1
The global averaging is IMHO ruled out. Mind that by using it you will include any 'additional' panel attached to the core wall in the equation and it will also influence results by adding up 'different' forces (imagine automatic direction and neighboring panels with different direction of local X axes while you open one of them in a new window to see only these results.
In your example this even changed the sign of forces:
Artur Kosakowski
Hi @Refaat
My personal conclusions from the test I made in the process of this discussion are that the best option seems to be to look at the not averaged (no smoothing) values of cuts done at the centers of the outer rows of surface elements. IMHO for this approach the size of the mesh should not be a critical factor.
I hope this helps.
Artur Kosakowski
Again I am saying Please dont make it related to mesh size It will have big effect on the results
Could you please Ask them to make it above or under the edge by 1 cm .
Hi Mustafa,
Thank you for your reply .
1. But do you think that is correct . The two shear forces (Nxy & Qxx) could be in the different directions since you have only supports at the bottom of core wall and all walls have the same stiffness as well all forces in case (1) applied on the core wall purely in one direction.
2. Why did you consider the sign of (Qxx) while as didn’t consider the sign of (Nxy) ?
3. As you see in the capture below (Artur’s capture) he added both forces and he used reduced forces instead of integral value since in some cases are not same values.
Thank you
Refaat
Dear Artur
Thank you for your always response .
Unfortunately RSA gives me 
different results when I change the mesh size even when there isn't any smoothing ( as in the captures ).
1. Please , could you point me to any rules to help me for mesh size selection ?.
2. May I know which value ( integral value or reduced force) shall I use it to calculate the applied forces and does the sign of force should be considered or not ? please , example .
I really appreciate some detailed help for this important issue.
Refaat
Hi @Refaat
I'm of the opinion that it makes more sense to compare results on the main force transfer direction which is along the shear walls rather across them. Looking at the test for double denser mesh the difference is IMHO not significant (less than 1%)
The cut is done at the center of the first elements of the mesh with no smoothing.
If you find your post answered press the Accept as Solution button please. This will help other users to find solutions much faster. Thank you.
Artur Kosakowski
Hi @Refaat
1. But do you think that is correct . The two shear forces (Nxy & Qxx) could be in the different directions since you have only supports at the bottom of core wall and all walls have the same stiffness as well all forces in case (1) applied on the core wall purely in one direction.
Yes this could be especially its finite element core and the result above show you that the integral value of QXX Comes sometimes +sign and sometimes - sign depends on the smoothing option the one you use any way it is small value so its still acceptable.
2. Why did you consider the sign of (Qxx) while as didn’t consider the sign of (Nxy) ?
It is considered and its related to automatic direction of the panels which is opposite of y global direction.
3. As you see in the capture below (Artur’s capture) he added both forces and he used reduced forces instead of integral value since in some cases are not same values.
Look at all the diagrams posted by me or Artur you will see the integral force of QXX have different sign So sometimes to be added and sometime need to subtract .
Anyway The QXX value ir really small compared to Nxy what is the important is That total value of reduced force should be very closed to actual not far.
This is for Reduced force the walls perpendicular to the direction will have small contribution of shear force but for reduced moment it will take more than the other walls (which they are parallel to the y direction ) So we cant take only 2 walls we should take the core wall as one element and one cut.
There is no question about that but my point was solely for the reduced horizontal force value vs. mesh size accuracy. If I understand the situation correctly the reduced bending moment will be based on vertical force distribution in these 4 walls instead.
Artur Kosakowski
Dear Artur
What is about when we have a core wall and two sides of it exposed to across shear? The mesh size will be effect on the results as appear in my captures of post 25.
I look forwards to hear your reply of the following inquiries :
1. Please , could you point me to any rules to help me for mesh size selection ?.
2. May I know which value ( integral value or reduced force) shall I use it to calculate the applied forces and does the sign of force should be considered or not ?
I really appreciate some detailed help for this important issue.
Refaat
Hi Mustafa , Artur , Refaat ,
As I only have an idea of how the program works ,I like (and need) to experiment . What really caught my attention on this thread was the very wrong ,almost 10% error (although on safety side) on the value given for FY by "diagrams for buildings" . It was supposed to be 1200kN and the result given was 1300kN.
For the original file , MX result is 2.2%(against safety) wrong which I say would be somewhat acceptable . Perfect result would be 23400kN.m
Then I tried changing the linear pinned wall supports into fixed. FY results were almost perfect , although MX results remained 2,2% wrong
Triple Refining ( this fixed model) only the last row of FE´s gives an almost perfect result with only 0,7% error. Double refinement would give a 1% error
Which for me would be a pretty acceptable results!!!
Of course you can get the perfect result on the base level using an workaround ,setting a central fixed node with full rigid links to all core wall bottom nodes.
Of course there´s the argument of changing the linear support from pinned to fixed. But in fact the difference is very negligible because even with pinned support the "fixed" effect is given by the restrained core wall retangular "section" . Comparing the Y displacements among the 3 different models this becomes evident.
As expected, linear fixed and fixed central node with rigid links give they same results .
So for me , this would be an acceptable way to get good results as I don´t think refining one row of FE´s would have a great impact on model performance .
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
Hi @Refaat
What is about when we have a core wall and two sides of it exposed to across shear? The mesh size will be effect on the results as appear in my captures of post 25.
This is exactly what I tried to focus your attention on. In your picture in the [post 25 you showed the part of shear taken by across walls (30 or 40 kN) whereas the along one take over 1100 kN. If so the accuracy due to mesh size for the across one is less important than the along ones for and for the along one the test made shows little dependence (which is good).
I look forwards to hear your reply of the following inquiries :
1. Please , could you point me to any rules to help me for mesh size selection ?.
I haven't made any deeper investigation on this subject but I think that for the global effects such we are discussing the mesh size is not the critical factor (and this is a sense was proven in my test scenario for two times smaller elements). I run a quick google search and found this:
http://peer.berkeley.edu/tbi/wp-content/uploads/2010/09/PEER-ATC-72-1_report.pdf
See 4.2.6.2 Influence of Mesh Size
2. May I know which value ( integral value or reduced force) shall I use it to calculate the applied forces and does the sign of force should be considered or not ?
I believe I answered this
My personal choice would be integral value with no smoothing at the center of the elements.
I really appreciate some detailed help for this important issue.
Refaat
Artur Kosakowski
Hi @Rafacascudo
You are very right about the support type.
But For refining the meshing I believe this is not practical because the core wall is not designed and one uniform section with base reduced forces.
Sometimes walls stops at certain levels . Sometimes thicknesses of walls change at certain levels.Sometimes there are outriggers connected to the core at certain level. Sometimes Designer want to change the reinforcement every few stories.
Do you think this easy for such core wall (this is only connecting elements in one section and there are many from the other sides )
Of course nothing is easy on a model like this giant building , but if you require such a precision , then you should try refining the mesh, at least , at the important spots.
I am pretty sure you are very able to do it fast and correctly
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
Hi @Anonymous
Thank you for your clarifications.
I am really convinced that the both shear forces (Nxy & Qxx) in your case should be always added (Not to be subtracted) . Because both of forces in the same direction. But the matter is how to select the suitable smoothing option for the case . We have to know precisely the effect of each of averaging settings in the analysis results . I completely uphold Artur’s suggestion to use (No smoothing) to avoid any conflict in the results especially when we have complete structural model ( slabs , walls ,…etc).
Refaat
What is the conclusion to be done by development team.
Hi All,
For me as A conclusion As we know that RSA is very rich software (many options giving you very good flexibility to choose what you want ) So I would like to Ask RSA development Team this issues :
1- Similar to panel maps and Panel cuts in order to solve this issue is to include the option of normalizing type on core walls diagrams as shown in the picture and to leave the choosing between of them to design engineer to choose the most suitable for him and This option also to be activated for calculating the reduced forces and moment in panel cut .
2- Adding another option allows you to specify where the most top and most bottom cut locations this option contains the following choices:
1-At center of top and bottom finite elements.
2-At top and bottom Edges.
3- 1cm above and under of lower and top edges.
I believe this will be the best solution..
One thing more which is also important
presentation of the core wall result when we choose one core wall and take to new window then view its results
The result will appear for all core walls which is should be present only for selected core see attached picture
Waiting for the feedback .
Many thanks.
What is the conclusion to be done by development team.
This situation is still under investigation but you will be the first person I will notify the moment I have got any new information
Artur Kosakowski
Thank you boss noted .
