Time step for implicit solver

Far · March 24, 2013, 10:27

I am simulating low pressure turbine unsteady flow and I have couple of questions:

1. How to decide time step for the implicit solver?

2. Do we need to satisfy CFL < 1 for implicit solver?

3. Why implicit solver is preferred in commercial codes and as I understand for unsteady flows explicit solver would be better option.

4. Or explicit solver is only efficient in LES where turbulent scales to be resolved are the same order as the time step dictated by CFL < 0.2 condition?

5. For the same geometry and flow, time step is inversely proportional to Reynolds number?

flotus1 · March 24, 2013, 11:10

A CFL number smaller than 1 is a stability constraint for some explicit solvers which does not apply to the implicit ones.

This partly answers question 3, because with the implicit solvers, the timestep size can be chosen according to the flow considered and does not have to fulfill a CFL criterium.

Since a CFL number below 1 is a good guess for the timestep in LES, the advantage of implicit solvers allowing for larger timesteps becomes irrelevant, since the timestep has to be small anyway.

According to Kolmogorovs turbulence theory, the timescale of the smallest eddies is proportional to $Re^{-1/2}$ . Yet this is only relevant for DNS.
In a LES, there are different approaches to estimate an appropriate time step size, like CFL<1 for example.

FMDenaro · March 24, 2013, 16:53

the choice is driven by the numerical stability as well as accuracy requirements.... often, implicit scheme are not "pure", in the sense they are mixed in an implicit and an explicit part. For example, a classical integration used in DNS/LES is the Crank-Nicolson for the diffusion and the Adams-Basforth for the convection. This result in a conditionally stable scheme that is not stable only under the satisfation of the cfl<1 condition. It exists a stability region in the cfl-Reh plane.

March 24, 2013, 10:27	Time step for implicit solver	#1
Far Senior Member Sijal Join Date: Mar 2009 Location: Islamabad Posts: 4,558 Blog Entries: 6 Rep Power: 54	I am simulating low pressure turbine unsteady flow and I have couple of questions: 1. How to decide time step for the implicit solver? 2. Do we need to satisfy CFL < 1 for implicit solver? 3. Why implicit solver is preferred in commercial codes and as I understand for unsteady flows explicit solver would be better option. 4. Or explicit solver is only efficient in LES where turbulent scales to be resolved are the same order as the time step dictated by CFL < 0.2 condition? 5. For the same geometry and flow, time step is inversely proportional to Reynolds number?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Time step size and max iterations per time step	pUl\|	FLUENT	31	October 23, 2020 23:50
Transient simulation not converging	skabilan	OpenFOAM Running, Solving & CFD	14	December 17, 2019 00:12
InterFoam negative alpha	karasa03	OpenFOAM	7	December 12, 2013 04:41
Upgraded from Karmic Koala 9.10 to Lucid Lynx10.04.3	bookie56	OpenFOAM Installation	8	August 13, 2011 05:03
Problems with simulating TurbFOAM	barath.ezhilan	OpenFOAM	13	July 16, 2009 06:55

March 24, 2013, 11:10		#2
flotus1 Super Moderator Alex Join Date: Jun 2012 Location: Germany Posts: 3,427 Rep Power: 49	A CFL number smaller than 1 is a stability constraint for some explicit solvers which does not apply to the implicit ones. This partly answers question 3, because with the implicit solvers, the timestep size can be chosen according to the flow considered and does not have to fulfill a CFL criterium. Since a CFL number below 1 is a good guess for the timestep in LES, the advantage of implicit solvers allowing for larger timesteps becomes irrelevant, since the timestep has to be small anyway. According to Kolmogorovs turbulence theory, the timescale of the smallest eddies is proportional to $Re^{-1/2}$ . Yet this is only relevant for DNS. In a LES, there are different approaches to estimate an appropriate time step size, like CFL<1 for example.

March 24, 2013, 16:53		#3
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,882 Rep Power: 73	the choice is driven by the numerical stability as well as accuracy requirements.... often, implicit scheme are not "pure", in the sense they are mixed in an implicit and an explicit part. For example, a classical integration used in DNS/LES is the Crank-Nicolson for the diffusion and the Adams-Basforth for the convection. This result in a conditionally stable scheme that is not stable only under the satisfation of the cfl<1 condition. It exists a stability region in the cfl-Reh plane.