Load Takes Difference

[AmeetEWP]

Hi,

I am modelling a very simple and symettrical 8 storey high R.C frame structure (square on plan) in Robot Structural Analysis 2015. The building has 275mm thick r.c slabs that are supported on columns arranged on a 3x3 grid spaced 5m apart (total 9No. columsn per floor). All columns have the same section size/properties as does the slab throughout the building.

The only load that is applied to the model currently is SELFWEIGHT.

We have produced two of the same models, one comprises indivdual floor slabs and the other comprising the full building model. We have been comparing the load take down results between the two models, so with the indivdual slabs we have been adding up reactions from each model and comparing it to the full building model. On comparing the two, we have found significant differences in reactions at GF level.

I have attached a screenshot showing the differences between the two models and the reactions that have been calculated.

Can you comment on what may be causing the difference.

Thanks 

[Rafal.Gaweda]

Check this:
http://forums.autodesk.com/t5/robot-structural-analysis/load-sequencing-construction-stages/m-p/4570...

If it is not the case please send us model

[AmeetEWP]

Hi Rafal,

Link below contains our model. I have included two folders, one contains the full building model and another for seperate floor slabs to form the comparrison.

A quick reminder. When we do a load take down to detemrine foundation loading, the values vary betwen the full model and the individually combined models.

My personal email if easier is: A.Mistry@elliottwood.co.uk

https://elliottwood.bigfilebox.com/lwt/88840-I4Odrt2Dk3Q1vUrwv8IexHDjJ

Thanks

[AmeetEWP]

Hi Rafal,

Can you confirm you have seen my previous message where I sent the link you thr link to our online Big Box file which contains our ROBOT files.

Please disregard all other messages I have posted on this forum and focus on this one only.

Many thanks,

Ameet Mistry

[Rafal.Gaweda]

It seems it is related to combination of two effects.

Effect 1 described here : http://forums.autodesk.com/t5/robot-structural-analysis/load-sequencing-construction-stages/m-p/4570...

Effect 2 : not all columns transfer loads on columns which are located below them

[AmeetEWP]

Hi,

I dont think this issue is related to phasing or construction phasing at all.

With regards to columns not taking load below, we have checked the full model which includes all 8 floors. We have done hand calcs to check that the reactions that ROBOT gives us is within an acceptable and realistic degree. We are happy with this. Further we have two different ROBOT models on which we have noticed this difference, so we are certain its not a modelling problem where a column does not take load. Also ROBOT display axial load increasing at each level going down the building.

The issue here is that we are getting two difference sets of results for something which should be the same. Whether we model individual slabs and collect the load under each column or model a full building structure, the total load under the columns in both models should be similar. This is not the case, and we are seeing significant difference. We have thought of factors such as column shortening, but for a building of this scale, with ONLY selfweight applied, this should be the case.

We are carrying out this exercise to determine foundation loadS. It is difficult to decide which model we should follow. The full building model shows perimeter columns to take more load than when modelled as individual slabs. However it also then shows interior columns to take less load compared to individual slabs. So which ever model we go with, atleast one of the column load types will be less than when modelled differently.

I look forward to your response

[Rafal.Gaweda]

Let's start from the beginning.
Model you described in first post and in pdf has nothing to do with models you have sent.
If you want us to check this model please send it.

If you want us to check model(s) you have sent please write precisely which support(node number) or column (bar number) to check.

[AmeetEWP]

Hi,

Apologies for the confusion. Hopefully this will clear it up.

We will focus on just the one model, which is the one we are designing – 8 storey R.C frame, with transfers at 6th, 7th floors. I have attached relevant ROBOT files and screenshots that will need to be referred to with the information below - see download link at the bottom of this message.

Differences in reactions

If you refer to the document attached “Reactions – Full Model Vs. Individual Slabs”, on page 1 I have displayed screenshots from ROBOT when I modelled slabs individually to collect reactions under columns at each level. Where we have transfers (i.e. 6th, 7th and 8th floor) I have modelled them together. At the bottom of the page you will see I have added up all of the reactions to determine the “would be” foundation loads. On page 2, I have displayed screenshots from ROBOT when I modelled the FULL building structure to collect reactions under columns, again at the bottom I have listed the “would be” foundation loads. On page 3, I have displayed a table which compares the “would be” foundation reactions that I got from individual slab models being added together and that from a full building structure model. You will notice there are significant differences in “would be” foundation reactions. It is this difference that I am confused by. What is likely to cause it?

Differences in axial force

The second area of concern, which is very much the same as the one above is a difference in axial load/punching shear load. If you refer to the document attached “Axial force – Full Model Vs. Individual”, I have displayed screens shots of the two models; full building model and isolated 6th,7th,8th floor model. Here we are looking at the axial load/punching shear load at the head of the column at 6th floor level. You will notice that the load is different in both models. This difference is noticeable at other column heads too. Again it is this difference that confuses me. What is likely to cause it?

Note the individual slab models originate from the full building model file. I used the full building model file and one by one separated floors to carry out this study. So the loading, properties, geometry etc is identical.

I would be grateful if you can review our models and let us know what is likely to cause this difference, and perhaps which modelling method is correct. Link below with ROBOT files and attachments

https://elliottwood.bigfilebox.com/lwt/89084-1E3gqNmuVd8wwrnVfcccwtVNJ

Many Thanks.

Ameet

[Rafal.Gaweda]

Differences in reactions

I think this is effect of plate stiffeness. In "load take down" approach ti is not considered.

See my results for SW only below. Excel table results = "load take down" + reactions in 4D and 5D nodes depending on rigidity of slabs. (Stiffeness reduction applied for all slabs \ all 3 slab thicknesses). So your decision is which apprach to consider or maybe both \ envelope...

Additionally I found error in your model : columns 35 147 153   not connected properly with upper columns (nodes 160 and 17969  should be in one place)

[AmeetEWP]

Hi,

I am a little confused by the results you have shown below. Is it that you have reduced the stiffness of the floor slabs in the full building model and as a result the reactions get closer to that calculated from the summation of reaction from individual slabs?

I am also confused why we would need to alter the stiffness, or reduce it as you have? Should load take down not consider stiffness as this is more realistic compared to a traditional hand calc or individual slabs which works closely on contribitory areas only?

Could you please clarify this and your reason for reducing stiffness. It is our understand stiffness should be kept as it is naturally in all models as this represents realistic stiffness behaviour.

I look forward to your response.

[Rafal.Gaweda]

 Is it that you have reduced the stiffness of the floor slabs in the full building model and as a result the reactions get closer to that calculated from the summation of reaction from individual slabs?

Yes.

I am also confused why we would need to alter the stiffness, or reduce it as you have? Should load take down not consider stiffness as this is more realistic compared to a traditional hand calc or individual slabs which works closely on contribitory areas only?

Exactly.

In load take down approach (summation of reactions) slabs stiffeness on which the columns "stand" are not considered because they are not existing in these models.

Simple models : level 4 only reaction + level 5 only reaction and compare with one model with these two levels (4+5) toghether and you will see the difference also.

[david.woodsZMDU3]

Hello, I am having the same problem. I have designed floor slabs from individual floorplates then I have added the reactions together for an accumulated load of the floors combined. I have also modelled the building in its entirety and compared the loads. After 9 levels the ultimate load for all the columns combined is approximately 10,000 kN greater than that of the full model (without providing the stiffness reduction suggested). By including a slab stiffness reduction of 99% I get loads similar (but short of) those taken from the accumulation of floor plate reactions. Is this the way we should produce loads for all structure modelled in Robot? How can we justify the slab stiffness reduction to building control? Can the loads from the full model without the slab stiffness reduction be used or would this be erroneous? Thanks in advance for your swift reply. Dave