The conclusion we draw from these and other experiences is that the sideshaft is either too strong, in that it doesn't flex enough, or too weak, in that it can't recover from a twist. There have been various approaches to this problem as far as we know, some have been to forge the shafts thicker in the middle (which then shear at the splines on either end), but mostly the accepted answer is to use Salisbury diffs (much, much heavier duty), Dana diffs, or as in the newer models, multi-spline diffs. To our mind this simply serves to transfer the problem at great expense to other elements in the drive train (to save relatively cheap sideshafts we now spin expensive gearboxes to death).

The approach my neighbour, Roy Maritz, has taken is to machine up - he's a precision engineer by the way - a set of sideshafts out of a good piece of stock (flexible but strong), which he intends machining flat on the sides to enable the shafts to accept a radial twist and then spring back to their former (or as close as possible) shape. The idea is to keep the cross sectional area of the 'new' sideshafts the same as the cross sectional area of the standard shafts.

Please note that although Roy has a fairly comprehensive machine shop at his disposal, we do not have access to computer modelling to test our theories. We rely solely on "common" sense, experience and somewhat educated guesswork. We are not doing this for profit or any financial consideration whatsoever. We are sharing this in the hope that we, and others, will benefit by being able to use sideshafts that don't let us down in embarrasing circumstances.

We have conducted a scaled experiment (using wooden dowels) to test the inherent structural strength of the different shapes (see photo on left - pardon the abysmal quality). What emerged was that the flatened dowel - on top in the photo - (with the same cross sectional area was able to accept a 270 degree twist before the material started giving, even then the cracks that appeared were longitudinal and did not affect the ability of the shaft to accept radial force. The round dowel (turned from the same stock and of the same cross sectional area) started splitting in the middle at about 180 degrees and twisted with hardly any resistance through to 360 degrees where it came apart entirely. You can see the effect on the bottom dowel in the photo.

This seems to bear out the theory. If you have something to contirbute to this experiment or the debate in general, please email me and I'll publish your comments on this page.

Lets hear from you all....
John Ousterhout wrote: I did a simple calculation on your flat axle idea, and had a comment about your experiment:

Assuming the max distance across the flats of 1.25 times the round shaft diameter (as someplace to start and I think it'll still fit into the housing), the flat axle will experience 1.6 times greater peak stresses than a round shaft of equal cross-sectional area. The region of max stress will be centered on the widest part of the flat, and will tend to shear through the web, leaving two rectangular sections. The advantage (if one can call it that) is that this will move the max-stress region away from the splines, into the shaft. Unfortunately, it will break at lower torques than the round one. It will also twist much further before doing so. I'll recalculate using a square shaft, to determine how big it would have to be to equal the strength of the round shaft. Regarding the experiment: Good thinking. Did you measure the forces (torques) required? If theory holds true, the flatted stick would have begun breaking by splitting along its centerline, allowing the remains to twist much further. The onset of splitting should have been about 1/1.6=0.62 times the torque of the round shaft. I'd like to have the dimensions you're planning to use to provide a better analysis. The math isn't intuitively obvious, but a description might be. JohnO

Hi John,
Thanks for your thoughts and calcs, that's the reason we wanted to open up this debate. Roy can machine up a couple of scaled down examples which we can then use a torque wrench on to test to destruction, but it'll make our job easier if someone can warn us off wasting time.
You're right about the splitting of the flattened dowel but our thoughts are that because both ends are captured (one in the hub and the other in the diff centre piece) this may inhibit the tendency for splitting somewhat.
John writes:

A quick calculation shows that a square cross section will need to be just slightly larger across the flats than the diameter of a round section to have the same maximum torsional strength. The additional section area doesn't add to the strength, but does take up some room, making the diagonal distance across the corners 1.5 times larger than the comparitive diameter. This means that, to get the same torque handling capability as a 1-inch diameter round shaft will require a 1.07inch square shaft that will measure 1.5 inches corner-to-corner. OTOH, square, super high strength stock, might be reasonably priced. The transition from round to square becomes critical, but not impossible. Don't give-up yet. You're developing a shock-resistant and limber (but possibly weaker) axle.

JohnO

Paul writes:

You mentioned before that the shaft would begin flexing sooner due to the flats, isn't this just what we're after? It seems to me that the longer the shaft resists flexing the more likely it is to twist off at the splines (the logical shearpin). The challenge, as I see it, is to find an optimally flexible and yet strong combination (of course we also have to make it such that it will also shear predictably enough so that we don't just move the shearpin effect to the long sideshaft!).

John writes:

Um, you're close, but the amount of force (torque) needed to twist the shaft may not be enough to move the vehicle like we want. Take an extreme case for illustration: connect the wheel hub to the differential with a thick steel cable. It should have the same total cross sectional area as the original shaft, but has to twist (wind-up) much further to get the same torque to the wheel. It is certainly limber enough to resist shock loads, but will still break because it is too weak in torsion. First the outer strands, which would carry most of the load, will break, leaving the (fewer) internal strands to carry the same load, then they'll break, leaving, eventually a single strand in the middle. This is almost a duplication of what happens with the solid axle. I think you're intuitively on the right track: trying to move the region of maximum stress away from the splines. The Rover axle is actually designed to do just that, which is why it is thicker at the ends than in the middle. The obvious problem is that it isn't quite thick enough and there's still too much bending going on right at teh end of the spines, so that's where the cracking starts. If it WAS thick enough, then we'd break axles in the middle, but (hopefully) less often. Making the same geometry out of tougher material is another alternative, but if it is still bending at the ends of the splines, the splines will suffer. I'd be interested in knowing how far an axle actually twists in use. Perhaps I should try to calculate it?

JohnO

Garret Scott writes:

Paul,

I too have been on the holy quest for unbreakable half-shafts. Since 1980 I have broken about 12 half shafts on Rover diff equiped series vehicles. I also have repaired many failed shafts on other vehicles. About 10 years ago I realized there had to be a better way. My father, being a machinist at a local nuclear weapons factory, consulted several metallurgists, and determined that if we made the shafts out of tougher material, they would not break.

So we made four shafts out of 4340VA (Aircraft Grade, X-rayed etc.) alloy steel. Had them individually and carefully heat treated to two different hardnesses.

They did last longer than the "standard" Rover supplied shafts, but they also did break.

After much research into this problem and two college courses in metallurgy, I have come to the conclusion it really doesn't matter from what kind of material, or what shape they are made.

I believe there are several fundemantal design problems which must be overcome. Most of these were realized and implemented by automobile manufacturers years ago. (about World War-II I believe)

THE BASIC PROBLEM: Mechanical contrivances (shafts) made from a metal will strain harden, become brittle, and break. This seems to become a big problem as the level of strain approaches the elastic yield point of the metal. So, if you can reduce the strain well below this point, it will likely never get brittle and break. I believe this is why most modern automobiles will NEVER experience a failed half shaft. Unfortunatly, vehicles which are used off-road, to tow, climb hills, or other rough activities tend to expose the shafts to much greater and varied stresses.

1. The Flexible Shaft.

A shaft flexible enough not to break, but flex instead, will do just that. It will flex, absorbing energy, maybe plastically deform, get out of balance, and probably introduce complex oscillations within the drive train due to flexing and unflexing. In a suspension system (such as torsion bars) these problems may be easily delt with by dampers (shock absobers), bushings, adjustments etc. But in a drive train, I don't think it will work well. I think you will find that actually the stiffer the shaft, the less it will deform, the less it will strain, and the better it will perform. Of the 12 half shafts I have personally broken, only 2 were short ones. This may validate the theory that less flexing is better, but I doubt it. 12 is really not a good enough number for sound scientific evidence.

2. Better Splines.

Careful analysis of failed shafts will usually indicate the failure occured at the point of maximum stress. This is usually the point at which the splines end and the solid shaft begins. The energy from the differential side gear is transferred to the shaft splines by directly acting on the side of the splines. The spline then transfers the energy to the body of the shaft. The spline will actually elastically deform a small amount each time force is applied. This force is then concentrated at the base (or root) of the spline as it is transferred to the body of the shaft. Like most metals, strain hardening and embrittlement will occur with repeated flexing of the spline root joint.

2A. Number of Splines.

If you have 10 splines, the torsional force will be transfered in a relativily even manner over the 10 splines. So, for example if you a transfering 200 lb-ft of force to the shaft, that equals 20 lb-ft per spline. Now, if you had 24 splines, the force concentrated at the critical base of the spline would be only 8 lb-ft per spline. A reduction of 60 percent or an increase in strength of 110 percent!

2B. Quality of Splines.

It is also known that a sharply cut joint at the base of the spline will concentrate the force at that point and enhance the strain-hardening effects. Unfortunatly, when splines are cut with machine tools, this sharp joint at the base of the spline often results. A far superior method is forging. Forging the splines allows the grain of the steel to be oriented in a much stronger manner, and also may provides nicely rounded (filleted) spline roots. Forging can increase the effective strength of the spline many times (100 or more percent). Note that Diesel engines always use forged crankshafts where Petrol engines rarely do.

3. Diameter of Shaft.

There are two basic concerns with the diameter of the shaft. One is the basic shear strength of the shaft. The other is the reduction of force applied to the splines (Moment or Torsion) as the shaft diameter increases.

3A. Shear Strength of shaft.

For example: a 1.11" diameter shaft (Rover Shaft) gives .97 square inches of cross-section. A 1.31" diameter shaft (DANA 60) gives 1.34 square inches of cross section. An increase of 34 percent for the DANA shaft.

3B. Force at Splines

The force applied at the outer diameter of the shaft (where the splines are) with 200 lb-ft would be would be 2,162 lbs for a 1.11" Rover shaft. For the DANA-60 shaft of 1.31" diameter, the force would be 1,832 lbs. This would give an apparent increase in strength of 18 percent for the DANA shaft.

The moral of the story is that lots of forged splines will increase the strength of the shaft many times. However increasing the diameter, the strength of material, the tempering process, or the shape will likely have much less of an effect on the overall strength of the shaft. If you have the original 10 spline Rover axles, I highly recommend replacing them with 24-spline 1.25" rover axles. You will have to replace the side gears in the differential (a standard Rover part) with 24-spline gears, and the drive flanges with 24-spline flanges (another standard Rover part). But the 24-spline shafts themselves, Rover doesn't make for a series vehicle. But KAM Dffferentials or England, or McNamara of Australia does. I think both of these companies will sell a kit of all parts neccessary. The cost really isn't too bad.

As far as a stronger axle causing damage to the rest of the drive train - pure nonsense!

Good Luck with your project and keep those (series) Rovers running!