Future of Connection Design

[jseng9]

Could structural steel connection design ever be 100% based on Finite Elements? I'm interested in hearing your thoughts.

 

[jayrod12]

Likely not. Honestly there are so many canned connection designs that meet so many of the typical situations that no one will be doing FEA on them. It's significantly faster to just use tried and tested connections and the potential savings by trimming it down doesn't seem worth it.

 

[skeletron]

There's probably some room for specialty connection design or connection research to gravitate toward FEA. But as a whole, there are so many more variables that the analysis time becomes a money pit. I think fabricators/constructors like the KISS method for familiarity and cost.

 

[JoshPlumSE]

I'm reminded of that Dallas Cowboys practice facility that collapsed. It wasn't designed by a civil / structural engineer but by some analysis guru that used all kinds of goofy analysis to justify why it should work.

Regardless of whether FEM is used, it connection design should always be supervised by a Civil / Structural PE. Because of this, I doubt if it will ever fully go FEM.

I have some experience with this during my time at RISA. The RISABase program (which has since been discontinued) was mostly an FEM approach to base plates. As such it was pretty useful for cases where hand calcs would be difficult. Think bi-axial bending, and / or misplaced anchor rods. That being said, there were aspects of FEM that really required engineering judgment to interpret what it meant. You'd get localized stress risers in some places. Very easy to interpret with engineering judgment, but really tricky to do programmatically.

 

[BUGGAR]

Watch some ironworkers put together a joint. Now imagine doing a fem on each piece as they assemble it by bolting and welding. Then they usually "bang" the pieces into place on the connected members. Then, when all the steel is up, they may come back to "rattle the bolts" for final tension. Feed that sequence into fem also. We're lucky that any analysis at all will work for a structural steel connection. We don't even use the stress area at the threads of the bolts we design for. If it's important seismic stuff, we require pretesting. My thoughts.

 

[ukbridge]

Hopefully not!

In my opinion its the launch of more comprehensive software - for example design of RHS/CHS joints to EN 1993-1-8.

I think theres a lot of scope for a decent developer to come along, create a good application that quickly and efficiently enables designers to sign off connection designs without having to scan through reams of black box calculations.

 

[Trenno]

Maybe...

Would be good to see a comparison between the output from programs such as Idea Statica Steel against tried & tested closed formed connection design solutions.

For stock standard connections I don't see there being much benefit using FEM, but think it would be a useful tool for complex connections in which standard connection design formulae don't cover.

Idea Statica Steel

 

[KootK]

Could structural steel connection design ever be 100% based on Finite Elements?

1) Given a long enough time horizon, I would say that yes, steel connection design could be 100% FEM. Recent developments in the Eurocodes seem to be pointing in this direction. However, I think that it would have to be FEM that:

1a) Captures the development of plastic mechanisms rather than straight elastic behavior.

1b) Checks stability issues associated with the assumed resistance mechanism as our hand methods do so that the solution can still be considered to be a lower bound plastic solution.

2) While I agree that present economic realities favor the continued use of hand methods, I could see that changing in the future. It's a bit of a chicken and egg thing I suspect. I think that we mostly stick to the stock configurations because, thus far, straying from them has been excruciatingly painful. Once the pain barrier is lowered, it wouldn't surprise me to see connection configurations become less standardized.