Seismic load combinations

[PatrickEC]

I am trying to define some load combination that include seismic cases and do not seem to be able to make it work.

In the attached pictures you can see how I defined everything. What am I doing wrong?

[Pawel.Pulak]

It seems that you run seismic analysis for non-linear model (tension-only bracings or uplift supports?).

If you define or modify combinations using the table of combinations the status of results changes to "out of date" and results for new created or modified combinations can be not available (N/A). In such case it is necessary to re-run analysis.

It can be avoided not using the table of combinations to to define or modify them - definingor modifying combinations using Manual Combinations or (for instance change to quadratic combinations) using Analysis Type window.

Such behavior results from non-standard operations on results made by Robot in such case. Because model is non-linear the superposition principle is not valid for it so theoretically all combinations should be iterated in non-linear way.

But modal and seismic analysis (using response spectra method) is performed for linearized model and results of seismic analysis are combined with results of static non-linear load cases without non-linear iterations. Of course performing seismic analysis for linearized model and then combining these results with non linearized (non-linear) results for static analysis is some approximation. The precise approach for non-linear model would be running seismic analysis using non-linear time history. More details in Rafal's answer to this post:http://forums.autodesk.com/t5/Autodesk-Robot-Structural/Barre-de-treillis-traction-compression-vs-s%...

---------------------------------------------
If this post answer your question please click "Accept as Solution". It will help everyone to find answer more quickly!

Regards,

[PatrickEC]

Yes, there are some tension-only elements in my model.

So I understand that setting aside accuracy of this solution, the way I defined it should give results? However, I made sure to re-run analysis after I added these combinations, so this is not the reason why I get N/A.

[Pawel.Pulak]

Can you send the model? Without results and compressed to ZIP to reduce the size.

Regards,

[PatrickEC]

Here is the reason why there were no results for those combinations:

1. The simple load cases that were included in these combinations were disabled (auxiliary).

2. A load combination including a seismic load case is always linear and the results are obtained using superposition. Hence, unlike in the case of nonlinear combinations all included load cases must be calculated independently.

Pawel:

"Stiffness matrix for modal analysis is taken from the last calculated case before modal case"

1. To make sure: Does a "last calculated case" mean either simple load case or load combination, or only simple load case?

2. I am going to stick to response spectrum method, treating tension-only elements as tension-compression for seismic analysis. I would like to avoid creating a separate model just for seismic design. Is it possible to define the load combination preceding modal analysis in such a way that no tension-only members are disabled?

My first thought was to include only some vertical loads in this combination, but even then there is also some small force in the braces and some braces will get inactivated. Maybe defining a load combination with no loads would work?

[Pawel.Pulak]

Hi Patrick,

answers to your questions illustrated by attached very simple model (4 bars) and screen capture made for it:

1. It is either simple load case or load combination.

2. For simple and regular models it is often possible to find loads resulting in all tension-only bracings working. For sophisticated and unregular models it is often practically impossible - for your model received by private message:

- using selfweight upside-down (in Z+ direction) resulted in about 30% of tension-only bracings still inactive 

- using negative temperature TX for all tension-only bracings (equal value for all of them) resulted in 2 such bars still inactive

The simplest solution to obtain dynamic analysis with no tension-only members disabled is to define modal analysis as the first load case - before any static load cases. Load cases and combinations can be renumbered in Load Types window (Modify button)

Just a reminder: if no tension-only members are disabled the stiffness of structure is overestimated.

---------------------------------------------
If this post answer your question please click "Accept as Solution". It will help everyone to find answer more quickly!

Regards,

[PatrickEC]

---

@pp2008 wrote:

The simplest solution to obtain dynamic analysis with no tension-only members disabled is to define modal analysis as the first load case - before any static load cases.

---

I renumbered load cases so that the modal analysis was first, but still some braces were inactive - I sent you the model. I ended up with creating a separate model, but I am curious what went wrong.