Discussion Groups

Robot Structural Analysis

Reply
Distinguished Contributor
bjur
Posts: 139
Registered: ‎09-14-2011
Accepted Solution

Response spectrum analysis - getting wrong results

1589 Views, 6 Replies
04-22-2012 11:13 PM

I can not get Robot to get the results I am expecting.  Could someone help me to understand why my reactions in Robot are so far off from what Risa3d or SAP2000 get.  I have attached the Robot file, along with the results for both Risa and Robot.  Not how far off they are.   The modal results are correct for both.  So I know I don't have an error with section properties or geometry.  I have matched their seismic criteria (S1, Sds...etc), yet still can't seam to get Robot to calculate the reactions even close to other programs.  I assume I have a setting incorrect.  Could someone help me in this matter?

Please use plain text.
Valued Mentor
tony.ridley
Posts: 652
Registered: ‎09-07-2011

Re: Response spectrum analysis - getting wrong results

04-22-2012 11:41 PM in reply to: bjur

Are you looking at reactions from the CQC, or for the first main mode for each direction? 

I'm not familiar with how Risa3d reports this, so only assuming it may be giving reactions for the main mode and you are comparing to CQC in Robot?

 

Also, does Risa3d automatically add gravity cases back to the seismic one? 

In robot, the simple cases there (for x, y and z) are lateral cases only and have no gravity load applied...

 

Tony

 

Please use plain text.
Product Support
pp2008
Posts: 471
Registered: ‎06-23-2008

Re: Response spectrum analysis - getting wrong results

04-23-2012 10:51 AM in reply to: bjur

Hello,

It seems that this post is some repetition of another post I have just answered:

 http://forums.autodesk.com/t5/Autodesk-Robot-Structural/Robot-Seismic-Combinations-Compare-to-Risa-a...

 

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

 

Regards,

 



Pawel Pulak
Please use plain text.
Distinguished Contributor
bjur
Posts: 139
Registered: ‎09-14-2011

Re: Response spectrum analysis - getting wrong results

04-23-2012 12:39 PM in reply to: tony.ridley

In Risa, I was compairing a CQC combination with Robot's combinations.  I was trying to make sure that the settings in Robot were correctly settup to give me the same results as I was getting in Risa (which I understand much better).  I was just compairing the lateral results (not dead load included).  One thing that came up in an answer posted to a previous question, was that I used a directional factor of 1.0.  It should have been 32.17 (unit conversion factor I guess).  I was unaware that I needed to scale this number.  I'm curious now, if I need to do this same procedure when doing an IBC based dynamic analysis.  A factor of 1.0 gives results that are too small, while a factor of 32.17 gives results too large.  I didn't see anything in the help file indicating that I should apply a scaling factor to the direction factor of 1.0.

 

I can get the correct (or close enough) results using spectral analysis now, but how do I get similar values using the seismic analysis type.  Do I have to apply a factor to the "Seismic Results" to get similar values as the spectral results?  Here is another file with both analysis types in the same file.  I would rather use the Seismic Analysis type as I can just plug in my site criteria, but I'm currently making a mistake somewhere with setting up the Seismic Analysis type where I don't get the same results as the Spectrum Analysis type.  I imputed what should be the correct Spectrum for the site using the same numbers I imputed into the Seismic Analysis parameters.

 

Sorry to post so many questions regarding this topic.  Bear with me as I figure this out.  Thanks again.

Please use plain text.
Product Support
pp2008
Posts: 471
Registered: ‎06-23-2008

Re: Response spectrum analysis - getting wrong results

04-24-2012 09:37 AM in reply to: bjur

Hello,

the main reason of difference between seismic load cases and spectral load cases in your last model was the behaviour factor R=8.5 used in seismic load cases. It results in reducing spectra for calculating forces, momenta and reactions.

After setting it to 1 the results aree close to each other - see below.seismic vs spectral.png

Some reason of difference is also in spectra - for instance the automatically generated spectrum for seismic load cases has the maximum value between 0.11 and 0.55 seconds. The spectrum defined by you for spectral analysis has the maximum value between 0.11 and 0.50 seconds.

It can be seen graphically in the  calculation note using Analysis>Calculation Report from the pull-down text menu.

 

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

 

Regards,



Pawel Pulak
Please use plain text.
Distinguished Contributor
bjur
Posts: 139
Registered: ‎09-14-2011

Re: Response spectrum analysis - getting wrong results

04-24-2012 03:22 PM in reply to: pp2008

I think I have this figured out now...

 

So Robot scales the Seismic results by dividing by (R/I), which is in the code (ASCE 7-05 Section 12.9.2).  When I entered the Static Base Shear into Robot for each direction, the reactions were scaled to the base shear value.  There are certain code limits that I need to make sure are being checked, but I'm feeling better about this.  Attached is a sample work flow to calculate the scaling factors for ASCE 7-05 (IBC 2006).  Note for cases where the dynamic base shear is less than the static base shear, the ASCE 7-05 sets a minimum of 85% of Static Base Shear.  Is this being taken into account?  Is there a way to view the unscaled dynamic base shear?  I would like to verify the dynamic base shear is being scaled correctly.

 

To get unscaled base shear results, would I enter 1.0 for R and leave the base shear section blank?

Please use plain text.
Product Support
pp2008
Posts: 471
Registered: ‎06-23-2008

Re: Response spectrum analysis - getting wrong results

04-27-2012 11:50 AM in reply to: bjur

Yes, to get unscaled values R should be set to 1 and "Base shear" should be set to "Inactive".

 

Regards,



Pawel Pulak
Please use plain text.