Description
Despite major advances in rocket technology, recent incident reports indicate that complete reliability remains an ongoing challenge. TFor launchers to achieve high reusability at the lowest cost with minimal environmental impact, resilience is a critical requirement for almost all launch vehicle subsystems. Among these, fluid valves remain a major concern; failures in sealing systems, such as the recent issues observed in Boeing's Starliner capsule, highlight the difficulty of maintaining effective seals under extremely severe working conditions, including cryogenic temperatures and aggressive fluids.
The aim of this research is to understand and characterize micro-leakage flows within such rocket sealing systems. To address this, a predictive model capable of quantifying leakage flows between the valve seal and seat is being developed and validated against experimental data provided by industrial partner Safran Aero Boosters. The flow physics in these sealing interfaces typically fall within the slip flow regime, with Knudsen numbers ranging from 0.01 to 0.1. To accurately model fluid behavior at this magnitude, the Lattice Boltzmann Method (LBM) is employed via the open-source software OpenLB. The LBM offers many advantages at such scales. It is a highly parallelizable algorithm, allowing for much faster calculations and the implementation of complex boundary conditions is straightforward. Additionally, it can capture fluid behavior at the mesoscopic scale, where the no slip conditions start to break down and classical NSE no longer hold true. A critical advancement presented in this work is the implementation of specific slip boundary conditions to correctly resolve the non-zero velocity at the walls, which is essential for accurate prediction in micro-channels.
A major methodological pivot from previous work concerns the geometric representation of the sealing interface. Early models relying on Gaussian statistics proved to be insufficient to capture the complex, multi-scale topography of industrial finishes. An innovative digital twin" generation method based on Spectral Synthesis has then been developed. Using Power Spectral Density (PSD) analysis of raw profilometry data from valve seats, we reconstruct 3D surfaces that statistically replicate the spectral signature of authentic production parts. This ensures that the numerical domain is equivalent to the physical microscopic geometries.
Given that the surface roughness of valve seals can reach values as low as 0.1\mu m to ensure leak-tightness, resolving fluid flow at this magnitude requires extremely fine meshes, requiring requesting high computation resources. Consequently, the workflow is implemented on the Tier-1 supercomputer LUCIA (Cenaero, Belgium). This paper presents the complete predictive framework: from the spectral generation of realistic rough surfaces and the determination of effective gap heights based on experimental data, to high-fidelity LBM simulations of rarefied gas flows. The model is still to be validated against leakage rate test campaign results.s Bbut it is hoped to provide a robust tool for designing the next generation of “zero-leakage” space launchers components.