What Would Be The Proper Choice Between Shell And Frame For Modelling 1.2M X 0.6M Vertical Element?

[WR1]

What would be the proper choice between shell and frame for modelling 1.2m x 0.6m vertical element? 

 

In a building I am studying expansion joints. All columns are 600mm dia circular. Except some vertical elements 1200x600mm.

 

1) The question is what system for seismic behaviour you would consider as per ASCE? shear wall or frame? 

 

2) How would you model 1200x600 as frame or as shell?

[UmarMakhzumi]

1200x600mm should be treated as column; It has a noticable weak axis response (espically as other members of LFRS are about same size in one axis).

I don't know about your framing and column count, but % of baseshear resisted by rectangular columns would be greater in one axis and almost equal in other. If you want to get more classy, compare % of base shear distribution ratio between your rectangular and circular columns to get a feel but with the information provided at hand, best bet would be to model it as a frame element. This answers your part 1 too.

Thanks.

[EngrUzair]

Aoa.

 

Keeping in view ACI 318-05 section 21.4.1.2 for special moment frames, a vertical member with 1200x600mm cross sectional dimension should be treated as a "column". Therefore,

 

1) a "frame' system should be adopted for analysis,

 

2) the 1200x600mm section should be modelled as a 'frame' or 'column'.

[WR1]

Uzair I know the clause you are talking about. But it is for special moment frames whereas I have ordinary. 

 

Umar, these large rectangular columns attract 70% of base shear when modelled as frames and 95% when shells. What would you do when you have 10mx20m large bulky column? Still model it as frame? Whereas in reality it has continuous support like compression tension couple making the base fixed?

 

I know we can model 200x200mm column as frame and then longer things as shells and then a mass by 3d brick elements. So whats close to reality? 1200mmx600mm vertical element as shell?

[Muhammad Imran Zafar]

Frame element

[UmarMakhzumi]

Uzair I know the clause you are talking about. But it is for special moment frames whereas I have ordinary. 

 

Umar, these large rectangular columns attract 70% of base shear when modelled as frames and 95% when shells. What would you do when you have 10mx20m large bulky column? Still model it as frame? Whereas in reality it has continuous support like compression tension couple making the base fixed?

 

I know we can model 200x200mm column as frame and then longer things as shells and then a mass by 3d brick elements. So whats close to reality? 1200mmx600mm vertical element as shell?

I am not sure that why you are getting different % of base shear for case of frame elements and shell elements. The analysis should give you the same results for both cases as stiffenes for frame and shell element wouldn't differ.

In general, the only advantage of modelling shear walls using shell elememts is that you can easily model wall openings and shell elements are much easier to model for a given length of wall compared to frame elements. In this case, I doubt if there are any openings.

Anyhow, Walls have 2D behaviour and Shell or Frame elements should suffice from analysis point of view. Once you get the forces you can resolve them in max compression and tension forces when designing for foundation or reinforcement in walls. This answers your "10mx20m large bulky colum" wall question too. I could see why you are asking this question.1200 x 600 mm columns is not that big.

Thanks.

[WR1]

Umar, there is a difference between stiffness of shell and a line element due to Poisson ratio although negligible.

 

I am getting higher forces because the line element was pinned and when modelled as shell...it is also pinned but now has 3 pin supports instead of 1 in line...so these 3 pin supports making the shell more stiff at base.

 

PS: This explanation is not mine..so I am not satisfied with this too.. because as I know the lateral force in rigid diaphragm should depend on stiffness of vertical elements and stiffness of line and shell dont differ much so the the base shear in these columns should not jump from 70% to 95%. Agreed. Btw does the above explanation about base fixity has any effect? I dont think so but just asking?

[UmarMakhzumi]

No, i don't agree with the explanation. You can mesh the wall finer and increase the number of supports from 3 to 6 and you should still the same base shear. Try it.

Thanks.

[Muhammad Imran Zafar]

Uzair I know the clause you are talking about. But it is for special moment frames whereas I have ordinary. 

 

Umar, these large rectangular columns attract 70% of base shear when modelled as frames and 95% when shells. What would you do when you have 10mx20m large bulky column? Still model it as frame? Whereas in reality it has continuous support like compression tension couple making the base fixed?

 

I know we can model 200x200mm column as frame and then longer things as shells and then a mass by 3d brick elements. So whats close to reality? 1200mmx600mm vertical element as shell?

Dear waseem As per ACI code 318-14, Horizontal dimension to thickness ratio is less than 3, so it should be treated as column. In my opinion it should be modeled as frame element

[WR1]

Well I dont believe in aspect ratios thing. ACI also clearly mentions in this clause commentary that it is the behaviour that determines if its a column or a wall not just merely the aspect ratios. Anyway one good thing to check is the moment diagram.

[Muhammad Imran Zafar]

Well basically code considers a member as compression member when primary stress is longitudinal compression. The differentiation between column and wall is based on the principal use However code permits walls to be designed using principles stated for column. With that just to differentiate arbitrarily code defines columns basing on aspect ratio. In combination these two aspects support the point of view that member under your consideration should be treated as column and may be modeled as frame. Moreover moment diagram may not be a decisive factor in your case because what i got from discussion your member is a compression member. So the question is how to model that compression member.

[WR1]

Thanks Imran Zafar, I will study this in detail and will update if I find something new.

 

 

Sir Umar,

unfortunately what we discussed (that shell and frame should give same base shear in those elements) is not matching with my models. I checked models again with Poisson's ratio = 0 so shell = frame stiffness. But still I am surprised to see that those few elements when modelled as lines carry only 329 kN shear and when modelled as shell carry 922 kN shear. What could be the reason?

 

Line elements have one pin support at base. Shell elements have 3 pin supports at base.

 

Now if I change shell pin supports from 3 to 1 (keeping the middle pin support) the shear on shells is almost the same as that of frames. It is now 342 kN (very near to 329 kN of frame elements)

 

So the base fixity is doing something here? What you say?

[Muhammad Imran Zafar]

Please go through the following document. It may help in analyzing the two different approaches and then you can check your model accordingly.

 

http://www.iitk.ac.in/nicee/wcee/article/14_05-01-0524.PDF

[UmarMakhzumi]

Thanks Imran Zafar, I will study this in detail and will update if I find something new.

Sir Umar,

unfortunately what we discussed (that shell and frame should give same base shear in those elements) is not matching with my models. I checked models again with Poisson's ratio = 0 so shell = frame stiffness. But still I am surprised to see that those few elements when modelled as lines carry only 329 kN shear and when modelled as shell carry 922 kN shear. What could be the reason?

Line elements have one pin support at base. Shell elements have 3 pin supports at base.

Now if I change shell pin supports from 3 to 1 (keeping the middle pin support) the shear on shells is almost the same as that of frames. It is now 342 kN (very near to 329 kN of frame elements)

So the base fixity is doing something here? What you say?

There are a number of things that require discussion here:

I am not sure if "Base Fixity" is the right term to use but clearly what is happening here is that with at least 2 base pin supports(the default when we model a wall in ETBAS) the shell elements "behave" as a true wall. Unlike for the case of single pin support where lateral resistance is through moment resistance only and the action is that of a frame instead of a wall.

To explain further, assume a model in which only shear wall is modelled(no other structural elements) by using a single pin node at the base. The structure is unstable by itself and any lateral load would cause the wall to rotate about base. Thus for situations in structural systems where shear walls are modelled using a single pin at base, we are enforcing a "frame action" rather than a wall action. That is why your base shear per support is coming close to that of column but slightly higher, the reason for which is given below in 4th paragraph.

If you increase the number of pin supports at base to more than two, you would not see any significant increase in the amount of base shear attracted by the wall. As already stated this is because minimum 2 nodes are required to enforce wall action any any further increase is meaningless other than offering a more even load for foundation design.

In addition to this, I would also like to mention the case when you modelled a single pin under wall and obtained a slightly higher base shear than that to when columns were modelled (342kN vs 329kN). The reason for this is the reduced length of the beams when walls are modelled result in a slightly stiffer system and increase shear per column.

Thus base fixity means nothing. The real jab lies in how ETABS calculates stiffness of shear walls and frames and interaction of shell and frame elements. What is apparent to me is that shell elements when modelled and meshed represent discrete structures rather than individual component of a structure.

For interest, you should compare the stiffness of single cantilever column to that of a cantilever wall fixed at base. Apply a 1kN load to both and note the deflection. Are results different. I bet they would be. Apparently ETABS might be using some assumptions related to wall geometry to justify a higher stiffness calculation.

Regarding your model, I would still suggest to model vertical elements as columns as your system is way too redundant than a shear wall system; I don't know about column count but naturally your number of columns would be >> than a typical wall system and I believe applying wall provisions in this case would be overkill.

Thanks.

[WR1]

Thanks, yeah thats a good point about reduced beam length in case of shell element or wall.

 

No I no. of columns is not more than these shell elements. They are almost equal. Ayway I am using frame resisting system but the question now is between shell or frame.

 

For more than 2 supports to capture wall action, let me verify this in a an example.

 

And for a simple cantilever shell vs frame, the result is same. I have done it previously many times, the only difference is that a shell is more stiffer than a frame of the same size because of possions ratio. As for shells flexural rigidity is EI/v instead of EI in case of frames.

[UmarMakhzumi]

Use frame elements, period.

[WR1]

Well, yeah I verified that if shell element has more than 1 support it again takes about 95% of total shear.

[WR1]

I am facing a similar situation again. What exactly is the difference between multi-storey frame with columns modelled as line element, and the same frame with same sizes with columns modelled as shell elements.

[UmarMakhzumi]

Rana,

 

Lets do a simple experiment. How about we model a cantilever beam with frame/line elements and then model the same beam as shell elements and in both cases we just assign one reaction support.

 

Thanks.

[WR1]

I would not have posted if it was that simple..anyway...im exploring it...when i will get an answer ill update

[UmarMakhzumi]

Actually I wanted you to do a comparison but never mind, I am doing it in SAP2000. I will post what I find.

 

Thanks.

[WR1]

i have done all the experiments before posting and the results are;

 

we will get the same displacements if we

 

1. model a cantilever column with fixed support

2. model a shell element of same size with pinned supports (more than 1 support will act as fixed) and dividing total horizontal tip load as work equivalent load with poison ratio = 0.

 

 this is logical.

 

now lets move on

 

if we model a frame with columns as line elements, and then replace these columns with shell elements of same size with above assumptions in step 2 we will get different results.

 

reason behind is the joint stiffness where beam frames into vertical elements. to get same results adjust the beam in shell models to the full length of shell elements (making joint more stiff).

 

this is again logical

 

now move to next step

 

replicate these models for 10 stories, you will need to adjust the beams framing into shell elements to get the same results as that of multi-storey plane frame.

go so forth

 

now the problem lies when you actually model a 3d building...with line elements as columns, and then using shell elements as columns of same size. thats where im stuck.