Solver errors
What each class of solver message means, what usually causes it, and what to do next.
Troubleshooting 2 min read
Solver messages fall into three classes, each needing a different response: the analysis could not run, it ran but the result is untrustworthy, or the environment got in the way.
The analysis could not run
| Message class | Meaning | Go to |
|---|---|---|
| Singular stiffness matrix / model unstable | A mechanism — the structure can move without load | Singular stiffness matrix |
| No supports defined | Nothing restrains the model at all | Add supports |
| Element with no stiffness | A member without a section or material, or zero thickness | Assign properties to everything |
| No loads in the selected case | Nothing to solve for | Check the case, and that loads landed on the model |
| Mesh generation failed | Surface geometry the mesher could not resolve | Simplify the boundary, check for tiny edges and self-intersections |
| Model exceeds plan limits | Over 500 degrees of freedom on the free plan | Plans and limits |
It ran, but the result is suspect
These usually arrive as warnings rather than errors. They are the more dangerous category, because there is a result to look at.
| Warning | Usual cause | Action |
|---|---|---|
| Poor matrix conditioning | Stiffnesses spanning many orders of magnitude | Remove very stiff springs; use fixity or rigid links |
| Large displacements | Wrong units, wrong modulus, near-mechanism | Check , , ; sanity-check one member by hand |
| Equilibrium error above tolerance | Load not reaching supports; conditioning | Compare reactions with applied load per case |
| Mesh quality warnings | Distorted elements | Mesh quality checks |
| Modal analysis: no mass | Zero density, no mass sources defined | Modal analysis |
| Buckling factors negative or absurd | Reference load case wrong or trivial | Linear buckling |
Environment problems
The solver runs in your browser, on your machine, so a few failures are about the machine rather than the model:
| Symptom | Cause | Fix |
|---|---|---|
| Solve fails on a large model but works on a small one | Browser memory limit reached | Coarsen the mesh; close other tabs; use a 64-bit desktop browser |
| Solve is much slower than expected | Threads unavailable, or the machine is busy | Check other load on the CPU; try a current browser version |
| Solve does not start at all | Very old browser without the required WebAssembly features | Update to a current Chrome, Edge, Firefox or Safari |
When to report it
Report it when:
- The message does not correspond to anything in this list.
- The model is stable, the units are right, and the result still disagrees with a hand calculation or a published benchmark.
- Something worked yesterday and does not today.
Send the project and, where relevant, the calculation you compared against, to [email protected]. During the beta these go to the person who wrote the solver, and fixes appear in the release notes.
Something wrong or unclear on this page? Tell us, and it gets corrected.