Housing deflection limits in design 1

[geesamand]

I've been reviewing the deflection of some older reducer housing designs for a design refresh. They have a pretty good track record for reliability, but based on the FEA and conservative load assumptions, the deflection of the housing is larger than expected. Our load assumptions are conservative in that the 1.0 service factor is never used (always >1.5) and also the external loads peak infrequently but we apply them equally and in the worst direction.

Looking through AGMA 6001, they recommend housing design deflections be limited to keep bearing misalignments within recommended ranges, and to confirm gearbox housing stresses are acceptable. But no mention of checking housing deflection w.r.t. misalignment at the various gear meshes. I.e. maximum in-plane misalignment and maximum out-of-plane misalignment.

1) Do your housing designs consider deflection at the gear mesh? What limits do you use?
2) Are these limits based on loads including peak external load in the worst possible direction?
3) We're using a lot of tapered roller bearings, and with the thrust induced by the radial load there is deflection that drives more clearance into the bearings. Has anyone seen a limit for that?

Thanks,

David

 

[geesamand]

I've answered my questions in the case of bevel gears, where recommended deflection limits are per AGMA 2005 section 15.

Helicals are still an open question though.

Thanks,

David

 

[tbuelna]

geesamand-

If you are doing a thorough design of a gear system, you must consider how structural deflections of housings, bearings and gearshafts affect the mesh contact conditions. You must consider both gearshaft bending and torsional deflections. The gearshaft bending and housing displacements are often addressed with crowning of gear and bearing roller faces, while torsional deflections in the gear/spline teeth are addressed with lead correction.

As for "acceptable" misalignment at the gear mesh, if you have oil-lubed gears designed for operating under hydrodynamic contact conditions, the amount of misalignment at the mesh contact that can be tolerated is actually quite small. You must consider that the thickness of the hydrodynamic oil film separating the gear flank contacts is usually less than .0001".

Regards,
Terry

 

[geesamand]

Thanks Terry.

I've been reading into my company's gear calculation methods. It does not currently handle housing deflection within the geartrain analysis, but it does account for the other factors. We then build the housing to support the remaining deflection budget after the fact. I can see how the total misalignment may be considered as a sum of sustained dynamic and static misalignments, and holding separate limits for each is not optimal. (But then, there are reports such as ISO 10064-3 / DIN 3964 that recommend housing machining accuracy alone.)

Do you know of any standards that cover the total misalignment calculation method? Without it, it seems I'm stuck trusting our calculation software to proscribe reality for us.

David

 

[gearcutter]

In my experience; it would be most unwise to totally trust FEA models when it comes to calculating the type and amount of tooth modification required to compensate for deflections.....especially if tapered roller bearings are being used on the parallel shafts.


Ron Volmershausen
Brunkerville Engineering
Newcastle Australia

http://www.aussieweb.com.au/email.aspx?id=1194181

 

[geesamand]

Well these housings have a good record of reliability and use tapered roller bearings, but I'd really like to know what you feel FEA models lack, and why tapered rollers make the risk worse?

Thanks,

David

 

[gearcutter]

FEA models can be a good place to start but you can't beat real-life tests.
I've lost count of the number of times where information gathered from FEA has not married up with what's actually occurring.
Perhaps the FEA technology in this country (Australia) is not up to speed with what the rest of the world is doing.
Load testing gearboxes and then performing a simple contact analysis is by far the quickest way of sorting out load distribution issues.

Around 3 years ago a company in Italy that I was asked to help was having this very same issue. They relied too heavily on computer models and it was costing them a great deal of money.

90% of the bearing-related gearbox failures that I've investigated have been as a result of issues with tapered roller bearings.
The single largest issue is the gradual lack of pre-load/running clearances through out the life of the box.

Ron Volmershausen
Brunkerville Engineering
Newcastle Australia

http://www.aussieweb.com.au/email.aspx?id=1194181

 

[geesamand]

I suppose I'm coming at this from the better direction: it's a fairly reliable product line that exists in several sizes, and we're refining some of the parts to bring the performance of various aspects into consistent proportions over the size range. We're using FEA to understand which sizes are less robust in various aspects, but all of them happen to use tapered roller bearings and exhibit good reliability. If anything, I'm concerned our current FEA studies are a little too conservative, as it's currently guiding us to add metal where experience with the product would not. And of course a good engineer knows to not "fix" a problem that a calculation suggests when in reality there is no problem.

FEA deflection studies typically are quite accurate for metal components. Cast materials are too. In your example case, do you think it was a load assumption issue? Were the thrust loads induced by the shaft bending loads accounted for correctly?

Overall I'm hoping to develop a consistent method of doing housing analysis, and using this first series to "tune" the assumptions a bit since we have field experience to compare against the calculation results.

David