Interesting bearing thought for main bearings.
 

Heres a thought I had the other day about replacing main and rotor bearings in our engines, now I'm not a bearing expert or engineer so feel free to tell me I'm stupid.

I pressed a set of main bearings out of a set of stationary gears yesterday and have been thinking of multi-peice e-shafts as well. I thought of replacing the journal bearing that normally presses into the stat gears and rotors with large roller bearings. Similar to this:

http://www.duckol.com/uploadFiles/up...r-Bearings.jpg

Obviously something thinner and wider, but something that could press into the stat gears and rotors. Then machine 5 sections for an e-shaft.
The first section would get pressed into the roller bearing in the front stat gear with your regular sized front shaft with key-way for the front cover assembly sticking out. On the inside part of the engine the front e-shaft section would be tapered and splined to accept the front rotor section of the eshaft which would be pressed into the roller bearing inside the rotor.

Heres a very crappy paint drawing:

http://i33.photobucket.com/albums/d8...earingidea.jpg

Thats just the front section of the engine, if it continued to the rear, there would be 5 e-shaft sections all coupled together by tappered splines. Including a main bearing in the center iron to prevent e-shaft flex.

Doing this you'd need to machine a complete new e-shaft but it would not have to have a perfect finish on it since the bearings would just be pressed onto the seperate sections.

Now obviously I haven't thought this right through because I dont really know what to do for oiling these bearings or how to hold the e-shaft all together (maybe a bolt down the center of it?)

But anyways....I'd like to hear other opinions on this idea. No harm in putting it out there!

Five points to build error?

I think you could go with the same idea used in 2-cycle engines and other equipment. The inner bearing race is the shaft itself. You could then use a 2-piece shaft.

Build error?? I wouldn't get a deaf blind man to do the machining LOL.

http://www.fc3spro.com/TECH/HOF/eshaft01.jpg
http://www.fc3spro.com/TECH/HOF/eshaft02.jpg

Do you really want to machine it smaller there to have it be able to fit into the pressed in end sections? That'll be a weak area in the rear, those pics are of the front, but it shows that the shafts aren't indestructable and they can and will shear off at that point

A better idea would be to use the shaft as a race and press in the bearing into the stat gear, if you can get one thin enough and use this with a pre-existing 2-piece eccentric shaft. You'll be seriously limited by the tight gap between the componenets, as it's designed to only accomodate the thickness of the bearing, and not seperate races and rollers. That sort of bearing with very skinny rollers is called a needle bearing by the way. I've heard them being referenced for use in really high rpm engines for the rotor bearings, but have no info on this.

Besides, you forgot to think about the most important thing, what is this meant to accomplish and can it be done in a simpler, easier, or already existing way and why will someone be willing to pay to have it done?

Why reinvent the wheel when its not broken? Sorry I don't see any reason for them. In the intermediate housing to eliminate flex ok, but I just dont see a use. Cool thinking though.

I can see wanting needle bearings with the idea to reduce friction for maybe more power and to possibly help fix wear or spun bearing issues with really high rpm's. But the shaft is supported by the oil pressure anyway, so there's not much friction to start with (it'll be small compared to that of the seals) and the bearings don't seem to be a problem for high rpms in particular (There's competition bearings that are meant to handle the high speeds better), or at least I don't see people complaining about them.

Good for you for thinking outside the box though.

I assumed you wouldn't, but I still think you would have a hard time keeping total clearance of all the pieces added together to an acceptable level. I also believe there would be strength problems. And how do you fasten them at assembly without excessive pounding, pressing, or beating of press fit pieces if you plan on no clearance between the pieces?

Actually I'm intrigued by the idea. I'm sorry if you thought I was scoffing at it. With a center bearing, 2-piece shaft, and using the shaft as the inner race it surely seems possible.

Yeah I know how journal bearings work. But I'm talking about revving waaay past 9 or 10K rpm. I've seen it happen too often with race engines, one missed shift from the top of 2nd gear back into 1st and bang...bearing failure.

So you guys think using the shaft as the inner race would be better than having a full bearing that just presses onto the shaft and into the stat gears/rotors? Hmmm well theres a thought... Maybe that way it could just be a 2 peice shaft instead of multiple.

Those pictures from above of the broken e-shaft are probably partially because of a bearing failure or excessive clearance. With a needle bearing design that shouldn't be a problem....or would it? Like I said, I'm not engineer or bearing expert so I'm just thowing ideas out there.

Maybe we need to concentrate on the source of the problem and beat the drivers who over-rev the engines ;) . Seriously though, some nice solid drivetrain mounts and some good solid shifter bushings should help reduce such occurances.

I think you'll HAVE to use the shaft as a race, not that it's a good idea to do so due to wear, but the severe space restrictions would seem to necessitate it, if one will even fit in the first place. It's a nice hard, smooth surface though which is good though. From what I've heard the mating surfaces for the bearings on the other side aren't exactly perfectly smooth, so you'd need a race there. Either way you could press the races on prior to the assembly of the motor and assemble the bearings while you're building the engine, then there's no need to do a 5 piece e-shaft or anything crazy like that.

The e-shaft picture was more to show that having it narrow down a bunch in there in order to fit the one splined bit into the other isn't a good idea as it'll severely weaken the part and that they're not exactly indestructable now. A hard drag lauch could probably shear it right off at the back of the engine. A needle bearing could fail too if the loads are high enough, it may not seize, but it could still be damaged.

Even if you can get the bearings to survive, if you over-rev it to, say 15k, then it should probably get looked at anyway because who knows what else might have been damaged right?

There's a problem here though, and it ties into why they use journal bearings. Using small diameter needle bearings like that, the velocities get silly really quickly. It's going to be a pain in the ass to keep them from just starting to skid, and because of the stress levels on them... they're going to wear to the point where they do that pretty quickly. And be more susceptible to over-rev.

That and the shaft is one of the things that's going to keep the rotors from skewing and contacting the irons, so doing anything to reduce the cross-sectional area of it would be counter-productive.

I'm thinking it would be hard to use a 2mm needle in the mains.

I have worked with engineers in the past who have worked with the formula one setups and have personally seen how cam roller bearings hold up in NASCAR cup engines at 4,500 rpm. Kenku is right you are going to increase performance at a sacrifice of longevity. After one race our cam roller bearings looked like a bag of ass and needed to be replaced. They would probably handle over-rev very well if the usage on them is within a certain limit, after that the amount of abuse they could take would be significantly less.

A roller bearing has less load bearing surface area. This increases the rate of wear at the points of contact. This is also why if you tear into a mazda pickup truck transmission and an RX-7 transmission of similar years you will find the truck has bushings and the RX-7 will have roller bearings instead. This is because the bushings can take more load and will last longer with the loads the truck would put on them.

Also a perfectly smooth surface is not always ideal, to keep the rollers working properly you will actually need a slightly rougher RA finish, otherwise the needles stop rolling and will skip across the oil eventually causing galling and bearing failure.

For these reasons it would never be economical and only have practical use on race circuit cars where frequent replacement would be acceptable.

Well obviously there'd need to be a more detailed study into this before any action is taken to make sure such things don't happen. Then the bearings would need to be selected based on sound engineering practice. Generally I don't see it happening, and don't see it being that helpful in making an engine live longer or better survive stratospheric revs, other things will break instead.

Just stick with a 2 piece shaft and lightened rotors (less force) then add some competition rotor bearings, some windowed mains, keep the oil pressure up and do the oil loop line to ensure good pressure to the front bearing (better support the force) and it should survive quite well. Having that center bearing will help reduce the incidence of spun bearings by keeping shaft flex in check so the shaft doesn't cock sideways in the bearing and it'll spread the loads out on a larger area, both of which will help to keep the shaft from making hard contact on the bearing.

It's been done. For reasons mentioned above, it doesn't work well. Back in the day, SCOOT used to do this but I don't think anything ever came of it. Journal bearings are highly reliable, and won't send needles through the engine if they fail...

^Interesting picture Aaron, is that a picture from Scoot?


So your saying that the roller bearings could possibly increase the performance because of less resistance? But wouldn't last very long...

The picture is from a random 13B drag engine I found. 2 piece Guru shaft...

Ah, center plate, the Guru shaft uses a roller bearing in the center plate, but notice the thickness of it compared to the main and rotor bearings.

I am an engineer at a company that designs and builds mud pumps for drilling. Our pumps go up to 3000 hp. We use roller bearings everywhere. We have a couple of competitors that use babit sleeve bearings. We tested a competitors pump vs one of ours in terms of mechanical efficiency. Ours tested at around 95% mechanical efficiency and the competitor without the roller bearings tested at around 75%. An oil temperature test showed where this energy was going. Our oil stays literally a few degrees above ambient temperature. It's amazing. We also aren't running through a hot internal combustion engine so heat is different from a car engine. The competitors oil temps were at about 180 degrees. Their oil is used as a bearing. That's what it does in a babit "bearing".

Bearings make a huge difference. Keep in mind we are turning at 120 rpms and not thousands so this doesn't mean that rollers are a good solution for a car engine. I've looked into using them but I am most worried about longevity. The rollers would be very small and would be turning very fast. Our oil is used to cool rotors too so oil temps would likely not change appreciably even if rollers worked good. I'm still willing to try it though just to see what would happen.

One problem has been finding a bearing that wouldn't require machining down the e-shaft. If you just press out the bearing from a stationary gear, you don't have much clearance. You need more. I just miked an e-shaft and got a usable diameter of 43.16 mm where the bearing rides. This may vary slightly shaft to shaft based on wear. 36 mm is as small as you can machine down the shaft for bearing room and still have threads for the e-shaft nut on the back. You can't really do much on the stat gears as they are fairly thin already. I pressed out a bearing and got 46.85 mm. That leaves us a little bit over 10 mm of room for roller bearing thickness.

I found a roller bearing in 2 widths. The relevant one is 20 mm. You'd need to use 2 of them side by side for a total of 40 mm. They have an inner and outer race and measure 37 mm inner and 47 mm outer which is within our machinable area on the e-shaft. That's the skinniest ones that I can find that have a decent load rating. These will give you a 100,200 psi static load rating. That's alot. The dynamic load rating however is 56,400 psi. Again that sounds like alot but I have yet to calculate what the dynamic load of the engine is at peak rpm. It may or may not be enough. Even if it is, you need a safety factor. For production pieces we shoot for 1.3:1 on things like belts but much higher on bearings. We like nearly 10:1. When our pumps fail it's usually a bearing that does it. Keep in mind that with our high safety factor, that's usually nearly 20 years down the line! I'll have to figure this all out in regards to a rotary someday. I deal with the same thing for pumps everyday.

Even with a bearing that will fit, you will have to machine the eshaft. That means rebalancing the rotating assembly. That's doable though. My biggest concern is internal engine clearance. I have yet to figure out how to set e-shaft play. For a good roller bearing to work and be reliable, it needs to be pressed on tight. If it's not, it may still work but wear will go up. If it is pressed on tightly, assembling the motor would be next to impossible and you can't set end play.

There is alot to think about with rollers in car engine. We don't have these concerns on mud pumps which makes things easier for us.

rotarygod, do you have a part number or anything? I seem to recall that once when I was looking out of curiosity I couldn't find anything applicable, but that was a while ago. I'm kinda tempted to try experimenting.

I forgot which catalog I saw it in. It's either timken, NSK, or URB. We deal with each of them (except our bearings are several feet in diameter!). I need to go figure it out again. I remember inner and outer diamters and the fact that they are needle rollers. That shouldn't be too hard to locate again. Keep in mind that you will have to turn down the eccentric shaft on a lathe to make them fit and do very minor lathe work on the stat gears. Just a few thousandths though.

If the problem of assembly and setting internal clearances can be solved, then it would also be possible to use rollers on the rotors as well. One problem at a time. I suspect this is why we don't see them. It's a pain in the ass!

Yeah, I've got a lathe and shitloads of spare 12A e-shafts, which is why I was pondering it. Like I mentioned before in the thread, running numbers through my textbooks say that it isn't going to last but... well hell, textbooks are an inexact representation of reality anyway. ;)

I have to wonder if they'd seal well enough to have any oil flow to the rotors.

Probably not worth the trouble, but hell it's just time and tooling.

Oil cools the rotors so you still need to have that. You also need to have a hole in the outer race of the rollers to get oil flow into the bearings. Although they aren't babit bearings, they do need lubrication.

If you could use a 3rd main bearing as the Guru shaft does, your bearing stresses go down as you now have more bearing area. That would make a difference in lifespan but creates other issues of course. You'll be limited to only one 20 mm bearing in the center though. You wouldn't get 50% more bearing but it's still more.

I am skeptical about longterm use. I have no doubt that it will work in the short term. Another issue with bearing life is engine rpm. Higher rpm's hurt bearing life. Rollers will probably work fine up to a certain rpm cap. I don't know what that is though. This rule holds true with all bearings though. More rpm's means less longevity. More power at a lower rpm is easier on the engine than less power at a higher rpm. Within reasonable limits of course.

I'll eventually try to calculate this all out to see what it "should" do on paper. Real world results can vary. That's why people test! I don't know when I'll get to it though. It's not really a priority.

I'm rather more than just skeptical about long-term use, but a lifespan of (to pull numbers out of thin air) 5-10 hours with a 9500rpm cap? For some purposes that might be enough to be useful if the efficiency gains are statistically significant.

Yeah this is the kinda info I like to hear, someone who deals with stuff like this. Thanks rotarygod. I talked to a couple bearing companies like NTN here in Canada and they told me the highest rpm roller/needle bearing they make is for 10,000rpm....then the cage that holds the needles in place break and would cause the same damage as in the pic Aaron posted above.

I've found bearings that can go as high as 11,000 that fit the needed size range. The problem is the load ratings. I can't find a heavy duty needle roller that will fit in this space. I can only find standard duty bearings. I doubt highly they would live long. I'm still looking. Next time our bearing guy stops by, I'll ask him if can find anything.

I have built loads of two stroke bike engines in my life and they all use needle bearings on the crank. These engines rev in excess of 13grand and the bearings are tiny. some go down to 40 mm in OD. I dont know the amount of load of a 2 storke compared to a 12A or 13B, but Ive seen and heard of 250cc v twin engines that develop 80 hp plus at 18k rpm.

I have been thinking of this for a very long time but never got to testing.

Keeping in mind, in a two stroke you dont have any oil pres at all! Oil is mixed with gas and this is the only lube the crank gets(apart from the main bearings that sit in gearbox oil)

I had a Suzuki RGV 250 with 15k on the clock and the only thing that failed was the cylinder liners. This engine saw 15grand every day.

I have a few dead 12A rotors, e-shafts and s-gears. all of these will need machining as they spun bearings.

If anybody can get part numbers on bearings I will gladly have these machined and build a test engine.

One thought though, are either the e-shaft or s-gears hard enough to use as a bearing race. If so, it will allow a larger needle witch inturn might increase life?
Saying this, on a 2 stroke the conrod is steel and this is used as the bearing race, aswel is the crank journal (witch I believe to be a lot harder than the con rod though). :bigthumb:

This would be really neat if we're talking about ideal conditions...

Now, when I'm talking about "ideal", I'm talking about a perfect...circular...motion.

The big problem is that the rotary engine does not do this.
Therefore, you get assymmetrical loads...which needle (hell, any roller / BB) bearings do NOT like.
There are diagrams out there that show eccentric shaft (and therefore bearing) loads.
I'd really question how long the bearings are going to last under such conditions.

Do we have an engineer qualified enough to answer those questions?


-Ted

But i`ve seen worn out con rods and they always wear on the top?

Just because it's not perfect circular motion isn't that bad, so what if it has asymmetrical loads? In "normal" uses there's almost always some sort of asymmetrica load. Centrifugal pumps load up one side due to the thrust loads from the outflow, horizontal things are loaded from gravity and so on.

I was thinking about this just now. I remember a very old suzuki 150cc 2 stroke air cooled engine that was used in a go-cart. This engine vibrated a lot because of the bigger clearences on an air-cooled motor. But stil, it had a rollar bearing crank in the same way the new ones do and it ran fine. keeping in mind that the bearing still coped with these vibrations. :Wconfused

So, Rotomax rotaries are base on Mazda 13B dimensions but use air cooled rotors instead of oil cooled (brings intake charge through the rotor before going into ports.)

Do they use needle bearings lubed by premix?

I know the Aixro XR50 (294cc 50hp wankel) uses a needle bearing rotor.
http://www.nova-racing.com/nova_gb/P...rig/XN0035.JPG

This is the bearing load info taken from the "Rotary Engine" book by Kenichi Yamamoto.


-Ted

We get the same types of asymmetrical loads on the bearings of our mud pumps. We run full rollers not only on the crank as mains but also in the connecting rods. There's no problem with it at all and our pumps get up over 2000 hp. I am one of the engineers there (and there's only 2 of us) so I guess I am qualified to answer the question. It's the same principle. The fact that it's a rotary vs a piston is irrelevant. Bearing loads do not need to be symmetrical. Babit style bearings are FAR less efficient. Bearings are the first things that typicaly fail in a pump but this is usually after continuous use for up to 25 years. A babit style bearing can not live this long in the same location. That's the average we aim for when determining bearing loads although bearing quality may shorten that from time to time. That's pretty damn good. The question isn't whether rollers are superior or not. There's no question there. They are all things being equal. The question is can we get one to fit in the small confines of the space available to have a high enough safet factor? Maybe is the best I can answer until I actually do more homework. I can get them custom built too along with an FEA stress analysis for verification but that wouldn't be cheap. The FEA would as I could run it myself. A custom made bearing isn't cheap.

Rotarygod, seeing that you have loads of experience with bearings, do you think the e-shaft and s-gears are hard enough to be used as a raceway?

One other thing to consider is what bearing speed can they handle? Since the bearing speed is rather high in a rotory.

I don't see a problem with the e-shaft or stationary gears being hard enough. You can always get them hardened to a different spec if that's a concern. However typically when we harden our eccentric shafts on our pumps (they can weigh up to 20,000 lbs!!!), we do it pre-machining of the eccentric lobes. We try to hold a +- .0015 tolerance over the entire machined surfaces and heat treating them after they are machined would alter this. Our tolerances are this tight because we use rollers. We could get away with less if we used babits.

Bearing speed is important to think about. Jsut because we can physically find a bearing that fits where we need it to doesn't mean it will do what we need it to. Bearing stresses rise exponentially with an rpm increase. There comes a certain point where just a little more speed is a ton more stress. We may find a bearing that fits fine but won't live long above a low rpm limit. Let's say 5000 rpm. It may be fine below this but not above. That wouldn't be too useful. Max bearing speeds that I have seen that may be functional are rated at up to 11,000 rpm. The rpm is fine but the next question is whether or not the stresses at the rpm limits we want exceed those of the bearing. I prefer to have a safet margin in there and I also prefer to calculate it based on continuous duty. Even though we won't be at redline constantly, it would be nice to have a bearing that could do it. There's alot to think about and then you need to calculate it all out.

I don't have the time to figure this all out. I guess I could sit down one night and create a virtual rotary engine in Inventor. I'd have to get the cross section of the rotors correct though but cutting one apart and miking it. That isn't hard. I've got spares. Then after a virtual is created complete with material specs applied to it all, I can use Finite Element Analysis to look into it more. This would take awhile though and I really don't have the time. If I used the FEA software in Inventor it would only tell me stresses in the e-shaft. For bearing stress in the bearings I'd need to use ANSYS but I only have that at work.

It's possible to do but time consuming. I have the ability. Just not the time. The easiest thing to do is to pick the toughest bearing you can find that will fit where you need it to and try it. Pray it works.

I'll try to get some pictures of a couple of our mudpump eccentrics and bearings here so you can get an idea of how really large these things are.

The main idea of this thread, roller bearings for performance, has been really interesting. I hope I'm not becoming obsessive but I find myself thinking about the idea a lot. As other people have stated, I've seen countless small piston engines (most two cycle) that use them at high RPM and no sump or oil pressure.

Here's a thought I want to throw out there.

If one was able to make a full roller rotary, two piece crank with a center bearing, couldn't you also decrease the clearances of the rotor. I'm not just talking side clearance. I got flamed once for suggesting that the rotor face to trochoid was about 1mm. What is it then? Does anyone know? But maybe you could use .25 mm for a face clearance and instead of .010" at the tips for side clearance you could bring it in to maybe .005"? Mainly because it would now be locked in and not flexing. This would really help high RPM operation since the seals become less effective at high RPM.

Of coarse this brings up the fact you would have to make your own rotor. Even so, the thought about this roller bearing thing led me to think you could really tighten it all up. I believe the limiting factor on face clearance would be the clocking of the rotor by the side gear when changing from front-lash to back-lash. You have to have clearance to run. This would prevent an even more precision fit.

I found the relevant needle rollers from Timken. Here's a link to their catalog along with descriptions, dimensions, and load ratings:

http://www.timken.com/products/beari...20Bearings.pdf

this might be a far strech, but why not use a planetary roller system like in the rotrex supercharger. Some radical modifications will need to be done to adapt it in there but might work

I think one of the main probs with the single peace shaft is it flexs to much , can you imagine how all those taper joints and splines would be under stress flex ,they would loosen up and cause other probs. I think just better material in a 1 pc shaft would help. note: yrs back porsche tried some roller brg engines they caused more trouble than all the expense involved. Ron

If a roller bearing is a problem, the real issue is that it was a poorly chosen bearing for the job. It's as simple as that. As I said, I'm not sure how easy it would be to integrate one in properly. This assumes that one can be found that works good and I'm not sure one can that will fit within the small space that we have.