Design of durability tests for transmissions 1

[pietro82]

Hi all,

I'm trying to design a durability test for a stepped transmission of an off-road vehicle. The maximum engine speed is 3000rpm. Our customer provided some field data collected for years and he asked to define a method to design a durability test for the transmission in function of customer usage. For the test, I'm considering to run the engine at the engine speed where the max torque occurs. Thus the max acceleration factor will be achieved. Moreover, considering the engine speed range is rather narrow, so the effect of the speed variation on the gear tooth damage (according to Miner) is very limited. May any effect be missed in running the test at a single engine speed?

Thanks,

Regards,

 

[BrianPetersen]

You asked a yes or no question, so I'll answer: Yes.

 

[GregLocock]

Slightly more helpfully, there may be speeds/loads/temperatures at which the oil film breaks down and so you get rapid wear. If your test does not inlcude that condition you'll miss it. Same with structural resonances etc.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[BrianPetersen]

A significant wear factor involves gear-changes, and those don't happen at constant speed or even the same speed. A constant-speed test will miss that. If your test involves constant speed and torque, it won't even capture torque reversals (e.g. from going downhill - or simply the driver lifting off the accelerator and re-applying).

 

[pietro82]

Thank you guy for your replies. We supposed to start the test when the engine temperature reached the target value calculated from the measured carried out from the customer. The target temperature is the 90th percentile of the relative frequency of transmission temperature. Regarding gear-changes, they are included in the test, because the test is composed of cycles reproduced for 30 times till the test duration is approached. Torque reversals are not popular in this type of machine because they mostly work at a constant speed, so the engine load or the selected gear ratio mostly change but of course it may occur sometimes.
How may I find the speeds/loads/temperatures where the oil film breaks down? Is the best testing the transmission through a driving cycle which includes accelerations, decelerations, steady conditions and so on and reproducing it till the same amount of damage for all failure modes is reproduced? Moreover, how many different tests are carried out in transmissions? Durability test, efficiency test, dynamic tests (?) ? Our customer has just started doing some tests in their transmissions.

 

[romke]

i know some car manufacturers do approach it somewhat differently. based on the foreseen use of the vehicle, they start out calculating how many hours a specific gear is used at full power representing the normal use under usually much lower load conditions. they thus arrive at the number of hours or km that each gear has to last under full power and then run test accordingly. that may result in a "needed life" for the reverse gear of a car where the total transmission needs to last for 300.000 km of only 15 minutes (because nobody will drive the car backwards at full power) and also relative short lived first and second gears, whereas higher gears may have to last much much longer.

the problem with this approach that you need to have a clear view of how the vehicle will be used - if actual use differs considerably the calculated outcome will not be very useful...for cars it usually is rather well known what the actual use will be based on known driving patterns, traffic conditions etc. for more utilitarian vehicles actual use may well vary a lot more over different customers and the method described may be much less suitable.

running tests at a specified speed need not be a problem, assuming that lubrication conditions as such are such that more or less elasto-hydrodynamic or hydrodynamic lubrication can be achieved. when the circumferential speed of the gears is too low, that condition may not be achieved and thus scoring might well be occur. the lubrication regime will depend om the circumferential speed - gears used at high rpm are easier to lubricate then slow turning ones.

results may also be heavily influenced based on type and viscosity of the gear oil used. EP additives may prevent fatal scoring while overdosing may lead to more rapid material loss, less viscous oils will be better for heat transfer but may offer less protection against wear, synthetic lubes may longer be able to withstand higher temperatures but may have detrimental effects on seals if those are not compatible with the fluid used.

all in all it will be a quite complicated task to get it right and even so, because the actual use will vary in practice, it still is a good idea to beef up gear sizing somewhat to accommodate unforeseeable operating conditions.

 

[GregLocock]

"How may I find the speeds/loads/temperatures where the oil film breaks down?"

Frankly, you can't unless you have a great deal of experience with that unit. You are only likely to find that sort of effect with a detailed investigation after a gearbx has blown up in service.

"Is the best testing the transmission through a driving cycle which includes accelerations, decelerations, steady conditions and so on and reproducing it till the same amount of damage for all failure modes is reproduced? "

Ideally yes.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[pietro82]

@Greg: Miner's law is a popular tool for the calculation of gear teeth (pitting and bending) and bearing damage. With this law, torque, and speed of gear wheels are the only parameters which affect the damage. So a constant high torque is more damaging than a variable one from 20% up to 100% of engine load. So damage cannot be a right descriptor/index to evaluate which driving cycle in the most severe for the transmission. So how could be ranked driving cycles according to their severity?

 

[pietro82]

@romke: The approach we are using is right the one you depicted and with the same results you wrote (i.e. short test for the rearward gear). My customer made many measurements just for understanding how his vehicles are used.
You spoke about driving patterns, what are for you? Just speed and accelerations profile or a more detailed "thing" ?(I'm sorry I'm not an English native speaker)

 

[GregLocock]

Well I'm afraid miner's law is being misapplied in this instance. If it worked we'd only durabilty test engines at one speed and load.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[pietro82]

@Greg: It's clear that is misapplied but Miner's method is widely adopted by engineers and it is also included in the ISO 6336-6. BTW, the problem is getting as complex as I supposed in the beginning where I should consider a series of maneuver/driving events. I spotted few that are recurrent for type of vehicle of my study, but I wasn't able to identify the most severe one because I have the speed, the torque and the engine speed profiles (I'm not even considering the case where gear shiftings occur) so the problem is a multivariate one. So what should I do?

 

[gearcutter]

ANSI/AGMA 6002-B93 Design Guide for Vehicle Spur & Helical Gears says this:

Designing gearing for vehicle use is unique in that the loading is not constant but highly variable. Furthermore, for reasons of space, weight and economy, it is usually not practical to design gears for maximum life based upon maximum gear load.
It is recommended that the cumulative fatigue damage criteria proposed by Miner be employed to ascertain the effects of variable loading on the life of the gearing.

It is recommended that overload factors be omitted from the standard stress calculation and applied to the duty cycle. The overloads and their duration are more correctly accounted for by modifying the duty cycle itself.

Because of the difference in the stress cycle curves for pitting and bending, the effective load for pitting and bending will be different. Thus, it is difficult to predict the mode of failure without the complete analysis of the duty cycle as indicated. Therefore, if the shape or magnitude of the duty cycle changes, the mode of failure may also change and this may influence the gear geometry design.

 

[GregLocock]

"it is difficult to predict the mode of failure without the complete analysis of the duty cycle as indicated"

Exactly. If you don't know what is going to fail you can't assume its stresses are proportional to load on the whole system so you can't use Miner's rule derived from the system load on the critical component. I repeat, that is why the durability test for an engine (in particular) cannot just be run at one speed and load and then just use Miner's rule.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[pietro82]

@Gearcutter: Thanks, I didn't know agma standard, in EU ISO 6336 is much more popular. I'll check out that standard.

 

[gearcutter]

You'll find that the AGMA load rating standards are more conservative than ISO.

I have little experience with vehicle transmissions; but of what I have had to do with them involved a prototyping phase that I was called into.
For various reasons; the major issue confronting the team was gear misalignment. The gear design was quite sound, but didn't take into account what components like the housing, shafts & bearings did under various load scenarios and how this affected what was happening at the mesh.

Machining of the housing proved to be far more difficult than first thought. Geometric tolerancing became a critical issue.

I also remember that there was a large difference between how the components performed when on the test-bench/dyno as opposed to when bolted to a chassis/dyno. The chassis seemed to offer far less support for the housing than the support that the test bench offered.

 

[romke]

@pietro82

"driving patterns" to me generally mean speed and load. usually car manufacturers try to find out which combination a vehicle may encounter during it's lifetime and then try to design a transmission that can allow that. it certainly will not work for everyone: people living in rural areas have less to cope with traffic jams and also may never drive on a motorway at high speed, when living in the mountain areas people will use lower gears quite a bit more and at more heavy loads etc. it thus is a statistics based approach that can be useful if most vehicles adhere to the calculated "average" - and some will encounter problems due to a quite different use.

a striking example where the Porsche 911 cars that were used by the dutch police in the sixties and seventies for highway patrol. because in those years they drove a lot backwards at high speed over the hard shoulder to warn upcoming traffic for slow and stationary traffic the reverse gear usually lasted about a year. the cars were clearly not designed for this type of use and failed completely. the move to another car maker did not solve the problem though, the cars that were used to replace the expensive 911's failed in the same way and in fact a lot more often then the more. as soon as the highways got better stationary signalization to warn for slow speed and stationary traffic jams and the need to drive backwards disappeared the problems were gone

the basic thought behind this approach to design automotive transmissions is the need for low weight - a restriction that applies less to industrial gears that are stationary. if you compare the physical size of a automotive gearbox with a industrial gear set that has to transmit the same power and torque, the difference is immediately obvious.

 

[pietro82]

@Gear: I'll check out, I'm searching the standard. In our case the test will be carried out on a full vehicle chassis (engine+transmission+axles), so I hope there is not the effect you mentioned

@romke: What we want to accomplish is exactly what you said. I still have not a clear idea on how considering all possible combinations of driving patterns. As you said a driver in rural areas use higher gear ratios for longer than a driver located in urban areas.

 

[IceStationZebra]

"Because of the difference in the stress cycle curves for pitting and bending, the effective load for pitting and bending will be different. Thus, it is difficult to predict the mode of failure without the complete analysis of the duty cycle as indicated. Therefore, if the shape or magnitude of the duty cycle changes, the mode of failure may also change and this may influence the gear geometry design."

Exactly what I was going to say, just more elegant. For a 1200hp gearbox test I conducted we knew the normal failure mode of the previous gearing was bending so we were careful to push the power level as high as we could without inducing pitting/lubrication failures. Another thing to consider is if the gearcase is a structural member - the deflections make a difference. ISZ

 

[dicer]

So what is the size of the vehicle?
What kind of loads are on the output shaft of the transmission?
Is it a large earth moving machine or a race car?
There are many transmissions that have been designed for most all applications.
Why reinvent the wheel?