I encountered the error 'NaN found in an element stiffness matrix' in my multi-layered piled foundation model, consisting of clay and sand (6-11 m UBC3D sand layer with a water table at 6 m). I suspected this was due to entering a unit weight of zero for the foundation, so I changed it to 0.1. However, this time I received the error 'NaN found during calculation, probably severe divergence.' I don’t think the issue lies with my earthquake input, as the same input worked without issues in a model where the water table was at 8.5 m. What could be the main issue?
GeoStudio | PLAXIS Forum - Error Code 36 and 39
Dear Burak,
Error 36 typically occurs when either one of the stiffnesses is very small, or when the ratio between the largest and the smallest stiffness in the model is very large. Typically we don't want the ratio between largest and smallest stiffness to be larger than let's say 1e5. If the ratio is larger, the risk of getting a ill-conditioned stiffness matrix increases, which could lead to error 36.
Error 39 is more general. Slow convergence can be because of a bad stiffness matrix (similar to error 36), but also due to for instance very small permeabilities in case of consolidation or groundwater flow, very large amounts of plasticity occuring (for instance during liquefaction).
In all cases it's best to inspect the output after the error occured and check for excessive displacements, stresses or excess pore pressures and then determine where that comes from.
With kind regards,
Dennis Waterman