Master thesis and internship[BR]- Master's thesis : Assessment of the Smoothed Particle Hydrodynamics formulations available in LS-DYNA for multiple-stage bird ingestion applications[BR]- Integration internship

Abstract

This master thesis presents a comprehensive assessment of the Smoothed Particle Hydrodynamics (SPH) formulations available in LS-DYNA for simulating multi-stage bird ingestion scenarios in aeroengine environments. Bird ingestion poses a critical threat to aircraft engine safety, particularly in the context of next-generation propulsion systems such as open-fan architectures. These configurations expose multiple internal components to sequential impacts, requiring accurate and robust numerical tools to model highly dynamic, nonlinear material behavior. Despite the widespread use of SPH in this context, no prior study has systematically compared the fifteen SPH formulations implemented in LS-DYNA to assess their relative performance for bird ingestion simulations.

This work addresses that gap by conducting a rigorous evaluation based on four stages: a physically representative single impact event, a sensitivity analysis to identify key numerical parameters, a double impact configuration to emulate multi-stage scenarios, and an experimental validation using high-speed test data. Each formulation was scrutinized based on physical realism, numerical stability, and computational efficiency. The single impact study provided a structured baseline that led to the elimination of eight formulations due to numerical limitations such as core instability, solver divergence, or inherent inconsistency. Six promising candidates were retained for further evaluation: formulations 0 (default), 1 (default renormalized), 5 (fluid particle), 6 (fluid particle renormalized), 15 (enhanced fluid), and 16 (enhanced fluid renormalized).

The sensitivity study revealed that the instabilities in formulations 1 and 6 could be partially mitigated by adjusting smoothing length, artificial viscosity, and particle spacing, but only at the expense of deviating from industrially accepted parameters. In contrast, formulations 15 and 16 remained consistently stable, with density filtering effectively suppressing the tensile artifacts seen in other variants. The double impact scenario further reinforced these conclusions, as only formulations 15 and 16 maintained stability, accurate momentum transfer, and physically realistic deformation throughout. Experimental validation confirmed the fidelity of both enhanced fluid formulations, with formulation 16 yielding the closest agreement with test data and demonstrating full stability even when the SPH domain was extended.

Overall, the results identify formulation 16 as the most robust and physically accurate option for industrial bird ingestion simulations in LS-DYNA, combining the benefits of renormalization with the stabilizing effects of density filtering. Formulation 15 remains a viable alternative when lower computational cost is a priority. Perspectives include large-scale industrial validation, generalization to other projectiles and constitutive models, and long-term exploration of alternative meshless methods for extended multi-impact simulations.