CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   k-omega for flow separation (http://www.cfd-online.com/Forums/main/92897-k-omega-flow-separation.html)

mike.mm September 28, 2011 13:43

k-omega for flow separation
 
dear all,

i am estimating pressure drop for flow across multiple cylinders with their axis perpendicular to flow direction. the pressure drop estimated by using standard k-epsilon model is deviating experimental value. After which i have used standard k-omega model and now pressure drop is matching closely with experimental value. how k-omega model is giving accurate estimate for pressure drop estimate for flow separation across cylinder ?
thanks in advance.

artemiss1984 September 28, 2011 17:28

Hi.

dear friend the k-w model employes integration to viscous layer without any damping function rather low Re k-epsilon model.I dont know about your k-epsilin simulation but generally the k-w model is better in prediction of separated flows.
in your problem the dominant effect is pressure drag which is (probably) obtained better with k-w model.
for more details about the modeling I suggest the turbulence modelling for cfd by wilcox.

mike.mm September 30, 2011 13:38

Thank u very much for u reply. Dear friend, today i have read paper by Wilcox [ URL ="http://www.scribd.com/doc/ 47919871/1988-Wilcox-AIAAJ-k-omega"]. I could not understand much but whatever i understood i have presenting here so that it will be beneficial for us.

paper says that Wilcox has done defect layer analysis using perturbation method for k-omega and k-epsilon models. he applied solution from perturbation method to adverse pressure gradient condition. He found that omega equation is better formulation in adverse pressure gradient and in k-epsilon model (equation 25), epsilon equation neglects some terms which significant for adverse pressure gradient condition. because of which over estimation turbulence length scale occurs. the model constants sigma and sigma* are chosen such that model accurately predicts flow separation. but i did not find that integration to viscous layer is responsible for better prediction.

If u can understand it better then please let me know u r thoughts.

artemiss1984 October 3, 2011 16:05

yes.it is correct.
try to obtain wilcox turbulence modeling for cfd.you will find your answer


All times are GMT -4. The time now is 05:33.