Stabilize the energy solve

In conjugate heat transfer, energy is often the most difficult linear system because one solve spans the fluid and conducting solids, including large differences in conductivity and cell geometry.

Confirm that energy is responsible

Inspect residual_temperature, temperature_linear_iterations, and temperature_right_preconditioner. Add minimum and maximum temperature monitors and an energy-balance monitor before rerunning.

Energy is always unrelaxed

The energy equation has no relaxation or pseudo-transient setting. Use the linear-solver and discretization controls below to address convergence problems.

Use robust advection and fallbacks

Return an overridden energy scheme to upwind. Then provide a multigrid-first fallback sequence with an alternative ILU method:

"numerics": {
  "stable": {
    "linear_solvers": {
      "energy": {
        "right_preconditioner": ["mg32", "mg64", "ilu55"]
      }
    }
  }
}

The standard presets already provide an energy fallback; adding another multigrid precision can give a sensitive system another route before the ILU alternative.

Check the thermal model and mesh

Confirm that heat has a path out of the model, powers and fluxes have the intended signs, and material properties use consistent units. Inspect high-conductivity solids, thin layers, abrupt material interfaces, and cells with large aspect ratios around the first temperature excursion.

If the linear iterations remain high, follow Improve mesh quality and rerun with the same energy numerics so the effect of the mesh change is clear.

See also

• Why energy is not relaxed explains how energy relaxation limits convergence of the energy balance.
• Preconditioners and fallbacks explains the fallback sequence.