Skip to main content

​
Overview

Simulation errors are inevitable. Solver failures, diverging residuals, highly distorted elements, negative volumes, unconstrained rigid body motion — each one can take hours to diagnose without knowing where to look. Nexus shortens the diagnostic loop significantly. Describe what you’re seeing, paste in the error message, and Nexus will identify the most likely cause, suggest a fix, and apply it on your confirmation.

​
What Nexus can help with

  • Solver errors — Interpret cryptic error codes, identify the failing step or element, and suggest targeted fixes.
  • Mesh failures — Locate problematic geometry, explain why meshing failed, and apply geometry repairs or local sizing overrides.
  • Convergence — Diagnose slow or non-converging runs (structural nonlinear, CFD, transient thermal) and suggest solver settings, relaxation factors, or stabilisation options.
  • Unexpected results — Flag results that look physically unreasonable, trace them back to incorrect BCs, material assignments, or unit mismatches.
  • Model verification — Check that units, materials, contacts, and loads are consistent before committing to a long compute run.

​
Example prompts

The solver failed with: "ERROR: Pivot too small — check for unconstrained rigid body motion (node 8241)." The model is a bracket with four fixed supports applied. What's most likely causing this and how do I fix it?

Meshing failed on the imported STEP file. The error says "unable to mesh body 3 — geometry defects detected." Identify the problem faces, describe the defects, and apply repairs to allow meshing to proceed.

My nonlinear static analysis is not converging after 12 substeps. The force residual is oscillating around 3%. I'm running large deflection with a bonded contact and a 5kN point load. What solver settings should I adjust?

The maximum von Mises stress is showing 1.2 GPa on a steel part loaded with 500N. That seems unrealistically high. Check the material properties, units system, and applied loads — tell me what might be causing it.

My Fluent run is diverging after 50 iterations. Continuity residuals are climbing. It's a steady-state incompressible internal flow case at Re ≈ 50,000. I'm using k-ε turbulence with standard wall functions. What should I check first?

​
The debugging mindset

The most effective troubleshooting prompts follow the describe symptoms rule from the Prompting Guide:
  1. State what happened — the exact error message, the step at which it failed, or what the result looks like.
  2. State what you expected — what a correct result or successful run would look like.
  3. State what you’ve already tried — so Nexus doesn’t suggest the same thing twice.
Here’s an example that does all three: 
The static structural run fails at substep 4 with "element distortion errors near node group 12400."
I expected the run to complete — it converged fine on a coarser mesh.
I've already tried reducing the mesh to 1mm in that region — no change.
The geometry is an imported STEP with a sharp re-entrant corner (R ≈ 0) in the failing region.
What else should I try?
Nexus will work through the most likely causes systematically and apply fixes directly when you confirm.

​
Tips for troubleshooting prompts

  • Paste the full error message — The exact text usually contains the node ID, element ID, load step, or iteration number that Nexus needs to locate the problem.
  • Describe the model briefly — Analysis type, material, loading, constraint type — enough context for Nexus to reason about the problem even if it can’t read the full setup.
  • Say what you’ve already tried — Nexus will skip those approaches and move to the next candidate fix.
  • Ask for an explanation alongside the fix — Understanding the root cause prevents the same error next time, especially for recurring model types.