Page 405 - Thermal Hydraulics Aspects of Liquid Metal Cooled Nuclear Reactors

368 Thermal Hydraulics Aspects of Liquid Metal Cooled Nuclear Reactors

Because of the trade-offs inherent in the choice between the overlapped and decom-
position approaches, current coupled models (such as those developed within the SES-
AME project) tend to be used in equivalent proportions. Both approaches may even be
used simultaneously: for instance, the STH/subchannel/CFD coupling developed at
Alternative Energies and Atomic Energy Commission (CEA) uses domain over-
lapping between STH and subchannel/CFD but domain decomposition between sub-
channel and CFD.
The next two sections will discuss practical implementations of both the over-
lapped and the decomposition approaches in two specific cases: when the coupling
boundary separates two domains undergoing fluid exchanges (Section 7.2.2) and
when the coupling interface is placed at a solid/fluid boundary (Section 7.2.3).

7.2.2 Coupling at hydraulic boundaries
The most common case of coupling boundary in a multiscale calculation is those of a
boundary between two fluid domains. In order to construct a coupling strategy at such
a boundary, the following list of requirements constitutes a good starting point:
1. The boundary should conserve mass, so that the flow exiting one of the domains (STH or
CFD) should be equal to the flow going into the other domain (CFD or STH).
2. The boundary should similarly conserve energy, with the transported enthalpy going out
from one domain being equal to the enthalpy coming into the other.
3. Finally, the boundary should ensure a consistent pressure field, so that both codes should
“see” the same pressure at the boundary. For incompressible flows (the most common case
for liquid-metal reactors), it is sufficient to ensure this equality up to a constant reference
pressure; hence, one should ensure the consistency of the pressure differences between
any two boundaries.
One should note that this “conservation” approach is sufficient, but not necessary, to
ensure a consistent multiscale calculation; indeed, one may prefer a coupling algo-
rithm in which one of these conditions would be relaxed but that would converge,
in time and/or space, to a consistent solution. Such nonconservative approaches
may be attractive in cases where the exact verification of these conditions would
require many iterations.
For domain decomposition couplings, conditions (1) and (3) can be ensured by
implementing matching inlet and outlet boundary conditions on the STH and CFD
sides, with a pressure boundary condition on one side matching a flow-rate (or veloc-
ity) boundary condition on the other: if code-to-code exchanges are used to ensure that
the flow rate (velocity) or pressure computed by one code at the boundary is used as a
boundary condition on the other side, then conditions (1) and (3) can be fulfilled
(Fig. 7.3, left).
In the case of domain overlapping, flow rates computed by the STH code can be
used as boundary conditions on the CFD side in order to guarantee (1) (for incom-
pressible systems, it is sufficient to impose flow rate at all but one inlets/outlets).
For the pressure consistency (3) to be verified, pressures can be imposed directly
on the STH side, yielding a coupling algorithm similar to those used for domain