Predicting Baffled Propellant Tank Slosh for Spacecraft

Overview

During flight, a spacecraft’s liquid propellant sloshes within the fuel tank in response to the vehicle’s movements. When the frequency of the propellant slosh coincides with either the dynamic control frequency or the fundamental elastic bending frequency of the vehicle’s body, the sloshing forces can interact with the spacecraft’s control system or structure. This could cause a failure of structural components within the vehicle or excessive deviation from its planned flight path. It is therefore important to provide adequate damping of the liquid motion and slosh forces in the fuel tank, and to develop methods to account for damping in the analysis of vehicle performance. For axisymmetric tanks, a common method of increasing slosh damping is to attach a series of ring baffles to the tank wall. The baffles provide a substantial degree of damping when the liquid surface is near them. The spacing between the baffles is chosen to ensure that the damping exceeds requirements, regardless of the fill level of the tank. Fuel-slosh damping by ring baffles in cylindrical tanks was investigated extensively in the 1960s, both experimentally and theoretically, but a survey of the 1960s damping measurements shows many apparent discrepancies. In this project, we aim to improve our understanding of the physics behind slosh damping, and to validate damping extracted from computational fluid dynamics (CFD) simulations against experimental data using the same parameters in a baffled tank.

Project Details

Accurately predicting the slosh damping value of a baffled tank is very challenging for any CFD tool, as the thin boundary layers near the tank wall and subscale vortices near the baffles must be resolved. In this study, a volume-of-fluid (VOF) module was linked to a production CFD solver, Loci-STREAM, which is being utilized at NASA’s Marshall Space Flight Center. This tool provides a fast turnaround time and a lower numerical damping using the high mesh density enabled by the massively parallel processing capability of NASA’s Pleiades supercomputer. Loci-STREAM was first validated against experimental data for slosh damping inside a smooth-wall tank. The slosh damping values for a baffled tank were then computed and compared to the experimental test data.

Results and Impact

The experimental data, which correlated with the industry standard for the smooth wall, were used as the baseline validation. We demonstrated that with proper grid resolution, CFD can accurately predict low damping values from smooth walls for different tank sizes. We then simulated damping due to ring baffles at different depths from the liquid surface, and very good agreement with the experimental data was observed. This work clearly illustrated the soundness of the CFD approach in modeling the detailed fluid dynamics of tank sloshing and the excellent accuracy in extracting slosh mechanical properties, especially slosh damping.

Application of the Loci-STREAM CFD modeling tool has great potential to aid in the design of propellant tanks with multiple baffles. In general, the practice of estimating the total damping for multiple baffles is done using a linear superposition of the estimated damping contribution from each individual baffle. However, experimental data involving multiple baffles are limited. CFD simulation could be used to determine the validity and applicability of such rules. Finally, CFD simulations could also be used to help develop a relationship between slosh frequencies and baffle parameters.

Why HPC Matters

Capturing the thin boundary profile and subscale vortices of the slosh wave requires a very fine grid resolution, which leads to huge memory requirements and intensive computation. Additionally, the complexity of solving for two different fluids with a high-density ratio (1,000:1 for water to air) causes the resulting equations to become very stiff (difficult to converge). Our simulations would not have been possible without the use of a supercomputer. All the simulations were performed on NASA’s Pleiades supercomputer using 500 to 1,000 processors. The Pleiades cluster allows us to solve complex, multi-phase flow problems through a high-bandwidth, low-latency network, providing highly parallel capabilities. Each computational run took 1 to three days to complete.