Checking stresses in stiffened axially loaded base plate - pipe column 4

[ST04]

Hello

I'm using BS code that is based on Ultimate limit state design, but I'm not sure about if I have to check the stresses under ultimate loads or service?

I am designing an axially loaded base plate for hollow pipe (CHS)column. It has to resist high compression and tension loads (alternating) around 3500 kN (Ultimate). So I had to provide stiffeners otherwise it won't work.

I used Ansys to check the stresses, but It was a real headache as the column face connected to the stiffener is always overstressed. I have checked several stiffeners arrangements and sizes but the final solution was to thicken the column wall by using 1 inch thick CHS which is almost double the required thickness to stand the axial load.

I was checking "Equivalent stress (von-Mises)".

Any suggestions?
Regards

 

[BAretired]

You should provide a sketch.

I don't know the British Standard, but in Canada we would use factored loads for all strength calculations. You should consider fatigue as well with alternating loads of such high magnitude.

What is the application? It seems a trifle unusual.

BA

 

[hokie66]

Why not run the stiffeners throught the column? Fabricate the stiffeners into a cruciform shape with the base plate, then slot into the column.

 

[ST04]

Thanks dear colleagues

@ BAretired
The pipe section is a part of cantilever truss. Although both tension or compression could occur with the above value, but the alternation is not expected to occur continuously to cause fatigue.

Thanks hokie66 for your valuable suggestion, I have tried this before writing the thread, I have tried to add internal plate to connect two opposite stiffeners, and tried to add two internal plate stiffeners normal to each others (cross shaped), but it didn't cause significant reduction (reduced only around 10 to 20 N/mm^2) in the other points connecting column to other stiffeners.

I don't think it is possible to use extra internal stiffeners as this will make it very difficult or impossible to be fabricated or fixed.

So I guess I have to provide the 1 inch thick pipe for the first portion connected to the stiffeners then go for thinner section.

 

[ST04]

Also one more thing,
I had my hand calculation for this situation, but I was wondering for future work if the FEM can be used solely to perform the design???

 

[SAIL3]

Typical local reinforcement for this condition would be to add a ring to the pipe col. located at a height from the base pl. that
would result in manageable stresses. Then vert stiffeners between the base pl. and this ring.
By varying the dist. between the ring and base pl and the no. of
vert. stiff. used, one can usually end up with stresses that are
manageable.
I tend to shy away from using FEM in cases where analysis by hand is possible. Someone commented on this site that FEA can give a
very accurate incorrect answer. I totally agree, since the accuracy of an FEM will depend on the boundary conditions you use
and other assumptions which are, in many cases, difficult to reproduce with enough accuracy and applicability in the model and
can give wildly varying results depending on what you choose.

 

[ST04]

Dear SAIL3
That just worked wonderful, I really appreciate your tip.
Best regards to all of you...

 

[TLHS]

If it's not just a case of your plates being too short vertically, it's possible that you have a problem with the assumptions inherent to your finite element model.

For instance it could be a problem related to the way your plate model acts. Many have infinitely thin plates in the model and you can sometimes have incredibly conservative or strange results with punching loads or moments applied by plates at steep angles to other plates if you aren't aware of the issue. You can also have problems if the normal assumptions you would use for hand analysis assume that small amounts of local yielding allow for balanced load transfer. Unless your finite element software supports modeling yielded material you'll just end up with overstresses and no redistributed loads.

 

[delagina]

tlhs, can you explain this further

"Unless your finite element software supports modeling yielded material you'll just end up with overstresses and no redistributed loads. "

thanks,

 

[TLHS]

When you approach/pass the yield point of most structural materials a small increase in stress results in a large increase in strain. Effectively, the material will hold the load it's got at that point but will deflect so much when more load is applied that other parts of the structure will be stiffer and will take the loads.

There's some software out there that, if you turn on the proper non-linear analysis, will apply the stress-strain curve of the material and actually model element yielding when it hits yield stress. (I have never actually used software that does this, personally)

Other software, however, just has a single young's modulus value that it will happily use from zero to infinite stress. It's fine if you're working in the usual structural realm, which is below the point where there's a significant yield, but if you get into small details, or connections, there is sometimes an assumption that the material will yield locally and loads will redistribute to match your assumed model. If the elements in your model don't hit a point where a small increase in stress will result in a large increase in strain this redistribution will never happen.

 

[ST04]

Thanks TLHS for your input, in my analysis I have used the non-linear choice.
Best regards