Speakers
Description
Thermal Vacuum Chamber (TVAC) qualification of large-scale hardware such as solar arrays increasingly relies on two-phase liquid nitrogen (LN2) cooling systems to achieve cryogenic transients from ambient to below -180°C. High-fidelity ESATAN-TMS thermo-hydraulic simulation of these systems is computationally expensive, with a single two-phase fluid-network transient exceeding 13 hours of runtime, limiting iterative test-plan development and control-strategy evaluation.
As ESATAN-TMS's default fluid library does not include nitrogen for two-phase modelling, a custom nitrogen thermophysical property library was developed from NIST real-gas reference data (4-0.1 bar, -190°C to +180°C), using a positive-definite enthalpy reference to eliminate sign-convention errors in energy-balance calculations. The library was validated through staged checks: known-power-input response, correct isothermal saturation behaviour across the two-phase regime, and round-trip consistency against the source pressure-enthalpy data.
Building on this library, a reduced-order model (ROM) was developed that eliminates the fluid solver entirely while preserving full closed-loop control for shroud heating and LN2 mass flow regulation. The fluid domain is represented as an imaginary node co-located within the existing thermal network, updated via energy-balance/enthalpy tracking against the custom library rather than a solved momentum/continuity field. A film-temperature-dependent heat transfer coefficient is applied within the node power calculation, mapped via a custom subroutine built from three correlations derived from initial two-phase simulation results across the loop's liquid, two-phase, and vapour regimes. The power calculation itself takes the minimum of a convective estimate ($Q = hA\Delta T$) and a mass-flow enthalpy estimate ($Q = \dot{m}\Delta H$), preserving the dominant physics without re-solving the flow field. Valve-induced pressure drop is represented as a calibrated saturation-temperature offset (-185°C at full line pressure, 3.15 bar, dropping to -190°C under throttled flow control), preserving the dominant thermodynamic effect of valve operation without solving for the pressure field directly. Mass flow boundary conditions were derived from full two-phase simulation results (lower bound) and empirical LN2 tank depletion during chamber operation (upper bound). The solar array payload was represented using effective front/rear thermal properties from manufacturer-supplied material test data.
The cooling-phase ROM described in this abstract was correlated and integrated with automated heating-phase control to form the predictive model. This predictive model was independently validated against a test not used in model development. The predictive model correctly predicted an anomalous heating behaviour during hot-dwell — one panel sensor failed to reach its target set-point — matching the observed deviation to within 0.262°C. A 40-45-minute cold-dwell timing discrepancy was attributed to overcooling-protection heater sensitivity combined with the simplified payload representation. The approach reduces simulation runtime from over 13 hours to under 1 hour — a >13x speedup — while retaining predictive accuracy sufficient for test-plan optimization and control-strategy trade studies ahead of future payload campaigns.
Keywords: ESATAN-TMS, Reduced-Order Modelling, Two-Phase Cryogenic Simulation, TVAC, Nitrogen Property Library, Heat Transfer Coefficient Mapping, PID Control