How to decide time step size, number of time steps

parmeshwar · June 3, 2016, 13:40

I am working on transient simulation. I am confused with following terms.

1. time step size= how to decide this, many users uses 0.01 or 0.1 or in this range only. What is the logic in deciding this time step size.

2. how to decide number of time steps

3. maximum number of iterations per time steps

thnks in advance

Sparow89 · June 6, 2016, 13:15

1. What kind of transient simulation are you running? If you are expecting a frequency response my general rule for calculating time step size is to take the inverse of the frequency (the period) and divide it by 20 for the maximum time step. This will give you twenty steps per period and wouldn't be a bad starting point. Depending on how the run goes you may need to decrease your time step further. I am fairly new to CFD so if anyone has a better method please correct me.

2. This one takes some experience I think to get a good idea of how long you should let your simulations run. You are trying to decide how long it will take for the wanted flow regime to form. This is very dependent on the simulation you want to run. If you are working through workbench and the GUI I would suggest setting a flow time much higher than you think you would need and monitor residuals and a surface output to check for convergence. Once your solution is converged you can stop the simulation and use the time step at that point for future calculations.

3. I would stick with the default 20 sub-iterations at first. If you google around or look through this forum you will find discussions of when and why your would change this.

parmeshwar · June 6, 2016, 13:30

[QUOTE=Sparow89;603600]1. What kind of transient simulation are you running? If you are expecting a frequency response my general rule for calculating time step size is to take the inverse of the frequency (the period) and divide it by 20 for the maximum time step. This will give you twenty steps per period and wouldn't be a bad starting point. Depending on how the run goes you may need to decrease your time step further. I am fairly new to CFD so if anyone has a better method please correct me.

Ans. I want temperature variation of water which is heated by solar radiation with respect time. that is how temperature of water is increasing with time in a day. I didn't get what you write aboe, can you elaborate it.

2. This one takes some experience I think to get a good idea of how long you should let your simulations run. You are trying to decide how long it will take for the wanted flow regime to form. This is very dependent on the simulation you want to run. If you are working through workbench and the GUI I would suggest setting a flow time much higher than you think you would need and monitor residuals and a surface output to check for convergence. Once your solution is converged you can stop the simulation and use the time step at that point for future calculations.

Ans.I got it now..........this is helpfull for me........thanks a lot......

3. I would stick with the default 20 sub-iterations at first. If you google around or look through this forum you will find discussions of when and why your would change this.

Sparow89 · June 6, 2016, 13:35

I was thinking about a fluid flow simulation not heat transfer. I haven't worked with heat transfer in Fluent, but I would think about it by looking at the rate of heat transfer to the water and dividing the time component by some division (20-50 or so) to get a reasonable time step.

parmeshwar · June 6, 2016, 13:40

Quote:

Originally Posted by Sparow89

I was thinking about a fluid flow simulation not heat transfer. I haven't worked with heat transfer in Fluent, but I would think about it by looking at the rate of heat transfer to the water and dividing the time component by some division (20-50 or so) to get a reasonable time step.

I am not getting this
dividing the time component by some division (20-50 or so) to get a reasonable time step

can you elaborate it

Sparow89 · June 6, 2016, 14:04

I would imagine for solar heating which is slow and over a long period of time you could use a large time step. Try taking your 24 hour period and dividing it by 100 to get your time step (around 14 minutes or 860 seconds) and see how it looks. Then try lowering your time step and see if it changes much. If you have an okay computer and your element size isn't very large it should run quickly.

parmeshwar · June 6, 2016, 14:14

Quote:

Originally Posted by Sparow89

I would imagine for solar heating which is slow and over a long period of time you could use a large time step. Try taking your 24 hour period and dividing it by 100 to get your time step (around 14 minutes or 860 seconds) and see how it looks. Then try lowering your time step and see if it changes much. If you have an okay computer and your element size isn't very large it should run quickly.

Thank you very much Sparow89.........

June 3, 2016, 13:40	How to decide time step size, number of time steps	#1
parmeshwar New Member Param Join Date: Oct 2015 Posts: 13 Rep Power: 10	I am working on transient simulation. I am confused with following terms. 1. time step size= how to decide this, many users uses 0.01 or 0.1 or in this range only. What is the logic in deciding this time step size. 2. how to decide number of time steps 3. maximum number of iterations per time steps thnks in advance

June 6, 2016, 13:30	Transient simulation	#3
parmeshwar New Member Param Join Date: Oct 2015 Posts: 13 Rep Power: 10	[QUOTE=Sparow89;603600]1. What kind of transient simulation are you running? If you are expecting a frequency response my general rule for calculating time step size is to take the inverse of the frequency (the period) and divide it by 20 for the maximum time step. This will give you twenty steps per period and wouldn't be a bad starting point. Depending on how the run goes you may need to decrease your time step further. I am fairly new to CFD so if anyone has a better method please correct me. Ans. I want temperature variation of water which is heated by solar radiation with respect time. that is how temperature of water is increasing with time in a day. I didn't get what you write aboe, can you elaborate it. 2. This one takes some experience I think to get a good idea of how long you should let your simulations run. You are trying to decide how long it will take for the wanted flow regime to form. This is very dependent on the simulation you want to run. If you are working through workbench and the GUI I would suggest setting a flow time much higher than you think you would need and monitor residuals and a surface output to check for convergence. Once your solution is converged you can stop the simulation and use the time step at that point for future calculations. Ans.I got it now..........this is helpfull for me........thanks a lot...... 3. I would stick with the default 20 sub-iterations at first. If you google around or look through this forum you will find discussions of when and why your would change this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Floating point exception error	lpz_michele	OpenFOAM Running, Solving & CFD	53	October 19, 2015 02:50
simpleFoam error - "Floating point exception"	mbcx4jc2	OpenFOAM Running, Solving & CFD	12	August 4, 2015 02:20
How to write k and epsilon before the abnormal end	xiuying	OpenFOAM Running, Solving & CFD	8	August 27, 2013 15:33
calculation stops after few time steps	sivakumar	OpenFOAM Running, Solving & CFD	7	March 17, 2013 06:37
IcoFoam parallel woes	msrinath80	OpenFOAM Running, Solving & CFD	9	July 22, 2007 02:58

June 6, 2016, 13:15		#2
Sparow89 New Member Stephen Join Date: Jun 2016 Posts: 8 Rep Power: 10	1. What kind of transient simulation are you running? If you are expecting a frequency response my general rule for calculating time step size is to take the inverse of the frequency (the period) and divide it by 20 for the maximum time step. This will give you twenty steps per period and wouldn't be a bad starting point. Depending on how the run goes you may need to decrease your time step further. I am fairly new to CFD so if anyone has a better method please correct me. 2. This one takes some experience I think to get a good idea of how long you should let your simulations run. You are trying to decide how long it will take for the wanted flow regime to form. This is very dependent on the simulation you want to run. If you are working through workbench and the GUI I would suggest setting a flow time much higher than you think you would need and monitor residuals and a surface output to check for convergence. Once your solution is converged you can stop the simulation and use the time step at that point for future calculations. 3. I would stick with the default 20 sub-iterations at first. If you google around or look through this forum you will find discussions of when and why your would change this.

June 6, 2016, 13:35		#4
Sparow89 New Member Stephen Join Date: Jun 2016 Posts: 8 Rep Power: 10	I was thinking about a fluid flow simulation not heat transfer. I haven't worked with heat transfer in Fluent, but I would think about it by looking at the rate of heat transfer to the water and dividing the time component by some division (20-50 or so) to get a reasonable time step.

June 6, 2016, 14:04		#6
Sparow89 New Member Stephen Join Date: Jun 2016 Posts: 8 Rep Power: 10	I would imagine for solar heating which is slow and over a long period of time you could use a large time step. Try taking your 24 hour period and dividing it by 100 to get your time step (around 14 minutes or 860 seconds) and see how it looks. Then try lowering your time step and see if it changes much. If you have an okay computer and your element size isn't very large it should run quickly.