How to decide time step size

caohan · September 6, 2010, 08:08

Hi all,
In a unsteady simulation, one thing always confus me which is how to decide time step size and how many time steps is enough, hope someone can explain for me
Regards
Han

tobytoby · September 6, 2010, 23:25

Hi Caohan,

Actually it does not have any requirement for the time step, yet I can share some exp. of myself to you.

1. base on the velocity equation: velocity = distance /time step
where the velocity is refered to the design velocity in your model; and the distance is the length between the inlet and the first contacted wall. So, it gives a rough reference for designing the time step.

2. Guess the time step first and modify it by experience (more trials within the same model). This saying seems rubbish .... I know ... yet it is always true. Jsut set a smaller iteration size first, and extend it after you confirm a suitable time step.

Gook luck to you,
Toby

pepi33 · September 8, 2010, 02:54

Hi all,

Firt of all, sorry for my poor english.

For me, it depends on what phenomenon you want to see.
For exemple if you want the very low scale near the wall region, you have to use this :
it is still the same concept but here time step= syze of the smaller mesh (near the wall region)/Local velocity in this smaller mesh
For exemple, at this moment i'm using a time step of 0.0001s.

Concerning the convergence of the time of the time step you have to use this concept (i think):
For your first case you have to calculate for a hihgt time (such as 1s, or more it depends on your model) . For all iterations you plot a characteristic (pressure for exemple) on one representative point (or more) of your mesh.
Basically, if your mesh an turbulence model is good you will have a sine curve (pressure/time) less and less tall (difference between +/-). Concerning to your criterion, wou will able to say at this time step it is enough converge for me.

I hope it will help you,
Best regards,
pepi

Philipov · September 8, 2010, 03:30

Time step must be lower then ratio of the minimum cell size and the expected velocity at that region. I am saying expected velocity because in general you may have reactions (for example combustion) ...
How many time steps? It depends on the total time you want to calculate
number of time steps=(total time)/(timestep) or something like this....

September 6, 2010, 08:08	How to decide time step size	#1
caohan Member tom Join Date: Feb 2010 Posts: 46 Rep Power: 16	Hi all, In a unsteady simulation, one thing always confus me which is how to decide time step size and how many time steps is enough, hope someone can explain for me Regards Han

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Transient simulation not converging	skabilan	OpenFOAM Running, Solving & CFD	14	December 16, 2019 23:12
Time step size, number of time steps and max iterations per time step	guido_88	FLUENT	4	August 30, 2012 14:49
critical error during installation of openfoam	Fabio88	OpenFOAM Installation	21	June 2, 2010 03:01
Modeling in micron scale using icoFoam	m9819348	OpenFOAM Running, Solving & CFD	7	October 27, 2007 00:36
Time Step Size for Airfoil Model - Fluent	Prateek	Main CFD Forum	0	November 4, 2004 18:42

September 6, 2010, 23:25		#2
tobytoby New Member Toby Cheung Join Date: Aug 2010 Posts: 3 Rep Power: 15	Hi Caohan, Actually it does not have any requirement for the time step, yet I can share some exp. of myself to you. 1. base on the velocity equation: velocity = distance /time step where the velocity is refered to the design velocity in your model; and the distance is the length between the inlet and the first contacted wall. So, it gives a rough reference for designing the time step. 2. Guess the time step first and modify it by experience (more trials within the same model). This saying seems rubbish .... I know ... yet it is always true. Jsut set a smaller iteration size first, and extend it after you confirm a suitable time step. Gook luck to you, Toby

September 8, 2010, 02:54		#3
pepi33 New Member cerisier Join Date: Sep 2010 Posts: 5 Rep Power: 15	Hi all, Firt of all, sorry for my poor english. For me, it depends on what phenomenon you want to see. For exemple if you want the very low scale near the wall region, you have to use this : it is still the same concept but here time step= syze of the smaller mesh (near the wall region)/Local velocity in this smaller mesh For exemple, at this moment i'm using a time step of 0.0001s. Concerning the convergence of the time of the time step you have to use this concept (i think): For your first case you have to calculate for a hihgt time (such as 1s, or more it depends on your model) . For all iterations you plot a characteristic (pressure for exemple) on one representative point (or more) of your mesh. Basically, if your mesh an turbulence model is good you will have a sine curve (pressure/time) less and less tall (difference between +/-). Concerning to your criterion, wou will able to say at this time step it is enough converge for me. I hope it will help you, Best regards, pepi

September 8, 2010, 03:30		#4
Philipov Senior Member Svetlin Filipov Join Date: Mar 2009 Location: United Kingdom Posts: 176 Rep Power: 17	Time step must be lower then ratio of the minimum cell size and the expected velocity at that region. I am saying expected velocity because in general you may have reactions (for example combustion) ... How many time steps? It depends on the total time you want to calculate number of time steps=(total time)/(timestep) or something like this....