Gusset weld design theory?

Date 05-27-2008 21:57

hello, welding engineering intern Metarinka here again.
We have a common problem in our shop, mounting plates for motors ( T made of 2 pieces of plate generally from .125-.5" Low carbon steel) that have gussets on them. As of right now the gussets are fully weld as well as the fillet joint on both sides of the plate.

There have been issues with distortion.

now I know from shop knowledge and my rudementary understanding of joint design that such a part is being overwelded however I don't have any information or numbers to back that up. I've already argued and convinced engineers to put a staggered stitch fillet instead of a full seam, and I'm hoping I can convince them that welding the full seam length of a gusset adds little to strength and just increases distortion. As my understanding goes gussets transmit forces in compression and as such the part would be prone to buckling or collapse of it's members before the welds at it's base would fail. I'm comfortable that 1-2" of fillet per leg would more than satisfy strength req's
Does anyone know where I could find some technical knowledge on weld design that would cover this? Would structural code contain this information. something akin to telling me how much strength I'm getting per length of fillet weld or specifically for gussets. Are there any general welding design theory books like that? if I remember correctly after a point in welding a fillet weld (both sides) the welded joint is stronger than the base metal and the weld should always fail at the toes of the weld. it would be nice to have a book that put numbers to these sort of things

I'll add a drawing if needed, the weldment is a T made out of 2 pieces of plate with a triangle shaped gusset flush with the outside edge. The gussets generally get a full fillet along their "inside" and a full sq groove (no bevel) along the outside.

By Bob Garner (***)

Date 05-27-2008 22:43

One thing that makes calculating gusset forces difficult is that gussets generally provide support against buckling of the main member. The force needed to restrain that buckling is very difficult to predict without going into all that finite element analysis with all the little grids and colors representing stress values.

There are definite design values of so many pounds capacity per inch of weld, but the problem is finding what the load to restrain buckling actually is. When we design braces for columns, we traditionally consider the required force to brace the column is 2 percent of the column axial load. But this doesn't translate as easily for gussets bracing plates.

Sorry I can't give a better answer.

Bob Garner

By ravi theCobra (**)

Date 05-28-2008 16:37

You might want to be carefull because these designs are a lot more than stress = force / area.

The most understandable work is Omar Blodgett from Lincoln . He's written several books -

By Metarinka (****)

Date 05-28-2008 17:08

I should add that the purpose of these braces are to prevent bending in the elastic region of the base metal and to hold tolerances against the (relatively) static loads they carry. The forces exerted on these parts are below the elastic limit. That's why I feel they are so overwelded. I don't have any hard numbers in front of me on the loads. but I'm assuming that if the member in tension is not plastically deforming (or failing for that matter) than certainly the same cross sectional area in compression should be much stronger.

I also know I'm starting to get over my head in terms of structural design (I'm a welding engineer not mechanical). But I know enough to know these braces and almost all our gussets are getting over welded.

By Metarinka (****)

Date 05-28-2008 20:57

thanks ravi thecobra,
I'll check out those books. I can see how quickly a simple joint can have complex forces on it.

By ravi theCobra (**)

Date 05-28-2008 21:22

I used to build oil refineries and their pumps to API Spec 610. One of the nightmares we had back in '81 - '82 was the flatness

of the motor pads -   We machined them to less than   .001 inches / foot flatness - to check this out you would get a motor

all snugged down ( bolted up ) and put a dial indicator on   your motor foot and loostened the bolt   -

If the spacing of the bolt pattern was 1 foot then   anything over .001   was rejected - it was more interesting on say a 500 HP

mator where the bolts were 30 inches on center - then the maximum   deflection   was   .0004.

This was a result of going to fabricated (from steel ) baseplates instead of the old cast (and much heavier - ) ones.