which option of segregated or coupled solver

- November 28, 2018 at 8:56 pm
  
  rm19g14
  Subscriber
  
  I am running a simulation of a car using fluent 19.1. the mesh has been generated with a polyhexcore scheme and the baseline works well with the SIMPLEC solving scheme and the k-e turbulence model. I get the issue that the 'turbulent viscosity ratio is limited to 1e+05' in xxx cells but this disappears after a few hundred iterations.
  
  The issue I'm having however is that when I add a front splitter, the turbulent intensity stays limited in a few cells and the residuals of k and e however around 4e-02. I haven't had this issue with the rearwing or the sideskirts I have added. I have tried increasing the under relaxation of k and e and ensured that my yplus doesn't go below 30, however I get the same thing during solve, the residuals of k and e drop to 1e-04 before abruptly jumping to 1e-01.
  
  I have read that using a coupled solver can help this and have given it a go, it appears to have solved the issue, however I am unsure about the impact this will have on the accuracy of the simulation as coupled solvers are usually used for compressible problems. Does anyone know of any reason why using a coupled solver for this case is fundamentally a bad idea and if it is does anyone have suggestions for how to solve the issue I'm having with the segregated solver.
  
  I have attached an image of my mesh, I'm fairly happy with its quality and don't think that this is causing the problem.
  
  Kind Regards
  
  Rikesh Mistry
- November 28, 2018 at 9:07 pm
  
  DrAmine
  Ansys Employee
  
  Coupled solver are superior to any segregated solver as it builds the stiffness matrix in different way incorporating the pressure correction within the momentum correction step. This would lead to quicker convergence and sometimes has some good influence on solving physically tough coupled cases like nature convection compared to segregated ones. Coupled Solver plus first order time marching algorithm for solving steady state is our recommendation.
- November 29, 2018 at 11:41 am
  
  rm19g14
  Subscriber
  
  so to clarify, when you talk about the first order time marching algorithm so you mean to use the pseudo transient option for the solver?
  
  regards
  
  Rikesh
- December 6, 2018 at 12:03 pm
  
  DrAmine
  Ansys Employee
  
  Sorry for the delay but yes I am talking about the pseudo-transient pressure based coupled solver.