'What about building 7'?

[LSSBB]

Convergence Procedure
The difference between external and internal loads is called the
residual. It is a measure of the force imbalance in the structure.
The goal is to iterate until the residual becomes acceptably small; less than
the criterion, where the solution is then considered converged.
When convergence is achieved, the solution is in equilibrium, within an
acceptable tolerance.

Yay, you've found a procedure definition (reference, please)!

In this instance for this particular algorithm, the residual (also sometimes called "error") is set to a difference between external and internal loads. What are the internal loads, and what are the external loads? What would cause them to not be in balance?

 

[pgimeno]

Are you keeping in mind that the girder between 79 and 80 isn't red, ie the end connections have not failed? The girder would not be removed at that point.

I suggest you look at figure 11-35 again.

 

[gerrycan]

Yay, you've found a procedure definition (reference, please)!

ANSYS

 

[gerrycan]

I suggest you look at figure 11-35 again.

[qimg]http://www.formauri.es/personal/pgimeno/xfiles/cache/NCSTAR1-9_fig11-35.jpg[/qimg]

Yeah, I see the girder with the end connections at 0% damage. That's why it is pink and not red.

 

[LSSBB]

ANSYS

ANSYS is an algorithm. You found how they measure the residual for each iteration until it finds an acceptable solution. It is using the delta between what it defines as internal forces, vs what it defines as external forces. Why do you think they would not "converge" on any particular iteration?

 

[pgimeno]

Yeah, I see the girder with the end connections at 0% damage. That's why it is pink and not red.

And pink means buckled, i.e. removed from analysis, i.e. state of connections irrelevant. The girder between 76 and 79 was then unopposed to push. What's so difficult to understand?

 

[gerrycan]

ANSYS is an algorithm. You found how they measure the residual for each iteration until it finds an acceptable solution. It is using the delta between what it defines as internal forces, vs what it defines as external forces. whhy do you think they would not "converge" on any particular iteration?

Hang on, you said I was wrong about the balancing of the model. Do you stand by that?

 

[gerrycan]

And pink means buckled, i.e. removed from analysis, i.e. state of connections irrelevant. The girder between 76 and 79 was then unopposed to push. What's so difficult to understand?

So NIST removed elements that retained vertical support?
Really?

 

[jaydeehess]

The question was raised " how do we know that the east displacement of col79, by the expanding girder between 79 and 76, occurred at the right time to contribute to girder walk off ?"

I have to ask when the wrong time would be? If col 79 moved before girder A2002 was pushed then that girder had to be pushed a lesser distance in order to have its web go beyond the col seat.
With girder connections at 76 and 79 failed, those two columns aren't going to re-straighten.

The only wrong time for this displacement to take place , wrt girder walk off, would be after girder walk off.

At least in the humble view of this non-engineer.

 

[jaydeehess]

So NIST removed elements that retained vertical support?
Really?

Beams and girders that buckled no longer offer lateral support to the girders and columns they are attached to. They may still be attached at both end but are useless, or worse than useless.
No?

 

[LSSBB]

Hang on, you said I was wrong about the balancing of the model. Do you stand by that?

Yes, because you defined solution convergence as load balancing, whereas in the algorithm used by ANSYS, the solution converges when the load balances. Understand now?

 

[gerrycan]

Yes, because you defined solution convergence as load balancing, whereas in the algorithm used by ANSYS, the solution converges when the load balances. Understand now?

You're struggling.
I thought you might nitpick about force balance.
So how would you go about locating such an unbalanced force?

 

[gerrycan]

Beams and girders that buckled no longer offer lateral support to the girders and columns they are attached to. They may still be attached at both end but are useless, or worse than useless.
No?

In the words of Dave, "oh dear"

 

[DGM]

In the words of Dave, "oh dear"

Explain? Has a buckled element not failed?

 

[gerrycan]

Explain? Has a buckled element not failed?

In this case no, because there is a particular damage pattern displayed by such elements in the analysis. eg 2 bolts fail in a 6 bolt connection = 33% damage etc. This element remains pristine at its end connections throughout the analysis.

ETA happy birthday !

 

[pgimeno]

So NIST removed elements that retained vertical support?
Really?

We've been through this already.

The beams east of the buckled girder between 79 and 80 were no longer there to restrain the girder from moving east. Fig. 11-35 shows that. Take another look if necessary.

With that girder free to move east, the state of that girder would not matter. Even if it was in a pristine state, it would not restrain the column. These beams were the only thing that prevented it from displacing east. Do you not see that?

But it was buckled. It could no longer perform any of its designed functions. Removing it from analysis was a necessity, as NIST explained.

 

[DGM]

In this case no, because there is a particular damage pattern displayed by such elements in the analysis. eg 2 bolts fail in a 6 bolt connection = 33% damage etc. This element remains pristine at its end connections throughout the analysis.

ETA happy birthday !

That's not right. How would it be "pristine" if geometry changed enough to fail 2 out of the 6 bolts?

ETA: Thanks

 

[Animal]

In the words of Dave, "oh dear"

You really do not know a thing about structures do you?

 

[gerrycan]

We've been through this already.

Correct, we have.

The beams east of the buckled girder between 79 and 80 were no longer there to restrain the girder from moving east. Fig. 11-35 shows that. Take another look if necessary.

On the floors directly above and below, these beams were for the most part still in place and connected, as were the corresponding girders (C79-80).
Just how strong do you think the C76-79 girder is?

With that girder free to move east, the state of that girder would not matter. Even if it was in a pristine state, it would not restrain the column. These beams were the only thing that prevented it from displacing east. Do you not see that?

What about the floors directly above and below. Wouldn't they restrain column 79?

But it was buckled. It could no longer perform any of its designed functions. Removing it from analysis was a necessity, as NIST explained.

And throughout all of that buckling that rendered the girder useless there was no damage to the end connections at all?

 

[gerrycan]

That's not right. How would it be "pristine" if geometry changed enough to fail 2 out of the 6 bolts?

ETA: Thanks

The 2 out of 6 bolts was an example of NISTs failure illustration method. There was zero damage to the end connections of the C79-80 girder throughout the analysis.