Build the perfect air-cooled engine

[wuzak]

We have had a long discussion, not always on point, about the merits of different liquid cooled piston aero engines.

Now we should look at the flipside - what is/was/could have been the best aircooled engine for all around service (ie used in bombers, fighters, attack, whatever). Or perhaps we should look at ones for specific duty.

I personally don't think production numbers give the answer to what was best, and there were several engines that didn't reach production, or only gained small production orders.

For example. The Armstrong Siddeley Deerhound. This was a 21 cylinder 3 row radial, with ohc and 2 valves per cylinder.

The Mk I had difficulty meeting the targeted 1500hp, and the chief engineer resigned as a result. Stewart Tresilian, formerly of Rolls Royce, was recruited to take over the job, modified the Mk II so that it was able to produce 1340hp, still short of the projected target. The modifications brought capacity up to 41l (2,505cid) from 38.19l (2,259.75cid).

Bore and stroke was 140mm x 127mm. This was carried over to the MkIII, which was a total redesign by Tresilian. He added reverse cooling to the design, and the engine was able to demonstrate 1800hp on the bench.

The engine was compact, with a diameter of 44 inches, giving significantly lower frontal area than the major radials of the period - the R-2800 was between 52 and 55 inches (depending on version), as was the R-4360, the Centaurus was 55.3 inches, the Hercules 55 inches. The smaller Taurus was slightly bigger at 46 inches in diameter. The R-1820 was over 54 inches in diameter, the R-2600 was 55 inches, and the R-3350 nearly 56 inches in diameter.

The R-2600 is probably closest in power output to the MkIII in its initial development status, and very similar in capacity. Had the Deerhound project been continued it may have been on par for power with the R-2800 by war's end. It was, however, cancelled mid 1941, and the company instructed to concentrate on gas turbines.

A further development, the 28 cylinder Wolfhound, never left the drawing board.

 

[Shortround6]

Lots of engines showed potential on the bench. A down fall of the Deerhound could have been cooling. They showed a lot of ingenuity in the cooling arrangements but they were complicated. Three cylinders in a row means air has to be directed between the rows, turned sideways, travel through the row (cylinder fins), get collected on the other side and then move back out of the space between the rows. Even an R-4360 used a simpler pattern than that. Another potential source of problems in cooling is the is the overhead cam. how much potential cooling area is blocked by the cam box?
Using the same size cylinders as an two row 18 a 3 row 21 gets a 1/6 increase in displacement but adds a heavier crankcase and crankshaft. The overhead cams may be heavier than push rods.

A sleeve valve Deerhound might have been interesting. The sleeve drives wouldn't have been much more coplicated than the overhead cams and might have even required less gearing than than the 14cylinder Hercules. It would also have freed up the head area for a more unobstructed airflow.

 

[Trilisser]

My perfect air-cooled engine would be (assuming at least 2000 hp power class)
-18-cylinder radial
-with sleeve valves and copper alloy heads
-two stage supercharger with infinitely variable drive plus aftercooler (i.e. RR type blower)
-direct injection

I see that Centaurus is the closest to that ideal.

 

[Shortround6]

All good ideas but the after cooler design is very hard to implement on a radial. on a V-12 the air comes out of one outlet on the supercharger case and you can put an after cooler between the outlet and the intake pipe/manifold that runs ito the V before it branches off to the individual cylinders. radial engines had multiple out lets from the supercharger ( the R-2800 may have had nine, see: R-2800 Overhaul Manual) which calls for either fancy duct work or multiple small aftercoolers.
I believe at least one experimental radial used one aftercooler for each cylinder or pairs of cylinders. It is not impossible but it is a lot harder than on a V-12.

Remember that the engine is only as powerful as the cylinder with weakest/hottest mixture

 

[Lighthunmust]

. He added reverse cooling to the design, and the engine was able to demonstrate 1800hp on the bench.

What is "reverse cooling". Is it the changes in direction of airflow that Shortround6 mentions?

You and Shortround6 are really making great posts here and on the water cooled thread. Thanks.

 

[Trilisser]

SR, excellent points. In fact, you got me pants down in this at first. But, when I think it more closely, it would be doable. First, the supercharger unit would not be integral with the engine. Instead, it would be driven off a short entension shaft and the aftercooler unit could be located between the engine proper and the supercharger unit. from that aftercooler unit the induction air would then be delivered to cylinders via brached piping. Of course, this would make direct injection mandatory.

 

[Readie]

The Centaurus was the logical end to power house aircooled radial design along with some of the American designs.Huge power available.
It is just me or do you find that the radial just does not have that the je ne sais quoi...the iconic sound of a liquid cooled inline?
It may just be the exhast arrangement, I'm not sure...
But, I expect Soundround will know.
Cheers
John

 

[Trilisser]

One annoying thing in radial sound may be when they start: the clanking due to crankshaft pendulum counterweights swinging to and fro. But it is a matter of taste.

 

[davebender]

It appears to me the benefits of air cooling disappear by the time you reach about 1,200 hp. After that air cooled engines become rather large and heavy relative to power output. They also become a lot more expensive to produce.

I nominate the R1830 as being close to perfect for an air cooled radial engine. Add fuel injection and it would be even better.
1,250 lbs. Dry weight.
1,200 hp.
Reliable and inexpensive.

 

[wuzak]

What is "reverse cooling". Is it the changes in direction of airflow that Shortround6 mentions?

Reverse cooling for a radial is taking the cooling air over the rear cylinders first and then moving forward.

That is, the air flows right to left in the picture above, not left to right as is convention.

The R-2800 and R-4360 both adopted fans to provide reverse cooling when specified in pusher configurations (XP-56 and B-36).

 

[wuzak]

Lots of engines showed potential on the bench. A down fall of the Deerhound could have been cooling. They showed a lot of ingenuity in the cooling arrangements but they were complicated. Three cylinders in a row means air has to be directed between the rows, turned sideways, travel through the row (cylinder fins), get collected on the other side and then move back out of the space between the rows. Even an R-4360 used a simpler pattern than that. Another potential source of problems in cooling is the is the overhead cam. how much potential cooling area is blocked by the cam box?
Using the same size cylinders as an two row 18 a 3 row 21 gets a 1/6 increase in displacement but adds a heavier crankcase and crankshaft. The overhead cams may be heavier than push rods.

A sleeve valve Deerhound might have been interesting. The sleeve drives wouldn't have been much more coplicated than the overhead cams and might have even required less gearing than than the 14cylinder Hercules. It would also have freed up the head area for a more unobstructed airflow.

In the case of the Deerhound, it had 50% more cylinders than other radials of similar capacity (R-2600, Hercules).

Mks I and II did have a lot of difficulty with cooling, hence the redesign and the move to reverse cooling. The cooling arrangement would be little different to an air-cooled in-line in that the air would be introduced on one side of the cylinder bank and taken from the other. It is also basically what the R-4360 did.

The cam boxes were no more of a problem than air-cooled in-lines. Though, it has to be said, such beasts tended to be small, low powered engines.

A sleeve valve version would have needed 7 drive shafts for the sleeves, maximuim, compared with 14 for the Hercules and Taurus, and 18 for the Centaurus. It may have been possible to do with 4 shafts - 3 servicing 2 cylinders and the 4th taking the last. The sleeves would then be driven by skew gears, as per Sabre.

 

[wuzak]

Here is an aircraft with a reverse cooling arrangement.

The de Havilland Albatross

The holes in the leading edge are for the cooling air. The air is forced (by ram air pressure) down the outsides of the engine, inside the cowling, then crosses the cylinder bank to provide cooling and ending up in the vee. It then exits via the panell beneath the engine. (at least that's the way I surmise the flow goes. (Often in the vee engine the air will go into a plenum in the vee and is pushed outwards to provide cooling, but the outlets beneath the engine look more likely to take the air from the vee.)

The engine used is the inverted V12 Gipsy Twelve

I couldn't find a picture of the Armstrong-Whitworth Whitley, but one did fly with the Deerhound installed.

 

[Shortround6]

It appears to me the benefits of air cooling disappear by the time you reach about 1,200 hp. After that air cooled engines become rather large and heavy relative to power output. They also become a lot more expensive to produce.

I nominate the R1830 as being close to perfect for an air cooled radial engine. Add fuel injection and it would be even better.
1,250 lbs. Dry weight.
1,200 hp.
Reliable and inexpensive.

It depends on how you measure power. A DB605D would be useless as a bomber engine. While it's peak 3-5 minute power may rival a R-2800 it's continuous, max rich, run it till the fuel runs out power level does not. A "B" series R-2800 can run at 1650hp at 2550rpm until it runs out of fuel as long as the cylinder head temp is within limits. There is no 5 min or 15 min or 30 min time limit. It does burn a tremendous amount of fuel doing this but it can be done. Cruising power in lean settings are going to be greater too.

A 1200hp R-1830 offers about 95hp per sq ft of frontal area, A 2000hp R-2800 offers 131 hp per sq ft of frontal area.

A 1200hp R-1830 also weighs over 1400lbs, not 1250lbs.

Adding fuel injection will give freedom from backfires and carburetor icing, I am not sure what else it does for the R-1830.

 

[davebender]

fuel injection will give freedom from backfires and carburetor icing, I am not sure what else it does for the R-1830.

Fuel injection is inherently better when a P-36 or F4F fighter aircraft is flying upside down.

 

[Shortround6]

Wuzak. You have the cooling arrangements of the de Havilland Albatross described perfectly.
While there may have been nothing wrong with the Gipsy twelve it does show a lot of the limits of aircooled V-12s. It was just a bit smaller than a R-R Kestrel in displacement (18.4litres vs 21.2 litres) it was wider, higher, longer, heavier and offered about 70% of the power. This doesn't include the size of the radiator although the cooling arrangements on the Albatross were a ways from simple light or compact

It sure does look sleek and pretty though.

The Gypsy 12 was under the disadvantage of using cylinder assemblies common to the company's 4 and 6 cylinder engines. While this meant that the cylinders were cheap and easy to provide spare parts for it also meant that the cooling fins were marginal for the duty they were asking of it. They were trying to get about 25% more power per cylinder than the unsupercharged 4 6 cylinder engines gave.
It also weighs 7lbs less than an 820hp Mercury radial. which, despite it's better streamlining probably explains why it wasn't used much.

 

[Shortround6]

Fuel injection is inherently better when a P-36 or F4F fighter aircraft is flying upside down.

Why?

 

[wuzak]

Here is another ai-cooled in-line - the Napier Dagger

You can see the inlets that lead to plenums between the cylinder banks for coling. The air exits to th eoutside of the cylinders.

Here it is in action in the Handley Page Hereford

It was also used on the Martin-Baker M.B.2

Data from Wiki

Specifications (Napier Dagger III MS)Data from Lumsden[3]

General characteristics

Type: Twenty-four-cylinder supercharged air-cooled H engine
Bore: 3.813 in (96.8 mm)
Stroke: 3.75 in (95.25 mm)
Displacement: 1,027 in³ (16.8 L)
Length: 80 in (2,032 mm)
Width: 22.5 in (584 mm)
Height: 45.125 in (1,146 mm)
Dry weight: 1,358 lb (616 kg)

Components

Valvetrain: One inlet and one exhaust valve per cylinder
Supercharger: Single-speed centrifugal type supercharger, 5.04:1 reduction
Fuel system: Napier-Claudel-Hobson carburettor
Fuel type: 87 Octane petrol
Cooling system: Air-cooled
Reduction gear: Spur, 2.69:1

Performance

Power output:725 hp (541 kW) at 3,500 rpm for takeoff
794 hp (592 kW) at 4,000 rpm at 5,000 ft (1,520 m)
Specific power: 0.77 hp/in³ (35.13 kW/l)
Compression ratio: 7.75:1
Specific fuel consumption: 0.43 lb/(hp•h) (261 g/(kW•h))
Oil consumption: 0.18-0.35 oz/(hp•h) (7-13 g/(kW•h))
Power-to-weight ratio: 0.62 hp/lb (1.02 kW/kg)

Less powerful than the Peregrine, but some 200lbs/90kg heavier. Although a later version was reportedly good for about 1000hp.

The sound was also, apparently, quite horrific.

A model of the Napier Dagger

The Dagger was developed from the H-16 Rapier.

 

[Lighthunmust]

Thank you for answering my question Wuzak, and with photos no less! Great stuff you and Shortround6 are posting. The rest of us should be paying you two for the education.

 

[wuzak]

Actually, to me the Pennine would be the best air-cooled engine. In early development it was capable of 1hp/cu.in, which the R-2800 could only do with a lot of boost and ADI, while the R-4360 only did so in VDT engines, which had to be continuously monitored and adjusted by hand (no control system existed at that time). The R-3350 could do more than 1hp/cu.in when in turbo-compound form.

In terms of power to weight it was about 1hp/lb, similar to most versions of the R-2800 and better than most other air-cooled engines.

Slightly longer than an R-4360 it could have fitted into an F2G (ie F4U with R-4360), and would have been a great bomber engine.

Data from Wiki

General characteristics

Type: 24-cylinder supercharged liquid-cooled 90 degree X layout aircraft piston engine
Bore: 5.4 in (137.1 mm)
Stroke: 5.0 in (127 mm)
Displacement: 2,791 in³ (45.73 L)
Length: 106 in (2692 mm)
Width: 39 in (991 mm)
Height: 37.5 in (952 mm)
Dry weight: 2,850 lb (1,293 kg)

Components

Valvetrain: Sleeve valve
Supercharger: Gear-driven centrifugal type supercharger, single stage, two speed
Fuel type: Petrol
Cooling system: Pressure air-cooled

Performance
Power output: 2,740 hp (2,043 kW) at 3,500 rpm, +12 psi boost at sea level
Power-to-weight ratio: 0.96 hp/lb (1.58 kW/kg)

 

[Shortround6]

The later Daggers were rated at 955hp for take off at 4200rpm and 6lb of boost. Also 1000hp at 4200rpm at 8,750ft. Cooling problems seem to be endemic though. Those air scoops came with the engine, Napier apparently not trusting the air frame manufactures to provide proper airflow for cooling. Most Herefords did not see combat (if any actually did?) and some accounts say a number of them were re-engined with Pegasus engines.