I'm modelling a simple, rectangular plate with ABAQUS/CAE, discretized with C3D8R elements. I'm adding two symmetry BCs and two geometry sets (which contain one edge of the plate) for a field output request. Also, I'm adding a point force. I have to discretize with a very fine mesh, so the final model could have up to 100 million nodes.
I'm observing a strange behaviour: From a certain size of the model (number of nodes), ABAQUS/CAE will define the element sets for the field output request with negative element labels, for example:
*Elset, elset=top_surface_nodes 5339, 26695, 48051, ... ... 67105891,-67090482,-67069126,-67047770,-67026414,-67005058,-66983702,-66962346,-66940990,-66919634 ... ...
I left out some parts, indicated by the dots. At some point, ABAQUS refers to the elements with a negative sign before the element labels. These elements cannot be found (even though an element with this label exists, just without minus sign).
Then, when trying to solve the model with ABAQUS/EXPLICIT, I get the following warning and the simulation fails:
***WARNING: The following entries for set TOP_SURFACE_NODES are not found and are removed: -67090482, -67069126, -67047770, -67026414, -67005058, -66983702, -66962346, -66940990, -66919634, -66898278, -66876922, -66855566, -66834210, -66812854, -66791498, -66770142, -66748786, -66727430, -66706074, -66684718 . This list is truncated to 20 entries.
Has anyone ever encountered such behaviour? I'm using ABAQUS 2019. The problem only occurs for a high number of nodes, for smaller models it works just fine.