Compound Fillets

[borjame]

I've used compound fillets for packaging and weight reductions in members under tensile load, but does anyone know of any guidelines or sources for using compound fillets under combined tensile and bending moments loads? Essentially I have a T section where the two "arms" of the T support the trunk as it's being pulled (I'm stuck with the configuration so please don't offer suggestions as how to improve the design, I know, I know...). Where the arms meat the trunk there are tensile and bending moments and how the compound fillet is really working isn't obvious to me.

 

[jetmaker]

borjame,

Could you clarify the loading configuration please. From the description, are you loading perpendicular to the T section?

Thanks

Jetmaker

 

[unclesyd]

I'm like "jetmaker" a little confused.
We always use an elipitical(3 to 1) fillet with the long leg in the direction the force is being applied. Usually the dressing of a fillet is to increae fatigue resistance.

Here is very good book

Design of Weldments by Omer Blodgett
This is a very handy and outstanding book and doesn't cost much ($12.50). 500 hardcover pages of formulas and diagrams

https://ssl.lincolnelectric.com/foundation/

 

[borjame]

A little clarity is usually needed. I have a T section with the arms of the T in the 3 and 9 o'clock position, the trunk of the T at the six o'clock position. The arms are rigidly attached to ground and the trunk is being loaded perpendicular to the arms. This is not a weldment, the T shape is being manufactured in detail so I can call out the fillet size. I'm familiar with what UNCLESYD is referring to, in particular "in the direction the force is being applied". In this case, however, that force is present along with a bending moment in the arms of the T. Will a benefit be seen by putting the larger fillet on the arms of the T to handle the bending moment?

 

[chili]

Blodgett is the bible when it comes to welding. If you don't have it, you need to get it. Don't let the cheap cost fool you, it is the best value you will ever get for your money. It's cheap because Lincoln subsidizes it. Blodgett covers pretty much anything I have ever seen when it comes to welding. He is especially good about weldments in torsion.

Have you considered full penetration bevel welds?

 

[borjame]

Chili, this isn't a weld, it's a fetaure on a manufactured detail like the radius on a forging.

 

[CWIC]

Apparently some of the replies were due to not reading borjames post thoroughly. He/she mentioned twice - this is not a weld. "Fillet" is also a term applied to inside corners of castings, shapes, etc. In this case it's a T.

Borjame, I can only speak of my experience with castings, compound fillets or extended fillet legs are used on the leg were a higher proportion of loading may occur. I do not have the guidlines you are looking for, just posting clarification.

 

[unclesyd]

You can consider a filet in weldmets, forging, or castings practically the same in the limits of achievability.
Echoing the post by “CWIC”. The object in either is to distribute the stress across the junction or connection favoring the member with the highest stress.
My came from an article comparing different fillets.

Post your question here.

http://www.forging.org/forum/forum.htm

You might also check the ASM Forging and Casting Handbook

http://www.asminternational.org

These handbooks are expensive but worth it. As mentioned in another post they have a $179 unlimited access fee.

 

[jetmaker]

borjame,

Let me first make the following assumption. You are wanting to use a compount fillet for fatigue reasons. Also, the T section is symmetric about the trunk. If this is not correct, please let me know.

The tensile load will cause a peak load to occur along the trunk at the intersection of the fillet tangency. The bending load resulting from the eccentricity of the bolts on the arms will occur midway in the fillet. The 2 will not overlap much. However, under linear-elastic properties assumptions (i.e. loading below yield point), the stresses should superimpose.

I would calculate the bending stress concentration factor for an angle under a bending moment. I believe this is available from ESDU. I would use the tightest radius that you call out for the compound fillet.

Increasing the fillet radius is always a good thing, as this will reduce the stress concentration. Another option is to install a radius filler on the arms to locally stiffen the arms and shift the load line closer to the trunk. I would suggest searching the web for tension clip analysis.

Hope this helps.

jetmaker

 

[GregLocock]

borjame - how are you determining the tradeoff? After all, if we fill the section shape with metal it will be very strong, as it is a rectangular rod, but that is scarcely a T section.

Also, what is your criterion of failure?

Also, how is the section made - eg if it is an extrusion then you could do some very fancy things with fillet shape.



Cheers

Greg Locock

 

[borjame]

Now we're getting into the nitty gritty...

The part is being cast(complexity of overall shape dictates this) but weight is an issue. My two criterion are weight and stress. FEA shows that the high stress point is actually in the fillet runout on the T arm portion which would say to me that there's where I should put the larger radius which is opposite of where you would place it for a simple member in tension. This is where I came to you guys to see if there were any general guidelines associated with this. I want to be sure I'm interpreting the model correctly. If these results don't jive with conventional wisdom the model may be providing bunk answers.

 

[jetmaker]

borjame,

Thanks for the added info. By all means, put the large radius towards the T arm. This will increase the material in the bending region, lowering overall stress.

However, as you have access to FEA, it should not be too difficult to do some quick checks to verify this. It seems that you have already undertaken some of this.

Good luck.

jetmaker

 

[GregLocock]

Sounds like a nice project for an optimiser. Sadly I haven't got one at home.

With a casting you could go to any fancy shape, I'd guess the arms and leg will end up tapered.

Is there any requirement that the top of the T be straight?

\   /
 \ /
  |
  |

is quite a likely solution.




Cheers

Greg Locock

 

[RPstress]

Where the peak stresses wind up can depend on the relative thicknesses of the arms and leg, and also on the stiffness of whatever is holding the arms down...the FE should probably include some of the "substrate" material and any fixings (e.g., are the arms bolted down? If so, modelling the pre-load is probably important).

Has anyone got any good sources of data for stress concentrations for compound radii?? I'm only aware of a small amount in Peterson.