Influence of flow behavior on residuals

MrNavierStokes · July 10, 2015, 08:41

Hello,

could someone give me an explanation why complex processes like separation bubbles have such an influence on the residuals of a RANS simulation?
I´m quite new to CFD and don´t really understand the reason for this, i mean
how exactly do high gradients and strong unsteadiness affect the solution process?

Thanks!

Regards

MrNavierStokes · July 16, 2015, 23:55

Could someone help me please?

David-G · July 24, 2015, 12:30

I don't really understand what you mean. The solution is presumably iterative, and as the iteration proceeds the residuals (hopefully) reduce. It takes many iterations for the regions of complex flow, e.g. separation bubbles, to be computed properly. So there will be contributions to the residuals from these regions.

MrNavierStokes · July 24, 2015, 20:59

I am talking about the 2D simulation of a flow through cascade, using a Transition Model for turbulence.
Since I can exclude modelling errors and bad mesh quality, I thought the convergence problems I am facing occur due to the complex features of the flow.
I tried to run the calculation steady-state as well as transient, but the residual behavior didn´t change. The solution converges well for first order upwinding, but not for second order.

Edeluc · July 27, 2015, 03:57

Hello NavierStokes,

Could you please get a little more specific regarding your "residual problems". Do you get undesired residual values in the continuity/momentum/energy equation or do you get the undesired values in the additional equations for your turbulence model? If you have a good post processing tool available, you can check the location of the "bad residuals" and see if they occur in your mentioned complex features or in the whole solution domain. This could give you a hint where the problems arise.

For second order upwind schemes oscillations in the solution can occur, depending on your boundary- and initial conditions as well as your mesh quality and size. You could try to use a TVD-scheme for the discretisation. These schemes have some sort of flux limiting function to stay stable during the whole solution process. In theory this gives you more than first order accuracy in your discretization process while staying stable during the solution process.

July 10, 2015, 08:41	Influence of flow behavior on residuals	#1
MrNavierStokes New Member Alex Join Date: Jun 2015 Posts: 22 Rep Power: 0	Hello, could someone give me an explanation why complex processes like separation bubbles have such an influence on the residuals of a RANS simulation? I´m quite new to CFD and don´t really understand the reason for this, i mean how exactly do high gradients and strong unsteadiness affect the solution process? Thanks! Regards

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
flow over a cylinder urgent!	kevin	FLUENT	8	August 11, 2015 13:00
Gate valve flow simulations...	nikesh	FloEFD, FloWorks & FloTHERM	5	January 28, 2014 01:31
Accuracy and convergence	pawan1989	FLUENT	5	April 15, 2010 17:17
potential flow vs. Euler flow	curious ...	Main CFD Forum	23	July 21, 2006 07:40
transform navier-stokes eq. to euler-eq.	pxyz	Main CFD Forum	37	July 7, 2006 08:42

July 16, 2015, 23:55		#2
MrNavierStokes New Member Alex Join Date: Jun 2015 Posts: 22 Rep Power: 0	Could someone help me please?

July 24, 2015, 12:30		#3
David-G New Member David Join Date: Mar 2014 Posts: 8 Rep Power: 12	I don't really understand what you mean. The solution is presumably iterative, and as the iteration proceeds the residuals (hopefully) reduce. It takes many iterations for the regions of complex flow, e.g. separation bubbles, to be computed properly. So there will be contributions to the residuals from these regions.

July 24, 2015, 20:59		#4
MrNavierStokes New Member Alex Join Date: Jun 2015 Posts: 22 Rep Power: 0	I am talking about the 2D simulation of a flow through cascade, using a Transition Model for turbulence. Since I can exclude modelling errors and bad mesh quality, I thought the convergence problems I am facing occur due to the complex features of the flow. I tried to run the calculation steady-state as well as transient, but the residual behavior didn´t change. The solution converges well for first order upwinding, but not for second order.

July 27, 2015, 03:57		#5
Edeluc New Member Lucas Join Date: Jun 2015 Posts: 20 Rep Power: 10	Hello NavierStokes, Could you please get a little more specific regarding your "residual problems". Do you get undesired residual values in the continuity/momentum/energy equation or do you get the undesired values in the additional equations for your turbulence model? If you have a good post processing tool available, you can check the location of the "bad residuals" and see if they occur in your mentioned complex features or in the whole solution domain. This could give you a hint where the problems arise. For second order upwind schemes oscillations in the solution can occur, depending on your boundary- and initial conditions as well as your mesh quality and size. You could try to use a TVD-scheme for the discretisation. These schemes have some sort of flux limiting function to stay stable during the whole solution process. In theory this gives you more than first order accuracy in your discretization process while staying stable during the solution process.