Topic: Query reg. definition of FE mesh


In modelling a beam in 2D, I want to alter the type of element used in FE mesh. I want to try 6-noded isoparametric triangular and 8 or 9 noded isoparametric quadrilateral elements.

In the definition of macro elements I chose mesh type as "quadrilateral". Now the further options available for quadrilateral elements are 'CCIsoQuad', 'CCQ10' and 'CCQ10SBeta'. I have gone through the explanations in ATENA Theory manual and ATENA  Input file document. Based on what is given, isoparametric elements are given as CCIsoQuad<xxxx> and the quantity in <> specifies no. of no. of Gauss integration points in the element.

I guess, I should be choosing the option 'CCIsoQuad'. But the options appearing in the window of macro elements doesnt give an option to choose no. of integration points in the element ( <xxxx> ). How do I do that?? By default is it linear or quadratic ??


Re: Query reg. definition of FE mesh

Dear Radhika,
1. what advantage do you expect from using quadratic elements?
We generally recommend to use linear elements for regions with expected strong nonlinearities like cracks. According to our experience, with cracks you get better results by refining to 1/2 element size than by switching to quadratic elements (while both use about the same memory and the quadratic mesh needs somewhat more CPU time).
You probably remember from some basic course on numerics or differential equations that the conditions for quadratic meshes to converge faster usually include smoothness of the solution of some degree, and this is typically violated for discontinuities like cracks. Therefore, it makes sense to use quadratic elements for elastic bending regions, but not for significantly cracked ones.

2. In ATENA ENgineering 2D, CCIsoQuad are linear, CCQ10 and CCQ10SBeta quadratic.
I must admit I am not sure if quadratic triangles are available in Egr 2D. However, as you are the first one asking for them I know about in many years, I suggest to use other element types, or explain why you wish to use them such that we can try to help you find some solution.


Re: Query reg. definition of FE mesh


I am trying to model reinforced concrete beams of three different sizes and compare with experimental results. For large (1350x300x50mm) and medium (675x150x50) sized beams, my results from ATENA are matching with experimental observations.

But for small sized beam, the load displacement graph suddenly comes back to zero after a few load steps (load is being given as  prescribed deformation) due to cracking at the point of application of displacement when Isoquad quadrilateral element are used. With CCQ10SBeta elements also same problem is there unless I use quite a large mesh size.  I am not able to understand why this is happening. Can you help me with this??

I have mailed my model to