continuity-convergence problems during mesh independence study
 

Dear all,

I'm quite new to Fluent and this is my first thread. I hope someone can help me.


I’m facing continuity-convergence problems in my steady k-w SST simulation with refined meshes.


I'm trying to simulate a flow with one inlet and ten outlets (one main outlet and 9 smaller outlets, see picture Initial_Meshing). The flow rate at the inlet and the main outlet is known and I want to examine the distribution between the 9 smaller outlets.


My initial simulation, with a coarse mesh (750.000 nodes, 1.750.000 elements) converged for first and second order discretization (see picture CoarseMesh_SecondOrder_ResidualPlot). For a mesh independence study I used the same settings with refined meshes (1.050.000 nodes, 2.770.000 elements and 1.500.000 nodes, 4.560.000 elements) . With the refined meshes, the simulation only converged with first order discretization but not with second order (see picture RefinedMesh_SecondOrder_ResidualPlot). I monitored the flow at the outlets. The monitors showed a constant, almost oscillating, change in the outflow (see picture RefinedMesh_SecondOrder_MassFlowMonitor). This was not the case in the initial simulation with the coarse mesh.



I'm unsing the following settings with a tet-Mesh. For the flow I've calculated reynolds numbers of about 4000 at the inlet, about 3000 at the main outlet and about 1000 at the small outlets.



Model: k-w-SST, with Low Re correction

Boundary Conditions
Inlet: Mass-Flow-Inlet
Main Outlet: Mass- Flow- Outlet
Small Outlets: Pressure Outlet with 0 gauge pressure
Methods:
Scheme: Coupled
Gradient: Green-Gaus Node Based
Pressure: PRESTO!
Momentum: Second Order Upwind
Turbulent Kinetic Energy: Second Order Upwind
Specific Dissipation Rate: Second Order Upwind
with pseudo transient setting

Controls:
Pressure: 0,2
Momentum: 0,3
Density: 1
Body Forces: 1
Turbulent Kinetic Energy: 0,75
Specific Dissipation Rate: 0,75
Turbulent Viscosity: 1


So far I've tried to:
- lower the relaxation factors
- use differnt discretization methods
- use differnt initializations


I really hope someone can help me.



Best regards,
Peter