residual convergence

- January 24, 2019 at 4:10 am
  
  swathi
  Subscriber
  
  the convergence level of my residual comes to 1e-3 at around 100th iteration but then later if diverges and becomes steady at 1e-1 for almost 6000 iterations. What should i do to get it converged to 1e-6
- January 24, 2019 at 4:23 am
  
  Keyur Kanade
  Ansys Employee
  
  is it steady or transient? what are other settings? which models?
  
  can you please insert some detailed zoomed in mages of the mesh and your set up.
- January 24, 2019 at 4:40 am
  
  swathi
  Subscriber
  
  I am using steady pressure based flow between two cylinders within laminar regime.
- January 24, 2019 at 4:48 am
  
  swathi
  Subscriber
  
  Boundary conditions.
  
  inlet- no slip wall
  
  outlet- no slip wall
  
  inner cylinder- rotating , temp=320K
  
  outer cylinder- stationary,300K
- January 24, 2019 at 4:53 am
  
  swathi
  Subscriber
  
  I tried with both Piso and coupled pressure-velocity coupling, the solution was converging to 1e-6 but the value of Nusselt number was having about 40% deviation from the required one.
- January 24, 2019 at 9:46 am
  
  Rob
  Forum Moderator
  
  Because it's a closed domain the mass balance can be a bit misleading. Add a few monitor points into the domain and see how they vary with iteration (I'd look at velocity & temperature). Also, I think you need a few more cells between the cylinders: 7-8 may not be enough in this system. As there's no through flow you can also make the domain shorter to conserve cell count.
  
  Finally, check the definitions of Nu and reference temperature to see which factor is off. Fluent uses a reference temperature for HTC and this often confuses as it's not related to any of the boundary conditions.