You really need to read up on fatigue and understand the very wide statistical population that results from fatigue failure testing.

It's quite typical for two identical parts to have failure rates that differ from each other by a factor of 3.

A rotating part in a jet engine is designed to survive 60K cycles. The part is removed from service at 20K cycles and destroyed. This provides some understanding of the nature of fatigue and the impossibility of stating a fatigue life with any certainty.

Fatigue life is quite similar to bearing life. One bearing will survive 5K hours and another bearing of the exact same type will survive 25K hours. The statistical population is huge.

When parts in a jet engine fail from fatigue do they bend? Or do they fracture?

Tom W

Jet engine components are usually flat discs or hubs with a very stiff outer structure, and, therefore, bending of such components is highly unlikely. They cannot be compared to the relatively weak construction of a connecting rod with an attempt to draw some type of analogy.

Jim, unless your dealer replaced ALL SIX of the original rods, it's pretty likely the failure 13kmi later was due to one of the other four original rods failing. You did not specify if they did replace all six or not. If they did, and you have a receipt for the six new rods... I'm stumped.

:confused:

Didn't someone else post something about this in which all the rods WERE replaced and it still failed again? I vaguely remember reading something to that effect.

Not that I know of. I can't recally any 3.5L engine which failed after receiving six of the new, updated rods AND also had the cylinder sleeves repaired or replaced. Putting a new rod into an oval bore isn't going to fix the oil consumption. You need round bores, good pistons & rings, AND new/updated rods. Times six.

:stuart:

I understand fatigue, the failure modes, the causes, the statistical analysis of such, the point is that a design shortcoming affects all parts in all engines. A bad design affects 100% of the parts. Again looking at statistics and six rods per engine, you would need to have a fairly high percentage of failed engines to indicate a bad design of the rods.

For a process or other variation to be the problem, any number from 0% to 100% is possible.

If an engine suffers the failure of 2 rods (using Jim Smith's example), they are replaced, and the engine again has failures, this would indicate to me that I am looking in the wrong place for my failure cause or variable, and I should tear into that engine to find out what is different. It could still be the rods, but why two in one engine, or six, was this from a bad heat-treat batch? Bad forgings? Bad material? Maybe it is another variable in the engine.

Engineering as you know is a fairly repeatable process, and every part on the CAD screen is exactly the same when you're finished, within fifteen or more decimal points depending on where you have it set. You can produce a "perfect" part and run it through analysis that can exceed design limits by whatever you wish to find the weak point (don't know if you've ever done FEA but I'm guessing you're familiar with it) and strengthen it or determine that it is well within design limits. The variables come into play in the production process, this is very hard to predict and control (especially these days when everything is out-sourced in the automotive world). By the mid-'80s CAD was the standard for Engineering of such parts.

I feel that it can be a failure of the rod due to inadequate rod strength, but my opinion based on my experience and training tells me that if so, this is due to part production & process controls rather than design. I'm confident that (from M-B track record of diesel engine design) the design was completed properly and to normal M-B standards and practices to perform for the normal M-B diesel engine life and the normal testing was completed also.

Back to the Jim Smith example: Is it also possible that there is an issue with these blocks? Possibly with the crank? How about the piston? Could there be something else that allows the piston to travel in a non-linear path and put abnormal side-loads on the rod? How about the cylinder walls?

I'm not saying that the rods aren't the problem, but there's precious little data to go on so my normal analysis path is to eliminate or table the ones that can be considered less likely and focus on the remaining possibilities first. In the absence of any real data or parts to analyze, to look at the whole assembly, the engine in this case, and see what is unique to the failed engines, or to the failed engines' families (the 3.5 or the 60x in this case). It doesn't seem to manifest itself in the 3.0L, only the 3.5L 603, so what parts are unique to this engine: block? pistons? crank? rods? balancer? ... no theory can be proven without data that we don't have.

Next question is the hydrolock (and rock) theory: If it is a hydrolock, and the early 60x engines had a known leak problem with heads and head gaskets, why don't we see a high rate of bent rods in these engines when we find bad gaskets and bad heads? This might already have been addressed but I haven't seen it.

This is a dead-horse as far as I'm concerned, and I'm disappointed in myself for spending so much time thinking about it, mainly because no conclusions can be drawn without DATA. We have none, can theorize, but we must then accept others' theories as it is all that we have. Discussion, theory, debate, but no data, this is the process that we go through so that we know what data to gather, we have no data.

Jeff,

All the data I've seen points to the original rods being a flawed design (i.e., not strong enough to deal with the additional load from the modified bore/stroke). The updated rods are significantly stronger. And in almost all cases, when the engine is rebuilt (note: rebuilt!) with six new/updated/stronger rods, the failure doesn't re-occur. If the block was the problem (or the crank), the new rods would fail again. 99% of the time, they do not. Also, Mercedes engineers did not re-design any other parts to cure this... just the rods.

For the record, I don't subscribe to any of the theories about carbon, EGR, hydrolock, or anything else... these would affect all other OM60x engines as well, yet they have no problem (ok, at least it's like <1%, if not zero... statistically nil, anyway.) Only the 3.5L has the extended bore/stroke compared to the 2.2L, 2.5L, and 3.0L versions.

:deadhorse:

Perhaps I'm an idiot then, ... but why, if all of the original rods were of a flawed design, does a fairly low number of them fail? All are stressed to the same point, some have gone several-hundred-thousand miles, others fail before 100k. This is where I feel that the rod design is not as likely as the rod. The difference being the production of the rod and variations therein.

Just as some rods are statistically on the low end of the scale, some will be on the high end. Those engines with high mileage could have gotten six rods from the 90 percentile in strength. Thus the design could have been flawed, yet still give some good examples.

Correct. And I agree with you, Jeff, it could be production tolerances / variances on the rods. But it's still the rods at fault - somehow!

:stuart:

>>the point is that a design shortcoming affects all parts in all engines.

Yes, but that doesn't mean that all rods will fail.

If you take a statistical view, you could, with the load along the x axis, plot the distribution of the strength of the rods. You could (hypotheticaly!) also plot the distribution of the loads borne by the rods in each engine.

The ideal situation is that there is no overlap at all between the two distributions. However, real distributions tend to have very long tails, and, depending upon the actual margins, [i.e. the spacing between the some particualr point in the bulk of each distribution] there may be some degree of overlap, i.e., some real possibility of failure.

Specifying the margin is just as much a part of the design as is analysing the chosen load cases, and if the margin is not sufficient, you will get a situation where some rods fail, some fail early, some fail late, and some never fail. The difficulty in setting a margin is that you have to deal with uncertainty - by definition, you don't know how much margin to allocate to this cause.

It is true, that there's not enough available data to say anything with certainty. There are things which might be ruled out though. However, what's certainty got to do with it? - this is an internet forum, where we express views and opinions, we aren't writing a textbook!

I can't really believe the fatigue viewpoint, because a con rod with a sufficient enough crack to promote bending and buckling will not long at all in a running engine - you would typically find most rods would break and cause lots of damage rather than just bending.

I agree that there probably would be other evidence of gasket failure, a history of coolant loss, evidence of steam cleaned piston tops, etc, etc.

In answer to Tom, yes, I'm an automotive engineer by training. Here's shrewd!, I qualified just as Rover was sold to the Germans, then RR, then Bentley, and now Jag and Land Rover have been sold to Tata - we don't really have a car industry anymore in the UK!. I've worked in other fields, from submarines, trains, planes (A380 recently), and I'm currently working on the design, analysis and testing of an instrument for an ESA mission. I don't have any special MB experience at all, and as such, my views don't count for much.

My question about a plain design flaw, such as not being strong enough to deal with the actual operating condition, is why does it take a hundred thousand to three hundred thousand km of operation to cause a failure. This is what leads me to the fatigue scenario. Bending, if the engine works for a hundred thousand km, is not happening, at least not in the plastic deformation range from day one. Something is happening that changes the rod's mechanical properties after an accumulation of many millions of cycles so that what was an elastic event, suddenly becomes a plastic event. Once that change occurs the engine is toast.

I do believe the key to the failure is lack of margin to the actual operating conditions. And, the added bore and stroke likely do not add enough space to really address this issue as one might wish. It is a design that has been hemmed in by the existing design it was based upon.

So, unless the operating conditions are becoming more stressful as the odometer ticks past 100,000 km, I am having difficulty understanding how a connecting rod is failing by a mechanism other than a fatigue related one at that later stage in life. Unless you are saying the rod bent in the first few revolutions of the engine with the throttle open, and it just takes that long for the cylinder to get oval shaped and start burning oil. That would be a new one. Jim

The people who are in a position to have actual data are not speaking of it....the factory and the dealers.

I would like to see an uprated rod next to an original design rod.

Thanks to all who have posted with ideas.

Tom W

Someone posted a photo of at least the new rod, if not both, on some thread buried in this forum. Or somewhere on the web. I recall reading about it... but can't seem to find it, d'oh!

:stuart: