
[Sponsors] 
February 17, 2016, 02:39 
problem with turbulent model divergence in amg solver

#1 
New Member
arshia
Join Date: Feb 2016
Posts: 18
Rep Power: 10 
hi all,
i'm tring to simulate a moving object in a pipe by patching a high pressure and tempreture zone behind the object, i used a six dof udf for moving object by dynamic mesh scheme , i solved this by using fluent 6.3.26 and it work well without any problem but when i'm tring to use ansys fluent 15 , it diverge at the first time step with inviscid model it works well on fluent 6.3.26 and fluent 15 viscous model : ke standard density based transient  time step:1e6 all of the setup are the same so thanks to help me after one time step this error apears: divergance detected in amg solvercoupled coursening group size divergance detected in amg solvercoupled increasing relaxation sweep i tried to decrease relaxation factor and courant number to 1 but it didn't help also i tried to fine the mesh but it didn't affected the problem Last edited by hamed1987; February 17, 2016 at 05:21. 

February 17, 2016, 09:50 

#2 
Senior Member
Lucky
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 5,695
Rep Power: 66 
After patching, did you iterate a few times before proceeding to the next timestep? It's important to do a few iterations after patching so that the solution field is updated and consistent with the current state.


February 17, 2016, 10:40 

#3  
New Member
arshia
Join Date: Feb 2016
Posts: 18
Rep Power: 10 
Quote:
after patching the temp and press i set the time step to 1e6 and iretation per time step 20 so we have 20 iteration in each time step and so the first time step and after these 20 iteration for first time step it diverges. 

February 17, 2016, 10:44 

#4  
Senior Member
Lucky
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 5,695
Rep Power: 66 
Quote:
If you simply patch and then press the calculate, it will go to the next timestep and not iterate on the patched fields. You will go into the next timestep with uniform fields from the previous timestep, which may be "inconsistent" 

February 17, 2016, 11:15 

#5  
New Member
arshia
Join Date: Feb 2016
Posts: 18
Rep Power: 10 
Quote:
i tried to solve the problem for some time step with inviscid method without any dynamic mesh( dynamic mesh was off) and then i switched the method to ke standard and the dynamic mesh on but it didn't work so is it diffrent with the method you say?? how can i iterate into the TUI?? thanks a lot 

February 17, 2016, 11:56 

#6  
Senior Member
Lucky
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 5,695
Rep Power: 66 
Ok you did several things that all resulted in the same problem. In a transient simulation, after you make any changes you should always do some iterations before proceeding to the next timestep.
When you switch from inviscid to a turbulence model in a transient simulation, the turbulence fields are empty and not initialized. If you press the calculate button, then it will go to the next timestep but it won't have any turbulence on the old field to solve with. You should feeze all current fields and iterate only the turbulence fields if you want to ensure that no other fields are affected. Quote:
just type this in and press enter solve iterate 20 

February 21, 2016, 06:00 

#7  
New Member
arshia
Join Date: Feb 2016
Posts: 18
Rep Power: 10 
Quote:
i did what you do, after patching pressure and tempreture i typed solve iterate 20, residuals was reach : continuty e18 xvel e12 yvel e12 energy e19 k e16 epsilon e16 after that i run the calculation by time step size e6, it didnt diverge but i have a new problem , the residuals of continuty and xvel and yvel and energy goes to be constant at these values (like the picture)Untitled.jpg and the object didn't move until now (after 1000 time step) 

February 21, 2016, 13:19 

#8 
Senior Member
Lucky
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 5,695
Rep Power: 66 
It looks like your simulation is working.
It looks like you have stationary boundary conditions (constant pressure outlets and some sort of constant inlet) and probably also constant properties. After an initial short transient your velocity and turbulence are stationary and hence the residuals stay constant (because the velocities and turbulence do not change at all). It will take time for your patched temperature to convect downstream into your domain. You should have a good feel for this. If your timestep is very small then it will take many timesteps for the new inlet conditions to arrive in the domain. This is probably unrelated to your problem, but I forgot to mention that you should not switch models (from inviscid to turbulent) in the middle of a transient simulation. You may get the wrong physics because your initial state is inviscid and suddenly you activate a turbulence model so the flow evolves from inviscid and transitions (temporally) into a turbulent flow. Unless this is what you are trying to simulate... 

February 21, 2016, 13:59 

#9  
New Member
arshia
Join Date: Feb 2016
Posts: 18
Rep Power: 10 
Quote:
i did'nt understand completly what you say when i calculate my problem with turbulent model in fluent 6.3.26 or inviscid model, the object move from the first time step because we have high pressure(350e6) and high tempreture (3000) and the object weight is 0.4 , so there is a high force behind the object that it must move at the first time step but when i tried to do what you said( tui iterations) after that i calculate it dosent move after 0.001s after 0.001s it must reach the middle of the barrel i have two stationary in mesh zone 

February 22, 2016, 05:18 

#10  
New Member
arshia
Join Date: Feb 2016
Posts: 18
Rep Power: 10 
Quote:
i continued the simulation but the object didn't move at all. 

February 24, 2016, 14:24 

#11 
New Member
arshia
Join Date: Feb 2016
Posts: 18
Rep Power: 10 
Hi
eny comment? 

Tags 
divergence amg solver, dynamic mesh, turbulence analysis 
Thread Tools  Search this Thread 
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Divergence detected in AMG & FMG solver.....  devesh.baghel  FLUENT  2  July 30, 2018 22:39 
Error: Divergence detected in AMG solver: pressure correction  wanna88  FLUENT  19  April 6, 2016 02:57 
Divergence detected in AMG solver: pressure correction  xinquanzhoucn  FLUENT  5  July 21, 2014 04:49 
Error: Divergence detected in AMG solver: xmomentum/ epsilon/ temperature  bubuchacha  FLUENT  6  February 26, 2013 03:30 
Divergence problem  Smaras  FLUENT  13  February 21, 2013 05:03 