Fast multilevel multi-phase CFD-nodal model for cryogenic applications (043992)
Project Status: Current
Investigator
Description
We propose an adaptive method for combining the lumped model and multi-material high-fidelity simulation of the
cryogenic fluids. Through a seamless combination of the nodal analysis and CFD, accurate predictions of conjugate heat transfer in all
materials (gas, liquid, and solids) and phase-change dynamics can be achieved. The adaptive mesh refinement method and dynamic
multi-zonal switching between CFD and lump-model are proposed as a unique model for design, verification, and test of the transport
and storage of cryogenic fluids efficiently. The model will be carefully verified with experimental data for different cryogenic flows.
The nodal model is used for modeling the slow background mass/momentum/energy processes and covers the entire fluid domain.
The coefficients of the nodal model will be initially set based on prior data and subsequently updated with CFD results. The nodal
model provides the boundary conditions for the concurrent local CFD calculations at the physically complex parts of the domain. To
ensure proper information transfer between the models, the block-structured adaptive mesh refinement (AMR) will be used. The AMR
method is designed such that fine grids are only used in the regions of interests with the fast processes while the much coarser grid
is employed in other parts. The coarsest grids coincide with the lumped degrees of freedom of the nodal model and are dynamically
refined to the finest level gradually over a predefined transition layer. The refinement is done based on the physical quantities such as
vorticity or the interface curvature between materials. This new approach has many benefits compared to full CFD or nodal models
such as what follows: (a) The domain is defined on patches with different resolutions in space (form coarsest level 0 to the finest level
L), and order of accuracy in time (from 1 to 16). The solutions on these patches are connected seamlessly to each other by providing
the ghost cell data in space and time, which will be used for the connection of nodal and CFD model. A new coarse level -1 with the
same resolution to the level 0 will be created in which the lumped model instead of the CFD will be used for the calculation. The solidfluid interface and a thin fluid layer next to it will always be represented with the CFD high-fidelity model. (b) The adaptive level
refinement criteria will be designed for switching between a CFD-based patch at level 0 to a nodal-model region at level -1. (c) The
resultant model can handle multiple materials (gas, liquid, solid) and can accurately reconstruct the interfaces using the state-of-theart moment-of-fluid method. (d) The CFD model of the fast processes is based on the conservative first-principal physical relations and
has already been verified for phase-change dynamics in microgravity. Those capabilities will be adopted here. (e) Multiple purposely
designed experiments on cryogenic helium pipe flow and vapor bubble nucleation will be conducted at the Cryogenics Lab in the
National High Magnetic Field Laboratory to assess the model accuracy and performance. These test studies are intimately related
to two canonical problems in space science and applications, storage of cryogenic propellants in microgravity and transfer of these
propellants through pipelines. The final demonstration of the model will be done with the NASA tests archive.
The proposed model will be the first effort to leverage the commonly used AMR technique of CFD model to reach a hybrid nodalCFD multi-phase model of the cryogenic flow that includes gas, liquid, and solid phases. The lumped model of the system, updated
automatically with the localized high-fidelity simulations and carefully validated, can be used as a design tool for many different
applications involve multi-phase cryogenic flows.