Duke Engines' compact, lightweight valveless axial engine | Hacker News Reader

Duke Engines' compact, lightweight valveless axial engine

gizmag.com

82 pointsanovio56 comments

This isn't a new engine, it isn't a new idea, and it isn't a particularly good idea.

Duke Engines have been around since 1993, and built their first prototype in 1996[1]. Axial engines themselves date back to 1911; Their practical use is limited to torpedoes, where the cylindrical form-factor is an advantage.

Axial engines have inherently high reciprocating mass compared to conventional piston engines, which is a catastrophic flaw in a performance engine design. Higher reciprocating mass increases inertia (reducing throttle response) and increases the forces at the end of the stroke (reducing maximum RPM). They offer no meaningful advantages in terms of fuel efficiency, and are likely to be less efficient in many applications due to the difficulty of implementing existing efficiency technologies (VVT&L, valve deactivation etc)

Both the current Duke engine and their hypothesised next-generation engine offers poorer specific power than current naturally-aspirated designs. The cylindrical form-factor is more difficult to package than a traditional piston engine; Camshafts offer enormous flexibility in terms of layout, allowing the engine to be squeezed into a multitude of shapes and sizes. Axial engines are inherently balanced, but balance is practically a non-issue in modern engines, even for layouts with very poor inherent balance.

[1]http://www.dukeengines.com/technology/overview/

Do the pistons rotate, or does the disk? The disk would have less mass I suppose.

The pistons rotate. Not just the pistons, but the cylinders and the connecting rods as well. It's basically the entire engine block rotating against a stationary cylinder head.

The large rotating mass problem is basically what killed off the pre-WWII-style rotary engines, and the Duke engine suffer from the exact same problem.

So why not connect the disk to the shaft, and have it rotate instead? Not a mechanical engineer, but that would have way less mass, way less clever mechanical linkages for valving, spark.

The disc contains systems that are not easily reciprocated: the intake and exhaust ports, as well as the ignition system. You can be sure that if it were possible to simply swap the rotation, Duke would have done it. The issue of reciprocating mass is well known by engineers who work with axial engines.

I guess I still don't understand. Lots of engines don't put the valves and ignition on a disk - so don't. The idea of valve-less cylinders via rotating ports can be done either way - rotating cylinders or rotating shaft - so reverse them.

I know, that's not a Duke engine. Just wondering who got it wrong the first day and went down this path. Like the old 'drum memory' systems that rotated the heads and left the magnetic memory stationary. Didn't take but 2 years to turn that around and invent disk drives.

So, you believe that sitting in your armchair, you've identified some fundamental flaw in the Duke engine design that they overlooked from day 1?

Yes, ported intake/exhaust solves the problem of rotating the disk instead of the cylinders, but porting comes with its own set of drawbacks. Ask any engineer who has worked on Wankel Rotary engine design and they'll tell you all about it. Ported engine designs include the Wankel Rotary design, as well as 2-stroke, reciprocating, piston-in-sleve (traditional 2-stroke ICE) engine designs. Both have issues meeting emissions requirements because of inherent limitations of ported engine designs.

Cam operated valves have some very specific advantages that play a large role in the ICE's ability to reach current specific output levels. With a ported engine, you cannot vary the intake/exhaust profiles; with a cam, you can. Variable overlap in intake/exhaust, as well as variable intake/exhaust opening area are key aspects of state-of-the-art ICE design. You give up both of these with ported engine designs.

Wondering about these kinds of things is great, but be conservative with your assumptions, and generous in your interpretation. It's condescending and narcissistic to assume that you can take a cursory look at the Duke engine, wave your hand, and solve a massive design flaw.

Oh get off your high horse! I'm speculating. You condemned the design in an earlier post, right? There are all sorts of possible geometries, and many have been tried; some are found wanting. Elsewhere I speculated about using such an engine for constant-speed operation e.g. a generator. It might be a win there, where the ICE responsiveness is not a priority.

Its great to hear from an expert, though one that's clearly invested in the current popular technology to the exclusion of admitting any benefit to this one. I thought you'd be right on board with speculating about where this went wrong. Sorry to have misjudged.

I'm a novice, not an expert. I have a great interest in cars and engines in general, and I've pursued them to some degree, but I'm not an engineer. I have no where close to the level of expertise that the Duke engineers do.

> I thought you'd be right on board with speculating about where this went wrong. Sorry to have misjudged.

You've misread me. There's no thing wrong with asking questions. For example, take the question:

"Why can't we rotate the disc instead of the cylinders?"

Versus your statement

"The idea of valve-less cylinders via rotating ports can be done either way - rotating cylinders or rotating shaft - so reverse them."

You state this as if it's obvious, and it is obvious. It's also obvious (to someone with domain knowledge) that it's not a simple matter. I took us part way down the rabbit hole, and I'm always happy to do that, but it really gets under my skin when questions are states as presumptuous declarations such as, "so reverse them." Not to mention this one:

"Just wondering who got it wrong the first day and went down this path."

This presumes that the Duke solution is the wrong one, and your solution is the right one, but you aren't even familiar enough with ported engine designs to know the basic drawbacks. Again, I'm not an expert, but I know enough to know that I can't make these kinds of presumptions. You don't, and my suggestion is that you should recognize and start with questions rather than insulting someone else's work.

I know this comes across as a scolding, but I genuinely don't mean to be harsh. I just felt like I should say something because of the way your writing came across.

Cool, I apologize if I offended. But if we're talking radical engine design, then nothing is off the table. Including solving the existing problems with the Duke design. They haven't had 2 centuries of tweaking, so criticisms about where soot collects are getting ahead of things.

And also including turning the problem on its head. If shaft momentum is such a well-known and pernicious problem, then its not out of line to question why somebody went down that road at all. Obvious really. Its called returning to fundamentals, and anybody can do it.

So, what's the problem with using rotation to port air and exhaust, if the disk is turning instead of the now-stationary cylinder block? A mechanical linkage between the disk and a ported sleeve should do the trick. If its still desirable at all - a stationary cylinder block lets you do it the valve way too if desired.

Anyway, I'm actually astonished that anybody ever thought it was a good idea to rotate essentially the entire engine. Centripetal forces, heavier bearings, linkage issues - it invents all sorts of problems. Whereas the idea of pushing against a tilted disk doesn't require that at all. I can't get over that fundamental notion, and I'm sure its fair to ask "where did they go wrong" without being accused of backseat driving. If there's some obvious need to NOT rotate the disk, I'm all ears. But I didn't read that anywhere.

Ah! Good point about the reciprocating mass. I did wonder why they suggested them for marine and aviation (semi-static rpm) applications but not cars etc.

Coupled to a automatic or CVT gearbox it may get around this problem.

I do think that your point about it lacking VVTL is somewhat amusing though.

Maybe useful for a stationary generator? Currently my options for a natural-gas generater are one: Honeywell makes a line of 5-12KW device. But they are god-awful noisy; its all any of the reviewers talk about.

But this gadget would shine there! Quiet, no vibration, run it at a single optimum speed all the time. Where can I get one!

No mention of meeting fuel economy and emissions requirements. Both of those are big topics that engine builders have to consider, with the 35mpg fleet-wide average that cars sold in the US must meet in about a decade. Having grown up in an age filled with car exhaust from carburated engines, I don't miss it a bit -- I like having clean air.

One of the comments to the article mentioned the difficulty in keeping high-pressure seals working. It was in conjunction with Wankel rotary engines, which have always had problems with the rotor tip seals leaking (ask any Mazda RX owner...) This engine has the same problem on the intake/exhaust end, as the piston carrier rotates past the openings.

Thank you. I'd wondered how this would compare to a rotary engine. It's interesting how enthusiastic the rotary-engine crowd is, since my takeaway was always "well if it's so much better why isn't it more common".

> "well if (rotaries are) so much better why isn't it more common".

That's a complicated question with a complicated answer, but it basically boils down to there having been hundreds of billions of dollars (trillions?) spent on R&D for traditional (pison, conrod, crank, block) engines, while only a tiny amount has been spent on R&D for rotaries and other "different" designs.

As an RX8 owner (and enthusiast), I can tell you Wankel Rotaries aren't categorically better, or worse. They have strengths and weaknesses:

- The vast majority of rotaries will need a full rebuild before 100k - worse

- They rev to 10,000 RPM with an unbelievably smooth power delivery - better

- I'm lucky to see 22 mpg - worse

- The engine is so small and light that the RX8 has perfect 50/50 weight distribution - better

- If you turn the engine off before it's warm it can flood - worse

- 231 bhp from a 1.3 litre engine? - better

Oh, and the noise - BETTER!

I do believe with more time and money invested the Wankel Rotary could be better than a standard Otto Cycle IC engine in every way. But electric is the future now, so that won't realistically happen.

Learned stick on my second gen RX-7...

-Almost no torque at low rpm - worse

-Easy to flood, then once flooded major work - worse

-Feed it a quart of oil every 1,000 miles - worse

No torque at low RPM is true, but I've not found it to be an issue. I use the engine in the same way as I do on my motorcycle - pretty much always above 4000.

The flooding issue I mentioned. It may have been different with the RX7, but the RX8 has a de-flood procedure that works 9 times out of 10. And failing that a bump start usually does it. Although I've never flooded mine, I'm always careful not to shut her down cold.

Oil usage in the real world is not much worse than many other performance cars, in fact it's better than most Honda S2000s. But for people who are not used to having to regularly top up oil I could see it being a ball ache.

> I've never flooded mine

Neither had I...just never let a valet, mechanic, or relative try to start your old RX-7 ;)

We have a bunch of old rotary race cars in the showroom at the 9-5, you wouldn't believe how much noise and oil they put out the back whenever they need to be moved around :)

wank-wank-wank-wank-BANG!-wank-wank-wank-wank

no torque at low rpm is pretty common among the smaller inline 4 cylinder japanese cars as well ( I really learned stick on my integra GS-R - 189hp 4 cylinder). basically if you don't keep the rpms up when you shift you're just pissing away acceleration power. It was probably more noticeable with the honda engine tho (that vtec).

You missed one - perfect balance. The CG of a properly designed rotary does not move. It's the reason it can rev so high. And flat torque across most of the RPM range is the reason for the power density since torque*speed=power you increase power by spinning faster. Oh wait, you did say smooth power delivery. It does go together with the others.

The short answer is that they're difficult to seal and the combustion chamber shape is awful (it looks like a flattened figure eight in cross section).

It's super tempting as an engine designer to take something that works well as a fluid pump and try and turn it into an internal combustion engine. But sealing and lubricating a combustion engine is a much more difficult problem than sealing and lubricating, say, an oil pump. Plus, in order to meet emissions standards, you need to worry about things like flame propagation and avoid little nooks and crannies where unburnt fuel mixture is likely to hide.

Usually the limited success of these sorts of engine designs isn't a case of a mad genius being too far ahead of his/her time, but rather that the advantages, however compelling, don't make up for one or more fatal flaws.

In this case the sealing issue is probably in better shape than the Wankel, since the seal is located outside of the combustion chamber. I could see something similar to a piston ring in a groove with a wave spring behind it being used no problem.

There is a brief mention of efficiency, but that doesn't seem to be a big point. The port structure does make me think it might be a bit dirty--kinda two-stroke-y.

But not all engines go in cars. Might be a nice one for aviation--though aviation engines are maybe even more handicapped against change than those in automobiles. The experimental aircraft market ain't that big. Other markets where Rotax lives might like it; pumps and such.

Anyway, that's probably another one of the reasons they're not expecting automotive applications in the first generation.

Axial engines go back a long way, to at least 1917.

https://en.wikipedia.org/wiki/Axial_engine

That format is widely used in hydraulic systems, and is the basis of continuously-variable hydraulic transmissions. Classically, it has problems at high RPMs, but is well behaved at low ones.

It's an idea that might be worth looking at again. With better materials and controls, it might work. The geometry is more flexible than with Wankel engines. The elegant Wankel geometry means there aren't many parameters that can be adjusted to improve combustion. In a piston engine, you can design piston face geometry, cylinder head geometry and fuel and air injection points for better combustion. With a Wankel, you're kind of stuck with the geometry. We'll have to see how this new approach works on pollution control.

The only ICE advancement that really seems interesting to me is adding an extra cycle to inject distilled water into the cylinder for an extra power stroke.

The idea being that the heat from the prior explosive power stroke is used to turn the water into steam. Now you're using wasted heat energy and removing the need for cooling components.

One downside to that is that cooling everything will add to NOx in the emissions. A nice hot exhaust is needed for the catalyst to complete clean combustion. Not that it's a bad idea, just that it has an emissions tradeoff.

Distilled water is corrosive to an engine. Otherwise an interesting idea.

I saw an article about that idea a few years ago, and wondered if it might be possible to combine it with a rotary engine. I ended up creating this rather odd-looking model:

http://sandbox.mikepurvis.com/design/engine.svg

Completely impractical, but an interesting experiment. There's some more general info on Wikipedia: http://en.wikipedia.org/wiki/Six-stroke_engine

Seems convenient to use a pretty lame reference engine to make yourself look good. 3 liter engines have made near 1000HP for years now.

How about hp per pound? Isn't that the advantage of this thing?

Like other engines where the pistons rotate (e.g. Gnome) it does have the weight advantage of not needing a flywheel (since the engine is its own flywheel). Other than that, probably not much.

But its supposed to be 1/3 the weight of an equivalent engine. Sounds like a lot?

Without seeing what engines they are comparing it to, I can't say for sure. My guess would they are comparing at similar RPMs, or have some other restriction. I'm sure it weighs a lot less than a traditional 3L engine, but I would guess it generates a lot less power at redline.

Sure its different, and comparing it via the normal means (litres or rpms) won't work because its so different - those means assume a substantially identical engine or they don't work at all.

The best naturally aspirated piston engine is 166 hp per liter. The best naturally aspirated Pistonless rotary engine is 188 hp per liter [1]. The Duke is at 71 hp per liter; it has a long ways to go but it's already better than a naturally aspirated Diesel engine (44 hp per liter).

[1] http://en.wikipedia.org/wiki/List_of_automotive_superlatives...

I always find it interesting that when hp/l stats are quoted they often neglect to include motorcycle engines... which have been in the 180-210 hp/l realm for about 5 years now.

And a high hp/l isn't necessarily the best piston engine as it's often at the cost of massive valve train losses resulting in poor fuel efficiency.

It's difficult to compare the specific output of engines of dramatically different displacement. As a general rule, displacement and specific output are inversely correlated. That's not to say that motorcycle engines aren't an absolute marvel of engineering and technology, but it's not possible to compare them directly to engines with twice the displacement because you can't simply scale up the motorcycle engine and get the same results at the larger displacement.

I agree that Duke has done relatively well, considering their limited resources. That is to say, Comparing superlatives isn't entirely fair to Duke.

BMW makes for good comparison here because they make a lot of 3.0L inline-6 engines. Their N52B30 engine [1] was (mass) produced in a variety of specific outputs, the nadir of which was the 200kW (272 HP) version used in two of their sporty SUVs. This engine makes/made 90 HP/L. The most common variant (used in the mid-tier 3-series) was the 190 kW version, making 86 HP/L. Both of these engines have a better specific output than the Duke engine, but BMW has considerably greater engineering resources, as well as the benefit of nearly 100 years of development.

1: http://en.wikipedia.org/wiki/BMW_N52#N52B30

The money quote, for the TL;DR crowd:

"Duke Engines' 3-liter, five cylinder test mule is already making a healthy 215 horsepower and 250 lb-ft of torque at 4,500rpm – slightly outperforming two conventional 3 liter reference engines that weigh nearly 20 percent more and are nearly three times as big for shipping purposes. With an innovative valveless ported design, the Duke engine appears to be on track to deliver superior performance, higher compression and increased efficiency in an extremely compact and lightweight package with far fewer moving parts than conventional engines."

215hp out of a 3 litre is shockingly low except for a naturally aspirated carburator engine from the 80's.

For example the 2 litre 2015 Mondeo Turbodiesel gets 210bhp.

The article is interesting but reads like a marketing release, the weight saving is the interesting take away I think if reliability matches a modern diesel.

I do agree with you but a turbo diesel isn't a valid comparison because it is:

1) turbo charged 2) diesel fueled

An inter-cooled turbocharger setup on this engine would likely net an additional 100 HP without much issue (presuming the sealing issues aren't horrific). Generally 100 HP per litre of displacement is the gold standard for naturally aspirated engines (see Ferrari).

Curious: what's the alternative to naturally aspirated engines? Space-bound rockets?

The other side of the coin is forced induction, using a turbocharger or a supercharger or even both. "Natural aspiration" just refers to engine with intake air at atmospheric pressure.

I used diesel as historically cc for cc they have lagged behind petrol engines (though that has largely ended).

The turbo is irrelevant though, the IC has a turbo, they work and are reliable the rotating engine doesn't but that doesn't rule out a basis for comparison.

It comes down to fuel efficiency, emissions, maintenance and cost and I can't see this engine winning (for cars).

The turbo is not irrelevant because it has a direct impact on the ratio of power delivery / fuel consumption due to the recaptured energy from exhaust gasses. The volumetric efficiency (ratio of cylinder pressure pre-compression to atmospheric pressure) of engines with turbo-chargers is almost always >1 while the volumetric efficiency of naturally aspirated engines is always (excluding rare high-performance vehicles in specific ranges of operation) <1.

I don't have time to do more thorough research right now, but the Ford Duratec 30 series of engines [1] is roughly comparable in both displacement and cylinder count and produces between 200 and 240 HP depending on configuration which puts this engine slightly ahead of normal engines from a performance perspective (smaller form factor and lighter).

Is there any apples to apples comparison that you can find that actually makes this development look substantially inferior? I get that you used a diesel because historically that gave the ICE an advantage but with modern diesel technology and forced induction they aren't the same animal.

[1] http://en.wikipedia.org/wiki/Ford_Duratec_V6_engine#3.0_L

Edit: changed 'pre-combustion' to 'pre-compression' for clarity on volumetric efficiency definition.

In this context it is irrelevant since my comparison was purely one of power output against displacement.

The Turbo allows a given displacement to output more power for a given displacement, indeed volvo have 2 litre putting out 450hp (uses staged turbo's one of which is electrically driven), I wasnt comparing the nearest ICE equivalent to the rotary.

It seems to share some things in common with the rotary, like moving the ignition chamber past the fuel/exhaust ports, rather than opening/closing valves. So it will need some way to seal the cylinder during compression/ignition... will those seals tend to break up at 100,000 miles like a rotary's apex seals?

I assume the application most people have in mind is cars. As others have said, electric seems to be the future there, so while this may be a revolutionary engine, it's possible the timing won't help it.

What about other applications? Would this work well for marine, aviation, factory, or other purposes?

Looks interesting, but I'm guessing there is a lot of stress on the recipricator mechanism that the pushrods connect to, pushing against an incline to cause the output shaft to rotate.

I agree, but in a typical combustion engine there is also a comparable amount of stress on connecting rod bearings and the typical design of a crankshaft also puts a lot of side-force on the entire piston and rod assembly.

I can't find enough information on their "reciprocator" to see how it differs from other swash/wobble-plate based axial engine designs.

Yeah, I'd be interested in seeing a life expectancy for these. If they're really all that's being said, I'm hopeful for the future.

It can be as beefy as it needs to be. There is no constraint there to limit the design.

It is actually an extremely flexible concept from what I can tell. Need a longer stroke? Add more angle to wobbler. Need more Torque? Expand the wobbler diameter.

There is a limit on the available angle that can be "wobbled". Once you exceed ~20 degrees IIRC, the displacements perpendicular to the stroke start causing problems.

I think the more interesting interaction is between the number of pistons and the size of the pistons.

Wow. That looks inspired by a helicoptor's swashplate driven from the rotor side. Always fun to see an alternative engine design.

Reply on news.ycombinator.com