Weld Strength (Elastic Method) 3

[LeonhardEuler]

Hello,

I am looking for guidance on using the elastic method with different size and type welds in the weld group. For example an HSS connected to an HSS at a right angle with a 3/8in bevel weld on the top and 1/16in fillet welds on the sides. Elastic method goes ahead and determines a size needed for the whole weld group, but the welds are different sizes and experiencing different stresses.

Thanks

 

[JAE]

I'd just ignore the meager 1/16" welds and see if the two bevel welds works.

But to answer your question - the elastic method uses "lines" as representative of the welds - all at the same nominal 1" (or pick your unit) size.
This allows you to generate the required moments of inertia to then calculate stresses based on those 1" lines.

Then you downsize to the required weld size that still meets maximum stresses in the welds.

If you have different relative sizes, I think you can still generate Ix and Iy and Ip with those variations in size and then do the same to back calculate the required size.

Check out Eng-Tips Forum's Policies here:
faq731-376

 

[Lomarandil]

Star for JAE

----
The name is a long story -- just call me Lo.

 

[CANPRO]

Also, a 1/16" fillet is not a structural weld, sounds like a seal weld. One more reason to ignore it in your strength calculation.

 

[msquared48]

Blodgett gives a good example or two of the analysis procedure JAE describes.

Mike McCann, PE, SE (WA)

 

[LearningAlways]

Why don't we assume the actual weld size to find the actual stresses in a weld group instead of using a unit throat t[sub]e[/sub] = 1". Using an effective throat = 1" is recommended in a few online sources but I can't figure out why. It seems like it is more direct to find the actual stresses based on actual geometry, rather than figuring out how to downgrade a computed stress from a 1" weld to a 1/4" weld for example. It is probably easier than I'm seeing it but still this befuddles me. Can someone please provide an example and shed some light on this? Because the polar moment of inertia of a weld with a t[sub]e[/sub] = 1" is much different than a moment of inertia with a t[sub]e[/sub] = 1/4" which results in much different stresses.

My next question is what do we compare our applied stresses against? It would seem obvious to compare against reduced capacity of phi*0.6Fexx. But in terms of design, how can we find the maximum capacity of a connection if we have an applied weld stress, multiplying stress by Aw (=lw*a(cos45)) seems like it excludes the overall geometry of a weld group and would apply only to a straight line weld. Or does that assumption, the assumption that Elastic Method uses (that welds are all lines), take into account the geometry?

Lastly, what is this Blodgett book that I've read about on this site?

Sorry for the article, thank you for the help!!!

 

[Celt83]

LearningAlways:
Blodgett Book do yourself a favor and spend the $25 on that book, I find it to be an invaluable resource even beyond steel design.

I bought a new copy about a year and half ago and from what I can tell the Lincoln foundation prints and binds a new copy when you order it.

give me a few weeks an I'll convert one of my spreadsheets to python for the elastic method and the predefined weld configurations in Blodgett and put it on my github.

Open Source Structural Applications:

https://github.com/buddyd16/Structural-Engineering

 

[DSB123]

Celt83,
I thought Anaconda was the latest way rather than Python!!

 

[dik]

Also, a 1/16" fillet is not a structural weld

I didn't know they came that small... a wind blown rain would penetrate that...

Dik

 

[CANPRO]

dik, it isn't just the size of your fillet that counts, its how you angle your electrode. But seriously, size matters.

LearningAlways, you could use the actual weld size in your calculation, but often you're going through this exercise to determine the weld size...so if you use the actual weld size in your calc you would need to re-do your calc for every weld size. Using the unit width produces a value of kN/mm (or kip/in if you're a savage) and you can compare the demand on the weld to tabulated weld resistances. In your example with the polar moment of inertia - that property is still linearly dependent on the width of the weld, which means you can use a unit width. If you were to calculate the actual area of the weld for the group properties, which area would you use - the effective throat or the contact area with the base metal? Depending on the direction of load and material properties, either one could be your limiting factor.

 

[LearningAlways]

Thanks for the help everyone!

 

[Whoopdedo_LSU]

I would look at where the max stress is applied. At the bevel weld? Size it based on that. The fillets are nothing more than serviceability at that point.

 

[Shu Jiang PEng SE]

It is better to refer AISC 360 Chapter regarding HSS connection. The stress in weld is influenced by many factors and weld elasticity assumption may be not appropriate.

 

[Celt83]

It took me a bit more that a "few weeks" but had some spare time last night. The linked python script will take in a list of linear weld segments and return the elastic weld group properties. There is a function to perform the force analysis and return the resultant unit shear and to determine the required effective weld throat from the resultant. Like I did for the wood stud wall I'll probably get a GUI banged out for this over the weekend.

It's mostly set up to be unitless, as long as you use consitent units accross the various inputs the results will be accurate, except for the effective throat function where I caved and locked in imperial units.

https://github.com/buddyd16/Structural-Engineering/blob/master/Steel/welds_elastic_method.py

Open Source Structural Applications:

https://github.com/buddyd16/Structural-Engineering

 

[LeonhardEuler]

Celt83I am very uneducated in coding. Can this code be ran in Matlab?

 

[Celt83]

To my knowledge it cannot, there isn't any very complicated math going on so may be trivial for someone familiar with matlab to replicate. Truthfully this could really be handled in excel the only benefit for me with python is the fluidity with the number of weld segments and I can implement load casing fairly easily.

Open Source Structural Applications:

https://github.com/buddyd16/Structural-Engineering

 

[Celt83]

Have a fully functional GUI built

A compiled version for windows can be downloaded from here:

https://github.com/buddyd16/Structural-Engineering/blob/master/Steel/welds_gui.zip

its about 16mb when unzipped

full source code for the GUI and Methods are here:

https://github.com/buddyd16/Structural-Engineering/tree/master/Steel

Open Source Structural Applications:

https://github.com/buddyd16/Structural-Engineering

 

[UcfSE]

...and weld elasticity assumption may be not appropriate

Agreed. For hollow sections welded to hollow sections, wall flexibility can make this assumption invalid.

 

[Celt83]

if anyone's already downloaded it, I've posted an updated version:
-added in a circle helper - creates segments at 1 degree intervals from a specified center point
-the outputs are all text boxes now so you can copy the results and have the ability to paste them in word, etc. (there were also performance issues that this corrects)
-the window will no longer resize after running the design, design output is a fixed height now with a scrollbar to see the full extent of the results.

Open Source Structural Applications:

https://github.com/buddyd16/Structural-Engineering

 

[Agent666]

Thanks for that code Celt83!

A question came up recently on weld groups between some friends, should we be taking the line of the weld as being on the face of the material or at the centroid of the weld after determining the size (doing a second iteration after initial sizing.

I'd always taken it to be at the face, but the purest in me feels taking it to the centroid is perhaps more theoretically correct, even if the answer is potentially only slightly different.