Page 346 - Thermal Hydraulics Aspects of Liquid Metal Cooled Nuclear Reactors

Core thermal hydraulics 315

authors, a.o. Pfrang and Struwe (2007), Chandra et al. (2009), Chen et al. (2014), and
OECD (2015). Therefore, care should be taken when applying such correlations. If no
other confirmation is available, application of the correlation of Kazimi can be rec-
ommended for heat transfer in wire-wrapped fuel assembly geometries. Descriptions
of these correlations can be found a.o. in OECD (2015). Recent experiments with
heated bundles (Pacio et al., 2016; Di Piazza et al., 2016) focus on local temperature
measurements providing data for CFD validation. The approach used in the past to fit
local measurement values into a global correlation valid over a wide range reduces the
information of the detailed fields within a complex geometry as the wire-wrapped fuel
bundle.

6.2.3.3 Simulation of operational behavior

6.2.3.3.1 CFD benchmarking
A comprehensive literature study on simulations performed for wire-wrapped rod
bundles was performed by Roelofs et al. (2013). Starting from this review, an updated
overview is presented in Table 6.2.3.2. By this time, many researchers have studied
the influence of RANS turbulence models on the results and have concluded that the
applied turbulence model only has a small effect (see, e.g., Pointer et al., 2009;
Merzari et al., 2016). One can also notice that much effort is being put in validation
of the RANS CFD methods, either by experimental data or by high-fidelity (LES or
DNS) numerical data, for example, by Pointer et al. (2009), Doolaard et al. (2015),
Shams et al. (2015), Merzari et al. (2016), and Obabko et al. (2016). Similar to the
observation in experiments, one can notice a gradual increase of the number of pins
in the simulations, up to complete fuel assembly simulations, for example, by Pointer
et al. (2009), Peniguel et al. (2010), Rolfo et al. (2012), Cadiou and Saxena (2015),
Pacio et al. (2017), Naveen Raj and Velusamy (2016), and Jeong et al. (2017a,b).
Recently, Doolaard et al. (2015) and Merzari et al. (2016) reported blind bench-
mark analyses of RANS models against 7- and 19-pin LES reference data produced
by ANL. Conclusions from the comparison are as follows:

l Simulations, both with an SST k-ω turbulence model and with a cubic k-ε turbulence model,
show good comparison with the LES reference data. Although it should be noted that the
SST k-ω turbulence model seems to perform a little bit better, no strong conclusion can
be drawn from this. For instance, the cubic k-ε turbulence model seems to perform slightly
better at higher Reynolds numbers.
l Especially, the regions in the wake of the wire or just upstream of the wire reveal deviations
between the RANS models and the reference LES. The region in the wake of the wire, where
shedding is likely to be present, is difficult to predict with traditional turbulence modeling. It
is however a region of considerable interest since it is also the region where the peak clad-
ding temperature is located.
While code-to-code comparisons such as the ones described in Doolaard et al. (2015)
and Merzari et al. (2016), of which a RANS result is shown in Fig. 6.2.3.3, are essen-
tial for continued code development, validation through experiments remains