USA (Ranger, Menasco), France (Renault), Italy(Isotta-Fraschini), Germany (Hirth, Argus), Czechoslovakia (Walter), and likely Japan and the USSR produced air-cooled in-lines. They tended to be heavier per unit power than opposed engines or radials.


As an aside, radials' diameter did not cause greater drag; nose shape is largely irrelevant to drag at the speeds WWII aircraft could achieve. Total wetted surface is far more important. Cooling drag is important, and getting low cooling drag is harder with direct vs indirect (liquid cooling in the engine and a liquid to air heat exchanger someplace) air cooling. Because air-cooled inlines were pretty much restricted to low performance aircraft, less attention was paid to optimizing their cooling system designs

 

[Shortround6]

Twin Daggers on a Westland Whirlwind anybody? - let the Whirlwind fanboys commence……..

Whirlwind fanboy here...........................Why ruin a perfectly good airplane?

The Peregrine was about 250lbs lighter than the edit>Whirlwind Dagger, which should cancel the weight of the liquid cooling.
It's peak power was over 6,000ft higher which pretty much cancels the difference in power at higher altitudes.

Right side air duct for the 520hp Gypsy King engine.
Double it for a Dagger, remember to double the left side duct.

Not seeing a reduction in drag here.

 

[Shortround6]

As an aside, radials' diameter did not cause greater drag; nose shape is largely irrelevant to drag at the speeds WWII aircraft could achieve.

Well, nose shape may be irrelevant to drag but nose size may not be. Also wetted area for the radial may be larger. Larger circumference vs a bit shorter length? Length is governed by the aircraft CG and not the length of the engine for some planes.

Drag of the Ki-100 was higher than the Ki-61.

 

[z42]

The Peregrine was about 250lbs lighter than the Whirlwind,

Huh? Presumably you meant that the Peregrine was about 250lbs lighter than the Dagger?

It's peak power was over 6,000ft higher which pretty much cancels the difference in power at higher altitudes.

At a first guess, that would be due to supercharger differences, liquid vs air cooled doesn't seem pertinent here?

Well, nose shape may be irrelevant to drag but nose size may not be. Also wetted area for the radial may be larger. Larger circumference vs a bit shorter length? Length is governed by the aircraft CG and not the length of the engine for some planes.

Drag of the Ki-100 was higher than the Ki-61.

The interesting question would be to know how much of the difference in drag between the Ki-61 and Ki-100 was form drag due to the blunter shape, how much skin friction drag due to increased wetted area, and how much due to the cooling drag.

 

[Shortround6]

Huh? Presumably you meant that the Peregrine was about 250lbs lighter than the Dagger?



At a first guess, that would be due to supercharger differences, liquid vs air cooled doesn't seem pertinent here?



The interesting question would be to know how much of the difference in drag between the Ki-61 and Ki-100 was form drag due to the blunter shape, how much skin friction drag due to increased wetted area, and how much due to the cooling drag.

1, yes, corrected it, thanks.
2, yes it is a supercharger difference, but it is also a gross vs net power difference. If you you use a small or lower capacity supercharger you can make more power at lower altitude.
Please note that the US radial engine makers took advantage of this with the two speed superchargers on the R-1820 & R-1830 and later engines. A 1200hp R-1830 was only good for 1050hp at 13,100ft once the supercharger was in high gear.

3. I have no idea of what was doing what. What we do know is this.
Plane..................................Altitude...............Speed...........................power
Ki-61 I................................sea level................302mph.......................1175hp (take off rating, speed maybe at 2400rpm instead of 2500rpm)
Ki-100...............................1000 m...................317mph.......................1500hp (take off rating.speed maybe at 2500rpm instead of 2600rpm)
Ki-61 I...............................5000 m...................360mph.......................1100hp (2400rpm at 4200 m)
Ki-100...............................5000 m...................352mph.......................1250hp (2500rpm at 5800 meters)
Ki-100...............................6000 m...................360mph.......................1250hp (2500rpm at 5800 meters)

The Ki-100 does not have the big radiator box under the rear fuselage.

That should reduce drag somewhat. But trying to figure out frontal area, surface skin area and drag through cooling ducts is a bit too difficult for me.
What it does look like is that the air cooled radial had around 10-12% more drag than the inline.
The exhaust system on the Ki-100 looks pretty good, 7 individual pipes in the 'duct' on each side, very FW 190 style.