Capacity of a round weld 2

[anchorengineer]

Hello All,
I'm looking for a formula to calculate the capacity of a fillet weld for a round post with a horizontal load at the top inducing a moment.
Thanks!

 

[racookpe1978]

I'm wondering if the original poster ever thought about what he was asking ...
"I have a pipe. How thick does it has to be? How long does it have to be?"

Do they actually graduate engineers that are that ... "basic" today?

 

[medeek]

You will note that my weld calculator has shear, moments and torsion. For the situation described by the OP, you would need to enter in an appropriate moment and a shear load.

A confused student is a good student.

 

[racookpe1978]

Concur. But does the original poster understand enough about design to use it safely?

 

[KootK]

Half of the post is in tension and half is in compression....it is a couple from the moment.

We discussed this option pretty thoroughly during round one Ron: Link. Perhaps you've reviewed that and are sticking to your guns anyhow but, just in case, I thought that it would be prudent to bring it to your attention.

However, if you apply a simple thought experiment perhaps it might become more clear. Assume we have a hollow pipe welded around its perimeter, at the other end of the pipe we have a load evenly distributed around the pipe perimeter that is perpendicular to the pipe's long axis. Now imagine that we slowly decrease the length of the pipe until it is almost zero length, everything else remains the same. You would find that the shear load around the perimeter of the pipe is now evenly distributed along the weld. There may be a flaw in this thought experiment but for now it makes sense to me.

The flaw is that, for loads applied further from the support, the stresses that comprise the resistance to those loads have time to reorganize themselves to reflect the inherent stiffnesses and flexibility of the cross section. Your mental experiment could also be applied to a wide flange section. In that case, the result would be the same. However, I think that we can all agree that the welds along the web should be designed to constitute the bulk of the shear resistance.

Now you have to consider that the horizontal load must be resisted in shear. The whole circumference resists this.

I've quoted Ron but this seems to be the sentiment of pretty much everyone other than myself. Try this "proof" to the contrary on for size:

1) The forces in the welds at the support are simply the stresses in the circular hollow section at the support multiplied by the wall thickness.

2) The shear stresses in the circular hollow section at the support -- or anywhere else -- are not resisted by all segments of the cross section equally. This can be verified through VQ/It analysis and by one our own threads on the topic Link.

3) IF (ARGUMENT 1 = TRUE) AND (ARGUMENT 2 = TRUE) THEN (Welds more parallel to the applied load will be disproportionately loaded in shear).



I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough that I want to either change it or adopt it.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[Brad805]

Interesting discussion and background info. I tried a FEA to take a get an idea. The model consists of a 6x6 piece of solid steel (base) and a 36" tall CHS 4"x1/4" pipe gaped 1/16" from the steel base to avoid full bond between the the CHS/base. There is a 2,250lb load applied to the top of the CHS. It is giving me a headache right now looking at all the various stress reports, so I have not decided which side I am on. I thought I would post it anyway.

 

[KootK]

Now we're having fun. Could you rerun the second plot to show shear in the direction of the load Brad? Better yet, can you generate the same stress component on a tube cross section located about 2D from the support?[pre][/pre]

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough that I want to either change it or adopt it.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[desertfox]

Hi

Here is an example very similar to the OP {-

http://blog.mechguru.com/free-cacul...ation-example-for-bending-moment-application/

 

[Tmoose]

"Could you rerun the second plot to show shear in the direction of the load...."

yes, I'd like to see that too.

regards,

Dan T

 

[BUGGAR]

Are you considering the different strengths of a fillet weld loaded axially vs. transverse to its axis?

 

[slickdeals]

Brad,
While you are at it, can you also run a square and rectangular tube?

 

[Ron]

The FEA model clearly shows the couple.

KootK...I agree that the weld will not take shear uniformly. The weld more parallel to the line of shear force will have lower shear stress because it likely has a larger cross section than the weld perpendicular to the line of force. It should be noted that the only truly parallel force is the tangent to the circle on either side of the ring.

If the half-circle were divided into 3 sections, the middle third would be the perpendicular force resistor and the two outer thirds would be the "parallel" force resistor.

In any case, the weld will be under both tension and shear on the side where the force is applied, so a unity check should be done.

 

[Brad805]

I have attached a few more stress plots. You cannot isolate the effects of different forces in this software, so I think I will shorten the column and increase the shear forces to show the effect of the shear better since we all agree on the effect of the moment. I did that last night, but I need to add a stiffening element around the top of the pipe since it started to deform excessively. For some reason solidworks uses the Y direction as the vertical. I don't know why exactly, but it is annoying. I will make a few other changes as well. I can see the 1/16" gap is influencing the stress distribution, and the fine point of the weld is indicating higher stresses than it should.

Engineers, egad. I only started the solid simulation a short time ago, so this is a good case study. The problem with solid modeling is the sheer amount of data.

 

[KootK]

The FEA model clearly shows the couple

To me, the second plot shows an elastic, M/Sx style stress distribution with the peak stresses at the extreme fibres as one would expect. Those stresses could not be represented, in statically equivalent fashion, by tension and compression forces located at the centroids of two half circle weld groups. I suspect that there is some "noise" in the plot as we appear to be seeing Von Mises stresses rather than purely axial stresses. Also, for the sake of this argument, it might be better if the peak stresses were not so close to Fy. Yielding will muddle the issues here.

If the half-circle were divided into 3 sections, the middle third would be the perpendicular force resistor and the two outer thirds would be the "parallel" force resistor.

We are in similar ballparks here. I would divide the entire ring into quarters and say that the two side quarters take most of the vertical shear and that the top and bottom quarters take almost none. That mechanical thread that I linked above (Link) yielded two interesting conclusions:

1) The max vertical shear stress in the tube will be located at the neutral axis (big surprise) and will take on a value if 2P/A.

2) If one plotted vertical shear force resistance over the height of the tube, the graph would be a straight line at a uniform value. Since there's more section available per vertical unit height at the top, I interpret that to mean lower stresses at the top and bottom of the section and higher stresses at the sides. And that's consistent with the 2P/A estimate above.

As for a practical weld sizing strategy, I'd probably just design the weld for the vector sum of 2xPxt/A and MxCxt/I. And not bother with the fact that the maximum stresses do not occur at different locations unless I find my weld size very objectionable.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough that I want to either change it or adopt it.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[KootK]

There is another point that was made in that mechanical thread that I feel is significant. To paraphrase:

For weld design, the end game should be only two force components in each unit segment of weld:

1) A force parallel to the longitudinal axis of the tube, reflecting bending stress.
2) A force tangential to the tube representing VQ/It stress.

That makes sense to me. There should be no component of weld force in the radial direction as radial flexibility in the tube walls will tend to relieve that force component. This would lead one to assume that the welds at the very top and bottom of the section resist none of the applied shear. And that would match the expected VQ/It result.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough that I want to either change it or adopt it.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[Brad805]

Attached are the results of a much shorter column. Small weld edge radii were added to reduce the appearance of stress concentrations and the weld mesh was reduced to .1". The load in the Z direction was increased from 10kN to 40kN (9kip). The Txz plot does show higher weld stresses where the weld is more parallel to the load. I attached the stress direction diagram from the software in case anyone is interested. I have things to do now.

 

[Hokie93]

The procedure outlined by spats is the method I employ in my practice and I believe it is supported by various structural steel textbooks, including publications by Omer Blodgett/Lincoln Electric. For example, this method is used on page 12 (Example 12D) of "Solutions to Design of Weldments", a publication of the Lincoln Electric Company. In Example 12D, a 9'-0 long, 12" diameter cantilevered pipe is subject to a 10 kip concentric, concentrated load at the free end. The solution uses the entire perimeter of the circular weld when determining the shear stress due to direct shear.

 

[Ron]

Hokie93....I would also use the entire circumference to compute the shear stress....my example above was in response to Kootk's comments, which would perhaps yield a more precise stress distribution but probably not necessary in the whole scheme of things.

Blodgett's examples are all done with "slide rule accuracy", as his work preceded commonly available computers by many years, thus the tendency to group things together and estimate the distribution of stresses to be more universal.

All the arguments aside, Blodgett was a welding design guru and his principles are valid and useful.

 

[paddingtongreen]

I think the line weld method is correct. If the fillet were a straight line there would be no question. In this case, the weld throat is longer than the face against the pipe.

Considering the moment and assuming that it is the major load, different story if the shear is the major load.
The problem that I see is in the materials, we check the shear on the weld/pipe interface area against the strength of the pipe material, we check the throat against the weld material.

Considering the shear force, I think the it is transmitted through the sides of the pipe, these are stiffer than the front and back faces.

Michael.
"Science adjusts its views based on what's observed. Faith is the denial of observation so that belief can be preserved." ~ Tim Minchin

 

[KootK]

I find myself in my least favorite of positions: I no longer believe that I am right but, at the same time, I also do not understand why I am wrong. Now I'm just... confused.

The Blodgett example that Hokie dug up is shown below (thanks Hokie). It does indeed corroborate Blodgett's intention of uniformly distributed shear for this situation. Ditto for the website that Desertfox linked us to above.

Equally damning, in my opinion, is page G-9 of this document: Link. There, the shear capacity of a round HSS is calculated using the the entire cross sectional area (0.6*Fy*Ag). Just. Like. The Blodgett example. This, even though the previous example for a square HSS utilizes only the webs in shear. Maybe the intent is to examine some heavily plastified ultimate strength condition, I'm not sure.

I like to debate structural engineering theory -- a lot. If I challenge you on something, know that I'm doing so because I respect your opinion enough that I want to either change it or adopt it.

The greatest trick that bond stress ever pulled was convincing the world it didn't exist.

 

[CELinOttawa]

I have a concussion... So if you need to, please take this with two grains of salt and correct me in the morning.

One of my favourite references for teaching students connection design is the (very) old Ketchum engineering manual. The other is an all-inclusive timber-steel-concrete analysis and design text from the late 1940s. I can't remember the second text's name at the moment, but it has one of the most Wooten like statements when you start the welding design chapter.

I paraphrase: Like all states of stress in steel, the true state of stress in a weldment is an incredibly difficult analysis, leaning towards an impossibility. As such in practice we apply conservative values to simplified analytical methods and this has proven to be acceptable in practice.

My point being: You're absolutely right, Kootk. There is a great deal of simplification in connection design. I simplify a lot of my connections a great deal, often producing something a great deal stronger than needed. It doesn't matter. Trying to apply a more refined approach to a sufficiently solved problem is like trying to make concrete more grey. I need a reson to care...

I am VERY interested in the answer. Don't get me wrong; I am keen to have the *right* answer. I just don't want any EITs, or anyone else for that matter, thinking this actually matters.