Inconsistent boundary/shape factor

[DRW75]

Hey folks.

Question: - can you (mathematically) have a second order shell element (say created by 8 node quadrilateral - serendipity element) which also has one edge also defined with two 1st order beam elements - one extending from a corner to the mid-node, and the other extending from the mid-node to the other corner node - SEE SKETCH BELOW.

I know that the shape function along the shell edge is a parabolic function, whilst the shape functions along the same beam side are two linear function - but will this non-compatibility result in serious math/convergence problems? Or will it simply introduce slight discontinuities between results on the elements.

I've tried to sketch the geometric formulation below:
o?.?o
? ?|
. .
? ?|
o?.?o

where o is a primary corner node,
. is an intermediate node
and | represents a beam elements connecting a node to an intermediate node.
and ? or ? represent the edges of the serendipity element

thanks

DRW

 

[DRW75]

just noticed that the ASCII characters that I had used didnt come out too good... but hopefully u get the picture.


DRW

 

[rb1957]

you can do this, particularly if you're aware that it is bad practice.

i'd suggest that you run a simple patch test and look at the results for endload, shear, and bending. looking at the results should let you know how sensitive this is to certain types of loading