Draft. This manual is new and still growing. If something here does not match what you see in the software, the software is right — tell us and we will fix the page.
The run will not start
The application refuses a run rather than launching one that will fail in the solver, where the reason is much harder to find. These are the messages that do the refusing, and the Problems entries behind them.
Every heading is the literal string. Where a message names something from your
case, that part is shown in angle brackets — <solver>, <patch> — and the
real message has your own value there.
Back to Troubleshooting.
The refusal itself
"1 problem(s) must be fixed before this can run."
In the New run dialog, in red, above a Start button that is greyed out. The count is the number of error-severity entries in the Problems tab; a warning does not stop a run.
The dialog does not say which problems. Close it and read the Problems tab in the bottom dock — each row names the step it belongs to, and clicking a row jumps to that step.
The literal (s) is not a typo in this page. The message is written once and
does not switch to a singular form.
"Cannot run: 3 problems must be fixed first."
The same refusal as a Log line, written when a run is started from somewhere that has no dialog — the toolbar, the command line, an automation step. The singular form is "Cannot run: 1 problem must be fixed first."
Each blocking problem is written out after it, so the log is a complete account even where there is no Problems tab to look at.
"The Free tier runs one job at a time. Wait for this run to finish, or stop it."
A Log line, and it names your tier rather than always saying Free. It is not about the case: something is already running. Stop it from the run's pane, or wait.
Nothing to run on
"No mesh has been generated."
A Problems entry against Mesh. Mesh the case first.
"The mesher is set to import a mesh, but no mesh file has been chosen."
A Problems entry against Mesh. The mesher is set to import, so nothing will be generated, and no file has been named to import. Either choose the file or set the mesher back to one that builds a mesh.
"Patch 'inlet' has no condition assigned for fields U, p."
A Problems entry, and its step is the patch itself rather than "Boundary conditions" — look for a row named after the patch. The field list is the set this case solves for, so it varies with the analysis.
A patch with no condition is the commonest first-session blocker: importing geometry creates the patches, and each one still has to be told what it is.
The solver cannot do what the case asks
These fire against Models, and they are all resolved by changing one setting rather than by fixing the mesh or the geometry.
"<solver> has no turbulence model, so it will not read k, omega or nut. Set the turbulence model to Laminar, or choose a solver that has one."
The chosen solver has no turbulence at all, and the case has a turbulence model set. Nothing is wrong with either on its own; they cannot both stand.
"<solver> has no steady form. Set the time dependency to Transient."
And its mirror, "<solver> has no transient form. Set the time dependency to Steady-state."
A third variant fires from the analysis type rather than the solver, and names both: "<analysis type> has no steady form. <solver> integrates in time; set Time dependency to Transient."
"<solver> cannot be set up completely by this build: it needs <what is missing>. The case will be written and the mesh will build, but the solver stops at start-up."
Read this one carefully — it is filed as a warning and it is a blocker. Because it is not an error, it does not stop the run: the case writes, the mesh builds, the job starts, and the solver exits at start-up. The message says so, but by then it is several minutes back up the Problems tab.
There is no setting that fixes it. Choose a different analysis type, or a different solver for the same physics.
"'C3H8' has no built-in thermodynamic data. The case will still write, but constant/thermo carries a placeholder comment instead of real coefficients for it, and the solver will stop as soon as it tries to use it."
The other warning that is really a blocker, on a reacting or combusting case. The species is named in the mixture, the application has no thermodynamic coefficients for it, and what gets written is a comment where the numbers should be. The plural form lists every unknown species at once and reads "…have no built-in thermodynamic data…".
Use a species the application knows, or supply the case's thermodynamic data yourself in the case directory before running.
Dimensionality
"This case is set to two-dimensional but no patch is Empty (2D), so the solver will treat it as 3D. The side faces have to be empty for the direction to be dropped."
A Problems entry against Boundary conditions. Two-dimensional in
OpenFOAM is not a switch — it is a pair of empty patches on the faces normal
to the direction being dropped. Set both side faces to Empty (2D).
"The thin direction of a 2D case cannot be the flow direction - that is the direction the answer is in. Choose a different axis."
A Problems entry against Mesh. A related rule, fired from the mesh side, names the axis: "This case is two-dimensional along Z, which is also the flow direction. The thin faces would be the inlet and the outlet rather than the empty pair, so nothing about the mesh would be two-dimensional. Choose a thin axis across the flow."
"The flow axis is vertical on a free-surface case. The surface is horizontal, so the inlet on the bottom face drives the flow straight through it. Set the flow axis to X or Y."
A Problems entry against Mesh; it says top instead of bottom when the flow runs the other way. The free surface is horizontal by definition — an inlet above or below it pumps the light phase through it at the free-stream speed, and the solver diverges. Set the flow axis to X or Y.
Phases
"Both phases are called 'water'. The names become field names, so they have to differ."
A Problems entry against Materials, on a two-phase case. The phase names are written out as field names, so two identical names describe one field rather than two phases. A multiphase case with more than two phases raises the same rule worded slightly differently: "Two phases are both called 'water'. The names become field names, so they have to differ."
Another rule on the same pane catches a dispersed phase that is not one of them: "The dispersed phase 'air' is not one of the two phases."