Computational & Run Time requirements for CAA

danillocafaldo · June 16, 2010, 02:36

Hi,

I'm a undergraduate student and I'm about to start a CAA research in CFX. I've no previous experience in this area, so in order to write my project I need to know the computational and run time requirements for a simple geometry. How many nodes would I need for a LES method?
Please, any comment will be welcome.

Thanks

ghorrocks · June 16, 2010, 09:22

It depends on what you are modelling, of course.

But any LES model will involve large meshes and long run times. LES is very computationally expensive.

danillocafaldo · June 16, 2010, 11:16

Thank you Glenn Horrocks for your answer.

My first idea is to model an airfoil-slat geometry, but I thing I won't have enough computational resource to explore this problem in CFX, mainly because its iterative solver. Maybe a CAA designed software should allow me to do this.

So in order to write my project I need to know what kind of geometry I will be able to explore in CFX.

Thanks

antonio · June 16, 2010, 11:21

Basically in a LES simulation you try to represent directly the large scales and the effects of the small scales are modelled. The required grid spacing h is proportional to a specified filter width that should be smaller than the size of the smallest energy-containing motions. you can see an overview of all modeling approaches in Pope for example.

Regards

ghorrocks · June 16, 2010, 19:38

Quote:

My first idea is to model an airfoil-slat geometry

What Re and Ma numbers?

danillocafaldo · June 17, 2010, 19:07

Re~9e6 e Ma=0.3

ghorrocks · June 17, 2010, 20:40

This is quite a high Re number, meaning that the turbulent structures will be quite small. On top of that you will have compressibility effects with Ma=0.3. This is definitely going to be a challenging model and will require a compute cluster of some size and lots of run time.

danillocafaldo · June 21, 2010, 20:01

Thank you Glenn Horrocks and Antonio for your answers.

I will change my initial idea and simulate a Mixing Layer, that can be interpreted as a very simplified model of slat.
I'm sure that this will be possible to do.

Best regards

June 16, 2010, 02:36	Computational & Run Time requirements for CAA	#1
danillocafaldo New Member Danillo Cafaldo dos Reis Join Date: Jun 2009 Posts: 12 Rep Power: 17	Hi, I'm a undergraduate student and I'm about to start a CAA research in CFX. I've no previous experience in this area, so in order to write my project I need to know the computational and run time requirements for a simple geometry. How many nodes would I need for a LES method? Please, any comment will be welcome. Thanks

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Transient simulation not converging	skabilan	OpenFOAM Running, Solving & CFD	14	December 17, 2019 00:12
Hydrostatic Pressure and Gravity	miliante	OpenFOAM Running, Solving & CFD	132	October 7, 2012 23:50
Differences between serial and parallel runs	carsten	OpenFOAM Bugs	11	September 12, 2008 12:16
Total run time	Dominic	Main CFD Forum	0	August 25, 2006 00:12
computational time dyn-mesh	mange	FLUENT	0	November 19, 2004 13:34

June 16, 2010, 09:22		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	It depends on what you are modelling, of course. But any LES model will involve large meshes and long run times. LES is very computationally expensive.

June 16, 2010, 11:16		#3
danillocafaldo New Member Danillo Cafaldo dos Reis Join Date: Jun 2009 Posts: 12 Rep Power: 17	Thank you Glenn Horrocks for your answer. My first idea is to model an airfoil-slat geometry, but I thing I won't have enough computational resource to explore this problem in CFX, mainly because its iterative solver. Maybe a CAA designed software should allow me to do this. So in order to write my project I need to know what kind of geometry I will be able to explore in CFX. Thanks

June 16, 2010, 11:21		#4
antonio Senior Member Join Date: Jan 2010 Posts: 110 Rep Power: 16	Basically in a LES simulation you try to represent directly the large scales and the effects of the small scales are modelled. The required grid spacing h is proportional to a specified filter width that should be smaller than the size of the smallest energy-containing motions. you can see an overview of all modeling approaches in Pope for example. Regards

June 17, 2010, 19:07		#6
danillocafaldo New Member Danillo Cafaldo dos Reis Join Date: Jun 2009 Posts: 12 Rep Power: 17	Re~9e6 e Ma=0.3

June 17, 2010, 20:40		#7
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	This is quite a high Re number, meaning that the turbulent structures will be quite small. On top of that you will have compressibility effects with Ma=0.3. This is definitely going to be a challenging model and will require a compute cluster of some size and lots of run time.

June 21, 2010, 20:01		#8
danillocafaldo New Member Danillo Cafaldo dos Reis Join Date: Jun 2009 Posts: 12 Rep Power: 17	Thank you Glenn Horrocks and Antonio for your answers. I will change my initial idea and simulate a Mixing Layer, that can be interpreted as a very simplified model of slat. I'm sure that this will be possible to do. Best regards