CFD of general missile body

cfd seeker · June 15, 2011, 05:25

hi all, I am new fluent user and i m doing CFD analysis of a general stepped missile body(diameter is different at different locations) in fluent. My aim is to find the CL,CD,CM at M=2 and AOA ranging from 2 to 10 degrees. I am using coupled implicit solver with steady state. I have made the mesh in Gambit and it has 2.6 million cells. Currently i am doing a run at 2 deg AOA and it took 14,000 iterations to converge(not fully converged,still the cofficients has some fluctations). My mesh is very large so my computer is taking 1 min/iteration. I have kept the courant no to 1 because solution diverges as i increase it beyond 1. So my solution is converging very slowly and taking a lot of time. I have heard that there are some MULTIGRID settings which aid in fastening convergence but i am totally illetrate in their knowledge. Can somebody please help me in these MULTIGRID settings or suggest something else which helps me in faster convergence?????
Any suggestions will be welcomed as my solution is taking alot of time and i have limited time to perform analysis at other AOA's.
Thanx in advance

skullCFD · June 15, 2011, 17:06

I have a similar (actually, worse) problem. My grid is 6 million cells. Each iteration is taking 2 min on a quad-core, 64-bit xeon machine and a total of about 30000 ite to converge. I wonder what flow model you are using. One thing I did not understand is why you have a Courant number in a steady simulation?

cfd seeker · June 16, 2011, 01:42

helo, i am using Coupled Implicit solver which uses Courant number even in steady flow and mate do u know anything about Multigrid settings which helps in faster convergence??? if you know anything about them then any ideas from your side will be welcomed.
thanx

madhuvc · June 16, 2011, 13:06

http://my.fit.edu/itresources/manual...g/node1007.htm

http://www.cfd-online.com/Forums/flu...multigrid.html

if this what you guys are talking about....

skullCFD · June 17, 2011, 19:19

Hi cfd seeker and madhuvc, All I know about multigrid methods is from what I read from Fluent manual few years ago. It says the solution progresses from zones with fine grid to zones with coarse grid. So this can be faster compared to solving the equations on the entire grid in a unique way. So you may not see an improvement in convergence time if your entire grid is mostly of the same size. I tried it one time and it did not speed up convergence for my case. Next time it could not obtain a converged solution. So, I think its efficiency and ability to obtain a converged solution depends on the problem i.e., can the problem be solved by such a progression of solution from finer grid to coarser grid, for example, what if you have coarser grids in a problem in the area where the physics is becoming important as the solution progresses but the multigrid solver tries to solve that important zone after solving some physically unimportant zone where the mesh is finer. But I am sure some people might have had success stories with this method. I request anyone who is more knowledgeable about this method in Fluent to kindly correct my understanding if I am wrong. Thanks.

cfd seeker · June 24, 2011, 02:26

hello skullCFD thanx for taking out time, mate that's exactly my understading but as for as theory is concerned this is OK. But i want to know what parameters needs to be modified in that Multigrid panel which helps me. For my your post i can infer that u are not pretty much clear in this regard but you are talking about 1 or 2 expriences, kindly let me know about them.
Thanx

skullCFD · June 24, 2011, 16:59

Hi, The meaning of each of those columns in the panel are given in: http://my.fit.edu/itresources/manual...igrid_Controls. I only changed the 'Max Coarse Levels' when I worked with it previously. Higher number for 'Max Coarse Levels' requires more memory but speeds up convergence. As I experienced before, it can diverge as well. If so, it might work if we slightly reduce this number. BTW, my case is still running

cfd seeker · June 28, 2011, 02:54

thanx skullCFD for replying, i will study this topic in detail and one more thing, i have posted a topic named "Turbulent Boundary Conditions" few days back but i didnt get any reply on that. Kindly take out some time and reply in that thread so that i can get some knowledge regarding that topic.
thanx in advance

June 15, 2011, 05:25	CFD of general missile body	#1
cfd seeker Senior Member Join Date: Mar 2011 Location: Germany Posts: 552 Rep Power: 20	hi all, I am new fluent user and i m doing CFD analysis of a general stepped missile body(diameter is different at different locations) in fluent. My aim is to find the CL,CD,CM at M=2 and AOA ranging from 2 to 10 degrees. I am using coupled implicit solver with steady state. I have made the mesh in Gambit and it has 2.6 million cells. Currently i am doing a run at 2 deg AOA and it took 14,000 iterations to converge(not fully converged,still the cofficients has some fluctations). My mesh is very large so my computer is taking 1 min/iteration. I have kept the courant no to 1 because solution diverges as i increase it beyond 1. So my solution is converging very slowly and taking a lot of time. I have heard that there are some MULTIGRID settings which aid in fastening convergence but i am totally illetrate in their knowledge. Can somebody please help me in these MULTIGRID settings or suggest something else which helps me in faster convergence????? Any suggestions will be welcomed as my solution is taking alot of time and i have limited time to perform analysis at other AOA's. Thanx in advance

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Lifting Body CFD Analysis	Doug Gard	Main CFD Forum	2	August 18, 2006 17:41
General CFD Question	sam kashan	Main CFD Forum	15	June 18, 2002 09:21
CFD JOBS and Expected Salary....	Noel Harrison	Main CFD Forum	11	November 22, 2000 07:15
ASME CFD Symposium, Atlanta, 22-26 July 2001	Chris R. Kleijn	Main CFD Forum	0	August 1, 2000 10:07
public CFD Code development	Heinz Wilkening	Main CFD Forum	38	March 5, 1999 11:44

June 15, 2011, 17:06		#2
skullCFD Member Join Date: Apr 2011 Posts: 30 Rep Power: 14	I have a similar (actually, worse) problem. My grid is 6 million cells. Each iteration is taking 2 min on a quad-core, 64-bit xeon machine and a total of about 30000 ite to converge. I wonder what flow model you are using. One thing I did not understand is why you have a Courant number in a steady simulation?

June 16, 2011, 01:42		#3
cfd seeker Senior Member Join Date: Mar 2011 Location: Germany Posts: 552 Rep Power: 20	helo, i am using Coupled Implicit solver which uses Courant number even in steady flow and mate do u know anything about Multigrid settings which helps in faster convergence??? if you know anything about them then any ideas from your side will be welcomed. thanx

June 16, 2011, 13:06		#4
madhuvc New Member MadhuVC Join Date: Feb 2011 Posts: 28 Rep Power: 15	http://my.fit.edu/itresources/manual...g/node1007.htm http://www.cfd-online.com/Forums/flu...multigrid.html if this what you guys are talking about....

June 17, 2011, 19:19		#5
skullCFD Member Join Date: Apr 2011 Posts: 30 Rep Power: 14	Hi cfd seeker and madhuvc, All I know about multigrid methods is from what I read from Fluent manual few years ago. It says the solution progresses from zones with fine grid to zones with coarse grid. So this can be faster compared to solving the equations on the entire grid in a unique way. So you may not see an improvement in convergence time if your entire grid is mostly of the same size. I tried it one time and it did not speed up convergence for my case. Next time it could not obtain a converged solution. So, I think its efficiency and ability to obtain a converged solution depends on the problem i.e., can the problem be solved by such a progression of solution from finer grid to coarser grid, for example, what if you have coarser grids in a problem in the area where the physics is becoming important as the solution progresses but the multigrid solver tries to solve that important zone after solving some physically unimportant zone where the mesh is finer. But I am sure some people might have had success stories with this method. I request anyone who is more knowledgeable about this method in Fluent to kindly correct my understanding if I am wrong. Thanks.

June 24, 2011, 02:26		#6
cfd seeker Senior Member Join Date: Mar 2011 Location: Germany Posts: 552 Rep Power: 20	hello skullCFD thanx for taking out time, mate that's exactly my understading but as for as theory is concerned this is OK. But i want to know what parameters needs to be modified in that Multigrid panel which helps me. For my your post i can infer that u are not pretty much clear in this regard but you are talking about 1 or 2 expriences, kindly let me know about them. Thanx

June 24, 2011, 16:59		#7
skullCFD Member Join Date: Apr 2011 Posts: 30 Rep Power: 14	Hi, The meaning of each of those columns in the panel are given in: http://my.fit.edu/itresources/manual...igrid_Controls. I only changed the 'Max Coarse Levels' when I worked with it previously. Higher number for 'Max Coarse Levels' requires more memory but speeds up convergence. As I experienced before, it can diverge as well. If so, it might work if we slightly reduce this number. BTW, my case is still running

June 28, 2011, 02:54		#8
cfd seeker Senior Member Join Date: Mar 2011 Location: Germany Posts: 552 Rep Power: 20	thanx skullCFD for replying, i will study this topic in detail and one more thing, i have posted a topic named "Turbulent Boundary Conditions" few days back but i didnt get any reply on that. Kindly take out some time and reply in that thread so that i can get some knowledge regarding that topic. thanx in advance