Stress due to bending of fillet weld joint

Join the discussion
Registration is free. Start your own thread to ask a follow-up.
5 replies · 15K views
Mech King
Messages
69
Reaction score
0
Evening everyone,

i have been referring to Shigleys "Mechanical engineering design" (pages 349 to 351) regarding a welded joint I am looking at. wanted to compare my manual calcs to sum FEA.

My joint is a circular shaft which is fillet welded to a base plate. A load is applied to the end of the shaft.

Do i compute the shear stress in the joint, mutlipy that value by 3, and the max bending stress in the joint, square them both, add them together and then square root them to get the von mises stress?

Also, Shigley lists the Unit moment of inertia for the joint as: pie*r^3.......... so, if I am treating the weld as a circular line, can i just divide the unit moment of inertia by the radius (the distance to the centroid) in order to get my section modulous Z? I will then divide M/Z to get my max stress?

It's been a very long time since i looked at welded joints, and am a tad confused,

any help much appreciated,

Cheers
 
Last edited:
Physics news on Phys.org
Mech King: No, square the shear stress before multiplying it by 3, not vice versa.

Yes, you are computing the unit section modulus, Zbar, correctly, not section modulus, Z. No, do not compute M/Zbar. First, multiply Zbar by the weld effective throat width t, then divide M by this result.
 
Thanks nvn,

i got confused with the text as it didn't explain the reasoning for multiplying by the weld throat by Zbar?

Just to clarify:

So for a circular weld in bending, i mulitly the Unit Moment of Inertia (Z bar) by the throat area (1.414*pie*h*r)?
 
Last edited:
Mech King said:
Do I compute the shear stress in the joint, multiply that value by 3, and the max bending stress in the joint, square them both, add them together, and then square root them to get the von Mises stress?

Mech King: No, don't use that method. Don't compute von Mises stress. Just compute and use the following stress. sigma = M/(Zbar*t), where Zbar = pi*r^2, and t = 0.7071*h.
 
OK nvn,

but shouldn't Zbar be pi*r^3 as in Shigley?
 
Last edited: