In the beginning of the Cz execution finish all settings and controls described in the section before. Then select the variables group

**Temperature**and start the computation. On this stage only forward temperature calculation, forward temperature calculation together with inverse modelling or forward and inverse model together with phase tracking can be executed. Usually no convergence problems are occured during this phase stage of Cz computation. As a result the global temperature distribution, heating powers and the equilibrium shape of the crystal meelt interface without account of the melt convection are obtained.In the second step switch to the variables group

**Convection**. Control settings in the automatically created tabs of the**Convection Parameters**dialog window. The automatic tabs have names of materials in the regions meshed structurally.In each automatic tab control the number of iterations on the structured mesh in the dialog field

**stacked iterations number**. For details see section Setting convection on structured mesh parameters. The table rows of the activated transport eqautions should be activated and their numerical parameters should be accessible for input. Meaning and details of the numerical parameters on the structured mesh can be red in the same section Setting convection on structured mesh parameters.The heating powers regulation can be temporally stopped by changing of the heaters type to a

**Fixed Power**. The inverse modeling can be activated again after the melt convection has achieved the convergence in the single enthalpy equation mode.### Note

The switching from the

**Convection**to the**Temperature**variables group doesn't delete the computation results for convection on the structured mesh as it would be the case on the unstructured mesh. Therefore the decoupled temperature computation with frozen values of the convective mass fluxes is possible and may be used for acceleration of the thermal computation or if difficulties in the deep convergence of the Navier-Stokes equations occur.Now start the computation. The melt convection and the convection in liquid encapsulant and gas if available will be computed.

If the convergence problems occur such like oscillating residuals, the increase in the underrelaxation of the transport equations on the structured mesh and in the underrelaxation of the forward computation for the global enthalpy equation may be required, i.e. their values should be reduced.

Another possible reason is, the mesh quality and mesh resolution is not satisfactory somewhere in the melt volume. The structured mesh should be improved, the velocities, temperatures and pressures on it should be initializated to default values, the previous step of the temperature computation should be repeated again. Then a new attempt of the melt flow computation can be done.

Or there is no existing physical solution for the laminar flow and selected material and process parameters. Then the melt viscosity should be increased until the laminar stationary solution can be achieved.

If the laminar stationary computation is converged and the selected material parameters of the melt coincide with its real data, the task is successsfuly solved. Otherwise in case of the increased viscosity, reduced specific heat capacity etc. one may either continue to approach the realistic material data in the next laminar computation (may be with increased mesh density and reduced underrelaxation of the transport equations) or try to compute the flow using the turbulence model.

### Note

The material parameters ramps are available only for computation of fluid flow on the unstructured mesh.For computation of the turbulent flow activate the turbulence model in the melt regions. Then reduce underrelaxation factors for all equation on the structured mesh and for the global enthalpy equation to 0.1.

The selected variables group may be either

**Convection**again or**turbulent_Convection**. The transport equations of the k-epsilon turbulence model will be solved in each case together with Navier-Stokes equations. The only difference is in the online visualization of the computed results. For the selected**Convection**the velocity vectors are accessible for visualization. In the**turbulent_Convection**mode no vectors are accessible, but the variables of the turbulence model (turbulent energy, eddy dissipation and turbuleent viscosity) can be visualizated.Then start the computation of the turbulent convection.