Rolling along

Thursday, November 10, 2011

Tags :  bearings

bearingsIn the search for ever lower engine friction, it is surely a wonder that the rolling element bearing hasn't featured very highly in recent years. So while ball or roller bearing technology is commonly seen in many engine ancillaries - for example, pumps, starter motors/generators, timing belt tensioners, rocker arms and now even turbochargers - the largest source of rotation friction, that of the crank and camshaft, have largely been avoided. I know two-stroke engines have, and continue to use, needle roller and ball assemblies throughout - principally, I would guess, because the application makes it a necessity (as in the lube requirements of the little end) and technically it is not that challenging - but in the world of high-performance four-strokes, rolling-contact bearings have never in recent years been truly favoured.

If you go back in history, ball or roller technology was a common sight. Before World War I, the leader in four-valve overhead cam engine design, the Peugeot L3, used ball races on the three main bearings of this four-cylinder unit. The big-end bearings were reserved for plain white-metal bearings but by this action the design intention, I think, was made clear.

Later on, in the mid-1950s and even when 'modern' thinwall steel-backed shell bearings were readily available, Mercedes-Benz still committed to a system of roller bearings on all main and big-ends. Here the multi-piece crankshafts necessary to take the roller system were assembled together using the Hirth process - a system of radial serrated journals clamped together axially by large screws. Even in 1970, legendary engine designer Mauro Forghieri is reported to have preferred roller bearings on the four main bearings of the 312B but eventually settled for just two, one at either end of the one-piece crankshaft.

Clearly the issue is the crankshaft and the strength of multi-piece units to enable assembly with the traditional roller bearing. In theory at least, the rolling point contact of the roller bearing over the sliding technology of a more traditional approach should give a significant advantage, and is perhaps why bearing manufacturers are now beginning to look at the concept again. The major difference over earlier designs is the development of split roller technology, which can be readily assembled around traditional one-piece forged or machined-from-solid crankshafts.


Analysis of friction within the traditional internal combustion suggests that around 20% comes from the main and a further 10% from the remaining big-end bearings. This comes from the drag induced in the oil film within the bearings and the high oil pressure needed to generate it. Rolling contact bearings, by their very nature, have considerably less friction, and since they do not require anywhere near as much lubricating oil they should therefore have a significant advantage.

It has been calculated that the friction of a roller bearing is around 10% of that of a traditional bearing at about 3000 rpm. If this holds true on the test bed then the FMEP (Friction Mean Effective Pressure) due to the bearings of an engine of 0.4 bar on an engine at 6000 rpm could be reduced to much nearer 0.1 bar or less - a small but nevertheless useful increase of around 2% in net power output.

In practice, however, no such advantages have been observed yet. Even with significantly reduced oil flow, the action friction measured is equivalent only to that of the best designed shell systems, which is particularly surprising. Clearly, modern shell bearings are much more efficient than many of us are aware and, after factoring in the extremely high levels of durability now expected, these apparently simple devices are highly underrated.

Fig. 1 - Typical Split bearing cage

Written by John Coxon

Leave your comment



I would disagree that the relative mechanical losses of a rolling element bearing are only 10% that of an equivalent journal bearing.  You must consider several factors when comparing the two.

Firstly, you must consider the relative load capacity versus journal size between the two bearings.  While both rolling element bearings and journal bearings ideally operate with hydrodynamic contacts, the journal bearing will have a much smaller journal diameter than a rolling element bearing for an equivalent load/fatigue life capability.

Secondly, the losses in a rolling element bearing are due to friction and viscosity.  The viscous losses increase at roughly the cube of the bearing's DN value.  

Finally, the oil flow requirements for either journal bearings or rolling element bearings are based on heat rejection.  Either type of bearing must have sufficient oil mass flow to ensure that the journal and bearing materials are kept within their allowable limits.

Rolling element bearings work OK for crank mains that have a fairly constant rotational velocity.  But for conrods, rolling element bearings are problematic, due to the varying rotational velocity and direction of load.

As you say, journal bearings are highly underrated.

You say that rolling element bearings are problematic when used in big ends.  How then do you explain the sucess of roller bearings used in the con rods of two strokes, some running at more than double the RPM of a four stroke using tradional shell bearings?
You both have valid points, however I think you have over looked one of the main reasons we don't see very many roller crank configurations in mass produced engines, cost! I manufacture crankshafts & have been involved in producing one piece billet two stroke V-6 crankshafts. Controlling distortion from the carburizing/quench/ temper process requires unique tooling, remember the journal hardness must exceed 60HRC. As for the bearings themselves, they are comparatively heavy & expensive to produce. Due to there physical dimensions the two piece bearings require larger con rod big end I.D & main bearing bores.

The use of rolling element bearings in N/A loop-scavenged two strokes is due to necessity, and not preference.  With a crankcase scavenged two stroke engine, there is not the possibility to have a recirculating oil lube system and journal bearings.  So rolling element bearings with combined air cooling and total loss oil mist lubrication is the only practical option.

You also need to consider the fact that two stroke cycle engines do not produce reversing dynamic loads on the conrod bearing like a four stroke does.  Reversing loads are not much of a problem for journal bearings.  But reversing loads are a serious problem for the particular type of rolling element bearings used for conrods.  These rolling element bearings must have some small amount of installed radial operating clearance.  And if there are any cyclic reversing loads acting on the bearing, it will quickly fail due to dynamic impacts between the rollers and races as the radial clearance is taken out.


I design gearboxes, and I can appreciate the issues involved with quench distortions of carburized steel parts.  You might want to take a look at some of the recent developments in carburizing steel alloys.  Questek/Latrobe have some proprietary carburizing alloy steels (C64/C69) that are designed for nitrogen gas quenching.  The gas quenching results in much less quench distortion.