How pseudo-transient method helps convergence in fluent

 User Name Remember Me Password
 Register Blogs Members List Search Today's Posts Mark Forums Read

 March 12, 2019, 12:37 How pseudo-transient method helps convergence in fluent #1 Senior Member   ali Join Date: Oct 2009 Posts: 318 Rep Power: 13 Does anyone have any documentations discussing how pseudo-transient method helps convergence in fluent, from theoretical perspective? Thanks,

 March 12, 2019, 13:29 #2 Senior Member   Lucky Tran Join Date: Apr 2011 Location: Orlando, FL USA Posts: 3,681 Rep Power: 45 Do you have a specific question? The pseudo-transient method is a form of implicit relaxation (as opposed to explicit relaxation). In explicit relaxation you solve the equation but limit the variable change a posteriori. In implicit relaxation, you don't solve the original equations but relaxed equations. The relaxed equations limit the variable change because you inject the current solution into the equations. You can think of implicit relaxation as lowering the local cell flow courant number. You can think of a steady state solver as a transient solver with a very wonky but large (almost infinite) time-step and a transient solver as having a small finite time-step. Last edited by LuckyTran; March 12, 2019 at 19:33.

 March 12, 2019, 16:23 #3 Senior Member   ali Join Date: Oct 2009 Posts: 318 Rep Power: 13 Thanks LuckyTran. Yes, I had seen that, but as you can see it does not give a good discussion why this would help. Reading you comment was helpful. Can we think of it the following way too? Adding transient term, makes the equation hyperbolic and it is no longer elliptic, hence it would be easier for the solver to reach convergence. Is this reasoning correct?

March 12, 2019, 19:32
#4
Senior Member

Lucky Tran
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 3,681
Rep Power: 45
Quote:
 Originally Posted by alinik Can we think of it the following way too? Adding transient term, makes the equation hyperbolic and it is no longer elliptic, hence it would be easier for the solver to reach convergence
Not really. Hyperbolic/elliptic/parabolic is a classification referring to the characteristics in one dimension only (either only time or only space). Introducing a transient term doesn't change the characteristics in space.

Also the characteristics has nothing to do with convergence. A parabolic/hyperbolic/elliptic equation can all be solved. It also doesn't explain how under-relaxation relates to stability, which is what the pseudo-transient option does. The pseudo-transient solver applies implicit under-relaxation to a steady solver.

What's hidden-under-the-hood is that the solution update of a steady solver vs iteration has a correspondence in time. So that a steady solver can be interpreted as a transient solver with a very wonky time-step. It's not so wonky when you realize that it is related to the local flow Courant number of each cell. But rather than the entire solution advancing the same time-step, each cell advances at its own local time-step. You can experience this most readily if you use the COUPLED pressure-velocity coupling scheme which is available in steady or transient solvers and has a flow courant number as an input.

Implicit relaxation slows down the solution so that it advances in time slower. It reduces the effective time-step of the simulation. As I mentioned already, a steady solver is like transient solver with a wonky time-step. The pseudo-transient option makes this wonky time-step smaller.

 Tags convergence, fluent, pseudo-transient

 Thread Tools Search this Thread Search this Thread: Advanced Search Display Modes Linear Mode

 Posting Rules You may not post new threads You may not post replies You may not post attachments You may not edit your posts BB code is On Smilies are On [IMG] code is On HTML code is OffTrackbacks are Off Pingbacks are On Refbacks are On Forum Rules

 Similar Threads Thread Thread Starter Forum Replies Last Post chary FLUENT 20 November 6, 2017 10:12 khaham FLUENT 13 March 15, 2016 09:18 roi247 FLUENT 4 October 15, 2015 14:41 saisanthoshm88 FLUENT 2 December 18, 2012 10:51 colopolo CFX 13 October 4, 2011 23:03

All times are GMT -4. The time now is 15:51.

 Contact Us - CFD Online - Privacy Statement - Top