Turubulent viscosity limited and then floating point exception

- June 19, 2023 at 9:10 am
  
  Agung Limowa
  Subscriber
  
  Hi, i have a really big model around 2km in length and 6.5 m in height. The model has around 8 million mesh and i think the quality is not that bad. I use the k-epsilon turbulence model and every time I calculate it, it show viscosity is limited to 1e05 in 23402 cells and then the floating point point exceeded. I think imporving mesh quality is not helping. what should i do
- June 19, 2023 at 1:43 pm
  
  Rob
  Forum Moderator
  
  To confirm 2000m long and 6.5m high? 2d or 3d?
  The problem may be mesh related but potentially not in the way you think. Please post images of the turbulence and fields at a point a few iterations before the solver fails.
- June 19, 2023 at 3:31 pm
  
  Agung Limowa
  Subscriber
  
  yes, that's right the length is 1890m and the 5x5 tunnel size and simulation on 3d. The simulation here are the image of what it says before the error. I am using K-epsilon and it seems didn't help. I tried to reduce URF also didn't help. Is it possible to do such simulation?
- June 19, 2023 at 3:57 pm
  
  Rob
  Forum Moderator
  
  Can you post some images of the mesh? How many cells do you have covering the tunnel cross section in both the "open" and "cluttered" parts?
  Mesh quality is actually two parts. Cell quality (the commonly checked metric) covers the cell shape. However, most new users miss the mesh resolution check: do you have enough cells to actually predict what's going on?
- June 19, 2023 at 4:10 pm
  
  Agung Limowa
  Subscriber
  
  how can i check the quality and what basepoint to tell that the mesh it's good. And also if i run in pressure outlet it has no error, so it still a mesh issue or maybe somethjng else ?
- June 19, 2023 at 4:24 pm
  
  Rob
  Forum Moderator
  
  What had you set if not pressure outlet?
  Cell quality is checked in the meshing tool. Whether you have enough cells is a bit more of a judgement: you need to balance resolution with compute resource. Posting images will mean I can comment.
- June 20, 2023 at 2:45 am
  
  Agung Limowa
  Subscriber
  
  Here is my mesh quality., my processor is a core i7 with around 16 GB RAM,
- June 20, 2023 at 8:45 am
  
  Rob
  Forum Moderator
  
  If that's a min of 1e-4 then you have a problem - the sensible lower limit for ortho is about 0.1, with 0.05 being viable IF you know what you're doing and the cell isn't anywhere with any flow gradients.
  However, that still leaves the mesh resolution.