Wobble-plate axial engine design project.

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.


Well-Known Member
Dec 26, 2021
Reaction score
Kawerau, New Zealand
This is based on the Duke and the Bristol designs, and is intended to be 3 cylinder air-cooled, rotary valve.
4-stroke, swept volume of 90 to 100cc, intended to have an RC model aircraft propeller drive, about 23 inch prop diameter.
this takes 4.5 to 5 hp basic to spin at 6,000 rpm. 6500-7000 rpm output with the test prop could be a more advanced aim.
The Bristol engine was quite a good bus engine in the 1920's and early to mid 30's.
Presumably designed to fit under the bus floor.
Only about half a dozen variations were made and used.

The Duke uses a variation on the wobble gearbox as a synchroniser. This also takes some of the torque thrust.
The Duke also has universal joints on the wobble-frame end, and standard wrist pins at the piston.
The wobble bearing setup looks quite robust.

The Duke is a modern version, but seems to have gone quiet since 2016.

Presumably no-one wanted to take them over. Their engines have been extensively tested.
It is a tough market out there for engines!
The Duke engine has an internal engine block that rotates.
I think a stationary block and a double-sided rotary valve like the Bristol, should work ok.

Duke made a 950cc 125hp version suited to motorcycles, and a 3 litre version for cars.
I don't think they got as far as emissions certification, which is a rather expensive moving target.

I could use standard sealed UV joint centres and grease-packed rollers, which should be good for 6000 rpm.
I propose to use two-stroke premix and draw it through the bottom case.
This avoids the need for using 3-ring sets and an oil circulation system, on a small engine.

A valve plate ratio of 1.5:1 may work. I have seen that on another design.
If the mixture is drawn up alongside the axial shaft, it could help cool the inside "faces" of the cylinders.
Lengthways fins are probably best for a straight shot from the cooling fan.
Images of both designs are included here.
A weight-power aim is for 4 kw for 2 kg. That is better than the Duke, but the air cooling should help.
Sliding faces on the rotary valve could be Nikasil on aluminium, as well as in the cylinders.
should everything be made of eutectic silicon-aluminium- rotary valve. pistons, barrels?
The wobble frame could be a stronger, more ductile aluminium, or aluminium-bronze?
The more ductile heat treated aluminiums don't do well with elevated temperatures.
There is one used for forged race pistons, but it is softer than eutectic Aly.


  • bristol wobble axial engine 1b.gif
    bristol wobble axial engine 1b.gif
    184.8 KB · Views: 254
  • duke thrust transfer mech- replaces gimbal.jpg
    duke thrust transfer mech- replaces gimbal.jpg
    64.8 KB · Views: 184
  • duke cross-section.jpg
    duke cross-section.jpg
    64.5 KB · Views: 180
Last edited:
I am still following the wobble plate layout.

A 3 cylinder version should be fairly well balanced.

I am looking at a peg through a block, sweeping through a sector slot, as the anti-rotation
This is similar to the DSE guide mechanism.

This is easier to sort out than angle-mating face gears , or a fabricated/machined array such as on the duke engine,
which used 5 sockets and 5 rollers in a similar manner to the face gear pair.
The peg would need to be precision-ground, heat-treated, and retained in a socket in the wobble disc assembly.
This can be more of a "spider" layout.
Next, I need to examine the x 1.5 speed valve system, to se how it would translate into a double-parallel valve plate,
and how the passages will clear the gear and bearing arrangement.
I have a porting diagram here, from the Rotocom.
I think this should still work with a stationary cylinder block.
An advantage of spinning the cylinder block is that I don't need an extra flywheel, but it needs more bearings and
more outer casing.

It would be good if the premix would pass through the centre passage and into the valve from below.
Protection of bearing and gears from the exhaust also needs to be considered.
This design need several substantial thrust bearings:
1) main shaft;
2) wobble spider;
3) rotary valve plate.

You could consider that there is no real gain overall with this design, as it is just replacing the standard valve and cam mechanism with a lot of different expensive parts.

An in-line 4 with a drum-disc valve setup would be simpler.
the only real gain is a reduction in frontal area.

I am not sure whether plain bearings can be used with premix.
That may need a fully pressed-up crankshaft, and all needle-roller bearings.


  • rotocom 2 valviing cycle.jpg
    rotocom 2 valviing cycle.jpg
    32.3 KB · Views: 84
  • dse gude device.jpg
    dse gude device.jpg
    8.5 KB · Views: 107
Last edited:
Next stage to be done- sort out a piston rod connection to the wobble plate.
The best choice is a universal steering joint cross.

These are 36mm across, though.
End shells are 6mm inner, 10mm outer, so that fits in with 36mm total.
this is very large, considering the entire stroke is 25mm, bore 35mm.

How much side angle do I need? I was thinking 30 degrees. A Heim type joint has 27 degrees in total, or 13.5 degrees each way.
I will check historical drawings, and see what angles they use.

The entire crankcase on a typical engine is around 70mm in diameter.
A radial 5 cylinder engine case will be slightly bigger.
The history of wobble-plate engines goes back to 1910, so it has had a good run.

One has been proposed as an aero engine for military aircraft in the united states in the 1930's, but they were beaten out first by
air-cooled radials, then by the v12 liquid cooled engine.

I have been looking at some bore-stroke combinations.
38mm bore and 28mm stroke doesn't allow enough room for centre ballraces.
This drops the centre bore down to 27mm
More than 3/4 of the bore can be finned.

The Bristol version has a bore to stroke ratio of 23 bore, 33 stroke, but close to square may be a good compromise.

The wobble plate angle seems to be 23 degrees, buy is more likely 22.5 degrees, being half of 45 degrees.
The overall diameter of the engine looks to be around 5.5 inches, (140mm) allowing an extra 10mm for finning.

I will draw up a version with the 36mm universal joint crosses, and se how everything fits.
The bristol has a gear case at the rear, and a valve shaft running forward to pick up a ring gear on the rotary valve.
the distributor sticks out the side at the rear, running off a skew gear set.
With a slightly smaller bore and longer stroke, 5 cylinders can easily be fitted in the same overall diameter.

The Bristol has an odd multi-disc device in the center of the disc valve.
This may be involved in water-cooling the rotary valve.
there are plent of seals between the inlet housing at the front and the rotary valve.
It looks like it has ring seals on the cylinder ports, but no specific provision for leakage between inlet and exhaust.
This seems also to be a feature of overhead drum valves.
A tight clearance or possibly a number of synthetic slipper pads could be used.
I am not sure whic bearings take up the rotary valve end thrust.
the end thrust of the wobble plate is taken by angular contact ballraces.
The large ballrace right in the middle could be for thrust.
A tiny version doesn't need all the roller bearings.

There are some strange shaft and housing details in the interior of the rotary valve, with some kind of flex plate and a splined shaft.
This could be intended to make the rotary valve self aligning on the sealing face.
There will be a provision for oil dribble between the two faces.
there are still 3 bearings radially supporting the rotary valve, and one nose bearing probably to centralise the nose oil seal.
When they made these, all those seals were most likely packings.
A simpler mechanism that works in a similar way is the cam-plate engine.
This has found application in torpedoes, running on "rocket fuel". These only have to run for a few minutes.
the idea is that you have very long pistons, no piston rods, and pairs of guide wheels in the middle of the piston.
Part of the piston must be cut away to clear the cam plate.
The cam-plate is shaped to maintain clearance to the guide wheels, so motion is not quite sinusoidal.
maximum use has a 5 cylinder with firing at both ends of the piston, but it could be adapted to 3 cylinder, firing at one end.
You can leave out the piston crown and rings at one end, and have a simple plate head.
this eliminates piston rods, wrist pins, wrist pin beatings, bottom end rod bearings, wobble plate bearings,
and adds guide pin axles, guide rollers, and guide roller bearings.

The mechanism creates a bit more friction than the wobble plate design.

A cnc-cam machine centre is needed to make the cam plate, with a synchronised horizontal chuck stepper drive and shaft centres.
I will look into that.
I will take a close look at piston designs as well.
add images:
The dynacam looks to be the best example.


  • alfaro 1a.gif
    alfaro 1a.gif
    21.9 KB · Views: 96
  • dynacam 1a.jpg
    dynacam 1a.jpg
    35.4 KB · Views: 90
  • dynacam 2a.jpg
    dynacam 2a.jpg
    30.6 KB · Views: 88
  • dynacam test 1a.jpg
    dynacam test 1a.jpg
    19.2 KB · Views: 91
Last edited:
1)There is possibly a small torque on the piston. Maybe a slider on the outer edge of the cam wheel would help with this?
2) I think operating the cam in a single plane is best for balance, and using a step-up gearset for the rotary valve.
this causes the piston centers to orbit in a single plane.

I think in this case, all torques on the block from piston movement cancel each other out.
The Dyna-can looks like it uses two piston cycles per revolution to avoid using a gearset for the rotary valves.
This also has the advantage that opposite pairs of cylinder firing balance the end loads on the rotary valves, cutting out
two thrust bearings.
Does 2 cycles per revolution cause more, less, or no imbalance? Can anyone work this out?
If piston numbers are even, then there are always pairs of pistons balancing out their strokes.
Does having odd piston numbers make a difference?

What is the maximum slope that can be used to propel the pistons? I would think 30 degrees would be OK.
close to maximum slope, piston accelerations are fairly low.
combustion forces are fairly moderate, also.

Race engines tend to use higher than normal piston side-thrust, and it causes a lot of wear, but it is considered an acceptable
sacrifice to make the engine smaller and lighter.
They may also use a more ductile piston material, which is softer and less wear resistant, but is less likely to crack.
- this is where piston coatings are an advantage.-resin-graphite-.

There will be rocking moments from the cam if it is in a single plane, which need to be counterweighted.

The pistons follow the same path as a wobble disc.
I think a wobble disc does need crank counterweights. These could be just counterbalancing the offset ends of the center wobble shaft, and bearings, though.

Can anyone explain the balance dynamics of these systems?
Another interesting problem. Work out the stroke as a ratio to the mean diameter, if the maximum tangential angle is 30 degrees,
and there are 2 cycles per revolution.

Is there always line contact with the piston wheel?
if we are displaced 30 degrees on the piston wheel, contact is no longer square on, and motion is skewed.
Can we cut the cam face to maintain square contact?

what is the thrust effect of the skewed sliding contact?

are plain thrust bearings sufficient?

Should the wheel bearing be plain as well?

Can this be lubricated with 2-stroke mix?

Does the piston wheel need a slight barrel form?

What are the best material, surface treatment and surface finish for durability?

Possibly a reason why they never advances from torpedoes?

An intake manifold could be at the non-firing end, with a branch into each cylinder. (2-stroke lubrication).
Last edited:
I prefer a single cycle per crank revolution, as it suits my rpm requirements to be a propeller drive for a 23 inch propeller.

6000 rpm would be too high with 2 throws- this is getting into race-car breathing territory, and quite high piston speeds.
Not that high, though, with a 30mm or so stroke.

A single throw barrel engine can also be a smaller diameter.

This is basically a swashplate engine, with the edge machined to use the two piston wheels or rollers.

Piston mass is another issue- these pistons are about 4x or more the weight of an equivalent 4-stroke piston.

I may have to go up a pin size with the roller pins, or "wrist pins".

Also, loading looks unsuitable for needle rollers, with a lot of skew and side loading.
The rotational speed of the bearing is quite high. How can I ensure enough oil gets in there?
- slot the bearing housing as per needle roller practice, and use a press-in bronze bush?
Surface finish requirements are likely to be fairly high, to the quality of a ground crankshaft?

I will look into the rotary valve ratio required, and a suitable layout for gears, bearings, and seals.

At some stage I need to do a material quantity survey, to see how it stacks up against an equivalent
radial poppet valve four-stroke. I am looking for less than 2.5 kgs for 90cc
The heavy pistons are a negative.
1) balance is simple, and can be done with weights attached to the centre shaft.
2) both swash plates and wobble plates have similar imbalance.
3) departure from sinusoidal is small, and has no noticeable effect on balance.
4) heavier pistons act as a flywheel, so any additional flywheel can be reduced.

Axial Vector engine corp became Axial Vector Energy corp, then went broke about 2017.

They were trying to market smaller diesel generator sets, using their engine design.

The took over all the assets and patents of Dynacam engine corp, but their last engine is quite a radical change.

There are no internal diagrams available, but I have some photos.
the squarish engine is the diesel, and the longer engine is a proposed aircraft engine.


  • axial vector engine.png
    axial vector engine.png
    406.6 KB · Views: 98
  • AxialVectorEngine-0706a.jpg
    20.8 KB · Views: 107
The piston roller wheels can be open deep groove ballraces.
They are likely to spin at 30,000 rpm - a 12.7 x 22.2 skf is good for 39,000 in oil,
so would probably work OK in premix.

Proposed rotary valve layout.
The drawing is a bit rough.
I will make a tidier one later.

Flat based conical, based on the Bristol layout, but simplified.
It also needs two trigger inputs.
One for main shaft angle, one for rotary valve angle.

I need to research 4-stroke ignition systems a bit more.
The one off the Saito R3 would do the job.

A problem area is suction from exhaust into the inlet, when the valve openings are closed.

The Cross-type drum rotary valve has a similar problem.

Possibly a slipper face pad separate from the face seal rings, with a silicone backing, to give it a little spring?
A problem area is combustion deposits sticking to the disc face of the rotary valves.
The face rings on the cylinders will get most of this off.
Rotary valve face lubrication is via the premix.

There needs to be a top seal for the exhaust collector area, and lubrication to the top gears driving the rotary valve.
this is not on the normal premix path, so additional oil distribution will be needed.
I will look into constant loss metering pumps.
This could be fed from the premix, and drain back to the inlet system.

The rotary valve could be eutectic Al-Si with Nicosil plating and surface ground.
The conical surface doesn't need a coating.
I am working out locations for bearings and thrust bearings.
I have added a proposed location for the gear reduction drive for the rotary valve.
Everything is held in place be steps in the main shaft, "D" keys, and a threaded nut as part of the main shaft extension.
This is usually 10mm.

I am not sure how efficient running the inlet through the engine centre is.
This is one of the things that restricts rpms on sidevalve engines.
- a contorted inlet passage-

Most sidevalves tend to top out around 4500 rpms, as opposed to 2-valve 4-strokes, that can do 9,000 rpm fairly easily.

Pro stock draq-race large V8 engines have been known to run to 15,000 rpm, but they must be a bit breathing restricted.
There doesn't seem to be a lot between hemis and wedge heads, though.
They get rebuilt a lot. Don't try this on a street engine!

I have marked up my changes on this image of the Bristol barrel engine cross-section.
It is very close to full size.
this give a diameter of 140mm and a length of 242 mm.
Stroke is 33mm and bore is 30 mm.

This shows the gearing in more detail, and the internal passages for the premix mixture.
This could be run on "Wasted spark" , so only one timing sensor is needed.

Valve timing of 10 degrees, 60 degrees is likely to long for this engine. 5 degrees, 50 degrees would be better.

Total ignition advance could be around 12 degrees, which is plenty for a small engine.
It needs more detailing on the mainshaft diameter steps, to end up with 10mm at the far end.
More joint detailing is needed, too:

1) the head plate to barrel;
2) the cover plate for the conical valve;
3) the gear chamber at the front.

The headplate will need recessed bolts to clear the rotary valve.
These bolt heads could be covered by plugs, or by a floating slider plate.
The slider plate would need to be fairly flat.
Maybe ground stainless sheet, supported by beads of silicone sealer,
The main head plate should give it a good backup.
the flat face seals for the cylinder heads can't be too recessed.
Possibly the inner surface can be partly exposed?

They just need a lip so they don't drop into the cylinder, and a wavy spring to hold them in contact with the rotary valve.
an ideal cross-section would be 2mm square. but that may be a bit flimsy.

Provision is needed to balance the rotary valve as well.
Can this be done with just the original material?
It is difficult to draw so that it will be close to balanced.


  • rotary valve layout 22-02-22.jpg
    rotary valve layout 22-02-22.jpg
    236.9 KB · Views: 87
  • marked up barrel engine layout 22-02-22.jpg
    marked up barrel engine layout 22-02-22.jpg
    271.4 KB · Views: 90
Last edited:

1) First cut at inserting an angular contact ballrace as a thrust bearing for the rotary valve.
This places load back into the main shaft.
Adjustment by shims.
This also supports the rotary valve away from the main shaft, and keys into the valve gear.
Retention by m10 thread on the main shaft.

2) different parts are now coloured to make them stand out.

3) actual size m10 sparkplugs seat shown.

4) detailing needed for head plate.

I will try to get the valve opening more than 50% of the bore.

5) locations added for shaft counterweights.
Piston counterweighting is the same as for the swashplate, but moved along a bit.

6) possibly the head plate can be trapped between the barrel and the rotary valve cover, so that it doesn't need
a separate outside bolting flange.

7) the sparkplug seat and the valve face can be at different levels.

8) should the head plate be made of cast iron instead of plain eutectic Al-Si.?
How heavy would that be?
Is it a better match for the head face ring seating?
The face ring gaps should be pinned.

Next stages:
1) axial view drawings.
2) tidier main drawings.

3) details drawings at 2x full size.
The head ring seals and slipper plate area need finer detail.
possibly the slipper plate can be laser cut and ground flat.
There needs to be raised circular lips around the cylinder face seals.
Are you going to model the project in 3d before you build?
I haven't got any plans for that.
I have been checking bearing size and availability
1) deep groove , 2 needed.
12x32x10 skf 6201. ok.
drawn 10x22x8, piston rollers, 6 needed.
10,22,6 skf 61900
10,26,8 skf.6000
I need to check loading. 26x8 may be more suitable.

2) angular contact:
drawn 12x32x10 angular contact, 2 per, skf 7201 BEP.

Drawn 20x42x10 angular:
nearest 20x47x14, 7204 BECBP- in blue- skf may not make these?
next 20x52x15 7304 BEP - shows in black.
I know B is a 40 degree contact angle, but I am not sure what the other codes stand for.
possibly race offsets, corner radii, ball size, abutment sizes, cage construction.?
The 20mm one takes the load from the rotary valve, but doesn't need to be that big.
The internal size is a result of getting clearance for the gears.
If I have to go up from 42 to 47, I will need bigger gears to get round the bearing.

I could use quite small gearsets:
18:36, and 27:27 teeth.
Possibly straight steel gears would be adequate.

They don't seem to be ground, so may cause excessive friction at 6000 rpm, steel on steel.
Any ideas on off-the-shelf solutions?

I was thinking metric module 1 by 8mm wide.
I would need to bore them out and add keyways.
Minimum bore is maybe 8mm to 10mm
Cast iron gears sound like they would wear better, and be easier to machine.

Any other ideas on stacking the gears and the bearing in a different fashion?
The advantage of this layout is the the bearing is adjustable from the outside with shims.
This sets the rotary valve clearance as well.
I have added a shoulder to the shaft for the nut to bear against.
Last edited:
Scheme for using a smaller angular contact bearing.
1) The bearing is reversed, and adjustment is via a long tube on the shaft.
2) the inner gear for the rotary valve is attached to a tubular extension of the bearing cover, which is screwed in place with
m5 socket-head screws.
possibly this needs to be a firm location of the gear.

3) the inner gear off the main shaft is driven off the sliding tube, which is also driven by the main shaft.

4) a peg-in-slot drive may be appropriate, as the sleeve is too thin for a keyway.
The actual required torque on the gears is quite small.
The peg could engage in a wide keyway in the gear.
The peg could be "drive-through" for disassembly.- maybe a large spring C pin. (or roll pin).
Any opinions?


  • marked up barrel engine layout  r4 23-02-22.jpg
    marked up barrel engine layout r4 23-02-22.jpg
    290.7 KB · Views: 79
fatigue limit of the piston rollers is a problem.
If the piston assembly weighs 100g,
6500 rpm,
peak force = 1.5 kN
rms force = 0.35kN
closest ballrace is 10x26x12,
static = 4.62 kN,
fatigue limit -= 0.083 kN.
seeing as rpm, r = 38mm, , rpm = 6500 x 38/13 = 19,000
this is within the max rpm limit, but we are getting into ultimate fatigue limit territory.
I will look this up.
I would expect that the bearing would need replacement fairly frequently,
but for the use requirements of the engine, it should be good for several hundred hours.

Another factor is the outer race thickness.

A race intended to be a roller would have a thicker outer race.
The thinner race increases peak loading on the balls, and flex in the race, which will reduce expected lifetime.
I can get 26x12 race in the 30mm bore, as the bearing must go up the bore a little.
The actual bearing is 10x26x8, skf 6000
lifetime looks ok.
L in millions of revs.
Lh = 10^6/60xn
life in hours
n= rpm
^ = exponent, power
C = dynamic load rating kN
P = actual average load, kN
there are various load factors as well, but I will ignore them for now.
so L = 2499 = (4.75/0.35)^3
Lh = 2192 = 10^6/ 60 x 19,000
loading factors are:
1) temperature;
2) cyclic load;
3) impact; - contribution from combustion.
4) lubrication- oil mist.
looking at combustion
estimated power = 6 hp.
rpm = 6500 = 680 rad/s
kW = 4.56
torque = 6.7 N-m
spread over 90 degrees of 720 degrees = 53.6 N-m
calculate force average.
T = fr
r= 16.5 mm
so f = 3248N average.
and peak is probably 2x this or more, as we are expanding adiabatically by about 8 times.
so, say 6.5 kN extra on top of the cyclic loading.
This is in excess of the bearing static load rating of 1.96 kN
Possibly a plain bearing may be a better bet.
The application is not that suitable for a needle roller.
The loading is likely to be substantially off-center, with a side drag component.

I will check ratings for a 10x8 bronze bushing.
I think rpm limits may be a problem here.
sliding speed is 1900 x 0.005 = 9.5 m/s.
I will get back on this.
500 ft/min.
500x 0.3/60 = 2.5 m/s
I will check on rating for a needle roller, rounded ends, caged.
I cannot increase the inner pin, as this increases sliding speed.
we are looking at around 8kN as a peak loading.
The nearest needle roller is 12,15,13
ratings C, C0, = 6.16, 8.65 kN
c0 is the static load limit.
rpm = 32000- ok.

What do you think? what is peak load compared with the 90 degree average at the crankshaft?
There should be allowances for friction, and actual peak pressure x area.
area = 707 sq mm.
if pressure = 1500 psi estimate(peak) , then P = (x 6.9) = 10.4 MPa
F = 10.4 x 10^6x 7.07 x 10^ -4
= 73.5 x 10^2 or 7.35 kN
my previous estimate was 6.5 kN

I would have to get into adiabatic expansion to calculate peak force more accurately.
4x 5 mm bolts can hold down 45mm diameter worth of this quite easily.

If bolt stress is say 250 MPa, area = 12.6 sq mm per bolt, total force = sxa = 250 x 10^6 x 12.6 x `0^-6 = 3.1 kN x 4 = 12.6 kN
area = 1.6 x 10^-3 sq mt

at 10.4 MPa, F = Pxa = 10.4 x 10^6 x 1.6 x 10^-3 = 16 kN

I don't think the load on the bolts is that great.
They are not torqued to their maximum amount, but at grade 8 , they are good for
400 MPa, I suppose.

What do you think?
Am I in the ballpark with my loadings?

1) How does a high cyclic loading affect expected life?

2) do I need to use special automotive needle rollers rather than standard SKF ones?
Who sells these?
Last edited:

selection of needle roller bearings.
1) my current engine, 45mm bore, has 10x14x15, as far as I can see.
The nearest SKF one is 10x14x10, or 10x16x12.
rating is 7.65, 7.2, 0.85, 3200
this is good for 7.2 kN static.
10.4 MPa gives 16 kN, so this is bearing good for 4.6 mPa , / 6.9 x 1000 = 659 psi. peak.
this sounds really low for a 9:1 real compression ratio.

15 mm wide instead of 10mm wide makes a big difference.
2mm is cage, so we are going from 8mm to 13mm, an x1.65 increase.
in psi, this goes from 700 to 1138 psi, which is more like a realistic number.
static rating needs to be around 11.7 kN

Mechanical load is only 1.5 kN peak, and a lot less for a standard 2-stroke.
My figure from actual torque should be divided by 3, and add a compression allowance, say 20% of the total?
6.5kN , divide by 3 = 2.2 kN, then add back compression, = x 1.2 = 2.64 kN

Is my estimate of peak pressure being twice the average a reasonable estimate?

this is a triangular ratio, where expansion is more a "square" ratio.

I will have a look at the adiabatics, but I think peak pressure is well over 700 psi.

I will see if I can find more needle race specs with a 10 x 14x15 to a 10 x 16 x 15.

An outer race but no inner race would suit me, as I can use an unhardened roller, and a standard gudgeon pin, or wrist pin.
I can also use wire circlips for retention.
Last edited:
Hello Owen
I just discovered this thread - I haven't been on in a long while (pandemic, life, etc).
I have a soft spot for weird engine topologies and find the wobble-plate design particularly fascinating. You probably know theses related design history facts.
1.Some automotive AC compressors use wobble-plate geometry. I think its the only contemporary use, but I'd be pleased to learn I'm wrong :)
2. The wobble plate geometry was used in some later steam engines (many design ideas that made their way into IC engines were originally used in steam - two-stroke is essentially uniflow, etc.

If you don't know it, this site is a treasure-house of engine designs (and other oddities)
http://www.douglas-self.com/MUSEUM/museum.htmhe puts wobble-plate engines as a sub-class of 'axial engines'
http://www.douglas-self.com/MUSEUM/POWER/axialsteameng/axialsteameng.htmthis is his axial IC page - lots to look at here
http://www.douglas-self.com/MUSEUM/POWER/unusualICeng/axial-ICeng/axial-IC.htmI hope this is interesting/useful.
Sad to hear the dormancy of the Duke project :(
Hello Owen
I just discovered this thread - I haven't been on in a long while (pandemic, life, etc).
I have a soft spot for weird engine topologies and find the wobble-plate design particularly fascinating. You probably know theses related design history facts.
1.Some automotive AC compressors use wobble-plate geometry. I think its the only contemporary use, but I'd be pleased to learn I'm wrong :)
2. The wobble plate geometry was used in some later steam engines (many design ideas that made their way into IC engines were originally used in steam - two-stroke is essentially uniflow, etc.

If you don't know it, this site is a treasure-house of engine designs (and other oddities)
http://www.douglas-self.com/MUSEUM/museum.htmhe puts wobble-plate engines as a sub-class of 'axial engines'
http://www.douglas-self.com/MUSEUM/POWER/axialsteameng/axialsteameng.htmthis is his axial IC page - lots to look at here
http://www.douglas-self.com/MUSEUM/POWER/unusualICeng/axial-ICeng/axial-IC.htmI hope this is interesting/useful.
Sad to hear the dormancy of the Duke project :(
I have had a look through douglas-self.
That is where I got most of my images.

Part of where companies go wrong here is:

1) Insufficient starting capital. They need at least $ 20 million usd just to get started, to show a product, and to start production and distribution.
They can go public after that.

No-one wants to subsidise a company that has no track record, and ownership depends a lot on initial capital.
even $5 mil euro is not enough to get going.

2) Targeting the wrong market.
I think Axial energy had the right idea looking at diesel generators in India and other areas, where power generation is not widespread.
They wasted a lot of money with initial fluffing around, and generally pissed off all the investors.

I think the new Triumph motorcycle company started with around 20 mil UKP, and it has done very well.
Software companies have got started on less, but they are a special case.

3) Duke had a lot of intellectual property, but no capital. Not a good start.
They also don't seem to have found a good market.

1 litre motorcycle engines and 3 litre general purpose engines have a very small potential market, and
there are many companies already selling- Rotax, S&S , Caterpillar etc.

4) The small Utility engines market is already well saturated.

I have found a suitable SKF bearing set that may do:
caged needle, outer race:
12-19-16 skf NK 12/16
this is 12mm bore, 19mm OD, 16mm wide.

specs C, C0, P , rpm
9.13, 12, 1.43, 30,000
C = dynamic rating in kN
c0 = static rating
P = fatigue limit- unknown number of cycles.
My previous calc with over 2,000 hours will have very many millions of cycles.
I will check - 2499 million cycles, 2192 hrs.

I need to find a wrist pin of this size and the correct length.
The outer edge of the piston roller only needs to be 8mm wide
Possibly a 10x14x14 could be a little marginal.
I can get a few likely cage-needle sets ex Ali Express.
the 2.5mm rollers look to have 3mm borders on the cages.
The static load looks like the important one in this case.

I need to recheck this, as I am mixing 2 different bores.
1100 psi x 6.9 = 7590 kPa

area = 15mm, 707 sq mm, x 10^-6

F = PA = 7.59 x 10^6 x 707 x 10^-6 = 5.4 kN

add 1.5 for the mechanical load = 6.9 kN
recheck SKF manual for 10mm:
10-17-16 would be OK for this at 5.94,8,0.9, 32,000
This has an outer race.
Last edited:
peak combustion pressure:
low = 1000 psi, high = 1500 psi
in bars, 70-100.
1 bar = 10^5 Pa (N/sq m), also close to 15 psi.
100 bar is about 14:1 cr.
70 bar = 1050 psi.
so 1100 psi is a good figure for me to use, for up to 10.5:1 cr.
adiabatic formulas:
gamma (g) = Cp/Cv, = 7/5 for air.
P2 = P1(V1/V2)^g
W =( P2V2 - P1V1)/(1-g)
this is work done by one expansion cycle (Joules)
T2 = T1(P2/P1) ^(g-1/g)
^ = exponent, powers.
g = 7/5 = 1.4
discharge pressure = 70 x 10^5(1/10) ^1.4 = 152 x 10^6, which is too high.
I will look at this.
* exponents not commutative!
(0.1)^1.4 = 0.04 x 70 bar = 2.8 bar, or 41 psi, -15 = 26 psi gauge.
This makes a very explosive sound, which is strange. I would have expected a noise like that at 200 psi.
I will check combustion temperature.
1) this is not readily available, but can be worked out from pre-combustion conditions and pressure change.
It will be slightly different due to effects of ionisation and other chemical products lowering peak pressure and temperature.
( this will increase heat capacity temporarily).

2) pressure at tdc before combustion.
p2 = 1x 10^1.4 = 25 bar.
t2 = t1(P2/P1)(g-1/g) - (0.29)
= 273 x(25)^ 0.29 = 694, - 273 = 421 degrees c
3) now for p to increase from 25 to 70, t has to increase by T2 = T1 (P2/p1)^ 0.29.
I think this formula should work at constant volume
694 (70/25)^0.29 = 935 K = 662 degrees C

I would have thought it would be much higher- maybe this formula is no good for constant volume heating.
This is how hot it would get if you kept compressing the gas.

4) maybe PV= nrT is better here?
This will handle heat transfer into the system.

P1V1/T1 = P2V2/T2 if V2 = V1 then the term can be removed.
P1/T1 = P2/T2, so T2 = P2x T1/P1 = 421 x 70/25 = 1179 K or 906 degrees C
This sounds a bit better.
release temperature is adiabatic again:

if release pressure = 2.8 bar, then 1179 x (2.8/70)^ .29 = 464 k = 191 deg C.

I would expect that something else is going on, as exhaust valve temperature is not fully due to combustion peak temperature,
as exhaust valves do get up around 650 degrees c - a nice bright red glow.

This indicates that muffler temperature cannot get over 200 deg C with totally unrestrained expansion.
In that case, temperature will not drop with expansion. - but air resistance will cause some temperature drop.
possibly the sound released is related to the work done on exhaust release.
Catalytic convertors can get hotter, as there are some exothermic reactions going on in there.

I would have thought that combustion peak temperature would be pretty hot, as a propane flame at room pressure can get up to 700 degrees C. I an sure 10:1 compression is going to make it much hotter.

Basic compression will get up to 421 degrees C without any heating.

simple addition of 421 plus 700 = 1,121 degrees C, which is higher than the 906 degrees calculated previously.

Possibly energy absorption is increased at the higher density, accounting for lower temperature than otherwise expected.
Last edited:

Latest posts