floating point exception [invalid operation]
 

Hello Everyone,

I am trying to simulate a single phase turbulent flow inside a square array sub channel for Pressurized Water Reactor rod bundle application. During meshing of subchannel geometry, I have used Surface Remesher and 1. Polyhedral+Prism Layer 2. Trimmer+Prism Layer. And also I have tried with different base size and different values for prism layer thickness, no. of prism layers, relative minimum size to obtain different quality of mesh from coarser to finer.

For Physics, I have selected following models: Constant Density, Gradients, Gravity, High y+ wall treatment, standard K-epsilon turbulence, liquid (water), coupled flow, coupled energy, steady and three dimensional.

Boundary conditions are as follows: inlet: Mass flow inlet (0.4779328kg/Sec), Outlet: Pressure Outlet (15513203.9025 Pa), Symmetry and Wall: Heat Flux (600000 W/m2).

My initial conditions are as follows: Velocity: 2.49 m/s, pressure: 101325 Pa, temperature: 300 K

For solver settings I have tried with different values of Courant No. ranging from 50-2000 and different under relaxation factor and also activated Linear Ramp for Courant No. Grid Sequencing for Expert Initialization, Expert Driver for Solution Driver, Continuity Convergence Accelerator, Linear Ramp for Under Relaxation Factor Ramp, And Different Cycle Types and Relaxation Schemes for AMG Linear Solver.

I have also tried with different Turbulence Model. But in all cases, the following error is shown:

" A floating point exception has occured: floating point exception [invalid operation]. The specific cause can not be identified. Please refer to the troubleshooting section of the User's Guide. "

Can anyone please tell me what is the solution to this problem and what can be the recommended solver settings for my mentioned Physics set up and boundary conditions?

Thanks.

I had the same problem but I solved reducing the number of prism layer and changing the prism layer stretching

floating point exception
 

Thanks jano. But could you please say what value you selected for no. Of prism layers and prism layer stretching?

I think it depends on geometry and base size of the mesh. I had a pipe with 485 mm diameter and I used a base size of 0.005 m. N prism layer is 3 and stretching is 2. I used prism layer thickness 100% relative to base. I'm not an expert in cfd, but I simply noticed that those values solved my problem...

Thanks a lot jano. But interestingly i found that if i change my turbulence model from k epsilon to reynolds stress it also solves the problem.

1. Why are you using the coupled solvers? These are not meant for single phase forced convective flows.
2. Why are you using constant density models with the gravity model?
3. Why do you have initial conditions already initialized in the flow?
4. Are you sure your mesh is properly resolved in the near wall regions?
5. Are you sure you have your first cell within the appropriate near wall y+ value?