Page 401 - Thermal Hydraulics Aspects of Liquid Metal Cooled Nuclear Reactors

364 Thermal Hydraulics Aspects of Liquid Metal Cooled Nuclear Reactors

codes; however, only system codes include the complete array of models required to
describe the complete reactor, such as core neutronics or pump models. Thus, no exis-
ting code can be used “out of the box” to provide a description of the reactor while
accounting for 3-D phenomena: additional developments are required.
It should be remarked that this need for new developments is the consequence of
two-way interactions between scales. In other cases, one may need to simulate mul-
tiple scales in the absence of such interactions: for instance, the assessment of safety
criteria during a given transient often requires knowledge of the local cladding tem-
perature maximum (which can only be obtained from a 3-D calculation).
However, in many cases, this local scale does not influence the system scale; hence,
a system calculation is sufficient to compute the overall behavior of the reactor. The
local calculation may then be performed in isolation by using the global evolution
computed by the system code as boundary conditions: such “one-way couplings”
can usually be performed using existing code capabilities.

7.1.3 Simulating multi-scale phenomena
Starting from the available thermal-hydraulic tools, two main directions can be pur-
sued to model phenomena such as those described above:
l One may choose to construct a model of the entire domain at the smallest scale required to
describe all the phenomena of interest (usually, a coarse CFD scale).
l One may choose instead a “multiscale model” approach, in which each part of the reactor is
modeled at the coarsest scale able to describe the local phenomena of interest: for instance,
one may wish to use a coarse CFD scale for large plena, a subchannel scale for the core, and a
system scale for the rest of the reactor (where no complex interactions with local phenomena
are predicted).
These approaches have both been used, including in the framework of the SESAME
project (Roelofs et al., 2015). Some of their advantages and disadvantages are listed
below:
l For the “single-scale” approach
+ The existing numerical framework of a CFD code can be used (with its associated ver-
ification/validation matrix).
– The whole domain (typically a reactor) must be modeled at the CFD scale, including in
regions where no local phenomena of interest are present (which can lead to extraneous
numerical cost).
– Existing models must be “ported” to the new code: these can include point-kinetics neu-
tronics (for the core power), pump- and coarse-scale heat-exchange models (from system
codes), and subchannel pressure drop/mixing models (from subchannel codes). Once
developed, these models must in principle be verified and validated to a level consistent
with those of their “original” code.
For the “multiscale” approach
l
+ Existing codes can be used to model each scale, thus obviating the need to implement
new models.
– However, modifications to these codes may be necessary so that they can be used in con-
cert as part of the multiscale model. This sometimes might entail the development of a