A data-driven based concurrent coupling approach for cryogenic propellant tank long-term pressure control predictions

Abstract

The design and optimization of cryogenic propellant storage tanks for NASA’s future space missions require fast and accurate predictions of long-term fluid behaviors. Computational fluid dynamics (CFD) techniques are high-fidelity but computationally prohibitive. Coarse mesh nodal techniques are fast but heavily rely on empirical correlations to capture prominent three-dimensional phenomena. A data-driven based concurrent coupling (DCC) approach has been developed to integrate CFD with nodal techniques for efficient and accurate analysis. This concurrent coupling scheme generates case-specific correlations on the fly through a short period of co-solving CFD and nodal simulations, followed by a long-period nodal simulation with CFD-corrected solutions. This paper presents the approach development, stability analysis, and efficiency demonstration, specifically for modeling two-phase cryogenic propellant tank self-pressurization and periodic mixing phenomena. Linear regression is employed to derive the data-driven correlations. The self-pressurization experiments of Multipurpose Hydrogen Test Bed experiments and K-Site tank are used to validate the approach. The DCC approach accurately predicts temperature stratifications while reducing computational time by as much as 70% compared to pure CFD simulations. Additionally, the DCC approach mitigates the risks of numerical instability and correlation loss inherent in current domain decomposition or overlapping-based coupling methods, making it a flexible and user-friendly approach for integrated CFD and nodal analysis of cryogenic tank behaviors.