problem with second order backward Euler scheme

shahpar73 · September 28, 2010, 08:39

Hi
I want to solve my problem transiently which has already been solved in steady state.
I specify the steady state result as initial file and get the solution with "First order backward Euler". But with "Second order backward Euler" the solver crashes at the start. I change the time step but no success, help please...

ghorrocks · September 28, 2010, 18:54

Second order schemes are always less numerically stable than first order. Usually you can get around this with a small initial time step which you ramp up later. Alternately you can start with first order and switch to second order once things have settled down.

Usually it is only the initial transient which causes the convergence problems and as once you have got past the initial bit it works fine.

September 28, 2010, 08:39	problem with second order backward Euler scheme	#1
shahpar73 Member Shahid Parvez Join Date: Jul 2009 Location: Pakistan Posts: 38 Rep Power: 16	Hi I want to solve my problem transiently which has already been solved in steady state. I specify the steady state result as initial file and get the solution with "First order backward Euler". But with "Second order backward Euler" the solver crashes at the start. I change the time step but no success, help please...

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Divergence Only with Second Order Discretization	RyGuy	FLUENT	2	August 11, 2022 17:08
Submerged fin, Convergence problem	supermouniette	FLUENT	10	July 6, 2009 10:47
Backward Euler	smagmon	CFX	3	April 16, 2009 08:10
Runge Kutta Method	CFDtoy	Main CFD Forum	12	May 22, 2005 13:00
Periodic flow boundary condition problem	sudha	FLUENT	3	April 28, 2004 08:40

September 28, 2010, 18:54		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Second order schemes are always less numerically stable than first order. Usually you can get around this with a small initial time step which you ramp up later. Alternately you can start with first order and switch to second order once things have settled down. Usually it is only the initial transient which causes the convergence problems and as once you have got past the initial bit it works fine.