37 replies to this topic

[unclematt]

Hey All,

Some time ago it occurred to me that you could employ silicon nitride ceramic roller bearings on a roller crank design. There is already widespread use of this material in ball bearing cranks in motorcycles, but I was looking for a way to apply that to larger engines, and perhaps even diesel applications. Bearings made from this ceramic are lighter than their steel counterparts, and are also much tougher. They are even used in the turbo-pumps on the space shuttle's main engines. They also require much less lubrication, and have superior wear characteristics. I have some pdf papers on this, but still no way to post them here (working on that). Cerobear and other companies produces these bearings. I wanted to utilize rollers instead of ball bearings to spread the combustion loads over a bearing contact "line" for each roller instead of the "point" contact that exists with ball bearings. Timken Bearings was working on roller crank designs a few years ago (link below), but haven't heard that went anywhere (using steel roller bearing, not ceramic).

Does anyone have any relevant info or experience on this front? My sources show that as much a 5.4% increase in engine efficiency when a roller crank is employed. Again, sorry I have no link, but working on that. Here is th epaper I am referring to:


Tiemann, C. , Orlowsky, K. , Steffens, C. , Bick, W. , Kalenborn, M.: "The Roller Bearing Engine - A Cost Effective Contribution to CO2 Reduction", Internal Combustion Engine Division 2006 Spring Technical Conference, 08.05.2006 - 10.05.2006, Aachen, Deutschland

And the Timken Roller Crank:

http://www.engineeri...lerized-Engines

And their supporting paper on the topic (using steel bearings, not ceramic):

http://www.osti.gov/...8h6q/958085.pdf

Edited by unclematt, 05 May 2011 - 16:01.

[John Brundage]

Didn't the engine in the OSCA have a roller bearing crankshaft?

[gruntguru]

Love the way the Timken mouthpiece says the system is 3 times more efficient than the original. What he didn't say is the original sleeve bearings only contribute a very small fraction of the total friction of the engine.

Now if they could come up with rollerised pistons and rings!

[cheapracer]

Love the way the Timken mouthpiece says the system is 3 times more efficient than the original. What he didn't say is the original sleeve bearings only contribute a very small fraction of the total friction of the engine.

Grasshopper, you are learning well!

[Kelpiecross]

Didn't the engine in the OSCA have a roller bearing crankshaft?

I think some of the 1970s F1 flat 12 Ferraris had a roller bearing crank.

[unclematt]

Grasshopper, you are learning well!

Yea, Timken has a long reputation of snaking investment dollars away from unsuspecting rubes...
(rolls eyes)

Don't know why this is news, but everyone engages in marketing these days, no exceptions as far as I can tell.

[unclematt]

And from one of the papers I quoted initially:

"Bench mark studies quantifying the sources of energy losses in internal combustion engines have shown that friction in the cranktrain, which includes the crankshaft, bearings, pistons and connecting rods, accounts for as much as 50% of the energy losses at low engine speeds and up to 70% at higher speeds. Those studies, which indicate that half of those friction losses may be attributed to the connecting rod and main bearings, have shown the potential to improve vehicle fuel economy by as much as 5% through the application of roller bearing technology."

The relevant graph is shown on page 14 of the last paper I quoted.

As always, any contrary information is welcome, please post a link to it. Thanks.

Edited by unclematt, 06 May 2011 - 15:02.

[Magoo]

And from one of the papers I quoted initially:

"Bench mark studies quantifying the sources of energy losses in internal combustion engines have shown that friction in the cranktrain, which includes the crankshaft, bearings, pistons and connecting rods, accounts for as much as 50% of the energy losses at low engine speeds and up to 70% at higher speeds. Those studies, which indicate that half of those friction losses may be attributed to the connecting rod and main bearings

That is fabulously incorrect.

Source: Me. Autosport Technical Forum; 6 May 2011; Silicon Nitride ceramic roller bearings on a crank?

[unclematt]

That is fabulously incorrect.

Source: Me. Autosport Technical Forum; 6 May 2011; Silicon Nitride ceramic roller bearings on a crank?

Yea, "me" doesn't quite cut it... :-)

[Magoo]

Yea, "me" doesn't quite cut it... :-)

That was sort of my point. Your source doesn't substantiate its claims either. But as everyone knows, if it's in a paper it must be true. So I gave you a handy citation you can use in case someone asks.

I read lots of papers. Some weeks it seems that's all I do. If I believed everything I read in them, I would have to believe a great number of conflicting things at the same time. To be honest with you I just don't have the intellectual horsepower to pull that off, though I have worked for people who could.

[unclematt]

That was sort of my point. Your source doesn't substantiate its claims either. But as everyone knows, if it's in a paper it must be true. So I gave you a handy citation you can use in case someone asks.

I read lots of papers. Some weeks it seems that's all I do. If I believed everything I read in them, I would have to believe a great number of conflicting things at the same time. To be honest with you I just don't have the intellectual horsepower to pull that off, though I have worked for people who could.

Did you bother reading the sources they cited at the end of the paper? If you read a lot of papers, wouldn't that be your first inclination? I HARDLY claim that anything in a paper must be true, but there is a process of verification that seems lost on many people these days. And as always I am open to contrary information. I am not a "faith based" engineer...

Edited by unclematt, 06 May 2011 - 18:26.

[Canuck]

Well...with a v-twin engine partially assembled on it's roller-bearing'd crank, pistons on, cylinders installed with no rings, I can spin the engine over with just a light hand-twist and it will continue to spin for several revolutions. Add the ring packs and I suddenly need tools to move it at all. A plain-bearing 4-cylinder crank with nothing on it is harder to turn than the aforementioned ringless roller crank, but is still substantially easier to rotate than the ring's v-twin. It's not scientific or apples to apples, but based on my experience I have to side with Magoo - bearings are unlikely 50% of the bottom-end's friction.

[cheapracer]

but based on my experience I have to side with Magoo - bearings are unlikely 50% of the bottom-end's friction.

If I ever bothered which I rarely did, around 15 ft'lbs used to be a non written sort of a rule to turn a reground crank in new mains. A good twist with a strong hand is enough. As some of you know when stripping down an engine with some miles on it free spinning the crank is quite easy.

[nodrift4me]

FWIW I've seen some ball-bearing turbos with silicon nitride bearings used. I'm not sure about any frictional advantages claims, but they claim (from memory) the bearing works better at higher temperatures and so lasts longer, better alignment, etc. I don't have any evidence for/against it working sorry.

[Magoo]

Did you bother reading the sources they cited at the end of the paper? If you read a lot of papers, wouldn't that be your first inclination? I HARDLY claim that anything in a paper must be true, but there is a process of verification that seems lost on many people these days. And as always I am open to contrary information. I am not a "faith based" engineer...

I could refer you to a number of SAE papers, but I'm guessing you don't have access since you are requesting info here, which indicates a degree of desperation. The NACA papers are in public domain, historically fascinating and still valid in their way, but seriously dated.

Assigning mechanical losses is rather fraught. There are essentially two ways: dyno motoring and modeling, and both their have limitations. Dyno motoring is performed with a mule engine in which various parts/systems added or subtracted while the change in absorption is recorded, and you can see all the troubles there. One problem with assigning mechanical losses is they vary with engine type (no of cyls, S/B ratio, etc) and application (race, passenger, industrial). So first, beware of global assertions. Next take careful note as to whether the percentages are FHP or FMEP. These are two rather different properties.

All this said, now I will make my own global assertion: In a gasoline ICE, over half the losses will be found at the piston, rings, and cylinder wall. In the gross output tableau (minus water pump, alternator, etc) it's more than that. Everything else might be considered the loose change in the cushions: Cumulatively they are significant, obviously, but taken individually each one doesn't amount to terribly much. A 5+ percent improvement in fuel economy is absolutely huge, and in my view, you are not going to find it in crankshaft rotating friction. It's not there.

[nodrift4me]

All this said, now I will make my own global assertion: In a gasoline ICE, over half the losses will be found at the piston, rings, and cylinder wall.

This is why I have seen a few race-only* pistons with a single compression ring and one oil ring. They make good power but the compression drops off far more quickly than a three-ring type.

* And on some very small engines, such as lawnmowers, etc.

[bigleagueslider]

unclematt,

Depending upon how your engine is configured and operated, using roller bearings for the crank mains and rods may or may not be worth the small gains in mechanical efficiency.

First, there are the obvious practical issues of cost and complexity. The roller bearings require case hardened inner and outer race surfaces. So the crank journals must be carburized as well as the rod big end bore, which makes these parts more expensive than their journal bearing counterparts.

Second, roller bearings must operate with some radial clearance. Thus they don't like to have dynamic load reversals, which produce impact loads on the roller and race surfaces.

Third, there is the issue of fatigue life. A roller bearing will have a much shorter fatigue life than an equivalent size journal bearing.

Fourth, regarding your proposed hybrid roller bearing design (ceramic rollers with steel race surfaces), it would actually have less load capacity than an all steel roller bearing. The ceramic roller complement would be much lighter than steel, obviously. But the reason these hybrid ceramic/steel bearings have less load capacity has to do with their relative modulus of elasticity. The ceramic roller material is much stiffer than the steel race material, so the steel race surface experiences more contact stress. The problem is also made worse by the way the bearing is loaded. In a rod bearing, the same sector of the inner race surface (ie. the crank pin) is repeatedly subjected to the peak loads. Thus this steel surface is the limiting factor in fatigue life, and not the ceramic rolling element.

slider

[unclematt]

I could refer you to a number of SAE papers, but I'm guessing you don't have access since you are requesting info here, which indicates a degree of desperation.

Desperation? LOL, yea, that simply MUST be it.

Please refer us to any and all relevant info you have access to. Don't concern yourself with my level of access to them, just list them here, or provide links to them.

Edited by unclematt, 07 May 2011 - 13:34.

[Canuck]

I think perhaps you missed slider's point. As he noted, the ceramic bearings have a higher modulus than steel. Higher than their crank pin and rod surfaces. More of the shock load is going to be transferred to the crank pin and rod end because less will be absorbed by the rollers, as opposed to a plain shell which acts as a sort of shock-absorber.

Then again, I'm operating on a tremendous lack of sleep and could have completely missed the boat there.

Edited to add that we shouldn't be goaded into posting our purchased SAE papers because you demand it - buy (or oherwise source) your own.

Edited by Canuck, 07 May 2011 - 13:42.

[unclematt]

I think perhaps you missed slider's point. As he noted, the ceramic bearings have a higher modulus than steel.

I re-read slider's post after I read yours, and I think I misunderstood him, so I deleted my response to him. Thanks for pointing out my error.

And asking for sources to back up people's opinions is hardly demanding anything. Why are people so threatened here by simple requests for verifiable info? Post the title of the studies you claim support your position and I will be HAPPY to buy them for myself. If you think I am trying to get others to pay for my research , you would be very incorrect. Its a little frustrating when you are trying to focus on the technical aspects of a concept, and instead get responses from people who want make assumptions about your motivation, and question your ability to do research.

In the engineering world I am familiar with, you don't make claims without supporting evidence unless you want to embarrass yourself. You also seek out alternative perspectives to make sure you are not fooling yourself into believing something that the evidence does not support. You know, the idea being you don't allow your EGO to overcome SCIENCE. Often people suggest new ways of approaching things one may not have considered initially. That is valuable to me as an engineer, sorry if it isn't to you guys. I guess we operate in different worlds...

Edited by unclematt, 07 May 2011 - 14:01.

[Canuck]

Magoo did the research and presented his conclusions and his (general) source. You dont like his conclusions (you claim unbiased position but your tone and posts indicate otherwise), so you demand his "proof". Go to SAE and NACA and find it yourself - it's your project. A less dismissive and condescending ego might get the cooperation (and paper titles) they wanted.

Your entire last paragraph (presuming you haven't edited it yet again while I'm writing this) is insulting, condescending and dismissive of everyone who's answered your post...all of the people in this thread. They should help you why?

[unclematt]

Magoo did the research and presented his conclusions and his (general) source. You dont like his conclusions (you claim unbiased position but your tone and posts indicate otherwise), so you demand his "proof". Go to SAE and NACA and find it yourself - it's your project. A less dismissive and condescending ego might get the cooperation (and paper titles) they wanted.

Your entire last paragraph (presuming you haven't edited it yet again while I'm writing this) is insulting, condescending and dismissive of everyone who's answered your post...all of the people in this thread. They should help you why?

No, Magoo did not do as you claim. He posted a bunch of unsourced text. No link, no paper title, nothing.

And you say I am being dismissive/condescending, but fail to hold anyone else here to that standard. My ego isn't the problem here, its people who want the reader to accept what they post as factual without providing any way to verify what they are claiming.

Edited by unclematt, 07 May 2011 - 14:21.

[Canuck]

On the contrary - it's people who want everyone to support their "next great idea", in the way that they've defined, in a matter that suits them. Magoo already pointed to SAE and NACA. Do your own searching from there instead of being all pissy because he didn't go back through his own library to find the titles specific to your quest.
I have a modest collection of about 50 SAE papers, a number of SAE tomes, countless little mathematical models I built for myself and a number of non-SAE automotive books and publications lining my shelves. I'd be damned if I can remember half of the topics, much less the titles and locations of them. I go back and re-read little personal "white papers" I've written on various ICE questions and find that half the time (or more) I have no idea what I'm re-reading - I'd have to start almost from scratch. In other words, digging up papers to support our posts takes time and effort that you have thus far been impolite enough to ignore. Once again, your project, your quest, your drive. Rather than having a hissy he didn't hand you the titles and/or publication numbers, you couldnhave thanked him for pointing you to the SAE library rather than Google. Ego.

[cheapracer]

My ego isn't the problem here, its people who want the reader to accept what they post as factual without providing any way to verify what they are claiming.

Actually your ego is the issue, you have been trained, work to and believe in a standard and are unwilling to accept that theres real world knowledge out there that no SAE paper has been written for, a University hasn't studied or a computer simulation hasn't been done for.

This may come as a shock to you but some people just know things.

[malbear]

Actually your ego is the issue, you have been trained, work to and believe in a standard and are unwilling to accept that theres real world knowledge out there that no SAE paper has been written for, a University hasn't studied or a computer simulation hasn't been done for.

This may come as a shock to you but some people just know things.

yes cheapy it starts when your dad chucks you some welding rods and says fix it or the harvestor has too much straw in the sample , why. or pulling your bike to bits as a teen just because
some kids just play with the widjit and learn how and why. I am sure you and Ian know what I am saying.
I remember saving the day when a mechanic dropped a ball into the JD gearbox so I made a "switch on" electric Magnet out of a small bolt and insulated copper wire and went fishing.
malbeare

Edited by malbear, 07 May 2011 - 20:42.

[Greg Locock]

For a good survey of somewhat old but relevant data, Heywood's book is engrossing, easy to read, and cheap. Figure 13-14 suggests that at half rated speed the fmep due to the crankshaft (mains and seals) is less than 10 kPa.

[Magoo]

Food for thought....regard a typical 4 cylinder engine with oh, 80mm x 75mm bore and stroke, with a conventional three-ring, four-rail package on each piston. Roughly calculate how many linear meters of piston ring that represents, and how far all this mess must be dragged up and down the cylinder bores with each rotation of the crankshaft. And note that, unlike a crankshaft bearing, very little of this hardware is under hydrodynamic lubrication -- it's a mix of thin film, boundary lubrication, and direct metal-to-metal contact.


Which raises one of the important limitations of motoring testing to analyze frictional losses: with no combustion pressure to interfere, a nice film of lubrication sets up on the cylinder walls, which on a running engine isn't there. Fired testing (that is, with the engine fired or running), for example instantaneous IMEP analysis with load cells on the connecting rod, quickly reveals this. Yes, there are SAE papers on instantaneous IMEP analysis.

...In racing engines you might see the claim that up to 75 percent of the frictional losses are at the cylinder wall, just to throw a number out there. This is not so much because the ring/piston/wall losses are higher, but that the other losses are proportionally lower, accessory loads (alternator, water pump) being absolutely minimized, for example.


Obviously, reductions in FMEP are equal to increases in BMEP. Not so obvious, magnitude of friction reduction tends to be constant over a range of loads and speeds. Thus friction reductions by percentage tend to be maximized at low speeds and output increase is maximized at low load. Total friction levels are lowest at low speeds and total output is lowest at low load. I suspect that at lower speeds and loads, the difference in friction between plain and roller crank bearings might be too small to quantify with precision. This is very fine whittling. In this area, incremental improvements like piston coatings and crankshaft offset are practical because the investment is low. However, at this time roller bearing cranks do not represent a small investment.

[gruntguru]

The improvements seen will be greatest at high speed and light load and least at low speed and high load.

Unclematt, from previous posts I believe your bearings will tend to be large in diameter, short in length. This is an area (which is no doubt why you are considering rollers) where roller bearings are likely to have an even greater advantage over journal bearings.

[bigleagueslider]

I think perhaps you missed slider's point. As he noted, the ceramic bearings have a higher modulus than steel. Higher than their crank pin and rod surfaces.

Canuck,

Thanks. The higher MoE of the ceramic roller causes more strain in the mating steel race surface than a steel roller would, for a given load. Basically, the steel roller is a better match stiffness-wise than the ceramic roller, and the lower race surface strain at each load cycle improves fatigue load/life capability. SiN rollers also have very low ductility, which means they tend to fail more catastrophically than a steel roller would.

The difference in friction loss between journal bearings and roller bearings is not as great as you would first imagine. The first thing you need to consider is the moment radius where the bearing contact occurs. For a given radial load capacity, the journal bearing will have a much smaller inner radius than the roller bearing. Roller bearings are also sensitive to skewing, especially with rollers that have a high L/D aspect ratio like a conrod bearing would. This leads to high friction losses. There is also the cage rubbing frictions, skidding of the rollers in the unloaded sector of the bearing, and oil churning losses.

SiN rollers would work well where inertia is an issue, or where there are high temps or limited lubrication.

slider

[cheapracer]

Just a note, we have a PhD in materiels here and one of his areas is ceramics and he is investigating if we will or will not produce ceramic bearings for wind turbines so if you have any direct questions I can ask him. I actually play with my balls daily, a couple of ceramic balls sitting on my table.

This comes about because we manufacture Cold Isostatic Presses that forge ceramics amongst other alloy compounds.

I prepared this thread earlier! (pre our PhD actually)..

http://forums.autosp...amp;hl=zirconia

Oh ****! I just remembered where that thread went

Edited by cheapracer, 09 May 2011 - 05:11.

[bigleagueslider]

cheapracer,

Last time I checked, processed SiN was about $600/lb. So it would not seem practical for something like a wind turbine, which is not weight sensitive, but is extremely cost sensitive.

I work on aircraft gear drives, and we cannot even justify the cost of SiN for most aircraft bearings.

slider

[cheapracer]

cheapracer,

Last time I checked, processed SiN was about $600/lb. So it would not seem practical for something like a wind turbine, which is not weight sensitive, but is extremely cost sensitive.

Wind turbines sometimes fail because of overheating bearings setting the oil on fire and at 1 to 2 million bucks a pop a little extra for lubeless bearings is tangable.

My balls are made from Al2O3 (aluminium oxide) whereas Tony's are made from brass.

Edited by cheapracer, 09 May 2011 - 07:55.

[24gerrard]

Wind turbines sometimes fail because of overheating bearings setting the oil on fire and at 1 to 2 million bucks a pop a little extra for lubeless bearings is tangable.

My balls are made from Al2O3 (aluminium oxide) whereas Tony's are made from brass.

Magnetic bearings are the future IMO.
Cost may be a problem today.

Edited by 24gerrard, 09 May 2011 - 08:53.

[Tony Matthews]

Tony's are made from brass.

Cough!

[Kelpiecross]

[quote name='unclematt' date='May 6 2011, 02:17' post='5000062']

Nobody seems to have mentioned (or maybe I didn't see it) that roller bearings on a crank are very difficult to apply anyhow (except for the end main bearings). It needs a split crank or split bearings - neither of which is a great idea.
I think it all depends on how good Timken's split bearing system is. I also don't think the 100+ year old Timken company is trying to fool people into investing in the same way as OPOC (for example).
The only normal road application of roller bearings in a car I can think of is the S600/S800 Hondas (and only on the three main bearings). From memory I don't think the bearings lasted for the life of the engine or car generally.

[Canuck]

Not a car but Harley still uses them although it seems to me the later TC engines have gone away from the double tapered Timkens to a single ball on the mainshaft side. The rods are rollers and the crank is a 3-piece unit - two halves with a one-piece mainshaft/counterweight and pinionshaft/counterweight respectively and a press-fit pin. Gone are the days of the keyed and tapered 5-piece style which did not suffer aggressive launches very well.

[cheapracer]

Nobody seems to have mentioned (or maybe I didn't see it) that roller bearings on a crank are very difficult to apply anyhow (except for the end main bearings). It needs a split crank or split bearings - neither of which is a great idea.
I think it all depends on how good Timken's split bearing system is.

My first crazy engine I started to mod was a VW type 1 that I made SOHC heads and was going to roller the crank but couldn't afford the bearings which were available back then even - mid 70's. My point is if they were available then and with the improvement of alloys and machining techniques over the last 40 years etc then I would think they are trustworthy by now? I seem to remember you could buy roller cranks for VW's back then (SCAT?) and certainly the most famous is the 1955 Mercedes GP straight 8 but of course that was a sectioned built up crank as were some of the 1930's German GP cars.

[Tony Matthews]

The crankshaft from my 1921/22 GN, roller bearing big-ends too.