# Floating point error with Extreme velocities entering at junction

Dear all,

I work with 3D, Transient, VOF(Water and air), Open Channel problem with 'Flow past row of cylinders mounted at the bottom wall'.

I run fluent in parallel mode. The code crash with floating point error as shown below.

Divergence detected in AMG solver: omega

turbulent viscosity limited to viscosity ratio of 1.000000e+05 in 889546 cells

Divergence detected in AMG solver: omega

Divergence detected in AMG solver: omega

Divergence detected in AMG solver: omega

Divergence detected in AMG solver: omega

Divergence detected in AMG solver: omega

Error at host: floating point exception

Error at Node 0: floating point exception

Error at Node 1: floating point exception

Error at Node 2: floating point exception

Error at Node 3: floating point exception

Error at Node 4: floating point exception

Error at Node 5: floating point exception

===============Message from the Cortex Process================================

Compute processes interrupted. Processing can be resumed.

==============================================================================

Error: floating point exception

Error Object: #f

When i plot velocity vectors, i observe velocities of extreme values are entering at one of the junction location of cylinder and bottom wall.the snapshot is attached.

Kindly help. Thanks in advance.

Regards,

Prabu

## Comments

Case diverged so velocities are garbage. Add more details about the run.

Geometry: The problem has a 3d rectangular prism domain, with open channel vof, whose boundary condition are as follows

left end-pressure inlet(with water level and velocity), right end-pressure outlet.

front face-symmetry;back face-symmetry;top face-symmetry; bottom face-wall;

It has slender cylinders emerging from bottom wall to top symmetry.

Meshing: tetrahedral mesh; meshing quality: skewness less than 0.8, min orthogonality:0.2

solver: transient,rans, sstkw,methods: piso, body force weighted, implicit, second order discretisation(u,v,w,k,w)

the problem is solves a flow past vertical cylinders in parallel processing. The solver brings floating point error.

Probably wrong pressure field resulting in divergence. Do you know the pressure at inlet. You need to be careful here for open channel flow as presure inlet is just providing information about total height. f you know the height and other thing like velocity probably you might use open channel wave too (much more easy to interpret).

Have you started using a smaller tome step size

Actually, the pressure boundary conditions worked well for the no cylinder case. But it gives error in the presence of cylinder. The height of the water and the inlet velocity is known aprior. The time step is 0.001 s. what could be the reasons for getting this error.

Kindly help,

Prabu

What's the mesh like around the cylinders?

For the meshing around the cylinders, I used both with and without inflation, as shown below.

Can you run the simulation by reducing the relaxation factors?

Rahul,

These are the values i used for under-relaxation pressure 0.3; density 1; body-force 1; mom 0.7; mp 0.5; k 0.8; omega 0.8; turb-viscosity 1

Thanks

Prabu