Advantages of sleeve valves for H-24 engines?

[ThomasP]

Hey Shortround6,

By the way, thanks for all the little details you have been giving on the engine development and history. They have been very useful filling in a number of gaps in my info.

 

[Shortround6]

You are welcome.

You are quite correct about the sleeve valve vs poppet valve arguments.
An awful lot depends on other aspects of an engine's construction.
Just about every air cooled engine built (or at least any supercharged one) had it's peak power determined by cooling.
2nd was the actual strength of the engine.
You could put the best sleeve valve ever built in a two row radial with no center bearing and you were going to get crap for sustainable power

There are large effects from scaling, or at least cylinder size, which is why there were a number of rules over the years for car and motorcycle racing over the number of cylinders that could be used or different displacements for different number of cylinders.

The whole power per unit of displacement thing is carried to extremes.

For aircraft designers the more important metric is power to weight. I don't care if engine A gets 1.10 hp per cu in instead of 0.98 per cu in if engine A weighs 18% more.

 

[wuzak]

The only other well developed sleeve-valve engine was the Dagger VIII. It was an air-cooled H24-cylinder design with a displacement of 1037 in3 - very significantly smaller than the Merlin & V-1710 - with a CR of 7.5 and max boost of +5 lbs on 87 grade fuel.

The Dagger was a poppet valve engine, with 2 valves per cylinder.

It was also heavier than the Peregrine, though didn't need a cooling system.

An example would be if we compare the Merlin 25 to the Sabre. Both engines were liquid-cooled with 1-stage/2-speed superchargers. The Merlin 25 had a CR of 6 and a maximum boost using 130 grade of +18 lbs/66.5"Hg. The Sabre had a CR of 7. This means the best we should expect in terms of boost from the Sabre, everything else being equal, would be:

6 x 66.5 / 7 = 57"Hg or +13.3 lbs when using 130 grade fuel

What was the highest realistic boost used by the Sabre in service - when using 130 grade fuel? IIRC the boosts were:
Sabre IIB_____2400 BHP at 3700 rpm at +11 lbs (1944)
Sabre VA_____2600 BHP at 3850 rpm at +15 lbs (1946)

The Sabre IIB was 1944, per your numbers. When was the Merlin 25 using +18psi boost?

At what altitudes were these performances achieved?

Also, wasn't the Griffon capable of 2,400hp in 1944 too?

2. Is the supercharger more or less efficient (in terms of heat added to the charge)? As far as I know the Merlin 25 supercharger was considered more efficient than the Sabre supercharger. Greater heat rise due to the supercharger would result in a lower altitude rating and/or a lower boost rating and/or lower BHP.

The Griffon engines in my lists are the 65 and 85, both well developed and using the 'Hookerized' supercharger and inter-/after-cooler. As far as I know none of the Sabre series used an inter-/after-cooler. Yet the Sabre still had the advantage in output. If the Sabre had a supercharger and inter-/after-cooler similar to the Griffons we can assume significantly improved performance from the Sabre.

The Griffon was lighter and had ~25% less frontal area.

As a 2 stage engine the Griffon needed the inter-/after-cooler, the Sabre was a single stage engine and did not need that facility.

There are no (I think) reasonably developed sleeve-valve liquid-cooled engines in the 1650-1710 in3 range. This makes (I think) any direct comparison to the Merlin or V-1710 impractical.

The Sabre was, in reality, the only liquid cooled sleeve valve engine made in any quantity.

------
One has to wonder, what sort of power could the Vulture have made if Rolls-Royce had the resources and finances to develop it along with the Merlin and Griffon.

It already had tested at 2,500hp (per RRHT) by 1941, and surely would have been making that in service by 1943 had its remaining issues been solved. Perhaps more.

According to Lumsden, the Vulture at tested at 3,000hp by the time of its cancellation. But the RRHT could not confirm that.

Sure, the Vulture was 42L vs the Sabre's 37L, but in physical size and weight they are similar.

------
I recall one of the claims made for the superiority of the sleeve valve was that the engine could run higher compression ratio for the same boost, or higher boost for the same compression ratio.

This claim was not borne out.

 

[wuzak]

Ricardo converted a Rolls-Royce Kestrel to Diesel with a sleeve valve system ~1930. It was disappointing and broke a lot.

A year or so later one was converted to run petrol and the standard Kestrel supercharger.

According to Nahum, Foster-Pegg and Birch, The Rolls-Royce Crecy, Historical Series No 21, "the petrol version (known as the RR/P) produced 680hp at 3000 rpm, with the standard moderate supercharger - considerably better than the standard engine".

But that seems to be a match for the Kestrel V, no better.

Does that mean that the sleeve valve was better (the RR/P being smaller in capacity), or that the power available was determine by the properties of the supercharger, and not the valve system?

 

[z42]

I recall one of the claims made for the superiority of the sleeve valve was that the engine could run higher compression ratio for the same boost, or higher boost for the same compression ratio.

IIRC the early motivations for sleeve valves were mainly that

1. Sleeve valves allowed bigger openings, leading to less constricted airflow into the cylinder compared to the then common 2V configuration.
2. Poppet exhaust valves will get very hot and cause premature ignition.

The first issue was largely solved by 4V configurations. That being said, IIRC the first 4V engines were automotive racing engines already in the early 1910's and were rapidly adopted by aero engines, which cannot have escaped the attention of sleeve valve proponents. The second issue was then largely solved by sodium cooled exhaust valves, albeit somewhat later than the adoption of 4V technology.

Meanwhile the issue with the compromise between strength, weight and heat conduction with the sleeves seem like a fundamental issue with the sleeve valve design, with no engineering solution in sight. The issue with lubricating the sleeve was perhaps solvable with sufficient amounts of elbow grease, but it does seem like an inevitably expensive solution requiring very fine machining.

The poppet valve certainly has it's issues with needing to convert the rotary motion of a shaft into reciprocal motion with cam lobes, metallurgy for the valves themselves and the springs etc. But these issues were engineering issues that were solvable. And fundamentally the poppet valve is a pretty sound solution, it's not that sensitive to differential heat expansion of the engine block/valve seats and the valve itself, and the higher the pressure in the cylinder the harder the valve will seal against the valve seat, no need for the valve to glide against the seat while under high pressure etc.

While there are no examples allowing a truly apples to apples comparison between sleeve and poppet valves, as mentioned in this thread, ultimately I think the indictment against the sleeve valve is that it pretty much died out with the WWII Bristol and Napier aero engines (counting here only the engines that achieved some level of mass production). Meanwhile millions and millions of poppet valved engines have been made and successfully used post-WWII. It's of course true that there is a degree of historical path dependence here, but even so, if sleeve valves would have been seen as giving an advantage surely someone would have developed sleeve valve engines post WWII.

 

[ThomasP]

Hey wuzak,

re the Dagger VIII being a poppet-valve engine

Oops! Thanks for the correction. I meant to say "well developed air-cooled poppet-valve engine of note", but the late night caught up with me and I forgot to edit it. I have corrected my paragraph on the Dagger VIII.


re the Merlin 25 at +18 lbs boost - when and what altitude

Both the Merlin 24 and 25 were rated at +18 lbs boost from their introduction to service. Part of the modification to the 24 and 25 was specifically to allow the higher boost vs the earlier XX-23. The +18 lbs boost is recorded in the 1944 edition of the Mosquito Mk VI Pilot's Notes (month unknown) and in the Merlin XX-25 series instruction manual from 1945. I have never been able to find out exactly when in 1944.

There were Mosquito performance tests carried out in Sep'43 at +18 lbs, and performance comparison tests at +18 lbs and +25 lbs boost in Feb'44. And there were Lancaster tests with the Merlin 24 at +18 lbs at in May'44 or B4.

Maybe someone with access to Sqn or production records can answer the question of exactly when.

The critical altitudes for the +18 lbs was about 2,000 ft in Low gear and about 9,500 ft in High gear - both with no RAM.

 

[Deleted member 68059]

Here is a graph of boost (Y axis in PSI-Gauge), vs date (x-axis).

See if people can work out which one is the Merlin and which one is the Sabre (the Centaurus is nearly identical to the Sabre trendline)

(You can if you wish disregard the final data point as its for a sprint run so is an unfair item, the point before it is the Merlin-100 type test value)

The Merlin was a pretty run of the mill layout and the short stroke high speed design of the Sabre is certainly more forward thinking, but
the sleeve didn't do much for it.

These days normally aspirated poppet valves will be able to achieve well above 100% volumetric efficiency, so the cylinder filling argument just evaporates really.

In my view a poppet valve Sabre would have been very good indeed, although it weighs about 700lbs more than a Merlin...

 

[Simon Thomas]

What are the examples?

Or do you expect me to buy a book to read two pages?

I really don't care what you do. I made a statement, and provided my source. As I said, it is worth reading the whole book to get the big picture. Taking a snippet as a means to understand a general pattern will not work.

Is there proof? Or just suspicion?

I just paraphrased from the book. It was written by an ex-Napier employee who as I understand is/was the president of the Napier Power Heritage Trust.

You don't think sleeve valve engines had a fair trial in WW2?

What would have been a fair trial?

It didn't get a fair trial in 2023. Anyone who says the sleeve it too complicated has not spent the time to understand it.
At the end of the day, there is not much between the air cooled sleeve and the air cooled poppet radials.
For the liquid cooled side, the poppets are noticeably better than the sleeves. Had Hooker worked for Napier instead of RR, the situation may have been reversed.

 

[Dash119]

An awful lot depends on when. The R-2600 is a bit of puzzle in itself. There are three totally different engines that share the same bore and stroke. The 3rd engine (the BB) started production in the middle of 1943 and is the 1900hp for take-off version but that doesn't line up with the Military power even at low level. This engine was only used by the Navy with few, if any, commercial sales after the war. Wright may (guessing) have seen the hand writing on the wall and concentrated on the R-3350 and the R-1820 and not tried to split their effort 3 ways. They had enough problems with R-3350 in 1944/45 anyway.
So this makes it rather difficult to compare to the Hercules engine/s.

Engine..........................take-off...................................Military high gear..............................Max continuous high gear
R-2600 BB....................1900hp/2800rpm..............1450hp/2600rpm/15,000ft.................1350hp/2400rpm/14,800ft
Hercules XVI................1615hp/2900rpm.............1545hp/2900rpm/15,500ft.................1300hp/2400rpm/13,500ft
Hercules XVIII.............1725hp/2900rpm.............1600hp/2900rpm/15,500ft.................1300hp/2400rpm/ 9,500ft
Hercules 100...............1675hp/2800rpm.............1625hp/2800rpm/19,500ft.................1415hp/2400rpm/16,500ft

Most any Hercules after the 100 really blows the R-2600 into the weeds, but the R-2600 was pretty much frozen in time from the Spring of 1943 onwards.
Now if you want to go backwards a 1940 R-2600 with 1600hp (the A ) for take-off edges out the 1940 Hercules III
They built about 1900 R-2600 A's in 1940.

P & W tried to build a post war R-2180 14 cylinder engine of 1800hp take-off (wet) and lost their shirt. (1/2 an R-3360) Only one company (SAAB) bought any.

There is a lot of the Hercules saga that is not easy to find.
One book goes through the cylinder head something like this.

Development stage, single piece sand casting, got to 540 sq.in cooling area.
First production engine, die cast single piece, 581 sq in cooling area.
early war time (MKs not given) two piece with die cast bottom shrunk into the flange, 728 sq in.
later heads with better casting got 777 sq in with increase fin pitch. decrease in CHT of 15 degrees C and cut 3 oz of the weight.

A number of experimental heads followed with a machined copper based head being the head used after the war. It was at least the 3rd head incorporating copper for better heat transfer.

A minor point, but the P & W R-2180 was half of an R-4360, Wright made the R-3350.

 

[Simon Thomas]

The R-2600-A5B was a 1939-40 engine but that is probably what the US was willing to export at the time.
The R-2600-BA engine was the 1941 version (443 built in 1941) and this was the 1700hp take-off power version. Used a steel crankcase instead of the Aluminum of the A series engines.
Might very well have used more fuel than the A.
But the BA could run 100rpm faster and had about 3000ft more altitude at max cruise.
BTW the R-2600 A5B was supposed to be rated on 90-91 octane fuel, the R-2600 A5A was supposed to run on 100 octane (US 100 octane)
Hercules XI was supposed to run on 100 octane (British)?

I am all for comparing engines, as long as the comparison is fair.
The British ordered the A5B engines, they got A5B engines, that may be all they could get. They got the results they got.
However that doesn't mean that the test shows the results of a best practice poppet valve engine vs a best practice sleeve valve engine.
Things changed by the year if not sometimes a bit quicker.

The A5B was already being used in the DB7A & B and the Vengeance, so possibly the AM wanted to keep using the same engine.
The BA is obviously a better engine, however the US was still not allowing military engines to be sent to the UK.
The civil GR-2600-B5 was also a 1700 hp TO engine, but it only received its type certificate on Jan 9, 1941. It would take a while for the design information to be sent to Short, and then the installation designed and the engines installed. The first Stirling to be fitted with the A5B (N3657) was allocated to 19MU on Feb 2, 1941.

The C-W engine summary states the A5A uses 90 octane. The Hercules manual and Stirling manual I have are late war revisions, and nominate 100 octane. I don't know if the Hercules XI was earlier set up for a lower octane.

Back in 1940/1941 the AM were desperate. They performed this trial as they were concerned about Hercules delivery, so they took whatever engine they had that was closest in performance and ran it to see what would allow them to get the most bombs into Germany. They could have waited to get the R-2600-B5. It is easy in hindsight to judge, however at the time there would have been enormous pressure to get on with it.

 

[Shortround6]

A minor point, but the P & W R-2180 was half of an R-4360, Wright made the R-3350.

You are correct.
possible reasons
BF (brain fart)
Senility
FF (fat fingers/typo)

 

[pbehn]

Here is a graph of boost (Y axis in PSI-Gauge), vs date (x-axis).

See if people can work out which one is the Merlin and which one is the Sabre (the Centaurus is nearly identical to the Sabre trendline)

(You can if you wish disregard the final data point as its for a sprint run so is an unfair item, the point before it is the Merlin-100 type test value)

The Merlin was a pretty run of the mill layout and the short stroke high speed design of the Sabre is certainly more forward thinking, but
the sleeve didn't do much for it.

These days normally aspirated poppet valves will be able to achieve well above 100% volumetric efficiency, so the cylinder filling argument just evaporates really.

In my view a poppet valve Sabre would have been very good indeed, although it weighs about 700lbs more than a Merlin...


View attachment 730411

Was there any technical reason for sleeves being actuated in the way they were? Could they have travelled straight up and down the bore?

 

[Shortround6]

The A5B was already being used in the DB7A & B and the Vengeance, so possibly the AM wanted to keep using the same engine.
The BA is obviously a better engine, however the US was still not allowing military engines to be sent to the UK.
The civil GR-2600-B5 was also a 1700 hp TO engine, but it only received its type certificate on Jan 9, 1941. It would take a while for the design information to be sent to Short, and then the installation designed and the engines installed. The first Stirling to be fitted with the A5B (N3657) was allocated to 19MU on Feb 2, 1941.

First B-25 flew with a R-2600-9 engine in Aug 1940. Granted military prototypes (or next thing to it) don't have to use engines with type certification (civil).
Design work started in Nov 1938 (nearly 3 years after the A engine) and it was first run in Nov 1939.

I am not blaming the British for doing what they did or getting the results that they did. Just pointing out that the comparison is a bit skewed in timing in favor of the Hercules and/or sleeve valves.

The C-W engine summary states the A5A uses 90 octane. The Hercules manual and Stirling manual I have are late war revisions, and nominate 100 octane. I don't know if the Hercules XI was earlier set up for a lower octane.

Lumsden says the Hercules XI could run on either and gives power for both fuels. Of course in mid to late 1941 I tend to doubt that 87 octane would have been used in actual operations out of England.

Lumsden's numbers.............horse power/rpm/boost/altitude

Engine...........................take-off.......................climb rate medium S............... Climb rate Full S..............................Max power (emer) Medium S...................Max power (E) full S
Herc XI 100.............1505/2800/+6.75............1325/2500/+3.5/2500ft.......1200/2500/+3.5/14,000ft...................1575/2900/+6.75/ 500...........................1510/2800/+6.75/11,250
Herc XI 87................1590/2900/+5..................1260/2400/+2.5/4000ft.......1135/2400/+2.5/15,500ft...................1575/2800/+5/ 1500...............................1590/2800/+5/13,500

I have tried to copy the numbers as best I can. Some of Lumsden's numbers in the tables could use a proof reader. If these are correct it would seem that the XI engine didn't get that big a jump with the better fuel. There could be different numbers approved at a later date. Lumsden does not give dates in the table.

It would appear (to my eye anyway) that the Herc XI was at times trading rpm for boost to keep the power at certain levels ?

 

[Dash119]

You are correct.
possible reasons
BF (brain fart)
Senility
FF (fat fingers/typo)

PEBKAC - Problem Exists Between Keyboard And Chair...

 

[Shortround6]

Was there any technical reason for sleeves being actuated in the way they were? Could they have travelled straight up and down the bore?

two reasons against it.
Greater chance of scoring the walls. It something gets in between the sleeve and the jacket or between the sleeve and the piston the is a better chance of it working it's way back out rather than plowing a groove.

You may need a longer sleeve (?)

Both Bristol and Napier used the 5 port design and with a reciprocating sleeve you may need a 6th port to get the air flow you want.
Not sure what this does to either sleeve strength or to cooling. the back and forth action may get the sleeves to go over more cool areas?

 

[Simon Thomas]

re Heat conduction of sleeves

Ricardo stated:

(2) That a moving sleeve, provided that only a thin oil film was maintained, appeared to be almost transparent to heat.
"The High-Speed Internal-Combustion Engine", Ricardo, Fourth Edition, 1954 Reprint, p349. [Ricardo made significant changes between editions.]

Fedden's paper "The Development of the Mono-Sleeve Valve for Aero Engines" Feb, 1939 stated in response to a question:
"I must again emphasize Mr. Ricardo's remarks about the transparency of the sleeve and the rubbing motion of the oil film, which is the real crux of that problem."

I had to do a heat transfer calculation on a 1/2" pipe at 1300°F, and was surprised to see that the temperature difference across the carbon steel pipe was 0.84 K. I changed the conduction coefficient to that of KE965 and the thickness to 1/8" and ran calcs from 2500°F to 500°F. I found that the temperature difference across the sleeve varied from 8 K at 2500°F to 0.5 K at 500°F. For me there seems no reason to doubt Ricardo's claim, with the same caveat - the oil film must be maintained.

 

[WARSPITER]

so the cylinder filling argument just evaporates really.

Subtle humour at it's best. Very good.

 

[Deleted member 68059]

re Heat conduction of sleeves

Ricardo stated:

(2) That a moving sleeve, provided that only a thin oil film was maintained, appeared to be almost transparent to heat.
"The High-Speed Internal-Combustion Engine", Ricardo, Fourth Edition, 1954 Reprint, p349. [Ricardo made significant changes between editions.]

Fedden's paper "The Development of the Mono-Sleeve Valve for Aero Engines" Feb, 1939 stated in response to a question:
"I must again emphasize Mr. Ricardo's remarks about the transparency of the sleeve and the rubbing motion of the oil film, which is the real crux of that problem."

I had to do a heat transfer calculation on a 1/2" pipe at 1300°F, and was surprised to see that the temperature difference across the carbon steel pipe was 0.84 K. I changed the conduction coefficient to that of KE965 and the thickness to 1/8" and ran calcs from 2500°F to 500°F. I found that the temperature difference across the sleeve varied from 8 K at 2500°F to 0.5 K at 500°F. For me there seems no reason to doubt Ricardo's claim, with the same caveat - the oil film must be maintained.

The sleeve isnt running at 1300F, (704 C) - gas temperature is not metal temperature. If it was, the oil film would carbonise almost instantly.

The sleeve metal temp is probably at about 200C on the gas side, (400 F)

 

[Simon Thomas]

I agree, that why I did a range of temperatures to show the trend with temp. At 400°F the temp difference across the sleeve is essentially transparent, which aligns with Ricardo's statement.
You would know far better than I the huge variation in pressure/temperature/time going on during combustion. From what I understand the peak gas temp for a typical large A/C radial is around 2500°F at around TDC on the combustion stroke. The lowest temp would be the inlet charge gas temp just prior to compression start. The head/sleeve/barrel/rings/oil see a rapid cyclic variation in temp between these two points. Due to the very fast cycle, the peak gas temp is not fully transferred to these components.
Per Ricardo, the sleeve almost acts like the sodium in an exhaust valve by transferring some heat from the head to the barrel.

 

[don4331]

ne fuel, the R-2600 A5A was supposed to run on 100 octane (US 100 octane)
Hercules XI was supposed to run on 100 octane (British)?

two reasons against it.
Greater chance of scoring the walls. It something gets in between the sleeve and the jacket or between the sleeve and the piston the is a better chance of it working it's way back out rather than plowing a groove.

You may need a longer sleeve (?)
View attachment 730427
Both Bristol and Napier used the 5 port design and with a reciprocating sleeve you may need a 6th port to get the air flow you want.
Not sure what this does to either sleeve strength or to cooling. the back and forth action may get the sleeves to go over more cool areas?

Close but no cigar:

If sleeve is just moving up and down, there are 3 issues:
1. You are stopping and start the motion of the sleeve similar to what is happening with the piston. So, you need a similar crank and rods - the sleeve crank only needs to be running 1/2 of the main crank but they are still going to be substantial items.
2. You need sleeve inside of a sleeve (dual sleeves) if you want to be able to properly control the timing of the intake and exhaust cycles. If you thought it was challenging with piston inside of sleeve inside the block, add another sleeve. The inner sleeve needs to extend past the outer sleeve which makes it longer = heavier.
3. And to paraphrase D Deleted member 68059 from the Sleeve Valves at high rpms thread, the multiple oil films over the large surface area of the all the sleeves adds internal friction.

The rotating motion of the Bristol and Napier drives was both smoother and allowed ports to be "timed".