Page 270 - Thermal Hydraulics Aspects of Liquid Metal Cooled Nuclear Reactors

240 Thermal Hydraulics Aspects of Liquid Metal Cooled Nuclear Reactors

0
Fig. 6.1.1.10 Rod bundle DNS: Profiles of k, k θ , u θ , and ε θ across the unit flow cell border.
0
x
difference θ w θ on the unit flow cell are reported in Fig. 6.1.1.9, alongside with
streamlines of the cross-flow components. The buoyancy-induced flow in these geom-
etries is seen to be characterized by nonzero time-averaged secondary flows. The cir-
culation is driven toward the rods in the narrowest gap, while it is directed toward the
center of the subchannel in the largest gap. Profiles of relevant turbulent quantities
along the curvilinear abscissa γ are reported in Fig. 6.1.1.10. Overall, it is to be noted
that once again the low Pr value determines very small fluctuations of θ, and an anal-
ogously small dissipation rate of temperature fluctuations. Near-wall peaks of turbu-
lent kinetic energy and Reynolds stresses are instead found.

6.1.1.6 Conclusions

DNS is one of the tools that are being used for studies of turbulent heat transfer at low
Prandtl numbers. As shown in the present chapter, DNS (at all Prandtl numbers) is
used in rather simple geometries and at low Reynolds numbers. Nevertheless, this
is exactly the type of flows where DNS results provide the key added value. The
amount of information in even the simplest DNS database is extremely rich and
the extraction of statistics and interpretation of results is often a difficult part of
the job.
Aside from the computational cost, the use of DNS in real-world applications is
further discouraged by the huge amount of over-abundant data with respect to quan-
tities of interest. Furthermore, uncertainties on boundary and initial conditions, which
cannot be avoided in any numerical approach, do not justify the use of DNS in these
cases. Of course, DNS in simple geometries can be a very useful tool for development
of subgrid LES models and RANS transport equations in controlled numerical exper-
iments. In this respect, DNS of liquid metal flows does not make any difference with
respect to the DNSs of other fluids.
The simplest type of passive scalar DNS in near-wall geometry reveals some pecu-
liarities of the low Prandtl number ( 0.01) heat transfer: near-wall mean temperature