How to estimate time to complete simulation.

frankyd · May 4, 2010, 14:25

Is there a way to estimate how long Flow-3D will take to run?

Items that are always the same include ...
Linux OS, 4 Gb ram
Quad core machine. (set OMP_NUM_THREADS="4")

Items that vary ...
Number of time steps
Number of cells in X, Y, and Z

My assumptions are ...
1.) Doubleing the number of time steps should double the run time.
2.) Double the cells in X, Y and Z and it should take 2^3 or 8 times longer.

It seems like my second assumption is terribly incorrect.

Thanks,

Frank D.

MuxaB · May 5, 2010, 00:02

this is all very approximate, but typically, if the time step is controlled by convection, i.e, dx/U, then doubling the number of cells in all three directions will increase the run time by 2x2x2x2=16 times since the time step will also be cut in half.

If you have transient free surface then it makes the estimate more complicated because the time step can vary a lot during simulation due to the wave movement and splashing, plus the domain occupied with fluid (as opposed to void) and hence the amount of computational work may also vary with time.

May 4, 2010, 14:25	How to estimate time to complete simulation.	#1
frankyd New Member Frank D. Join Date: Jun 2009 Location: Oregon Posts: 5 Rep Power: 16	Is there a way to estimate how long Flow-3D will take to run? Items that are always the same include ... Linux OS, 4 Gb ram Quad core machine. (set OMP_NUM_THREADS="4") Items that vary ... Number of time steps Number of cells in X, Y, and Z My assumptions are ... 1.) Doubleing the number of time steps should double the run time. 2.) Double the cells in X, Y and Z and it should take 2^3 or 8 times longer. It seems like my second assumption is terribly incorrect. Thanks, Frank D.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Transient simulation not converging	skabilan	OpenFOAM Running, Solving & CFD	14	December 16, 2019 23:12
How to write k and epsilon before the abnormal end	xiuying	OpenFOAM Running, Solving & CFD	8	August 27, 2013 15:33
Hydrostatic Pressure and Gravity	miliante	OpenFOAM Running, Solving & CFD	132	October 7, 2012 22:50
Time step for combustion simulation	AdidaKK	CFX	11	December 3, 2009 11:47
IcoFoam parallel woes	msrinath80	OpenFOAM Running, Solving & CFD	9	July 22, 2007 02:58

May 5, 2010, 00:02		#2
MuxaB Senior Member michael barkhudarov Join Date: Mar 2009 Location: Sante Fe, New Mexico, USA Posts: 336 Rep Power: 18	this is all very approximate, but typically, if the time step is controlled by convection, i.e, dx/U, then doubling the number of cells in all three directions will increase the run time by 2x2x2x2=16 times since the time step will also be cut in half. If you have transient free surface then it makes the estimate more complicated because the time step can vary a lot during simulation due to the wave movement and splashing, plus the domain occupied with fluid (as opposed to void) and hence the amount of computational work may also vary with time.