Computational investigations of high frequency acoustics and instabilities in a single-element rocket combustor

Randolph Jedediah Smith, Purdue University

Abstract

The overall objective of the current research is to evaluate if and how computational fluid dynamics (CFD) can enhance our understanding of combustion instability. Specifically, the first objective is to examine whether computational fluid dynamics models can, with the computational resources that are available today, replicate high amplitude, high frequency combustion instability phenomena without using any kind of heat release response function, external forcing, or input from the corresponding experiments. As part of this, it is desired to identify the minimum physics necessary for unstable CFD solutions. Upon observing combustion instability, the second objective is then to develop an analytic framework in which the instability mechanisms of the engine can be investigated. Companion experiments have been performed and are used to guide and validate the computational model. The general configuration is a single-element, longitudinal rocket combustor with a choked exhaust nozzle. The oxidizer and fuel are introduced in a coaxial, non-premixed fashion. Decomposed hydrogen peroxide is used as the oxidizer, with JP-8 or methane as the fuel. This research shows that combustion instability can be simulated with the current level of modeling. Rayleigh index calculations have indicated conclusively that the unstable behavior is the result of coupling between unsteady heat release and acoustic pressure fluctuations. Steep-fronted pressure oscillations and oscillation growth have also been observed. Various computational studies have been performed, including the variation of oxidizer inlet, combustor step height, chemical kinetic rate, oxidizer post length, and combustor wall boundary condition. Where possible, the results of these computations were compared with experimental data. Experimental instability trends have been correctly predicted for the effect of back-step height in the combustor, changes in the upstream inlet geometry, and to some extent, the effect of changing the oxidizer post length. For the chemical kinetic rate and wall boundary condition studies, direct experimental comparison was not available, but the results are in-line with the general behavior of the experiment. Because flow conditions from the entire flow field can be extracted from CFD results, much more detailed analyses can be performed with computational data than with experimental measurements. Along with instantaneous and time-averaged flow field plots, several other diagnostics were developed to study the instability mechanisms, including power spectral density, Rayleigh index, mode shape, and phase difference analyses. The various types of analyses resulted in valuable insight into the causes and characteristics of instability. These diagnostics suggest that multiple flow phenomena affect the level of instability, including the acoustics of the system, the upstream conditions, vortex shedding and impingement on the combustor wall, the location of heat release, and the axial velocity oscillation and gradient upstream of the combustor step. In most cases the computations were not able to quantitatively match the level of instability of the experiments. In order to achieve results within a reasonable time frame and obtain enough computational data to provide statistically robust results, many assumptions were required, such as axisymmetric modeling, the use of a one-step finite-rate global combustion model, and the neglecting of any two-phase effects. As computational resources increase in the future, these assumptions can be relaxed and higher fidelity results can be obtained. The results of this research indicate that, when validated through experimental data, CFD can potentially be an important tool for future prediction and analysis of combustion instability.

Degree

Ph.D.

Advisors

Merkle, Purdue University.

Subject Area

Aerospace engineering

COinS