"Jets are for kids."

[280plus]

From a friend, obviously  :

Friends,

Below, you will find a great test stand video of the Pratt & Whitney 4360 Wasp engine introduced in 1944. Basically, the engine was 28-cylinder four-row air cooled radial engine. Each row of pistons was slightly offset from the previous, forming a semi-helical arrangement to facilitate efficient airflow cooling of the successive rows of cylinders, with the spiraled cylinder setup inspiring the engine's "corncob" nickname. A mechanical supercharger geared at 6.374:1 ratio to engine speed provided forced induction, while the propeller was geared at 0.375:1 so that the tips did not reach inefficient supersonic speeds. Initially, it developed 3,000 horsepower, later models gave 3,500 horsepower.
            The 4360 powered the B-50, successor to the B-29 and later the B-36, to name a few. And although reliable in flight, the Wasp Major was maintenance-intensive. Improper starting technique could foul all 56 spark plugs, which would require hours to clean or replace. As with most piston aircraft engines of the era, the time between overhauls of the Wasp Major was about 600 hours when used in commercial service.
       Why am I sending this to you? Because it looks and sounds neat. That's all. Just a test stand video of a newly restored engine. But turn the speakers all the way up. Bask in the roar of twenty-eight cylinders belching the sound of freedom. Also, the sound will most likely clear out the cat, your wife and any other extemporaneous people not accustomed to military sounds and the glorious roar of a Pratt & Whitney "round" engine. As the Warbird people say, "Jets are for kids."

http://vimeo.com/16117810

[birdman]

Trivia question:
Why do radial engines always have an odd number of cylinders per "row" (a single disk)

[vaskidmark]

Trivia question:
Why do radial engines always have an odd number of cylinders per "row" (a single disk)

1) Because they can.

2) To keep them from spinning off.

stay safe.

[CNYCacher]

Trivia question:
Why do radial engines always have an odd number of cylinders per "row" (a single disk)

There is probably something  bad associated with having cylinders directly opposite each other.

[AJ Dual]

I presuming to keep some sort of nasty harmonics from forming.

And the offsets and angles of all that machining are just amazing for something that was made pre-computer and pre-CNC.

[AmbulanceDriver]

Wow, that's purty.   Gotta love that throaty roar...

[birdman]

There is probably something  bad associated with having cylinders directly opposite each other.

Winner! 

[AJ Dual]

Winner! 

Do we mean the unworkability of a camshaft like that? Or is it torques and stresses?

I've seen some four cyl horizontal aircraft engines that look like they're pretty much in line. Like this Lycoming. http://www.verticalavionics.com/images/product/engine_series_320_4.jpg

Although I can't see the other side, so maybe there still is an offset?

[CNYCacher]

It actually has to do with needing two revolutions to complete the firing order.  The pistons travel to TDC position in order, like 12345.  But those can't all be compression strokes, because then you would need to follow it up with 5 exhaust strokes. (C = compression stroke, E = exhaust stroke):

1C 2C 3C 4C 5C 1E 2E 3E 4E 5E  That would not make for a very smooth running engine

Instead, the valve timing is made such that they fire every other cylinder.  After two revolutions, we are back to where we started:

1C 2E 3C 4E 5C 1E 2C 3E 4C 5E

If you tried this with an even number of cylinders, the best you could do would be

1C 2E 3C 4E 1E 2C 3E 4C 1C 2E 3C 4E 1E 2C 3E 4C 1C 2E 3C 4E

Now.  I SUPPOSE that with enough discs in the engine (the one in question might be a candidate), you could do all compression then all exhaust,  But offset them with other piston sets further down the crankshaft.  That's how all non-radial engines work when you think about them as if they were radials with one piston per disc.  But why do that?  You are adding torque stress to the crankshaft when you could just go with a odd number per disc and have each section of the crankshaft nice and happy and independent.

Thanks for that thought exercise, birdman

[Brad Johnson]

Yup, firing order like CNYcacher says.  Two stroke radials can have as many cylinders as the designer wants because they fire once per revolution.  For strokes fire every other revolution so you need an odd number of cylinders to offset the firing cycle and keep the power pulses even.

For a 9 cyl engine the firing order would be something like 1,3,5,7,9,2,4,6,8, repeat.

I've heard stories of early radial efforts where there would be one revolution that was all firing cycle, another that was all exhaust cycle.  In other words the same 9 cyl above would fire all cylinders in sequence on one revolution, then have a complete revolution for the exhaust stroke.  I can only imagine how odd it sounded.  Brrrripp..silence..brrripp..silence.  Radials tend to produce massive amounts of torque for their displacement, but in a relatively smooth fashion due to the constant firing cycle (so much torque that many pilots died when they poured on full power to abort a landing and the torque corkscrewed the plane into the ground).  The constant "all torque/no torque" cycling of a single-series firing order would be hell on engine mounts and airframes.

Brad

[birdman]

I disagree.  The firing order reason you stated would be workable with either even or odd, with alternating compression/exhaust strokes, regardless of even or odd, because camshafts rotate at half engine speed, so using your logic, I would just make the cam lobes offset by correct angle, and it would still work with an even number.  The reason is one of lockup, with an even number, the carrier assembly on the crankpin would be placed in a position of being resisted by two opposed cylinders in compression, preventing rotation.  If firing order were the logic, for the reasons of smooth power production, all current engines are built incorrectly--since odd engines by definition are out of balance compared to evens (regardless of configuration)--the question was WHY radials are odd.  It's not smooth power output, because an even engine would be smoother.

Now a caveat, the "radials are always odd per disk" is only valid if the number of cylinders per disc is 3 or more.  Note, an opposed twin (BMW motorcycle) is a two cylinder radial.  Now, if you imagine 4 cylinders on the same crankpin, it doesn't work due to interference.  However, you can make a 3 bank, (bugatti w-18 prototype), but not a 4,6,8 etc.

[birdman]

Further clarification--it has to DO with firing, but NOT due to firing order alone, but rather spacing of firing vs angle--with an even number, you would have an opposition to firing.  Also, firing order is not the largest contribution to balance, but rather mass balance is.  In a radial, symmetric (odd or even) mass balance is less important, and now the spacing of the firing matters, so in effect, everyone was right, just incomplete.

[Brad Johnson]

I disagree.  The firing order reason you stated would be workable with either even or odd, with alternating compression/exhaust strokes, regardless of even or odd, because camshafts rotate at half engine speed, so using your logic, I would just make the cam lobes offset by correct angle, and it would still work with an even number. 

Sure, but then you'd need crank and camshafts with multiple journals, making the engine deeper and adding mechanical complexity.  The beauty of a radial is the single-journal concept.  It gives the shallowest profile for the number of cylinders involved.  It also aids in cooling as all the cylinders receive equal airflow (cooling was a huge problem with the three and four row radials).  Unfortunately the single journal aspect of a radial forces the skip-cylinder firing order to even out the power pulses.  That dictates an odd number of cylinders.  Plus, if you're going to use a multiple journal configuration then the radial layout becomes cumbersome and ineffiecient from a packaging standpoint.  At that point an opposed or V- config is more practical and easier to design around/for.

The lack of complexity was one of the main reasons radials saw such widespread use.  No cooling systems to maintain (or get shot out in combat), simple mechanicals that could be easily repaired in the field, and cast-iron reliability in severe conditions.  Unless the master rod was gone, radials could, and many times did, function for extended periods with multiple cylinders shot away.

Brad

[HankB]

Anyone notice the caption under the picture? "This engine will go on a Corsair when finished."

Must be building or restoring an F2G . . .

[CNYCacher]

I disagree.  The firing order reason you stated would be workable with either even or odd, with alternating compression/exhaust strokes, regardless of even or odd, because camshafts rotate at half engine speed, so using your logic, I would just make the cam lobes offset by correct angle, and it would still work with an even number.

Ok then  here is the TDC order of a 6-cylinder radial disc:

1_ 2_ 3_ 4_ 5_ 6_ 1_ 2_ 3_ 4_ 5_ 6_ 1_ 2_ 3_ 4_ 5_ 6_ 1_ 2_ 3_ 4_ 5_ 6_

Show what the firing order will be by filling in the blanks with c and e for compression and exhaust strokes.

If firing order were the logic, for the reasons of smooth power production, all current engines are built incorrectly--since odd engines by definition are out of balance compared to evens (regardless of configuration)--the question was WHY radials are odd.  It's not smooth power output, because an even engine would be smoother.

You can't compare cylinder counts on a linear or V (or even W) engine to the cylinder count of a disc on a radial engine.  A straight 4 engine has 4 different spots on the crankshaft where the pistons interface, and they are balanced by putting 1 and 4 180-degrees opposed to 2 and 3.  Then they make the 1 and 4 come to the top of the cylinder at the same time, but 1 is on compression stroke and one is on exhaust stroke.  For every 2 rotations, each cylinder tires once, but they are evenly spaced.  This is analogous to a radial engine that has 4 discs of one piston each.  You can't compare it to a radial of 1 disc with 4 pistons.  There is no way to make that fire evenly.  The pistons HAVE to come up to TDC in order: 123412341234.

[roo_ster]

That motor is not only a technological marvel, it is a thing of beauty.

[birdman]

.
You can't compare cylinder counts on a linear or V (or even W) engine to the cylinder count of a disc on a radial engine.  A straight 4 engine has 4 different spots on the crankshaft where the pistons interface, and they are balanced by putting 1 and 4 180-degrees opposed to 2 and 3.  Then they make the 1 and 4 come to the top of the cylinder at the same time, but 1 is on compression stroke and one is on exhaust stroke.  For every 2 rotations, each cylinder tires once, but they are evenly spaced.  This is analogous to a radial engine that has 4 discs of one piston each.  You can't compare it to a radial of 1 disc with 4 pistons.  There is no way to make that fire evenly.  The pistons HAVE to come up to TDC in order: 123412341234.

Not my point.  I wasn't trying to compare cylinder counts on an inline or V multiple cylinder, only the number PER journal.  If the firing order were the case, and an even number were impossible on a single journal, an opposed twin wouldn't work, by your logic.  I was reducing your logic to the simplest case to show that firing order isn't the primary driver, of odd or even.

I'll simplify further.  A two cylinder radial would have the same firing problem you suggest as the reason even numbers don't work, however, an opposed twin is actually quite balanced and smooth. With even power delivery.

And to further clarify, a radial with four cylinders can also have an even, alternating firing order, 1c,2e,3c,4e,1e,2c,3e,4c....with simply a two-lobed cam, alternating cylinders actuated from opposite lobes, so that is an even firing order, but still doesn't work in a radial, because you can't still achieve a lockup condition where you have opposing pressure forces--my whole point, you can't have an even numbered radial due to angular spacing and resulting opposition--but firing order or cam timing are a secondary issue, NOT the primary.

[280plus]

My head is starting to hurt... 

 

[birdman]

I think the point of confusion (from my side) is the implication that the adjacent non-firing pattern is disruptive to power delivery, but in an 8 cylinder engine (for example, the biggest gap would be 135 deg--the c-e-e-c sequence or three 45 deg steps), while in a 5 cylinder radial, the biggest gap is 144 (two 72 deg steps).  So it can't be firing gap, that makes them unworkable (true, a 7 or 9 would be smoother than an 8, by 30-45 deg, but a 10 would be more smooth than a 5), but 5's are used and 10's are not--so since it is always easier to machine things in even divisions, there has to be a compelling reason for the choice (the gaps aren't that large relative to normal (+\-45deg from 90 on an 8 cylinder is the same unevenness as a cross-plane 90deg v8).

So to prevent further argument, I'll state my point again, everyone is correct, BUT it isn't JUST the firing order or unevenness that is the issue, it's opposed forces that can result in a lockup indeterminate stable condition, which an odd number avoids (AND adds to smoothness). So everyone is correct.  

Sheesh, I wasn't saying anyone was wrong, only that there is a bigger reason than the one cited (by this thread, and wikipedia)

[birdman]

Also, two stroke radials were mentioned, which, while you can make small engines this way, are difficult in large engines that don't have mechanical supercharging as crankcase scavenging is nearly impossible due to limited volume changes (which is also why opposed cylinder, single journalntwo stroke twins don't work as well as two journal parallel ones--the sliding or close-tolerance seal required to separate the crankshaft for appropriate scavenging is inefficient or unworkable, and merely tolerable in small engines)

[CNYCacher]

And to further clarify, a radial with four cylinders can also have an even, alternating firing order, 1c,2e,3c,4e,1e,2c,3e,4c....with simply a two-lobed cam, alternating cylinders actuated from opposite lobes, so that is an even firing order, but still doesn't work in a radial, because you can't still achieve a lockup condition where you have opposing pressure forces--my whole point, you can't have an even numbered radial due to angular spacing and resulting opposition--but firing order or cam timing are a secondary issue, NOT the primary.

1c,2e,3c,4e,1e,2c,3e,4c,1c,2e,3c,4e,1e,2c,3e,4c, is uneven.  No way to make it even in a 4-stroke setup.  2 revolutions per cycle means the number of cylinders must not be divisible by 2 if you want evenly-distributed power.  This ONLY applies to a radial because the cylinders share a single lobe on the crankshaft and therefore their TDC order is fixed and sequential.  This is in opposition to multi-lobe crankshafts where you set up different cylinders to have synchronized TDC timing, but opposite combustion cycle.

Here is away you could have a 2-lobe 4-piston-per-lobe radial engine, with an even power distribution as a whole, but unbalanced distribution per lobe.  You make the double-power and double-exhaust strokes cancel out

Disc1: 1c,2e,3c,4e,1e,2c,3e,4c,1c,2e,3c,4e,1e,2c,3e,4c,1c,2e,3c,4e,1e,2c,
Disc2: 1e,2c,3e,4c,1c,2e,3c,4e,1e,2c,3e,4c,1c,2e,3c,4e,1e,2c,3e,4c,1c,2e,

[Brad Johnson]

I'll simplify further.  A two cylinder radial would have the same firing problem you suggest as the reason even numbers don't work, however, an opposed twin is actually quite balanced and smooth. With even power delivery.

Unfortunately you can't have a two cylinder radial. Two cylinders on the same journal can only be a boxer or V configuration.  Two cylinders on seperate journals can also be placed in an inline confige.  You would need an additional cylinder on the same crank journal to be able to call it a radial.  Technical distinction, I know, but still applicable.

Here's another twist... if you have an even number of cylinders, say 10, on a split (offset) journal crank, you really DON'T have a single row 10-cylinder radial.  You have a two-row radial with five cylinders per row, only with the crank journals siamesed rather than being seperated by a main bearing.

Brad

[RevDisk]

A lot of the stories I heard from Pratt & Whitney matched a lot of my experiences at Sikorsky.  I was talking to an engineer doing material analysis testing on the F-35.  We remarked and pondered over how in the heck they ever built the J58-P4 engines used in the Blackbird.   

The amount of just plain ridiculously stupid stuff done on a daily basis would blow your mind.  Lemme put it this way, Six Sigma is the corporate religion.  It wasn't done at P&W, but someone in the F-35 program forgot to include the weight of wires during one point of development.   

[birdman]

1c,2e,3c,4e,1e,2c,3e,4c,1c,2e,3c,4e,1e,2c,3e,4c, is uneven.  No way to make it even in a 4-stroke setup.  2 revolutions per cycle means the number of cylinders must not be divisible by 2 if you want evenly-distributed power.  This ONLY applies to a radial because the cylinders share a single lobe on the crankshaft and therefore their TDC order is fixed and sequential.  This is in opposition to multi-lobe crankshafts where you set up different cylinders to have synchronized TDC timing, but opposite combustion cycle.

Here is away you could have a 2-lobe 4-piston-per-lobe radial engine, with an even power distribution as a whole, but unbalanced distribution per lobe.  You make the double-power and double-exhaust strokes cancel out

Disc1: 1c,2e,3c,4e,1e,2c,3e,4c,1c,2e,3c,4e,1e,2c,3e,4c,1c,2e,3c,4e,1e,2c,
Disc2: 1e,2c,3e,4c,1c,2e,3c,4e,1e,2c,3e,4c,1c,2e,3c,4e,1e,2c,3e,4c,1c,2e,

Exactly what I said in my post, my point AGAIN is that that discontinuity is not larger than other, smaller radials which work well, so it can't be JUST that.  Jeez, I concede your point, illustrate why it matters, state that it's that and more, and you still argue "against" the statement?

[birdman]

Unfortunately you can't have a two cylinder radial. Two cylinders on the same journal can only be a boxer or V configuration.  Two cylinders on seperate journals can also be placed in an inline confige.  You would need an additional cylinder on the same crank journal to be able to call it a radial.  Technical distinction, I know, but still applicable.

Here's another twist... if you have an even number of cylinders, say 10, on a split (offset) journal crank, you really DON'T have a single row 10-cylinder radial.  You have a two-row radial with five cylinders per row, only with the crank journals siamesed rather than being seperated by a main bearing.

Brad

Technically, a twin boxer is a radial.  I was using it as an example--even number of cylinders, evenly distributed in a full circle about the rotational axis, sharing a single journal.  Boxer twins have all of these features, so are technically a radial--an even numbered one with the same issues, hence my use of it as an example of of why it's more than the Quasi-big bang firing order causing issues that is the reason.