In retrospect, were the BMW radial engine developments a mistake?

[z42]

The Soviets were somewhat stuck. They had a very small aero industry before 1917 and then it collapsed. Many of the best and brightest fled the country either at the time or in several years of civil war that followed. The Soviet leadership wanted to modernize and the fastest way to do that was to import foreign technology will the domestic workers/engineers were educated/gained experience. Trouble was that some of the technology was constrained by what countries would deal with them and when (although the did wind up buying quite a few designs).

As an aside, the story of the Soviet industrialization is really something. Amtorg boss Saul Bron and American industrial architect Albert Kahn setup training for thousands of Soviet engineers and designers, and oversaw building of hundreds of plants and factories. Magnitogorsk, a massively huge steel works that AFAIK still is one of the larger ones on the planet. Modeled after the US Steel corporation plant in Gary, Indiana. The Stalingrad tractor works, that produced a zillion T-34 tanks during WWII? A carbon copy of the International Harvester plant in Milwaukee. Etc etc., designed together with Americans, often trickier parts/machines sourced from the US and the rest locally.

And what happened to Saul Bron after all this massive contribution to the industrialization of his country? Well, having dealt with those Americans so much he was considered potentially dangerous and corrupted by their pesky capitalist logic, so he was executed during one of Stalin's purges.

 

[special ed]

I have mentioned the Ford engineer who supervised the plant to build Model C trucks in the late 20s was sentenced to Siberia as a spy upon project completion. Those trucks are in many photos from WW2.

 

[Shortround6]

Any insight on why in late 1930s Germans should look at Hispano - of all engines - as an inspiration on how to make big (and bigger) V12 engines?

Only to see what not to do
It is instructive as to what it took for the Swiss and the Soviets to get more power out of it.
The German engines may have had much stronger crankshafts to begin with but most WW II engines needed bigger/heavier crankshafts to raise the rpm much (or at all), some engines would tolerate higher boost but not additional RPM.

 

[SaparotRob]

As an aside, the story of the Soviet industrialization is really something. Amtorg boss Saul Bron and American industrial architect Albert Kahn setup training for thousands of Soviet engineers and designers, and oversaw building of hundreds of plants and factories. Magnitogorsk, a massively huge steel works that AFAIK still is one of the larger ones on the planet. Modeled after the US Steel corporation plant in Gary, Indiana. The Stalingrad tractor works, that produced a zillion T-34 tanks during WWII? A carbon copy of the International Harvester plant in Milwaukee. Etc etc., designed together with Americans, often trickier parts/machines sourced from the US and the rest locally.

And what happened to Saul Bron after all this massive contribution to the industrialization of his country? Well, having dealt with those Americans so much he was considered potentially dangerous and corrupted by their pesky capitalist logic, so he was executed during one of Stalin's purges.

Deserved more than just an "informative".

 

[Shortround6]

And it seems for radials it's a bit similar, that a two row with 7 or 9 cylinders per row is the sweet spot, seems all the radials with >2 rows had serious problems with cooling the back rows as well as vibration issues.

For some reason 9 cylinder single row radials seem to be fairly easy to design. Just about everybody built them (except Armstrong-Siddeley).
Most people seemed to able to build 14 cylinder two row radials.
18 cylinder two radial engines seem to be a much harder proposition. A lot of people tried, most did not succeed, or at least the failures out number the successes. Or varying degrees of success. The Soviets went though 3 different 18 cylinder engines before they got one that worked (No 4) . The Italians went through 3 different ones (or more?) but you don't hear much about them, some were used in the Cant 509 floatplane and a few SM 79s had them but with 1000-1100hp from a 45.7 liter engine the Italian aircraft designers weren't beating down the door to get them.

The big boys figured out how to cool the rear rows. The real problem was cooling high power cylinders, you need a crap load of very deep fins spaced very closely together and you need baffles that force the air though the fins and not tootling on by the outer edge of the fins. The ability to manufacture the fin structure was sometimes beyond the capabilities of the engine makers/tool makers. It was one thing to make a few prototypes, it was another thing to manufacture hundreds if not thousands of cylinders per day.
P & W R-2800 cylinder could make 111hp from 2.55 liters ( at 2000hp) at 2700rpm and stay cool (for at least 5 minutes) in the thin air of 25,000ft. It make 90hp at 2550 rpm for hours (sucking down a crap load of fuel as coolant).
The Problem with big V-12s is that they have 2/3 rds the number of cylinders. Each cylinder has to make 50% more power. They have water/glycol to help them but you also have bigger pistons and a longer stroke. The volume of the cylinder increases faster than the area of the cylinder walls and head to help take the heat away.

 

[z42]

The real problem was cooling high power cylinders, you need a crap load of very deep fins spaced very closely together and you need baffles that force the air though the fins and not tootling on by the outer edge of the fins. The ability to manufacture the fin structure was sometimes beyond the capabilities of the engine makers/tool makers. It was one thing to make a few prototypes, it was another thing to manufacture hundreds if not thousands of cylinders per day.

How are these things made, BTW? Is it really possible to cast the cylinder barrel with all these thin fins? Or are the made from sheet metal cut and brazed/welded to the cylinder (I've never seen what could be a weld joint between the fins and the cylinder on an air-cooled engine so I guess this isn't it either?)? Or milled from a big chunk of metal (must be very time-consuming?)?

Is there any study on the cost (or at least number of man-hours) to produce a radial vs. a liquid cooled V-12 of equivalent power? Not to say that developing the manufacturing techniques to produce modern style monobloc inline engines (including the head gasket!) wasn't challenging, but I'd image once you figure it out, the man-hours spent on each casting is somewhat modest? And there's a relatively small number of such castings that make up the major parts of the engine.

Edit: This article suggests that air cooled cylinder heads are made by casting. Possibly lost wax casting for stuff that needs to be very precise and small dimensions like the cylinder head with it's large thin fins? Castings for Aircraft: Part 1: Wax and sand casting processes.

 

[Shortround6]

How cylinders and cylinder heads for air cooled engines varied from country to country and at any given time. A lot depended on the casting and forging industries of the time as many engine makers did not do their own casting and forging (some did) but subcontracted it.
At one point (late 30s?) the British were using forged cylinder heads because the British Aluminum industry could not produce castings of the required strength. The American Aluminum industry could supply casting of the required strength, casting was cheaper. Later on P & W and Wright both changed to forged heads on the higher powered engines.
Specialty machinery was also developed to handle large production runs. The US were the first to use large ganged slitting saws. The saw blades were mounted side by side but the depth of the cuts were controlled individual on each blade by cams running on a mold so each blade could follow individual contours.
Cylinder barrels went from all steel with the fins cut in them (which meant you started with the really thick cylinder) to a steel liner with an aluminum "muff" around it and Wright figured out how to machine groves in the steel and then roll/calk sheet metal fins into the groves.

At one point (and remember there were two different R-2800 cylinder heads) a single cylinder head started out as about 70lb forging and by the time they were done the finished cylinder head was 16lb (or under?) They used a drop forge on the original forging at least three more times and there may have been intermediate machining operations. The valve seats were certainly pressed into place.

Some engineers could look at a captured engine and figure out what was done but not how it was done. Like figuring out exact alloy/s, the surface finish, the hardness and sometimes even the grain structure. They could tell the difference between cast and forged. It was harder to to tell the steps taken to get to the end result.
Ford figured out how to use cast steel cylinder barrels instead of forged. But they were using a rotary table with 8 stations that spun as the metal was poured into the molds and the centrifugal force helped force out air bubbles and inclusions. Also had a different grain structure than regular castings.
Machines like that made sense if you were building hundreds if not thousands of engines per month.

 

[rinkol]

The practical reality was that the BMW 801 was the most powerful engine suitable for service use from 1941-1944. Both DB and Junkers had serious problems into 1944. The radial engined FW 190 ws also a preferable choice for ground attack than the versions with liquid cooled engines.

 

[z42]

The practical reality was that the BMW 801 was the most powerful engine suitable for service use from 1941-1944. Both DB and Junkers had serious problems into 1944. The radial engined FW 190 ws also a preferable choice for ground attack than the versions with liquid cooled engines.

The BMW801 also had serious problems for a long time after the initial introduction, largely due to the valve issues. BMW was quicker in figuring out the solution, but this isn't anything inherent to radial vs. inline engines. The real failure here is perhaps that the information wasn't disseminated from BMW to DB and Junkers, causing great harm to their war effort overall.

In the hypothetical scenario in this thread where BMW would have continued developing inlines instead of pivoting to radials, they could just as well have been the ones to first figure out the solution to the valve issues in that case too.

Admittedly a radial has some advantages wrt ruggedness. Is it enough to make it worth pivoting if your national R&D and production capacity otherwise is focused on inlines? Probably not.

 

[Shortround6]

Is it enough to make it worth pivoting if your national R&D and production capacity otherwise is focused on inlines?

In 1935-38 there was darn little production capacity for inlines, from any manufacturer.
Which would be smarter, introduce a 3rd line of engines (DB and Junkers both had projects, just not a lot of production) or take the money for the BMW inline factory and just build another DB or Jumo factory?
As soon as you change the bore and stroke, or even the spacing between the bores, you change the vibration pattern of the engine.
There were limits as to how big you could make cylinders, when you tried to go bigger it took more work. The twin row radial offered 16.6% more displacement for the same sized cylinder. Obviously it had a bunch of new problems but trying to scale up the existing V-12s also had problems. And if you were building a new factory anyway you aren't' saving much tooling.

 

[tomo pauk]

In 1935-38 there was darn little production capacity for inlines, from any manufacturer.
Which would be smarter, introduce a 3rd line of engines (DB and Junkers both had projects, just not a lot of production) or take the money for the BMW inline factory and just build another DB or Jumo factory?

Depends on what engine type is to be built. If BMW is to make a V12 that is 30-35% or greater displacement than the DB 601 or Jumo 211, and can offer 25-30% more power, then it is worthwhile to shell monies to BMW.

As soon as you change the bore and stroke, or even the spacing between the bores, you change the vibration pattern of the engine.
There were limits as to how big you could make cylinders, when you tried to go bigger it took more work.

You can note that nobody is suggesting outrageously big cylinders here.

The twin row radial offered 16.6% more displacement for the same sized cylinder. Obviously it had a bunch of new problems but trying to scale up the existing V-12s also had problems. And if you were building a new factory anyway you aren't' saving much tooling.

Twin row radial will be operating with just two valves per cylinder, historcally it took more time to get good and efficient supercharges on ww2 radials than on ww2 V12s, and the radial will make greater drag than a decent V12 installation. BMW managed to make fresh air intake either inefficient (that is for 'normal' installations) or draggy (for outside intakes).
BMW 801 consumed more fuel than DB 603 or Jumo 213, let alone DB 605; what it consumed was pricy and not something Germany was awash.

 

[marathag]

Depends on what engine type is to be built. If BMW is to make a V12 that is 30-35% or greater displacement than the DB 601 or Jumo 211, and can offer 25-30% more power, then it is worthwhile to shell monies to BMW.

Consider this development path

Year first run	Engine	Displacement	HP	weight	Note
1926	BMW VI	2865	640HP	1124 lbs	Direct Drive
1930	Mikulin M-17	2864	680HP	1200 lbs	License copy of BMW
1931	Mikulin AM-34	2864	800HP	1480 lbs	Monoblock construction
1939	Mikulin AM-35	2847	1350HP	1830lbs	Intercooler, glycol cooling introduced
1941	Mikulin AM-38	2847	1700HP	1940 lbs	low altitude
1942	Mikulin AM-42	2847	2000HP	2195 lbs	improved AM-38
	DB 601	2068	1330HP	1323 lbs	better grade fuel, plus fuel injection, higher compression ratio and supercharger design

 

[z42]

In 1935-38 there was darn little production capacity for inlines, from any manufacturer.
Which would be smarter, introduce a 3rd line of engines (DB and Junkers both had projects, just not a lot of production) or take the money for the BMW inline factory and just build another DB or Jumo factory?
As soon as you change the bore and stroke, or even the spacing between the bores, you change the vibration pattern of the engine.
There were limits as to how big you could make cylinders, when you tried to go bigger it took more work. The twin row radial offered 16.6% more displacement for the same sized cylinder. Obviously it had a bunch of new problems but trying to scale up the existing V-12s also had problems. And if you were building a new factory anyway you aren't' saving much tooling.

Fair enough, though I wasn't really thinking in terms of sharing bore, stroke and bore spacing. More in terms of the general engine development ecosystem with universities and technical schools, research institutes, companies making whole engines or components etc. Somebody who is an expert on vibration analysis of inline engines and how to solve vibration problems there will surely be able to apply that knowledge to radials, but the issues and solutions will be slightly different. The knowhow how to produce good inline monoblocks and head gaskets are different than producing cylinder barrels for radials. Cooling design and manufacturing is obviously very different. etc. Radials generally running hotter meaning there are different issues to solve wrt valves, sparkplugs, fuels, perhaps even lubricating oils.

As an aside, this AFAIU was one of the things that made developing sleeve valve engines slow. The rest of the engine industry had pretty much settled on poppet valves, and information about alloys, (sodium) cooling etc. etc. was disseminated through the industry. For the sleeve valves, the buck pretty much started and stopped with Fedden's team at Bristol (yes, later obviously Napier and RR got in on the game too).

 

[Shortround6]

Somebody who is an expert on vibration analysis of inline engines and how to solve vibration problems there will surely be able to apply that knowledge to radials, but the issues and solutions will be slightly different. The knowhow how to produce good inline monoblocks and head gaskets are different than producing cylinder barrels for radials.

There wasn't as much institutional knowledge as you might think. At lot of what you are talking about didn't even exist in the late 19 teens and was only being developed in 1920s.
They were evolving everything during WW II, the engines in 1944-45 were very different from the engines of 1939-40.
Some of the engine designers of WW I had never gone to school for engineering. It showed in the late 20s and early 30s as demands exceeded "intuition".
Henry Miller of Miller/Offenhauser fame bankrupted a company he worked for the late teens with a 6 cylinder inline that broke crankshafts. It was years before he designed another 6.
He built a lot of 4's and 8's.

for some idea See.
Crankshaft Design Evolution

For V-12s there seemed to be a barrier around 2400rpm, or at least for Hispano Suiza until the mid 30s. Racing engines excepted.
Different orders of vibration came into play.
And the constant fight between weight and strength. Car designers weren't a lot of help. They could just add weight to solve stiffness or strength problems.
Some the car guys were very good, but not all, just like some of the aircraft engine guys weren't that good.

 

[z42]

There were limits as to how big you could make cylinders, when you tried to go bigger it took more work. The twin row radial offered 16.6% more displacement for the same sized cylinder. Obviously it had a bunch of new problems but trying to scale up the existing V-12s also had problems.

Depends on what engine type is to be built. If BMW is to make a V12 that is 30-35% or greater displacement than the DB 601 or Jumo 211, and can offer 25-30% more power, then it is worthwhile to shell monies to BMW.

You can note that nobody is suggesting outrageously big cylinders here.

Another option for avoiding big cylinders, without going to the H or X layouts, or bolting two V-12's together like the DB 610 tried, would be a V-16. Say, take the same bore and stroke of the 36L BMW 117, and make it a V-16 would give you 48L. Not saying that would be trouble-free, but V-16 layouts were successfully used in racing engines, and are used to this day in large diesel engines.

 

[z42]

There wasn't as much institutional knowledge as you might think. At lot of what you are talking about didn't even exist in the late 19 teens and was only being developed in 1920s.
They were evolving everything during WW II, the engines in 1944-45 were very different from the engines of 1939-40.
Some of the engine designers of WW I had never gone to school for engineering. It showed in the late 20s and early 30s as demands exceeded "intuition".

All the more reason to focus efforts to build up that institutional knowledge as quickly as possible! Especially as Germany was trying to catch up, after the loss of WWI and the treaty of Versailles, the economic woes of the Weimar republic etc.

 

[mikemike]

So as a result of this decree and BMW realizing they were starting from zero wrt radials, they licensed the P&W Hornet to kickstart their development.

Some other posters also remarked that BMW had no experience with big radials, but don't forget that BMW had a shotgun wedding with BRAMO, also known as SEM or Siemens-Schuckert, who had been working on such contraptions. Bill Gunston writes that the Twin Fafnir, a 14-cyliner two-row, developed 2000 hp in October 1938. So there was institutional knowledge.

 

[tomo pauk]

Bill Gunston writes that the Twin Fafnir, a 14-cyliner two-row, developed 2000 hp in October 1938. So there was institutional knowledge.

Care to elaborate on the Twin Fafnir, eg. actual nomenclature, is there some footnote to this, any clues in German language, what fuel was used etc?

 

[Shortround6]

Not saying that would be trouble-free, but V-16 layouts were successfully used in racing engines, and are used to this day in large diesel engines.

A number of V-16s were actually a pair of V-8s placed end to end and drove the cams and other accessories through the center gear tower. The US Chrysler V-16 not only operated this way but used the center gear tower dive the driveshaft which ran through the "V" of the forward V-8.
V-16s were generally considered to be heavy for the power developed.
None of the V-16 racing engines were trouble free but often the troubles had nothing to do with the crankshaft.

In AIrcraft engines power to weight was very important. More so than in racing engines or large diesels (marine/locomotive). There is no artificial displacement on aircraft engines or on marine engines, etc.
large diesels were also designed for manufacturing economy. Being able to offer 6-8-12-16 and even 20 cylinder engines all using the same pistons, connecting rods and valve gear meant more than saving a few hundred pounds on engines that weighed tons. The Diesels also weighed several times per hp what the aircraft engines did. not a good comparison.
As quick snap shot of this, in 1951 Alfa Romeo was running a 1.5 liter straight 8. They had stated work in 1938 so there was a lot of development. An awful lot of development.
At the end of the 1951 season they were getting 420hp/9,300rpm from a 363lb engine running 98.5% Methanol fuel and 3.10 Atm manifold pressure.
BRM was trying to sort out their 1.5 liter V-16 engine which finally gave them 430 hp/11,000rpm from a 525lb engine. They were using 4.85 Atm pressure.

Now consider that the aircraft engines had economy of scale. They also were running on gasoline and not Methanol. The BRM engine was one of the ones that used the central gear tower and separated the engine into two 750cc V-8s (with a 135 degree angle between the Vs) and again, power was taken from the middle of the engine.

 

[z42]

Some other posters also remarked that BMW had no experience with big radials, but don't forget that BMW had a shotgun wedding with BRAMO, also known as SEM or Siemens-Schuckert, who had been working on such contraptions. Bill Gunston writes that the Twin Fafnir, a 14-cyliner two-row, developed 2000 hp in October 1938. So there was institutional knowledge.

Seems the situation with Bramo was somewhat similar to BMW. Back in WWI they did produce some advanced for the time rotary engines, but they didn't keep developing them (and while rotaries obviously have a lot in common with radials, it was clear time was running out for them), and when they realized they were behind the state of the art they licensed the Bristol Jupiter in 1929, a 1918 design. And further Bramo radials were developments of this design. After BMW bought Bramo they combined their efforts (or less charitable, BMW was the boss and cancelled the Bramo projects and moved the engineers over to the BMW projects?) which eventually resulted in the 801.