Are you sure about grid independence?

Vignesh2508 · August 23, 2019, 03:24

Guys,

I was thinking about the grid independence test, which is quite easy when having a single simulation. But when simulating multiple sets with varying initial conditions, can we use a single grid after conducting the independence study. Maybe that grid is independent for that set of initial conditions (how do we confirm?). Does that mean one should conduct grid independence study each time one changes the initial conditions (even though it is expensive to do so).

Thanks

Vicky

FMDenaro · August 23, 2019, 04:08

For a steady laminar problem the final solution does not depend on the guess solution. Your question is well focused in case of time-dependent problems where changing the initial condition drives to different evolution.

Vignesh2508 · August 23, 2019, 04:49

But even in steady-state problems, there could be different steady states based on the initial conditions right?

So when has to one stop doing the grid independence study?

FMDenaro · August 23, 2019, 05:02

Quote:

Originally Posted by Vignesh2508

But even in steady-state problems, there could be different steady states based on the initial conditions right?

So when has to one stop doing the grid independence study?

For a steady problem there is not a mathematical initial value condition but only boundary conditions. The “initial” condition you are talking about is, actually, a guess zero iteration.

Carlo_P · August 23, 2019, 05:34

Hey Vicky, I think you are talking about different BCs, not itial conditions, right?

Are they simply dfferent in value (same speed, but different values) or also in type (once you impose the speed, another time pressure, etc..etc..)?

You can give a look at my master Thesis, from page 60:
https://www.researchgate.net/publica...LEX_HULL_FORMS

In the end, if the speed doesn't change too much, the bahaviuor is the same. So, you can run a sensitivity at the highest speed. If the result at the highest speed is indipendent by the mesh, you can think that it is ok also for other speeds.

Cheers,
Carlo

Vignesh2508 · August 23, 2019, 05:48

Thank you Carlos and Denaro for the explanation. Yes I meant the BC's. Take an internal duct flow scenario with varying velocities for each case. You mean to say that it is enough to do the independence study for the maximum velocity case and use the same grid for other cases, right?

Carlo_P · August 23, 2019, 07:14

I'm not sure, obviously, but it is a good way.
I mean, normally with coars mesh you don't have good results becasue you miss some local pick.
The biggest pick are at higher speed. If you have good mesh at higher speed, it is good enoght for smaller speed.
Normally. It is not a maths rule. But in thesis works and the common sense tells that it is correct.

August 23, 2019, 03:24	Are you sure about grid independence?	#1
Vignesh2508 Member Vignesh Rajendiran Join Date: Aug 2016 Location: Chennai, India Posts: 62 Rep Power: 10	Guys, I was thinking about the grid independence test, which is quite easy when having a single simulation. But when simulating multiple sets with varying initial conditions, can we use a single grid after conducting the independence study. Maybe that grid is independent for that set of initial conditions (how do we confirm?). Does that mean one should conduct grid independence study each time one changes the initial conditions (even though it is expensive to do so). Thanks Vicky

August 23, 2019, 05:34		#5
Carlo_P Senior Member Carlo_P Join Date: May 2019 Location: Italy Posts: 176 Rep Power: 8	Hey Vicky, I think you are talking about different BCs, not itial conditions, right? Are they simply dfferent in value (same speed, but different values) or also in type (once you impose the speed, another time pressure, etc..etc..)? You can give a look at my master Thesis, from page 60: https://www.researchgate.net/publica...LEX_HULL_FORMS In the end, if the speed doesn't change too much, the bahaviuor is the same. So, you can run a sensitivity at the highest speed. If the result at the highest speed is indipendent by the mesh, you can think that it is ok also for other speeds. Cheers, Carlo FMDenaro likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Grid and time step independence study	ndabir	FLUENT	4	February 16, 2015 17:40
Grid independence check in transient simulation	abhi084	Main CFD Forum	1	December 4, 2014 16:55
Grid independence test issue	djordje8	STAR-CCM+	0	September 8, 2014 11:22
grid independence test issue	djordje8	STAR-CCM+	0	September 8, 2014 11:21
vof grid independence and validation	vof_grid	FLUENT	4	March 19, 2007 14:40

August 23, 2019, 04:08		#2
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,896 Rep Power: 73	For a steady laminar problem the final solution does not depend on the guess solution. Your question is well focused in case of time-dependent problems where changing the initial condition drives to different evolution.

August 23, 2019, 04:49		#3
Vignesh2508 Member Vignesh Rajendiran Join Date: Aug 2016 Location: Chennai, India Posts: 62 Rep Power: 10	But even in steady-state problems, there could be different steady states based on the initial conditions right? So when has to one stop doing the grid independence study?

August 23, 2019, 05:48		#6
Vignesh2508 Member Vignesh Rajendiran Join Date: Aug 2016 Location: Chennai, India Posts: 62 Rep Power: 10	Thank you Carlos and Denaro for the explanation. Yes I meant the BC's. Take an internal duct flow scenario with varying velocities for each case. You mean to say that it is enough to do the independence study for the maximum velocity case and use the same grid for other cases, right?

August 23, 2019, 07:14		#7
Carlo_P Senior Member Carlo_P Join Date: May 2019 Location: Italy Posts: 176 Rep Power: 8	I'm not sure, obviously, but it is a good way. I mean, normally with coars mesh you don't have good results becasue you miss some local pick. The biggest pick are at higher speed. If you have good mesh at higher speed, it is good enoght for smaller speed. Normally. It is not a maths rule. But in thesis works and the common sense tells that it is correct.