Modeling self-excited combustion instabilities using a combination of two- and three-dimensional simulations
Abstract
Self-excited combustion instabilities have been studied using a combination of two- and three-dimensional computational fluid dynamics (CFD) simulations. This work was undertaken to assess the ability of CFD simulations to generate the high-amplitude resonant combustion dynamics without external forcing or a combustion response function. Specifically, detached eddy simulations (DES), which allow for significantly coarser grid resolutions in wall bounded flows than traditional large eddy simulations (LES), were investigated for their capability of simulating the instability. A single-element laboratory rocket combustor which produces self-excited longitudinal instabilities is used for the configuration. The model rocket combustor uses an injector configuration based on practical oxidizer-rich staged-combustion devices; a sudden expansion combustion section; and uses decomposed hydrogen peroxide as the oxidizer and gaseous methane as the fuel. A better understanding of the physics has been achieved using a series of diagnostics. Standard CFD outputs like instantaneous and time averaged flowfield outputs are combined with other tools, like the Rayleigh index to provide additional insight. The Rayleigh index is used to identify local regions in the combustor which are responsible for driving and damping the instability. By comparing the Rayleigh index to flowfield parameters it is possible to connect damping and driving to specific flowfield conditions. A cost effective procedure to compute multidimensional local Rayleigh index was developed. This work shows that combustion instabilities can be qualitatively simulated using two-dimensional axisymmetric simulations for fuel rich operating conditions. A full three-dimensional simulation produces a higher level of instability which agrees quite well with the experimental results. In addition to matching the level of instability the three-dimensional simulation also predicts the harmonic nature of the instability that is observed in experiments. All fuel rich simulations used a single step global reaction for the chemical kinetic model. A fuel lean operating condition is also studied and has a lower level of instability. The two-dimensional results are unable to provide good agreement with experimental results unless a more expensive four-step chemical kinetic model is used. The three-dimensional simulation is able to predict the harmonic behavior but fails to capture the amplitude of the instability observed in the companion experiment, instead predicting lower amplitude oscillations. A detailed analysis of the three-dimensional results on a single cycle shows that the periodic heat release commonly associated with combustion instability can be interpreted to be a result of the time lag between the instant the fuel is injected and when it is burned. The time lag is due to two mechanisms. First, methane present near the backstep can become trapped and transported inside shed vortices to the point of combustion. The second aspect of the time lag arises due to the interaction of the fuel with upstream-running pressure waves. As the wave moves past the injection point the flow is temporarily disrupted, reducing the fuel flow into the combustor. A comparison between the fuel lean and fuel rich cases shows several differences. Whereas both cases can produce instability, the fuel-rich case is measurably more unstable. Using the tools developed differences in the location of the damping, and driving regions are evident. By moving the peak driving area upstream of the damping region the level of instability is lower in the fuel lean case. The location of the mean heat release is also important; locating the mean heat release adjacent to the vortex impingement point a higher level of instability is observed for the fuel rich case. This research shows that DES instability modeling has the ability to be a valuable tool in the study of combustion instability. The lower grid size requirement makes the use of DES based modeling a potential candidate in the modeling of full-scale rocket engines. Whereas three-dimensional simulations may be necessary for very good agreement, two-dimensional simulations allow efficient parametric investigation and tool development. The insights obtained from the simulations offer the possibility that their results can be used in the design of future engines to exploit damping and reduce driving.
Degree
Ph.D.
Advisors
Anderson, Purdue University.
Subject Area
Engineering|Aerospace engineering|Mechanical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.