How do I select solver options for external flow over an aircraft by fluent?

hadieliasi · April 20, 2011, 02:12

hi
I have several problems about external flow over an aircraft with fluent.
its mesh is tetrahedral(3D) with about 3000000 milions node.
These options is set in fluent:
pressure-based
steady-state
air: ideal gas
turbulent: sparlat almaras model
BC: 1- pressure farfield(mach number=.8, flow direction(1 0 0))
, 2- wall for aircraft body
coupled algorithm
cfl=200
I used hybrid initialization & FMG( Full Multigrid) initilization.
in these conditions, Cl,Cd have high scale( multiple by 10+8). the solution is converged but i know this solution is wrong.
please help me about this problem.
How do I select solver options for external flow over an aircraft by fluent?

alastormoody11 · April 20, 2011, 07:05

One suggestion is that you should use the density based solver for your simulation considering the fact that your Mach number is 0.8.

Also how are you judging the convergence of your solution.

Do your Cl and Cd values stabilize when your solution is converged?

hadieliasi · April 24, 2011, 00:54

hi. thanks for your attention.
I selected this method but CL & CD have larger scale than that one.
don't you think that this problem is about its mesh?

alastormoody11 · April 25, 2011, 02:42

Since you are getting such high values for your Cl and Cd please chk your reference conditions.

I think that you have specificied your operating pressure as 0, specify it as 101325 and then check again.

Please read up what the other reference values need to be, it will make a differennce whether you are using 2D or 3D.

Also for an accurate Cd you might want to consider using more established turbulence models.

Also in the FLUENT manual they have suggested some recommended ranges for the wall y plus values so please chk that you are ok on that front.

hadieliasi · May 2, 2011, 04:29

i check it but there are these problems.
i do mesh finer than that one and those scale was better than previous solution ( about 10e+2) but the convergence was very slow.
however, i think when its mesh become finer the solution become better. but in this way there is a problem that is mentioned above that it has very slow convergence and when the mesh is more finer ,my system ( PC) cant do it despite my system configuration is:
1-core i7
2- 18 GB Ram
3- win 7 , 64bit
what is your opinion about parallel solution?
i don't know how do i do system paralleling.
do you know? could you tell me more information about system paralleling, if you could?
thank you.

alastormoody11 · May 2, 2011, 04:54

I do not claim to be an expert on parallel processing so the following answers are from my experience and not definite know-how.

Firstly even if you have are not running the simulation in parallel, you can first check the utilization of each processor, if while running your simulation in serial mode, both your processors are running at full capacity, running in parallel is not going to make a whole lot of difference.

I do not think this will be the case for you since you are using a quad/8 core machine.

If your processors are not being utilized, you can give the simulation run in local parallel, the instruction as to how to do this are clearly mentioned in the manual.

While running in parallel, one of your major concern should be the efficiency you are getting. When you give the simulation in parallel, the CV is basically divided into a number of pieces equal to the number you mentioned as the number of processors given(in case of local parallel), What you have to ensure that time required for communication between these pieces is balanced by the added computational muscle courtesy of increased number of processors.

Also have you checked the wall y plus values, are they in accordance with the recommended values in the FLUENT manual?

Once you get your simulation up and running, it might be a good idea to do a sensitivity study.

April 20, 2011, 02:12	How do I select solver options for external flow over an aircraft by fluent?	#1
hadieliasi New Member hadieliasi Join Date: Apr 2011 Posts: 10 Rep Power: 15	hi I have several problems about external flow over an aircraft with fluent. its mesh is tetrahedral(3D) with about 3000000 milions node. These options is set in fluent: pressure-based steady-state air: ideal gas turbulent: sparlat almaras model BC: 1- pressure farfield(mach number=.8, flow direction(1 0 0)) , 2- wall for aircraft body coupled algorithm cfl=200 I used hybrid initialization & FMG( Full Multigrid) initilization. in these conditions, Cl,Cd have high scale( multiple by 10+8). the solution is converged but i know this solution is wrong. please help me about this problem. How do I select solver options for external flow over an aircraft by fluent?

April 24, 2011, 00:54	cfd	#3
hadieliasi New Member hadieliasi Join Date: Apr 2011 Posts: 10 Rep Power: 15	hi. thanks for your attention. I selected this method but CL & CD have larger scale than that one. don't you think that this problem is about its mesh?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
mach number and selection of the fluent solver	turbinesv	FLUENT	4	April 24, 2011 01:14
Flow Development in Fluent	sth	Main CFD Forum	0	November 30, 2010 12:26
On Setring Step Size in unsteady Solver of FLUENT	lzgwhy	FLUENT	0	August 19, 2009 23:44
Can segregated solver be used for compressible flow?	Steven-GY	ANSYS	0	May 14, 2009 11:37
fluid flow fundas	ram	Main CFD Forum	5	June 17, 2000 22:31

April 20, 2011, 07:05		#2
alastormoody11 Senior Member Join Date: Jun 2009 Posts: 100 Rep Power: 17	One suggestion is that you should use the density based solver for your simulation considering the fact that your Mach number is 0.8. Also how are you judging the convergence of your solution. Do your Cl and Cd values stabilize when your solution is converged?

April 25, 2011, 02:42		#4
alastormoody11 Senior Member Join Date: Jun 2009 Posts: 100 Rep Power: 17	Since you are getting such high values for your Cl and Cd please chk your reference conditions. I think that you have specificied your operating pressure as 0, specify it as 101325 and then check again. Please read up what the other reference values need to be, it will make a differennce whether you are using 2D or 3D. Also for an accurate Cd you might want to consider using more established turbulence models. Also in the FLUENT manual they have suggested some recommended ranges for the wall y plus values so please chk that you are ok on that front.

May 2, 2011, 04:29		#5
hadieliasi New Member hadieliasi Join Date: Apr 2011 Posts: 10 Rep Power: 15	i check it but there are these problems. i do mesh finer than that one and those scale was better than previous solution ( about 10e+2) but the convergence was very slow. however, i think when its mesh become finer the solution become better. but in this way there is a problem that is mentioned above that it has very slow convergence and when the mesh is more finer ,my system ( PC) cant do it despite my system configuration is: 1-core i7 2- 18 GB Ram 3- win 7 , 64bit what is your opinion about parallel solution? i don't know how do i do system paralleling. do you know? could you tell me more information about system paralleling, if you could? thank you.

May 2, 2011, 04:54		#6
alastormoody11 Senior Member Join Date: Jun 2009 Posts: 100 Rep Power: 17	I do not claim to be an expert on parallel processing so the following answers are from my experience and not definite know-how. Firstly even if you have are not running the simulation in parallel, you can first check the utilization of each processor, if while running your simulation in serial mode, both your processors are running at full capacity, running in parallel is not going to make a whole lot of difference. I do not think this will be the case for you since you are using a quad/8 core machine. If your processors are not being utilized, you can give the simulation run in local parallel, the instruction as to how to do this are clearly mentioned in the manual. While running in parallel, one of your major concern should be the efficiency you are getting. When you give the simulation in parallel, the CV is basically divided into a number of pieces equal to the number you mentioned as the number of processors given(in case of local parallel), What you have to ensure that time required for communication between these pieces is balanced by the added computational muscle courtesy of increased number of processors. Also have you checked the wall y plus values, are they in accordance with the recommended values in the FLUENT manual? Once you get your simulation up and running, it might be a good idea to do a sensitivity study.