Fluent5.5 / 6 : 2-layer / enhanced wall treatm.

Volker Pawlik

Due to difficulties with fluent6.0's parallel mode in combination with periodic boundaries (I could run the case only in single mode) I created the same case in fluent5.5. In both cases I used the two-layer wall model or the "enhanced wall treatment" (fluent 6.0) respectively.

I must admit that the restriction of at least y+=5 was not fulfilled everywhere. Therefore I hoped that the new treatment of boundary layers by fluent 6 would be superior to the one of fluent 5.5.

Both cases did not fullfill the enthalpy balance quite reasonable. But the fluent6.0 case in addition showed an up and down of the epsilon (turb. dissipation) residual (range from 1E-3 up 1e8 !!! over the iterations), which was not eliminated by refining all (!) boundary cells in order to achieve y+<=5.

Latest postprocessing gave only a few cells (about 13 from 600000) far away from the walls where the residuals where high. Instead they were all found at the inlet zone near cells of worse skewness (about 0.73 hexas). The cells themselves had a skewness of about 0.5.

Did s.o. observe the same or can give me a hint to avoid or explain such a behaviour.