Need a book? Engineering books recommendations...

Return to index: [Subject] [Thread] [Date] [Author]

Re[2]: modeling of 3d-frame with shearwall

[Subject Prev][Subject Next][Thread Prev][Thread Next]
     More often than not, the manuals that come with the software often 
     indicate the limit or the capacity of the program. Although, I would 
     say, that it does not say so much.
     
     In SAP90 5.18, it limits itself to 10,000 joints or 1,000 equations on 
     640KB memory and 30MB Hard disk.
     
     That appears to be overriden now as a result of the development of CPU 
     capacities and in the program itself.
     
     As to the refinement of the mesh, I would think it should be left to 
     the judgement of the user and the degree of exactness he/she wants to 
     achieve. This, from my point of view, can be achieved with familiarity 
     with the program (read:frequently using it) and from experience with 
     other programs as well. An aspect ratio of 1 is most desirable but if 
     unavoidable should be as much as possible near 1.
     
     Consider, too, the size of your structure. If you have a very big 
     structure and foresee a problem with running the software because of 
     the amount of data it need to process, then maybe its time to break it 
     down to several models. But in so doing, the user should be well verse 
     with the modeling options of the supports to achieved a near "exact" 
     analysis of his/her structure.
     
     Jerome
     
     
     


______________________________ Reply Separator _________________________________
Subject: Re: modeling of 3d-frame with shearwall
Author:  MIME:nanditan(--nospam--at)giasbma.vsnl.net.in at INTERNET
Date:    8/6/98 7:22 AM


This is my second attempt to get a response from members on a subject of 
some importance:
     
To Siamak's query regarding modelling of a 3D frame/shearwall building 
Jerome had suggested the following procedure:
     
>     You must prepare at least three or four nodes to identify the plate 
>     element. Then assign the material properties and thickness. You also 
>     have to identify it if its a membrane or a plate element. For walls, 
>     it should be plate or solid. I usually assign plate.
>     
>     Say you have a 26' long x 12' wide and 20' high (one storey is 3m) 
>     building and the West face is a shear wall.
>     
>     <-- 12'--->    <-----------26' ----------->      <---12'--> 
>     14        13    13/14        15/16       17/18  18        17 
>     +---------+    +++++++++++++++++++++++++++      +---------+ 
>     +         +    +            +            +      +         + 
>     +         +    +            +            +      +         + 
>     +    B    +    +            +            +      +         +
>     +8        +7   +7/8         +9/10        +11/12 +12       +11 
>     ----------+    +------------+------------+      +---------+
>     +         +    +            +            +      +         + 
>     +    A    +    +            +            +      +         + 
>     +         +    +            +            +      +         + 
>     +2        +1   +1/2         +3/4         +5/6   +6        +5 
>
>
>        West Face           North & South Face           East Face 
>        (SWall)
>
>       At the west face, the A (PLATE ELEM 1 ) Plate would be made of node 2 
1  
>       7 8; the B (PLATE ELEM 2) plate by 8 7 13 14. You can assign the 
>       thickness of, say 10" and E260 KSI
>
>        So the input would look like : 
>
>        SHELL
>        NM=1
>        1 E = 3300
>        1  JQ = 2 1 7 8 TH = 10
>        2  JQ = 8 7 13 14 TH 
>
>        You ca then attach your beams/girders and assign your columns. 
>
     
I had raised my doubts about the procedure with the following questions:
     
>	1. Is it all right to divide a shear wall into plate
>          elements which are one storey high and one bay wide 
>	   or should a more refined mesh be used ?
>
>	2. Is it all right to connect beam elements (13-15) and 
>	   (7-13) to the plate element (7-8-13-14) ?
     
Jerome responded with the following answers :
     
>     A more refined mesh would be better. The resulting stresses are 
>     reckoned from the c.g. of the element. So, if you have more elements, 
>     you will be able to get near "exact" values for every area of wall. 
>     That would mean , however, more time for data preparation and run-time 
>     for the analysis. But I can probably sacrifice that.
>     
>     You can attach the beam elements to the plate and model the 
>     connections as you deem necessary. There is a feature in the program 
>     that allows you to release any of the DOF's you want.
>     
>     So far, I have not tried modeling a shear wall embedded with columns 
>     yet. But what you can probably do, to idealize it, is to assign it as 
>     a solid or plate element with a different thickness.
     
     
I agree with Jerome's answers and would like to add that when one attempts 
to attach a beam or a column element to a shearwall using plate elements, 
it will be wrong to use (storey-high x bay-wide) elements simply from the 
consideration of an acceptable aspect ratio. Things are even more 
complicated if non-planar shearwalls are present.
     
This leads to the following questions :
     
1. What is an acceptable level of mesh refinement?
     
2. If the mesh size is going to be unmanageable, is it advisable to use
   plate elements at all?
     
     
Rudra Nevatia
Central Computing Facility