Node and elements in data check

1

image
image434Ă—175 2.44 KB

hello everyone,
when i run a simulation, I dont understand why the node and elements that i see in the datacheck (picture) are much more than real number of nodes and elements that are in the model(physical nodes)

2

Check your numbering and if there are any gaps…

3

Normally, I would associate it with expansion to solids but there are no 1D and 2D elements according to this screenshot.

1 Like
4

probably these models are plane stress or any 2D element with three dof’s per node, it’s being expanded to 3D solid element internally with MPC generated in Z directions.

5

My model is formed by 4 components in C3D10 and a lot of contacts. But the total nodes are 800.000 nodes.
Indeed in the screenshot that I send you 19.000.000 nodes.

6

My model is formed by 4 components in C3D10 and a lot of contacts. But the total nodes are 800.000 nodes.
Indeed in the screenshot that I send you 19.000.000 nodes

7

Contact elements could also be counted, maybe that’s why the number is so large.

8

Don’t worry about it. Those numbers are called “estimated upper bounds” because they’re not the actual numbers. CCX seems to use the space for nodes to store other types of data too but they don’t go into the system matrix so it doesn’t matter.

In allocation.f, it looks like ways to increase the reported number of nodes includes:

MPCs
1D and 2D elements
KINEMATIC
*PRE-TENSION SECTION
C3D8I
*BOUNDARY
*RIGIDBODY
COMPOSITE
*RADIATE

There could be other things too.

9

well it’s quadratic tetrahedral element C3D10 actually, interesting since a screenshot given show the number integration point as eight instead of four. Is this user element from somewhere, not default in distribution?

10

it is standard element c3d10 as in Abaqus

11

However, it looks very strange. I run a simulation with a model of 2.5mil of nodes and 40 contacts and it works. While for this model formed by 800.000 nodes (there are also more contacts and pretension) I have and error memory

13

How many nodes in the biggest *PRE-TENSION SECTION? That generates MPCs that might include all face nodes in an equation which would lead to a dense part of the system matrix (n^2 nonzeros).

Otherwise, I would try deleting features like that without reducing the true node count to see what affects the reported upper bound node count and memory error.