CFD Online Discussion Forums

CFD Online Discussion Forums (
-   STAR-CD (
-   -   Fully developed flow in Star-cd (

SMM September 7, 2011 03:53

Fully developed flow in Star-cd
Hi all,

Recently I uploaded a querry regarding fully developed flow in star-cd. I mentioned that the manuals clearly reveals that a fully developed flow in pipes and ducts can be modelled using uniform pressure boundary condition at the inlet and a specified mass flow rate at the exit which can be explicitly defined by OUTLET type of boundary condition in star. I further simulated the same problem in COMSOL as well and if I impose an inlet pressure of 1.9142Pa (calculated from the analytical expression available for steady laminar fully developed flows in pipes) and a zero pressure at the exit the COMSOL easily showed that the resulting flow is fully developed, however we can not use two pressure boundary conditions simultaneously in a fluid domain in star-cd, but still if we follow the manuals and impose a value of inlet pressure as given above and specify the mass flow rate at the exit the code should predict the fully developed flow but it is actually not and diverging significantly from the results of COMSOL.

Any experience on this issue??? May be the information in the manuals is insufficient I guess??? or may be entirely wrong.

vishyaroon September 8, 2011 09:28

Hello Syed, for a fully developed flow (I guess you are using cyclic BCs), you can either assume a mass flow rate through the duct/pipe or the pressure drop. In most cases the primary outputs desired are the pressure drop or mass flow rate and hence you fix the other.

Is there a reason why you want to fix the inlet pressure? The inlet pressure should not matter as you may be interested in the pressure drop for a fully developed condition.

SMM September 8, 2011 22:06

Hi, Aroon

Thanks Aroon for the reply. Actually I have read in the manuals that cyclic boundary conditions imposes equal scalars at the geometrically same boundaries. By the word scalar I mean to say that it will impose same velocity components as well as pressure at the cyclic boundaries, but as u know that in a fully developed flow the velocity components remains the same along the length but pressure drops takes place hence the pressure cannot remain the same. In the manuals it is clearly written that one has to impose a uniform pressure at the inlet and a outlet b.c at exit with prescribed mass flow rate in order to simulate a fully developed flow in ducts and pipes (see Methodology manual chap 5 under the heading Fully Developed Flows version 4.14.014).

I followed the manuals and found out that the code needs some minor adjustments of the relaxation factors and no of sweeps for pressure corrector equation in this case for obtaining converged and stable solution for fully developed flows, however the results does not match fully with the theoretical results for example the computed friction factor shows a 1.5% error with the theoretical friction factor (f=64/Re).

Another issue is with the mesh in STAR-CD usually in laminar flows to resolve viscous B.L one needs to increase the grid density near the wall but I have found out that in STAR-CD if we use an arbitrary refined grid near the wall by keeping the no of radial cells fixed the computed results deteriorates (shows high error). Therfore I think that in STAR-CD we have to satisfy the R/38 criteria for the first wall adjacent cell height for laminar flows and find an optimal no of cells in the radial direction that gives the lowest error. Infact I have generated the mesh like this and achieved good results for developing flows.The same strategy works with developed flows as well. Do u have any suggestions on this???

vishyaroon September 9, 2011 15:54

Hello Syed, Sorry about the misunderstanding. Never worked with this condition as we've always had some geometric feature in the streamwise direction preventing us from using this.

All times are GMT -4. The time now is 20:01.