Response Spectrum Analysis

Question:

Can you send me an example of Response Spectrum Analysis illustrating the difference of ticking or unticking the "Add Signs"  check box? (see below)  thanks



Answer:

Hello,

Kindly refer the image below on how the results would vary.  The details of how this works could also be referred in the online help manual (Link).
The model files are also attached.  Kindly let us know if further assistance would be required.

Regards,
Nandeep
Technical Manager, MIDAS


Hi, Nandeep

Thanks for your quick reply.


But what we have is 2018 V1.2, your models were created under higher version.  so I cannot open them.


Can you somehow save the models for lower versions?


thanks





Hi, Nandeep.


I create my own RSA models with and without "Add signs".  ... seems to be working....

Questions I am thinking:

1) when you do a modal combination, say SRSS, you will lose the signs automatically by the rule of SRSS.  What exactly do you mean by "add signs"? How do you know what sign to add?

2) What do you mean exactly by "along the major mode direction?" and what is the def of a major mode?

3) what do you mean by "along the absolute maximum value"

4) what do you mean by the checkbox "select mode shapes" and what do the mode shape factors mean?


Need more elaboration, please.


thanks



Hello Jack,

Glad you were able to generate models at your end!  Now, about the questions this time around, details on each of these is provided in our online help manual (Link).
I believe I shared this link before.  Kindly go through the explanation there for better understanding.  If however, you have doubts about any specifics, kindly let us know.

Regards,
Nandeep
Technical Manager, MIDAS


Hi, Nandeep,


Gone through the link, my questions remain.


Can you elaborate? with some examples preferably.



thanks




Hello,

I can't find better words to describe the feature!  However, I'll try...
1) when you do a modal combination, say SRSS, you will lose the signs automatically by the rule of SRSS.  What exactly do you mean by "add signs"? How do you know what sign to add?
A:  The methods to add sign are already mentioned in below questions.  Basically, its to provide direction to magnitude after obtaining SRSS or CQC results.

2) What do you mean exactly by "along the major mode direction?" and what is the def of a major mode?
A:  Major mode would be the one with maximum mass participation for the said direction.  So, if RS along X direction is being checked, then major mode would be the mode with maximum mass participation in UX.  Since its easy to co-relate the mode shape and deformation, to simply understand the sign convention, we can say that sign of eigenvector of this mode would be taken and assigned to the CQC or SRSS result of displacement and the force would be given signs accordingly, which would lead to this displacement.  Kindly refer the image below.

3) what do you mean by "along the absolute maximum value"
A: The signs are taken from the same table as shown above.  But, the method of taking it changes.  In this case, the sign would be taken for the vector with maximum magnitude, irrespective of the major mode.

4) what do you mean by the checkbox "select mode shapes" and what do the mode shape factors mean?
A:  That's quite simple.  Here you select which modes you want to consider for CQC or SRSS combinations.  Other modes would be ignored in the combination.  As for factors, if you'd like to scale up or down the mode influence, then that can be done here.

Regards,
Nandeep
Technical Manager, MIDAS


Thanks for your reply, Nandeep. Now I understand the issue better. But if I think deeper, I still have some issues:

1) Say I have a model with only 2 mode shapes, 1st being major at 51% mass contribution .  The sign of the 1st mode shape will be taken as the sign of the SRSS/CQC results if I tick "Along the Major Mode Direction".  I suspect, there will be some nodes that will have signs same as the 2nd mode shape, given the 49% contribution.  How does Midas handle this situation?  It sounds the graphic results can not ALWAYS be correct.

2) Presumably, I will get different results if I tick the other radio button.  Which one will be the CORRECT one?  How do I know which button to tick? keeping in mind there is ALWAYS some torsion in any model. 

Sounds like it is a work of art, not of science.


I have to say, the language used by Midas is vague either in the help or in the various dialog boxes.  I don't understand it most of the times.


You will see what I mean if you compare the language used by MS in EXCEL, for example.


Thanks


Hello,


Glad the response helped a bit.  To further clarify, kindly refer my comments below:
After performing RS analysis and combining modal results, signs are definitely lost.  Now, engineers are left with 2 choices.  Whether to add signs or not.
1)  If signs are not added, then the results can be very conservative.  This is not acceptable in many engineering communities.  If it is acceptable without sign, this resolves the issue.
2)  If signs are to be added, which most engineers do, then widely acceptable methods are to add sign for major mode or add sign for max value.  Both methods have their limitation.  What method is suitable for a particular structure is left up to engineers discretion.  Only the guideline on what method to be selected is provided in the online help manual.  Obviously, the results could be different with different method.
To summarize, this is a option where different engineers have different views.  As an engineering software solution provided, what we have to do is to give engineers flexibility and ease of usage for the method that they are comfortable with.  In that sense, I too would agree with you that its an art, but I believe, its not completely devoid of science.  Research papers supporting the methods to add signs could be located online.
Personally, I'd recommend not to add signs in case when different modes have very little difference in mass participation (51% & 495), unless you are absolutely sure of what is happening with signs.  It'd be better to go with conservative approach if that would be acceptable.  Though, such scenarios would be rare practically.
We will definitely try to improvise on our dialogue box language issues.  I'd appreciate if you could point our certain precise issues for me to intimate our development team.
Kindly let us know if further assistance would be required.

Regards,
Nandeep
Technical Manager, MIDAS


Thanks again, Nandeep.


In terms of the language, I can give you an example here:

Refer to the screenshot, I'd change "modal combination type" to "modal combination method".

I'd also change "select mode shapes" to "select modes for modal combination" and remove the "mode shape factor": it 's not clear what is a "mode shape factor" and I doubt anybody would bother changing them.

Cheers and have a good day.



Hello,

Thanks for the suggestions!  Personally, I am very used to what the options mean, so I tend to overlook such a part. A bad habit developed due to daily exposure!
I'll definitely update our development team about this.  Kindly let us know if further assistance would be required... :)

Regards,
Nandeep
Technical Manager, MIDAS


Hi again, Nandeep,


Refer to the attached model.


I understand:

1) By modal combination, the signs are lost, either for displacement or force.  But the absolute values of all analysis results should remain.

This appears to be true with displacement, and not with, say, reaction. See screenshot:


Pier reaction without tick at CBMAX 2475 load combo.



Pier reaction with check box ticked at CBMAX 2475 load combo


can you please explain?



2) I'd imagine Midas would do a modal combination on displacements to get the total displacement.  And similarly a modal combination on force to get the total force.  In another word, if I were to apply a set for forced displacement statically, I will not get the same force.  please confirm.


thanks




Hello,

The model file was not attached.  Kindly try sending the same again.

Regards,
Nandeep
Technical Manager, MIDAS

Hi, Nandeep,


I figured out the answers to my questions from yesterday. No worries.



Now one new question: (See model attached.)


When I check the reaction in spreadsheet format, the reported summary does not match my hand summation (8203  reported vs 20919 by hand).  Can you please shed some light here? thanks

see screenshot below:





Hi, Nandeep,


Refer to the same model.


For some reason, some beam element shear forces changed as I switch from "no sign"  to "add sign".  This cannot be true.  But why?

Specifically, I was checking the Fz of "Fixed bearing" at CBmax 2475 and CBmin 2475.


thanks





Hello,


First of all, about the reactions.
In case of RS analysis, the summation of reactions is not the linear summation of reactions of each support.   To obtain reaction at each support, CQC is done for reaction obtained at each support.  But to obtain the summation, the CQC is done for summation of reactions in each node.  Hence, this can never be same.  A sample  calculation of random value with SRSS method is attached for better understanding.
About the beam shear.  The reason for that is also quite simple.  When signs are not there, the minimum value would always be positive, but when signs are there, the minimum value would be negative as well.  So, if checking summation or envelops from load combinations , then obviously the values would be different.
Kindly let us know if further assistance would be required.

Regards,
Nandeep
Technical Manager, MIDAS


Thanks again Dandeep.

This is what I expected but good to have your confirmation.  This is what it should be for summation.  I don't see why it is the same for envelope.

On the shear force question, for example, I understand when I switch from "No sign" to "Add sign" and check the CBmax2475/CBmin2475, the signs will change and hence the max/min values will switch.  But the absolute values should not change.  Can you comment on this?

Numerically, for member ID48. When I have "add sign" Fz=-5399/822 for CBmin2475/CBmax2475... but when I have "no sign", Fz=4544/7009.

This cannot be true, right?

If true, why and what value should I use  for design?


Further thoughts:

1) To make it simple on the summation, I think we can say this is a matter of taking CQC before summation (which is wrong) or after (which is correct).


2) In my model, I defined a load combo "1.0X+0.3Y @2475" after the analysis and there is no need to rerun the analysis. Presumably, Midas simply combines the relevant results as the results of the combo is there right away.  True?  If yes, it is then taking CQC before summation (in this case, combination) and this would be wrong.  Or is Midas taking CQC after summation (in this case, combination), this should take some computing time and it doesn't seem to.


Looking forward to your comments...



Thanks



Hi, Nandeep,


Another simpler model attached for your perusal.

If you take a look at the reaction, you will see it makes a big difference selecting "no sign" vs "add sign".


Which is the correct one?



Thanks


Hi, Nandeep

Refer to the "Steel frame" model.


It seems to me for load combo "1+0.3", Midas simply adds up the relevant reactions to report the combo reaction.  This is wrong as it is taking  CQC first and then adding up.  Should take CQC last as we discussed earlier. 

Similarly with the other combo"0.3+1".


Comments?




Hello,


Your understanding is correct.  Actually this is pertaining to how the combinations are made.  Right now, the combinations are made with Type as Add.  For combinations, we have 4 options.  Add, Envelop, ABS and SRSS.  We do not have CQC here.  But for the results obtained after CQC of response spectrum, SRSS type combination should be used, if these have to be combined.  This would give results closer to what is expected.
Kindly let me know if you have further comments on this.

Regards,
Nandeep
Technical Manager, MIDAS


Hi, Nandeep.


Thanks for your reply/explanation.  With the  combination type switched to SRSS, I got some results that are seemingly good.


Additional thoughts:

1) By CQC/SRSS, we know the signs are lost in the 1st place.  There is no bringing back.  Giving users the choice of "add sign" seems to be a trick that gives plausible results which are never correct.  Better remove this feature in future releases. Agree?

2) Even with SRSS as the combination type, say for the 1.0x+0.3Y combination, the results are still wrong in theory (may be close enough practically).

Reasons #1, the code requirement of "100% in one direction+30% in the other direction" should NOT be interpreted as SRSS of the results from RSA in x and RSA in y directions, it should be: i)ADD within each of the modes with 2 RS's applied simultaneously and then ii) SRSS/CQC the results from i) between all modes.

Reason #2, this would require some computing time after the combo is defined.  It require to run another RSA load case where 2 RS's are applied simultaneously.  Midas does not seem to have this feature?  Good to have in the future.


The recommendation of using SRSS between RSA results in x and RSA in y directions appears to be a Band-Aid Solution. Agree?

Thanks very much.







Hello,


Glad the earlier suggestions helped!  About the current comments, kindly refer the response below.
1) By CQC/SRSS, we know the signs are lost in the 1st place.  There is no bringing back.  Giving users the choice of "add sign" seems to be a trick that gives plausible results which are never correct.  Better remove this feature in future releases. Agree?
A:  Actually, without signs, the design would be very conservative.  It is a well recognized method on how to add signs in case of RS.  The same has been followed in midas.  This was added on recommendation of practicing engineers.

2) Even with SRSS as the combination type, say for the 1.0x+0.3Y combination, the results are still wrong in theory (may be close enough practically).
Reasons #1, the code requirement of "100% in one direction+30% in the other direction" should NOT be interpreted as SRSS of the results from RSA in x and RSA in y directions, it should be: i)ADD within each of the modes with 2 RS's applied simultaneously and then ii) SRSS/CQC the results from i) between all modes.
Reason #2, this would require some computing time after the combo is defined.  It require to run another RSA load case where 2 RS's are applied simultaneously.  Midas does not seem to have this feature?  Good to have in the future.
The recommendation of using SRSS between RSA results in x and RSA in y directions appears to be a Band-Aid Solution. Agree?
A:  Again, this is something for which different engineers have different approach.  Some prefer linear addition while some prefer CQC/SRSS as we discussed.  We need to provide both options for engineer to choose from as there is no clear guideline in code as to how these combinations are to be done.  They just mention about 100% in one direction and 30% in other, but not about how these need to be combined.
About the second reasoning, i.e, applying 2 RS simultaneously, could you provide further insights/reference on this?  I believe I am clear on the method that you are suggesting, but I couldn't find any technical reference recommending this.  If you could share details and credible reference, I could pursue our development team accordingly.
Awaiting your response!

Regards,
Nandeep
Technical Manager, MIDAS


Hi, Nandeep,


Attached please find another model.


In the RSA's, I have 475x100% without signs and 675X100% with signs added.


Take a look at the deformed shapes and let me know what you think. (one is obviously wrong that yet Midas seems to suggest using.)

Thanks



Hellok,

As we discussed before, Midas can just provide methods that engineers want to use to suite their practical needs.  I totally agree with your observations that with sign and without sign would give completely un-comparable results.  As midas technician, I can only say that both the methods are correct and its up to the engineer which one they want to go for.
However, as an engineer, personally I feel that if I can be sure of the quality of execution, I'd prefer the method to add signs as that tends to give more realistic results, whereas the conventional method without signs tends to give very conservative results.  Again, I reiterate though, that none of the method is wrong.  Its just like how Working Stress Method is not wrong for design.  Its just way too conservative as compared to Limit State Method.  Thats it.

Regards,
Nandeep
Technical Manager, MIDAS
Creation date: 10/4/2018 1:19 AM (dklee@midasit.com)      Updated: 10/4/2018 10:55 AM (dklee@midasit.com)
Files   
Book1.xlsx
DataImage23.png
DataImage24.png
DataImage27.png
DataImage45.png
DataImage52.png
DataImage62.png
DataImage7.png
DataImage72.png
Model 3.mcb
Model 3[1].mcb
RS Sign 1.mcb
RS.mcb
Steel frame.mcb