Shell 1

[stefano74]

Hi all,
I'm struggling with shell elements, I read in this forum that it would be better to use simple models such as frames. However I'm giving a try, and this is my experience:
- I'm trying to link 2D shells with frames. Think about a balcony, linked to a beam and columns.
- I found the results very dependent on the mesh size.
- I found unphysical results when I use triangular elements.
- I found discontinuities within the shells.
- Even if I average there is a large V23 shear stress, which as far as I know it is unphysical (all the tex book say that the shear can be neglected).
- What's the difference between thick and thin plate?
I really would like to understand how to use this elements.
If someone has had similar experience and can suggest solutions it would be great!

Thanks

 

[popabf]

- I found the results very dependent on the mesh size.
- I found unphysical results when I use triangular elements.
- I found discontinuities within the shells.

Of course the results depend on the type of meshing. All joints formed by meshing the shell elements must be in connection with the joints resulted from meshing the bar elements. If not, discontinuities appear.

 

[myabc]

For thick plate, the effects of shear deformation and rotary inertia should be included. While those effects are neglected for thin plates.

Usually, try to divide the shell element like "square element".

Ta

 

[Diquan]

First of all if you are going to use shell element you need to know the basics of "finite element analysis" (FEA) if you don't know then dont' use it!!

1)To link the joint of the shell to the joints of the frame, select the shell then the frame (this is for the entire floor, not just that part) then "assing-> joint-> constraints -> on type of constraint select diaphragm then click on add new constraint, put a name check the Z axis if the gravity goes in the Z direction, and click OK on everything

2)Smaller the shell better results BUT takes more time to analyze

3)When you learn FEA you will see that the easy shape shell element to analize is a triangular shell because it has less degrees of freedom, so the result difer from a 4-nodes shell

4)When you do a mesh try that all the joints near the shell are all interconnected to transmit the forces (thats the discontinuities)

5)Form the V23 stress the theory for this complicated (more to me because i speak spanish not english) BUT depends on the analysis that your are doing of course the shear stress is big because is a cantilever and the moments, shear forces are big and this depends on long is the arm (lenght of the cantilever)

6)The difference is the type of analysis, read the SAP manual

All i can say is for shell elements on SAP you have to understand very well what the program is doing for that first and very important LEARN FINITE ELEMENT ANALYSIS and then READ THE SAP MANUAL, SAP is very complex FEA analysis and to understand what the program is doing you have to learn and read

Diquan

 

[popabf]

To link the joint of the shell to the joints of the frame, select the shell then the frame (this is for the entire floor, not just that part) then "assing-> joint-> constraints -> on type of constraint select diaphragm then click on add new constraint, put a name check the Z axis if the gravity goes in the Z direction, and click OK on everything

I would dissagree with you on this one. there is no need to insert constraints in the analysis when we want the program to transfer the forces from the shells to the beams. For this is necessary to be carefull on the conetion fo the joints (every joint resulted from the meshing of the shell must be conected to the joint resulted from the meshinf of the beam or column).
Constraints are used to simplify the analysis by not using shells.

A Diaphragm Constraint causes all of its constrained joints to move together as a planar diaphragm that is rigid against membrane (in-plane) deformation. This constraint can be used to model concrete floors (or concrete-filled decks) in building structures, which typically have very high in-plane stiffness

FROM SAP2000 manual: only for concrete floors (or concrete-filled decks) in building structures, which typically have very high in-plane stiffness. A balcony does not have that. And a floor bends under the loads.
If you have another opinion, please write...we all learn from others.

 

[masomenos]

Constraints are used to simplify the analysis by not using shells.

This is not correct for many cases. Constraints connect two or more joints in any or all degrees of freedom depending on the type of constraint. Constraints account for the moment differentials resulting from the joint offset distances. Rigid diaphragm is just one type of constraint. Assigning joint constraints or frame insertion points (which automatically creates joint constraints) might be required if the slab was very thick and/or if the beams were deep, in which case, the centerlines are far enough apart that the analysis results might not be accurate enough if connected at centerline joints.

Try and imagine a column coming into a 4 foot thick basemat modeled as a shell element. You might want to draw the column ending 2 feet away from the shell centerline, then connect the column joint the the shell joint using joint constraints (usually body type) to obtain accurate column moments acting on the basemat. Would it be "accurate enough" to connect beams, columns, and/or shells at centerlines? That's an engineering judgement call, depending on what you're trying to do.

 

[popabf]

I agree that are many different situation where the engineering call ca make you take the judgement to use constraints. But I don't think the floor of the balcony is a situation that require the use of them or that are not necessary to connect joint of the "floors" to beams/columns. It depends on the type of floor, as you said. if you use every time the constraints, the moemnts obtained on the "floors" (shell) will be much smaller, and therefor you'll obtain much smaller reinforcemnt area necessary to the floor.

 

[masomenos]

if you use every time the constraints, the moemnts obtained on the "floors" (shell) will be much smaller, and therefor you'll obtain much smaller reinforcemnt area necessary to the floor.

In most situations, the only constraint being assigned to a floor(modeled with shell area elements)is a rigid diaphram type, which acts on the in-plane forces only, not on the moments which would be used to calculate slab reinforcement. You're correct in that some constraints can potentially cause other problems that the engineer needs to be aware of. For example, if you use rigid diaphram constraints on floors, you can't check local axial forces in beams, or accurately check floor reactions in irregular shaped buildings either. Rigid diaphrams may also give you less accurate fundamental period, particularly with irregular structures. I agree that in some cases, particularly floor systems and walls, the engineer needs to consider the option to not use constraints such as rigid diaphrams, but instead use actual FE element stiffnesses based on material and section properties(aka 'flexible diaphram'). It's the non-centerline connections where I think joint constraints/frame insertion points have the most application and benefit. If beam joint connections are modeled more realistically using insertion points or constraints, that may affect calculated reinforcement, but it should be more accurate and realistic.

Consider these situations involving the possible application of joint body constraints or frame insertion point assignments: A thick floor slab modelled with shells with deep rectangular or Tee beams modeled as frame elements - If you model beams and floor all connected at centerline joints instead of using frame insertion points and/or constraints to "move" the beams below the slab centerline to where they are in reality. In many cases, this will affect calculated moments in the beams significantly.. not uncommon to change the amount of required calculated reinforcement in the beams by 15% or more. It may also result in your floor slab being more flexible since beams are no longer modeled on the slab centerline. But in such cases, use of insertion point frame assignments or constraints gives you more accurate reactions which can affect the design. Of course, it takes a little longer to model with constraints/frame insertion points, but it's still much faster than modeling these effects with with "dummy" rigid frames (dummy frames can look bad graphically too) or 2-point rigid links. Again, an engineering judgement call as to what is accurate enough.

Another situation which was discussed on another thread is in 'stepped' column configurations where you have concrete columns of different sizes (larger columns at bottom, smaller ones at upper stories) lining up at a common face of the building. In reality, this results in moments being cranked into the structure as a result of these non-centerline column alignments. How significant are those moments and to what extent are they resisted by the floor systems? It depends on the structure. But in order to accurately consider those effects, you need to use frame insertion assignments to 'move' I and J ends of the selected columns, in which SAP will automatically assign joint constraints internally.

This is a good discussion on use of constraints. Too bad it's in a thread entitled "Shells".

 

[jackyyau]

so we have done the shell anyalyses and obtained all the internal force resultants M11, M22, M12, V13, V23, F11 etc

How to proceed to design the shell slab, wall orll-beam in accordance with the design codes like BS8110 etc?

I found there is a gap between the design codes and shell analyses results.

 

[masomenos]

jackyyau, SAP2000 does not design shell elements (except for the new general concrete design feature under Display>Show forces/stresses>shells, which I'm not familiar with), only frame elements. But I see that my copy of SAFE says that BS8110 - 1989 is available for slab and basemat design. If you're stuck with just SAP and don't have access to SAFE, which is a specialized program for this kind of design, I suggest that you familiarize yourself with the SAP 'section cut' feature, which will give you resultant moments for your shells that you can use for design.

 

[jackyyau]

SAP2000 section cut may provide the resultant moment for a simple salb. But it fails to display the resultant or the figure when the cut is through shell elements of complex geometry for example a wall of a oval-shape shaft.

Also the resultant is difficult to understand.

From the shell analyses, the design moment may be conservatively taken as:

M1=abs(M11)+abs(M12)
M2=abs(M22)+abs(M12)

Can someone tell the difference between F11 and N11 in SAP2000.