Page 382 - Thermal Hydraulics Aspects of Liquid Metal Cooled Nuclear Reactors

(U)RANS pool thermal hydraulics 345

example, VOF in this case. The strategy to be employed is thus strongly affected by
the chosen numerical framework.

6.2.4.2.3 Building a numerical model
After identifying the domain of interest, the next step is to generate its discrete rep-
resentation. There is no real specificity of pool modeling with regard to the compu-
tational geometry and the computational discretization (grid, or mesh), apart that there
is a quite high level of complexity. Here, what makes the true difference is the capa-
bilities of the available tool combined with the effective ability of the single user to
wield it. The snappyHexMesh of the OpenFOAM package and the STAR-CCM+
meshers are quite different tools.

6.2.4.2.3.1 Meshing in STAR-CCM+

The most important thing is to maximize the information when working on the com-
putational geometry. The more we do on the geometry, the easier will be the meshing.
This is where you decide the higher level of discretization, deciding all conceptual
volumes that must have a distinct representation: core, heat exchanger, primary pump,
etc. For each volume thus defined, we must separate its envelop in different surfaces
chosen in relation with their later functionality: inlet, outlet, internal, contact to a wall,
etc. These surfaces are also best suited for the later model “instrumentation” as they
are entities where all physical quantities can be measured: mass flux, heat flux, tem-
perature, pressure, etc. They are also the seed for specific modeling: symmetrical or
periodic boundaries, localized pressure loss or thermal resistance, etc.
Most depends on the capabilities of the meshing tool and on the geometric splitting
of the domain. Before meshing, it is more convenient to ensure that all contacts
between volumes coincide exactly with existing faces defined on the volume, meaning
that it is important to have conformal interfaces. Where feasible, this conformal inter-
face can be the seed of a conformal mesh. And even when the mesh cannot be made
conformal, it remains so in a weaker integral way. While conformal meshes are suited
for any surface shapes, it is much better to limit nonconformal meshes to plane sur-
faces; otherwise, there will be a small discrepancy between the surface sizes at both
sides of the interface. It is normally feasible to have all fluid internal interfaces con-
formal and to restrict nonconformal mesh regions to planar surfaces. This is much
more difficult with fluid/solid interfaces, especially in the case of thin structures
because the respective meshing tools produce different mesh shapes. All in all, for
fluid/solid interfaces, the integral matching on a surface basis is a reasonable compro-
mise. Practically, the MYRRHA computational grid (Fig. 6.2.4.2) has been built in
many steps, making use of the following:
First, the polyhedral meshing tool where other meshing tools are unfit.
l
l Second, wherever possible, the directed meshing tool, essentially an extruder. When avail-
able, it takes as seed the surface mesh from the polyhedral meshing tool.