design one spread footing with three columns

[rigid_joint]

arsap has by default the ability to design one spread footing with two columns.

in a case of three (or more) columns how can it be simulated that?

i tried some simulations with rigid links and for uls combinations things work fine, but not for seismic combinations.

is there any suggestion?

 

[Pawel.Pulak]

What is wrong for seismic combinations?

And why have you defined it using pinned support? Rigid link is already defined to do not fix rotations in slave nodes - so columns are pinned at the bottom. Using pinned support in such case results in some unstable teeter-totter mechanism.

Regards,

[rigid_joint]

i don't know what is wrong. i turned it in fixed support, but still seismic combinations are not working correctly.

in fact, there is a great reduction of the Ex,Ey,Ez results in the 2nd case.

all other loads (g, q) are ok

*****************************************************************************************************************************************

1st example (3 supports)

G summary reactions of the three pinned supports -----> total Fz=33.184

2nd example (1 combined support)

G reaction of the one combined support -----> total Fz=33.184

everything ok in G loads between them

---------------------------------------------------------------------------------------------------

seismic loads

1st example (3 supports)

Ex ------------>  Fz=-662.585

for modal results

2nd example (1 combined support)

Ex  -----> total Fz=-16.737

for modal results

as you can see , same modal results, same uls results, totally different seismic results

------------------------------------------------------------------------------------------------------------------------------

Same thing happens with Ey and Ez.  Extremely great reduction.

But G, Q loads are ok in both models, which means that there is a simulation problem here.

Maybe this happens because in the first case we have 3 supports and in the second case 1 support.?

But we need 1 combined support for the 3 supports, in order to set the spread footing on it.

[Pawel.Pulak]

I cannot be 100% sure without the model but it seems that results for "1 combined support" are correct and not for "3 supports".

The table of reactions in "Sum of val." row displays the sum of reactions for supports currently displayed (filtered) in the table.

In case of seismic load case, when displaying reactions for quadratic CQC combinations, it corresponds to sum of quadratic combinations from specific supports. It means that when calculating quadratic combinations the information about possible opposite sign of reactions from various modes was lost.

It is related to the sequence of calculating combinations of reactions - in case of filtering 3 supports  "Sum of val." is calculated by calculating quadratic CQC combinations from all modes for each of supports and then adding CQC combinations from these 3 supports together in linear way.

What is necessary for design of one support for 3 columns is the reversed order of combinations: first linear addition of reactions from 3 supports for each vibration mode and then quadratic CQC combination of these sums.

I try to show it in the screen capture made for very simple model with seismic analysis. As you can see for each of modes there is no significant resultant vertical component (FZ reactions for mode 1 have opposite signs). But for CQC combination "Sum of val." has significant value for vertical directtion. It is not observed for ""Sum of reac." and "Sum of forc." because these parameters are calculated first adding results for each mode and then calculating CQC between them.

Similar topic was discussed in this forum thread (but there it was related to some issue):

http://forums.autodesk.com/t5/Autodesk-Robot-Structural/Seismic-Result/m-p/3101884/highlight/true#M4... 

If it is still not clear please attach your model.

Regards,

[rigid_joint]

thank you very much. you are very clear (and you remembered me the theory). i' ll check them in my model

[rigid_joint]

which is the right simulation?

or

or something else?

[Pawel.Pulak · Accepted solution]

In my opinion none of them.

It should be fixed support to give non-zero MX, MY moment reactions for the group of columns.

As concerns details of rigid link definition it depends whether bottom ends of columns should be modeled as pinned or fixed to the foundation.

Regards,

[rigid_joint]

yes, you' re right, it has to be fixed.

the column bottom ends are pinned so as the rigid link will be Rx, Ry, Rz free.

I'm a little confused today. thank you again