Brackets with plates in bending - Fatigue allowables and FEA

[trainguy]

Hi all.

I am designing brackets for railway equipment, which are resulting in plate bending (I usually avoid this as much as is practical, but I can't here). We are doing FE analysis to size these, and are ending up with surface stresses (as expected) that are higher than midplane stresses.

Now the fun part: I'm having a lot of trouble keeping these surface stresses (near some fillet welds) within AWS D1.1 fatigue allowables.

What is an accepted methodology here? Do I simply modify the design to get surface stresses (at high stress areas) within AWS limits? The limits are approx 10 ksi stress range - very low! I can see these brackets getting quite thick if I continue...

Any input would be appreciated.

tg

 

[rb1957]

i guess you're welding Al ... 6061 ? i don't know AWS, but i'd assume the fatigue allowable for steel would be higher than 10ksi.

what's your fatigue environment like ? How severe is the spectrum compared with the static loads ?

what're the consequences of a fatigue failure ?

how binding are the AWS guidelines ?

to deviate from a guideline you need some rationale, either the giudeline is very conservative, or a fatigue analysis is acceptable.

if you can't rationalise your way around the guideline, then you need to do something different ... instead of welding, machine the brkt; instead of 6061, steel.

how expensive/cheap is your manufacturing ?

 

[corus]

I'm not familiar with this code but if it's like others then it's the weld classification that's the concern. You can improve the weld by shot peening, for example, which may move the weld up a class.

corus

 

[IceStationZebra]

You didn't post a picture, but bent plates often fail along the edge where a notch concentrates the stress and gets the crack going. Process can be very important.
ISZ

 

[tolson]

trainguy

You may have found your answer weeks ago, but maybe I can offer some assistance. In general, you are correct in lowering the surface stress due to geometry. One of the issues with gusseting for rigidity is that essentially a welded gusset is only as good as the weld - which you are well aware of in your post. Can the piece be bent on the end to add a formed rib? I would rather deal with the endurance limit of steel than a weld if possible. 10 ksi is actually a fairly good allowable compared to a fillet weld at a million or two cycles.

One concern about AWS, or any weld code for that matter, is how well is the fabricator (in house or outside) prepared to back that up? Assuming they are following AWS D1.1 std and it is specified for them to do so, if this is not a pre-qualified joint or you have not qualified the joint, I suggest dropping the allowable even more than code weld classification. Simply, a more refined part geometry or more ribs may be required to reduce stresses at the welded joints.

T