Sy. Jaw et Cj. Chen, PRESENT STATUS OF 2ND-ORDER CLOSURE TURBULENCE MODELS - II - APPLICATIONS, Journal of engineering mechanics, 124(5), 1998, pp. 502-512
Turbulent flows, from simple to complex, are predicted using various t
urbulence models. Free shear flows; wall shear flows, with and without
separation; and more complex flows such as flow past an axisymmetric
body or flow past a three-dimensional ship hull are predicted by the R
eynolds stress model, k-epsilon-eddy viscosity model, low-Reynolds num
ber model, two-layer model, and so forth. Each turbulence model applie
s successfully to some turbulent flows, but predicts others unsatisfac
torily. It is concluded that at present, no unique turbulence model ex
ists that can predict satisfactorily all turbulent flows. Turbulence m
odel coefficients are a major concern to the model's predictability. U
sing different model coefficients for free shear and wall shear flows,
the existing turbulence model can predict satisfactorily more turbule
nt flows, including round jet and separation flows. Adopting the Reyno
lds stress model in general improves the Reynolds stress prediction. H
owever, the improvement on the mean flow quantities seems quite limite
d. improving the modeled epsilon equation, by either remodeling its pr
oduction and destruction terms or recalibrating its model coefficients
, seems to be the best way to improve the predictability of the existi
ng turbulence models.