Hi all,

Just a quick question. Does CFX use dual-time stepping method for the transient simulation? If not, what technique is being applied? How can I evaluate the Courant number in CFX? Can't find much information about it in the user manual. Thanks, Pete

Hi,

What is dual time stepping? I suspect the answer is no, it evaluates the transient terms with first or second order terms. It is described in the manual.

CFX10 reports Courant numbers of the transient as it progresses, and you can use adaptive timestepping based on Courant number (and lots of other parameters too)

Regards, Glenn Horrocks

If by dual timestepping you mean that an inner iteration is used to resolve nonlinearities within the timestep, the answer is yes. These inner iterations are the 'coefficient loops'.

Thanks a lot, guys. I read from one Journal paper that the authors use the dual-time-stepping AC mdthod of Merkle and Athavale(1987) in their transient simulation. The formula is:

[Q(l+1,n+1)-Q(l,n+1)]/tau+Gamma*[1.5Qijk(l+1,n+1)-2Qijk(n)+0.5Qijk(n-1)]/2t+Rijk(l+1,n+1)=0

tau=dual time increment;

l=dual iteration level;

n+1=new physical timestep;

Q(l+1,n+1)=l+1 dual-time iterate of the final solution at new physical time step n+1;

ijk=grid node in space;

I only found in pg233 about the discretization of transient term. Not sure about if we're using the same method in CFX. So just to confirm that we're talking about the same subject here... If this is exactly the same, please kindly let me know. Thanks a lot for your time.

Opps... please igore the '2' beside 't' in the second term .... typing mistake. sorry about that.

R - discrete approximation of the spatial derivative terms.....

Looks like you are talking about the transient discretisation. CFX uses second order backward Euler scheme to discretise the transient. This is not identical to what you typed in your last message.