Year: 2025
Type: conference-paper
Abstract: Rotating detonation engines are a type of combustion engine that utilizes continuously propagating detonation waves through an annular or circular combustor chamber. These engines aim to achieve superior efficiency and specific impulse over traditional deflagration-based systems. The nozzle design is pivotal in transforming the high-energy exhaust from the combustor into usable thrust, thereby making its optimization crucial for maximizing engine performance. One promising approach incorporates aerospike nozzles featuring a contoured ramp that leads into a spike, which may be truncated or not, at the center and is open to the atmosphere on the side. Due to the high unsteadiness and non-uniformity of the exhaust, the nozzle flowfield can undergo severe changes, affecting the engine’s maneuverability and performance. This study employs computational modeling and dynamic analysis to explore the flow dynamics and efficacy of a nozzle geometry sourced from the literature, characterized by a divergent design with a bent outer wall. In this study, computational fluid dynamics (CFD) simulations were performed using ANSYS Fluent to analyze the distributions of pressure, velocity, and temperature within the nozzle across both steady-state and transient phases. The numerical setup employed density-based solvers to address the compressibility effects inherent in supersonic flows. The CFD model incorporated conservation equations of mass, momentum, and energy, complemented by the ideal gas law and Sutherland’s viscosity equation. A systematic grid independence study was first conducted to ensure a balance between computational efficiency and numerical accuracy. This led to the selection of a mesh that captured critical flow phenomena while remaining computationally feasible. The characteristics of both steady-state and transient flowfields are subsequently detailed. To further explore and characterize the transient flow behaviors, including shock diamonds and shear layers, Principal Component Analysis (PCA)―also known as Proper Orthogonal Decomposition (POD)―was employed. PCA, an unsupervised machine-learning technique, provided a data-driven approach to extract dominant flow modes from instantaneous snapshots. This enabled the decomposition of complex flowfields into interpretable patterns, revealing significant insights into the flow physics. Specifically, it elucidated the evolution and interactions between shocks and expansion waves within the nozzle and their influence on thrust production and total pressure losses. By identifying the dominant flow modes, PCA highlighted critical areas where geometric modifications could mitigate losses and enhance nozzle performance. The integration of PCA in data postprocessing and design exploration has demonstrated its potential to enhance aerodynamic analysis, reducing the computational effort required to analyze large-scale CFD datasets while still preserving essential flow characteristics. The proposed methodology provides a robust framework for integrating CFD and machine-learning techniques in propulsion research, setting the stage for future advancements in nozzle development and other aerospace applications. The outcomes also reinforce the critical role of data-driven approaches in enhancing the understanding of transient supersonic flows, paving the way for innovative solutions to longstanding challenges in propulsion engineering.
DOI: 10.2514/6.2025-97997
Authors: Khushi Piparava, Dr. Liwei Zhang
Language: English
Subfield: Computational Mechanics
Field: Engineering
Domain: Physical Sciences
Open Access status: bronze
Award: G1570797360
OpenAlex ID: W4413190889