Ball bearing eccentric

[DetroiTug]

Bart,

Since you're not really following traditional steam engine design. For a crosshead I would look at maybe two hardened thompson shaft that parallel the piston rod on each side, Then a block with two linear sealed ball bushings on each end and the piston rod attached in the center. Clevis on the bottom for the connecting rod. If the shafting is heavy enough, it would not even need to be supported on the bottom.

-Ron

[Lopez Mike]

I've used those Thompson shafts in some lab equipment where the loads were fairly rough and intermittent (a giant microtome) with good service. They do have the advantage that they can be adjusted to zero clearance and have grease fittings.

The down side was that they were not cheap. But again the bill was being paid by an N.S.F. grant so it wasn't a major problem.

[barts]

This is pretty rugged service for linear bearings;20 G acceleration, and a peak piston speed of 3.8 meters/sec; the simulation program says the engine will do near 500 rpm at 10% cut-off. Whether that's accurate we'll see, but it seems prudent to design for that.

I'm leaning toward v-groove roller bearings running on 90 degree track; these sealed bearings would ride on either side of the forked con-rod arms. The con-rod itself uses greased needle bearings.

I'm working on a packing gland's eye view of the cross head...

- Bart

[Lopez Mike]

Those numbers are moderate compared to I.C. stuff.

I would look into the linear bearings anyway. You might be surprised.

[fredrosse]

"This is pretty rugged service for linear bearings;20 G acceleration, ........."

I would think the quasi-steady side thrust load from the connecting rod (based on maximum steam pressure on the piston at somewhere near 50% stroke) is far greater than the inertial part of this condition. The only time I have seen the inertial loads becoming significant in reciprocating engine work is when several thousand RPM is the operating regime.

[DetroiTug]

"peak piston speed of 3.8 meters/sec"

That is a consideration. The ball bushings have internal circulating passages that transfer the balls from one end to the other to traverse across the bearing surface. This requires the balls at the end of traverse to essentially make a U-turn in to the re-circulation tubes - quite different than a standard ball bearing. My experience with ball bushings have been slower speed higher load applications. They should have a specified surface speed rating like a ball bearing has an RPM rating which is essentially a surface speed rating. For an engine running 500 RPM, I would definitely give them a try.

I like the side slung rotating cantilevered valve design used in some of the German marine engines, instead of the Stephenson links which make the engine much longer. They typically have a complex casting with complex linear slides, I thought that would be a good place for ball bushing also. Much easier to make up, although not traditional.

V-wheels: Yes they are much better for high speed. The drawback is they run any foreign particulate over instead of pushing it away as the wipers on a sealed ball bushing and conventional crosshead do.

-Ron

[barts]

Well, I fiddled around with numbers today for the cross head while waiting for some builds to finish.... and cross-heads are a tough service. I checked the peak speed, which is ~750 ft/minute, and the 20G acceleration, and the peak load against the bearings could be 1000 lbs if the cut-off is long. hmmm.

That acceleration is too much for the rolling guide wheels for some reason.

Interestingly enough, it turns out good old sae 660 bearing bronze can handle this - just makes the peak speed of 750 ft/min, and if I use enough surface area we can handle the load. But perhaps plotting the side force on the cross head for a more realistic cut-off is useful... so I wrote a little program tonight to compute the cylinder pressure and cross head load, and plotted the result. You can see the two plots in the attached graph; the X axis is in degrees after TDC, and I've not added in compression load on the other side of the piston.

Next, I need to compute the velocity per degree so that I can plot PV vs crankshaft angle; since the velocity is going to be at a maximum at about 90 degrees, we can see that the load then is much lower, so the bearing can be a lot smaller than otherwise would be required if the cut-off could be made longer.

Still checking the numbers, but I found this interesting enough to share.
- Bart

[Lopez Mike]

I love all this stuff that we couldn't dream of calculating just a few years ago.

That said, try to remember how many estimates go in to this. I'd bet that the real world numbers could easily be off by 50 to 100 percent. We might hope that they will be in a good direction!

The one that I suspect would be most resistant to accurate simulation would be volumetric efficiency. Predicting port flow is a bugger.

Funny how plain bearing surfaces still have their strong points. As far as I know, the thrust bearings in the largest ships are still plain bearings immersed in lubricant and shaped to form a hydrodynamic wedge.

And all large telescopes use plain bearings with pumped lubricants to eliminate static friction. The Keck telescopes in Hawaii use plain bearings and with velocities in the vicinity of a few feet per hour, stiction is a serious issue. The moving mass is around 5000 tons and can be moved slowly by one person leaning against it.

[Dhutch]

If you can keep them well oiled, you're fine, but absent a little dip pan underneath, they tend to run dry from time to time.

Fair enough. We have a mechanical lubricator which oils everything which needs more than a daily oil, including the four eccentrics so have yet to have an issue with then running dry.


Daniel

[barts]

I love all this stuff that we couldn't dream of calculating just a few years ago.

That said, try to remember how many estimates go in to this. I'd bet that the real world numbers could easily be off by 50 to 100 percent. We might hope that they will be in a good direction!

The one that I suspect would be most resistant to accurate simulation would be volumetric efficiency. Predicting port flow is a bugger.

The good part here is that for these purposes we just assume 0 port losses, and we end up w/ a bigger safety factor
Estimating fuel consumption is another matter entirely, of course. Back to cross-heads...

If I go with plain bearings, I could put two round hardened shafts on the loaded side. The best approach might be a aluminum crosshead block that has seals and a oil/grease reservoir for all four bronze bearings. Visually I like this better than the roller cross head; it leaves the engine more open and it would make the piston rod seal easier to reach. The tricky bit would be the seals, though.

Of course, I could just chuck the idea of a sliding cross-head and go really old-school: a Watt's linkage. W/ sealed greased needle bearings all around, it would certainly be nice and clean, and would help satisfy my love of 'monkey motion'. This would raise the engine a bit, as apparently the optimum size of the center link is about 2/3 of the stroke, or 4 inches. It also does raise the amount of reciprocating weight. I've never seen an engine w a Watts linkage running at several hundred rpm .

Time to go cut some metal and see what other ideas appear.


- Bart