Search

The rotating detonation engine is increasingly favored as a viable pressure gain combustion technology for both propulsion and power generation applications. Practical designs involve the discrete ...

An experimental study was conducted to characterize fundamental behavior of detonation waves propagating across an array of reactant jets inside a narrow channel, which simulated an unwrapped rotating detonation engine (RDE) configuration. Several key flow features in an ethylene-oxygen combustor were explored by sending detonation waves across reactant jets entering into cold bounding gas as well as hot combustion products. In this setup, ethylene and oxygen were injected separately into each recessed injector tube, while a total of 15 injectors were used to establish a partially premixed reactant jet array. The results revealed various details of transient flowfield, including a complex detonation wave front leading a curved oblique shock wave, the unsteady production of transverse waves at the edge of the reactant jets, and the onset of suppressed reactant jets re-entering the combustor following a detonation wave passage. The visualization images showed a complex, multidimensional, and highly irregular detonation wave front. It appeared non-uniform mixing of reactant jets lead to dynamic transverse wave structure. The refreshed reactant jets evolving in the wake of the detonation wave were severely distorted, indicating the effect of dynamic flowfield and rapid pressure change. The results suggest that the mixing between the fuel and oxidizer, as well as the mixing between the fresh reactants and the background products, should affect the stability of the RDE combustor processes.

A simplified description of heat release oscillations associated with vortex-driven combustion instabilities is developed that can greatly reduce the amount of data required for determining the lin...

Combustion stability characteristics of a turbulent diffusion flame established between a center jet of gaseous oxygen and coflowing jets of gaseous hydrogen blended with different amounts of gaseous methane are studied in a rectangular combustor operating under atmospheric pressure conditions. A compression driver, mounted near the injector, is used to acoustically excite the flame from a transverse direction. Resulting flame perturbations are studied using OH* chemiluminescence imaging, dynamic pressure measurements, and high-speed flow visualizations. Both steady-state perturbations and perturbations as the acoustically forced flames transition from one fuel blend to another are studied. Simultaneous measurements of pressure oscillations and heat release oscillations are used to obtain local Rayleigh indices showing locations that drive or dampen the instability. Transient measurements associated with real-time in situ methane blending are used to obtain timescales associated with the suppression proce...

A dual-mode scramjet combustor featuring a fin-guided hydrogen injector and a cavity flame holder was operated in a transitional regime characterized by frequent combustion mode hopping and large flame front movements. In this regime, which occurred at equivalence ratios between 0.19 and 0.24, the combustor operated in scramjet mode initially, but transitioned into ramjet mode with time. The experiments were performed using a direct-connect vitiated-air facility which generated a high-enthalpy Mach 1.9 flow at the isolator inlet. High-speed schlieren visualizations, OH* chemiluminescence, and wall pressure measurements were used to study the alternating combustion behavior that occurs throughout the scramjet to ramjet mode transition. The results showed that the transition process was characterized by a series of flame oscillations driven by mode hopping cycles, consisting of transient thermal choking, flame flashback, cavity flame holding and detachment. These mode oscillations were initially sporadic and short-lived events of upstream-downstream flame propagation, but as the transition advanced and the presence of the flame in the cavity increased, they transformed into longer and more stable cycles. The characteristic duration of the entire transition process was determined to range between 0.42–0.61 s using a statistical analysis based on image processing of the schlieren visualizations. Increasing the overall equivalence ratio decreased the transition duration, but had no significant effect on the dynamics of this process. The results suggest that thermal choking due to wall heating downstream of the cavity drives the scramjet to ramjet mode hopping, while flame quenching and blow-off at the cavity caused by its initially cold walls reinitializes scramjet operation.

A method for modeling the internal flowfield in a rotating detonation engine is developed using shock-expansion theory combined with the steady two-dimensional isentropic method of characteristics. An analytical model using the oblique shock relations, the Prandtl–Meyer function, and the detonation jump conditions is used to determine the basic shock structure. Once the structure is known, a shock-fitted method of characteristics solution is marched out to generate the rest of the flowfield. Reactant injection is handled analytically by solving the conservation equations for a flow undergoing a sudden expansion along with the method of characteristics compatibility relations to provide a new boundary condition. A new solution is then initialized using information from the previous solution to calculate the new shock structure. This process is repeated until the solutions converge. The converged solution is the ideal steady-state solution of a rotating detonation engine in the wave-fixed reference frame. T...

The effect of density ratio, velocity ratio, momentum ratio, and fuel composition on the susceptibility of shear coaxial flames to periodic pressure waves is studied experimentally. A GO2-GH2 turbulent diffusion flame, established in an atmospheric-pressure, two-dimensional, shear coaxial combustor, is acoustically forced by a transversely mounted compression driver, and the level of flame acoustic interaction is quantified using OH* chemiluminescence imaging. The parameter of interest is varied by diluting the propellants with helium, argon, or methane by appropriate amounts while holding other parameters relatively constant. Flame response is found to vary exponentially with density ratio, whereas it is found to vary linearly with velocity ratio and fuel composition, at least within the limited ranges tested. Flame response is found to vary exponentially when momentum ratio is varied by varying density ratio, whereas it is found to vary linearly when momentum ratio is varied by varying velocity ratio. O...

The mixing flowfields of transverse walled fuel injection with and without the guidance of a fin in a Mach 2.2 flow were experimentally characterized and compared. This study evaluated the ability of fin-guided injection to enhance fuel–air mixing while reducing shock-induced stagnation pressure losses. The nonreacting gaseous injection experiments used helium as a hydrogen surrogate and simulated four mass flow rate conditions to investigate the performance of the proposed injection scheme with variable jet momentum. The analysis of schlieren visualizations demonstrated a 100–200% increase in jet penetration for the fin-guided cases over the baseline 12 diameters downstream of the injection point. Wall pressure measurements were correlated to the schlieren results which showed that the strength of the jet-induced shock in the baseline was reduced by 33–47% by using the fin. A planar-Mie scattering method used to obtain cross-sectional views of the injection flowfields revealed that the fin was not only r...

Dynamic interactions between shear coaxial injector flames and periodic compression waves are investigated under controlled density ratio variations. Turbulent diffusion flames are established between an oxygen slot jet in the middle and two hydrogen slot jets on the sides, representing a sectional view of a shear coaxial injector flowfield. Transverse acoustic excitation is applied using a compression driver mounted on one side of the combustor to force possible flame–acoustic interactions. While maintaining a stoichiometric fuel–oxidizer ratio, the density of each stream is controlled by diluting the propellant gases with inert helium and argon by appropriate amounts. Results show strong dependence of flame–acoustic interaction on the density gradient between the propellants. The observed trend is consistent with the vorticity production mechanism created by the baroclinic source term in the flowfield. It is observed that the flames become more resilient to acoustic coupling when the fuel–oxidizer densi...

The dual-mode scramjet isolator flow field is simulated experimentally and numerically in rectangular aspect ratio 1.0, 3.0, and 6.0 ducts with inflow Mach numbers ranging from 2.4 to 2.7. Often neglected in past research efforts, the focus of this work is on resolving the three-dimensional nature of the shock train front. Previous work utilizing multiplane shadowgraphy and focusing schlieren deflectometry revealed that the shock train front is hybrid oblique/normal in nature, with the low momentum corner flow separating up to one duct height upstream of the center flow field. This initial separation spawns oblique shock planes that transform into stronger, more normal shock structures near the center flow region. This study provides supplemental analysis of the three-dimensional isolator flow features through steady-state k − ω RANS simulation, experimental and numerical boundary-layer profile analysis, and quantitative global density gradient magnitude calculations by way of the Background Oriented Schlieren (BOS) method. Validation and verification of the fully started isolator simulations show good agreement to experimental data, while the unstarted isolator simulations match the qualitative features of the three-dimensional shock train front observed in the experimental aspect ratios tested. Experimental and numerical boundary-layer measurements indicate that the major axis (side wall) boundary-layer is more susceptible to separation than the minor axis (nozzle-bounded) boundary-layer due to a lower momentum thickness, influencing the formation of the shock train along the wall center lines. The application of the BOS method indicates that the shock train structure strength doubles between the outboard and inboard regions of the preliminary shock train element. A continued discussion regarding the importance of accounting for the often neglected three-dimensionality of the isolator shock train is provided.

Shock train structures in rectangular isolators interacting with a turbulent boundary layer were investigated using simultaneous flow visualization from both the major and minor axes. Experiments were conducted at incoming Mach numbers ranging from 2.2 to 2.5, and using two different duct aspect ratios of 3 and 6. The interactions were three dimensional in nature, and the shock train structures were more complex than they initially appeared from the single-axis visualization. Viewing through the major axis, the shock trains appeared oblique in nature. However, they contained a series of normal shocks when observed through the minor axis. The initial shock-boundary layer interactions were always observed with oblique shocks initiating from the corners, and they caused large flow separation at the interaction points. While these interactions appeared as symmetric oblique shocks from the major-axis view, the oblique shocks transformed into a normal shock near the duct center. As a result, hybrid oblique-normal shock train structures were observed in rectangular isolators at the Mach numbers tested. The number of individual shocks in the shock train structure increased with the aspect ratio. Lastly, dynamic wall pressure measurements were made parallel to the major axis. The results confirmed that the interactions initiate from the corner well ahead of the center boundary layer disturbance.

A method for modeling the internal flowfield in a Rotating Detonation Engine (RDE) is developed using shock-expansion theory along with the two-dimensional, isentropic method of characteristics (MOC). An analytical model using the oblique shock relations and the Prandtl-Meyer function is used to estimate the basic shock structure. Once the structure is known, the centered expansion fan is used to initialize a shock-fitted MOC solution. By modeling the flowfield in the steady, wave-fixed reference frame, computational requirements may be reduced while still providing sufficient information to perform fast, parametric performance estimates.

A method for modeling the internal flowfield in a Rotating Detonation Engine (RDE) is developed using shock-expansion theory along with the two-dimensional, isentropic method of characteristics (MOC). This work is an extension of research conducted at the University of Michigan in the 1960s and 70s on detonations with compressible boundaries. A simple analytical model using the oblique shock relations and the Prandtl-Meyer function is sufficient to estimate the basic shock structure. Once the structure is known, the resulting centered expansion fan is used to initialize a MOC solution. This analysis captures the bulk fluid flow occurring within the annulus of an RDE without the use of time-consuming numerical solutions. The goal of this model is to provide performance estimates one step up from a basic thermodynamic model while still being fast enough for large parametric studies.

An experimental study was performed to evaluate the e ame-holding and mixing enhancement characteristics of supersonic reacting e ow over acoustically open cavities. Several cone gurations of acoustically open cavities were placed inside a supersonic-combustion duct just downstream of the fuel injection ports. The resulting changes in e ame behavior and combustion characteristics were assessed using schlieren visualization of the uncone ned e ow and wall pressure measurements of the cone ned e ow along the duct. The results were then compared with the baselinecase, which used no cavity. Although thecavitiesimproved thecombustion performancefrom thebaseline, the amount of enhancement was dependent on the particular shape of the cavity as well as the e ow conditions. Certain cavity cone gurations that were strategically placed inside the combustion duct led to a faster increase in the axial pressure force. The data showed that the recovery temperature was higher and the total pressure proe le was more uniform at the exit plane, suggesting enhanced volumetric heat release and faster mixing associated with the cavity- ine uenced e owe eld.

Active instability suppression using periodic liquid-fuel injection was demonstrated in a dump combustor. The controller fuel, which made up 12%–30% of the total heat release, was pulsed directly into the combustion chamber, and the injection timing was adjusted with respect to the combustor pressure signal. Because the injection timing determined the degree of interaction between pulsed fuel sprays and periodic large-scale flow features, it significantly affected the spatial distribution of fuel droplets inside the combustion chamber. Simple closed-loop control of the pulsed injection timing was applied to two different cases that developed natural instabilities. In the first case, the instability frequency was unchanged at the onset of the closed-loop control, and this fact allowed up to 15 dB reduction in the sound pressure level. A detailed investigation showed that the pressure oscillation amplitude reached the minimum value when the start of the pulsed fuel injection was synchronized with the inlet vortex shedding process. In the second case, the same controller was applied to a higher output combustor, where the injection timing affected not only the oscillation amplitude but also the instability frequency. For the high output case, the controller was able to suppress the oscillations initially, but it could not maintain the suppressed amplitude, resulting in unsteady modulation of the oscillation amplitude and frequency. The intermittent loss of control was linked to the frequency-dependent phase shift, associated with an electronic band-pass filter. The present results open up the possibility of utilizing direct pulsed liquid-fuel injection for active combustion control in propulsion devices, but they also show the limitation of a simple phase-delay approach in completely suppressing the natural oscillations under certain conditions.

Turbulent compressible shear layers of round supersonic jets were excited by placing an open cavity near the nozzle exit, which produced pressure oscillations due to flow-induced cavity resonance. As a result of the excitation, large coherent structures were created in the highly compressible shear layers, making it possible to study the effect of such structures on turbulent compressible mixing. For this study, pressure-matched Mach 2 jets were used under both nonreacting and afterburning conditions. By varying the cavity dimensions, we could apply excitation over a wide range of frequencies and manipulate the coherent structure dynamics in the near field of the jets. Mie-scattering flow visualization images revealed that large coherent structures were generated when high-amplitude excitation occurred at frequencies close to the jet preferred mode, which was near the Strouhal number of about one-half. Growth rate of nonreacting shear layers was quantified as a function of excitation frequency; and, for afterburning jets, the change in global flame luminosity due to excitation was measured. It was observed that the growth rate was drastically increased in the near field when the coherent structures were organized. The afterburning characteristics were also modified by coherent structures and their dynamics. Under these conditions, afterburning intensity increased when the excitation frequency was higher than the preferred mode frequency and decreased at lower frequencies. The results suggest that the initial size of the coherent structures not only determined the rate of large-scale entrainment, but also modified molecular-level mixing by affecting the timing of large-structure breakdown.

Pulsating liquid-fuel sprays from automotive fuel injectors were characterized and analyzed in nonreacting and reacting experiments to develop active combustion control using liquid-fuel injection. Square waves with variable duty cycle were used as an input signal for injector operation. Using ethanol pressurized to 275 kPa, pulsating injection with 100% flux modulation was achieved at frequencies as high as 1 kHz. Spray size and spatial distributions were measured from instantaneous Mie scattering images. The results showed that the pulsating sprays contained a high number of relatively large droplets that would be unfavorable for direct combustion control applications. A dump combustor experiment using these injectors showed very little dependence on phase of fuel injection. To reduce droplet size, a new injection setup was designed by combining a fuel injector with a swirl-based atomizer. In the new setup, the fuel injector functions only as a solenoid valve with high-frequency response, whereas the atomizer is used to reduce droplet size. A substantial reduction in droplet size was obtained while maintaining a similar frequency response. Using the improved injection setup, nonreacting experiments were performed in an actively controlled dump combustor to characterize spatial distribution of fuel droplets as a function of control parameters. Fuel injection was synchronized with different stages of air vortex development, and the interaction between spray droplets and large-scale flow features was investigated systematically. The results showed that the synchronized injection yielded an effective way of controlling the spatial distribution of fuel droplets in the dump combustor flow field. Finally, a reacting experiment showed that combustor controllability was significantly extended. The results open up the possibility of actively controlling liquidfueled combustion in other practical combustors.

Recent interest in supersonic combustion (scramjets) and noise reduction for the high speed civil transport (HSCT) plane prompted renewed research in supersonic mixing processes and means to control them. The scramjet propulsion concept requires rapid mixing between fuel and air in order to minimize the size of the combustor and affect the performance of the entire vehicle system. Also, accelerated mixing of exhaust plumes with coflowing air has been shown to lead to jet noise reduction. Other examples of technological applications requiring control of mixing in compressible flows include thrust augmenting ejectors, thrust vector control, metal deposition, and gas dynamic lasers. The technological challenge of mixing enhancement in compressible flows stems from the inherently low growth rates of supersonic shear layers. Many mixing augmentation methods employed efficiently in sub­ sonic flows failed to work at elevated Mach numbers, and some were inefficient because they were utilized outside their effective range. Never

Quasi-steady shock trains in a rectangular isolator of the aspect ratio 3 were characterized with Mach 2.5 upstream flow and under slowly varying backpressure conditions. A time-history schlieren technique, accompanied by simultaneous static and dynamic pressure measurements, was performed to better understand the transient aspect of shock-boundary layer interaction leading to inlet unstart. Both spanwise and centerline static pressure measurements were utilized to characterize the propagation of shock trains inside the isolator, while the dynamic pressure measurements were further analyzed in an effort to find any precursor of unstart. A database showing time-resolved schlieren images with corresponding axial pressure profile and pressure gradient profile was constructed. The preliminary results revealed a clear difference between visible shock train length and the extent of the pressure gradient in the isolator. Investigation on the shock train dynamics revealed a highly oscillatory nature of the propagating shock trains. Additional research is in progress to characterize the shock train motion and the dynamic behavior with the downstream pressure disturbance characteristics.

Progress toward the development of an improved stability analysis tool, anchored by a combustion response model incorporating detailed flame-vortex dynamics, is presented through the modeling of a premixed, 2-D, rearward-facing step combustor. A global stability model was developed for this combustor using commercial finite element software, COMSOL Multiphysics, that integrated a 2-D reacting mean flow model, a thermoacoustic model, and a 1-D flame response model, derived from detailed experimental data acquired in this geometry. The accuracy of the integrated model was evaluated in the context of the 2-D combustor’s unstable 150 Hz mode. The model predicted the mode to be linearly stable despite experimental observations to the contrary, but further analysis showed that the prediction was driven by the assumed characteristics of the flame response model. Areas for improvement in the flame dynamic model and the overall integrated model were identified and will be the focus of ongoing development efforts. I. Introduction YNAMIC instabilities in aerospace combustors are marked by high amplitude pressure oscillations and result from a coupling between combustion heat release fluctuations and the acoustic modes of the combustor. High frequency dynamic instabilities (hundreds to thousands of Hertz) are often manifested in strongly separated flows, as created by a bluff body or a sudden expansion, and were first identified in practical configurations in the 1950’s. 1,2

Phase-locked measurements of CH* chemiluminescence associated with a twodimensional dump combustor were used to map local heat release rate from selfexcited combustion instability as a function of space and instability phase. The naturally unstable condition, targeted for detailed measurements, occurred at an inlet velocity of 45 m/s and an equivalence ratio of 0.67, and was dominated by periodic motion of large vortical structures at 150 Hz. The models for fluctuating heat release rate were empirically established by assuming a separation of variables and a sinusoidal temporal dependence and calibrating the coefficients using either the local CH* chemiluminescence or the broadband light emission measurements in the visible spectrum. Model reconstruction of the fluctuating heat release and the flame structure resulted in a good agreement with the measured data at various instability phases both qualitatively and quantitatively. In particular, both the local and global Rayleigh index calculations yielded excellent agreements with the experimentally determined values, suggesting the empirically constructed models can provide good insight into the driving and damping mechanisms as well as an accurate prediction of combustor stability. The results showed a good correlation between the local chemiluminescence and the periodic vortex motion, opening up a possibility of building a physics-based heat release fluctuation model by simply utilizing the understanding of the vortex dynamics. It was also noted that, for this particular operating condition tested, the broadband emission measurements can be used to build an approximate model of fluctuating heat release that can accurately predict the stability characteristics of the combustor.

Compressible shear layers in supersonic jets are quite stable and spread very slowly compared with incompressible shear layers. In this paper, a novel use of a cavity-actuated forcing technique is demonstrated for increasing the spreading rate of compressible shear layers. Periodic modulations were applied to Mach 2.0 reacting and nonreacting jets using the cavities that were attached at the exit of a circular supersonic nozzle. The effect of cavity-actuated forcing was studied as a function of the cavity geometry, in particular, the length and the depth of the cavity. When the cavities were tuned to certain frequencies, large-scale highly coherent structures were produced in the shear layers substantially increasing the growth rate. The cavity excitation was successfully applied to both cold and hot supersonic jets. When applied to cold Mach 2.0 air jets, the cavity-actuated forcing increased the spreading rate of the initial shear layers with the convective Mach number (M c ) of 0.85 by a factor of three. For high-temperature Mach 2.0 jets with M c of 1.4, a 50% increase in the spreading rate was observed with the forcing. Finally, the cavity-actuated forcing was applied to reacting supersonic jets with ethylene-oxygen afterburning. For this case, the forcing caused a 20%–30% reduction in the afterburning flame length and modified the afterburning intensity significantly. The direction of the modification depended on the characteristics of the afterburning flames. The intensity was reduced with forcing for unstable flames with weak afterburning while it was increased for stable flames with strong afterburning.

Supersonic shear layers have inherent low mixing rates due to compressibility effects. Their mixing rate relative to subsonic shear layers can be up to 5 times lower. Several important technological applications which require intense mixing of supersonic flows gave impetus to research aimed to develop methods to enhance mixing in compressible flows with minimal performance penalty. This paper reviews some of these methods applied to both planar shear layers and jets. The methods are arranged in several categories: passive and active control of shear layer instabilities, three dimensional jets, generation of axial vorticity and shock interaction with shear layers. The paper concludes by discussing the importance of the wide range of length-scales in which turbulent mixing occurs.

An experimental investigation was performed to characterize liquid fuel injection into a Mach 2.1 air ow simulating a scramjet combustor ow eld. In particular, ethanol fuel penetration and dispersion characteristics were compared for a transverse injection con guration with and without a mixing enhancement n. In previous studies involving gaseous fuel, the n-guided injection not only improved turbulent mixing but also served to reduce the stagnation pressure loss associated with the fuel injection shock. To characterize the mixing between liquid fuel and air, laser-sheet visualization of the cross-sectional view and schlieren visualization of the side view were obtained at various time intervals. Wall pressure measurements along the axial direction were used to help interpret the visualization data. The side views of fuel injection revealed that the fuel penetration height in the neareld was twoto six-times greater with the n-guided injection. Also, downstream evolution of the liquid-fuel cross-sectional area suggested a greater vaporization rate associated with the n-guided injection. The present results showed a signi cant improvement in key injection parameters and mixing characteristics associated with n-guided liquid-fuel injection.

Reacting flow experiments conducted in a dump combustor with an inlet-tocombustor length ratio of greater than 10, resulted in self-sustained, combustion driven oscillations at various longitudinal acoustic mode frequencies of the setup. Under certain operating conditions, vortex-driven combustion instabilities gave rise to pressure oscillations at the 15 th harmonic. Since a natural selection of such a high harmonic over the lower ones is very unlikely, plausible physical mechanisms explaining the selective amplification of the higher modes were investigated. Analytical results show that the observed instability mode could be naturally selected under properly timed vortex shedding and vortex bursting processes if those processes took place at specific locations with respect to the shape of the acoustic mode that was eventually amplified. To explain the physics of this unlikely mode selection process, three specific steps of frequency analysis are described. The first two steps have to do with the well-known duct acoustic resonance and the Rayleigh criterion. The last step is associated with the vortex shedding and bursting locations. It is shown that in order to sustain a vortex-driven feedback mechanism, vortex shedding has to occur at or near a velocity anti-node, while vortex bursting has to occur near a pressure anti-node. Equivalently, it is shown that for such vortex driven modes in a dump combustor, a velocity anti-node has to occur near the dump plane and a pressure anti-node has to occur near the exit nozzle. When applied to the observed experimental conditions, those frequency selection steps accurately analyzed the 15 th harmonic as the main mode of oscillations.

Computational Results are presented from an ongoing study to simulate the flameacoustic interaction behavior observed in a model shear-coaxial injector experiment. Acoustically forced turbulent mixing between gaseous hydrogen and oxygen jets was simulated at various forcing frequencies and amplitudes using the Loci-Chem CFD solver. A grid was generated that provided a reasonable approximation of the acoustics within the combustor chamber. Previous experimental work demonstrated that the flow field depended heavily on both the frequency and amplitude of forcing. Qualitative examination of the simulated flow fields showed similar dependence to the forcing amplitude and frequency, as well as a similarity in the general behavior. Although these results are promising, further work simulating reacting flow fields in Loci-Chem is required before any conclusions can be drawn on its usefulness in modeling combustion instabilities.

Non-premixed and premixed modes of Colorless Distributed Combustion (CDC) are investigated for application to gas turbine combustors. The CDC provides significant improvement in pattern factor, reduced NOx emission uniform thermal field in the entire combustion zone for it to be called as a isothermal reactor, and lower sound levels. Basic requirement for CDC is mixture preparation through good mixing between the combustion air and product gases so that the reactants are at much higher temperature to result in hot and diluted oxidant stream at temperatures that are high enough to autoignite the fuel and oxidant mixture. With desirable conditions one can achieve spontaneous ignition of the fuel with distributed combustion reactions. Distributed reactions can also be achieved in premixed mode of operation with sufficient entrainment of burned gases and faster turbulent mixing between the reactants. In the present investigation two non-premixed combustion modes and one premixed combustion mode that provide potential for CDC is examined. In all the configurations the air injection port is positioned at the opposite end of the combustor exit, whereas the location of fuel injection ports is changed to give different configurations. The results are compared for global flame signatures, exhaust emissions, acoustic signatures, and radical emissions using experiments and flow field, gas recirculation and mixing using numerical simulations. Ultra low NOx emissions are observed for both the premixed and non-premixed combustion modes, and almost colorless flames (no visible flame color) have been observed for the premixed combustion mode. The non-premixed mode was also provided near colorless distributed combustion. The reaction zone is observed to be significantly different in the two non-premixed modes.

Colorless Distributed Combustion (CDC) is being investigated as it shows great potential for lower NOx emissions, noise reduction and uniform thermal field for gas turbine combustors. CDC is characterized by distributed reaction zone which leads to uniform thermal field and provide significant improvement in pattern factor, reduced NOx emission and lower sound levels. Basic requirement for CDC is mixing between the combustion air and product gases to form hot and diluted oxidant prior to its mixing with the fuel. This leads to spontaneous ignition of the fuel and distributed reactions. This requirement can be met with different configuration of fuel and air injections with carefully characterized flowfield distribution within the combustion zone. In the present investigation four sample configurations have been examined. They include a diffusion flame configuration and three other configurations that provide potential for CDC mode combustor operation. For all four modes same fuel and air injection diameters are used to examine the effect of flow field configuration on combustion characteristics. The results are compared for flowfield and fuel/air mixing using numerical simulations and global flame signatures, exhaust emissions, acoustic signatures, and thermal field using experiments. Both numerical simulations and experiments are performed at heat load of 25kW, using methane as the fuel at atmospheric pressure using normal temperature air and fuel. Lower NOx emissions, better thermal field uniformity, and lower acoustic levels have been observed when the flame approached CDC modes as compared to the baseline case of a diffusion flame. The reaction zone is observed to be distributed over the combustor volume in the CDC mode.

The oblique and traverse configurations of injecting gaseous fuel in a low-aspect-ratio supersonic combustor are characterized and compared numerically using a validated code. Non-reacting conditions are considered, where fuel is simulated by helium. The combustor, which has a rectangular cross-section of constant span, is attached to a Mach 2 nozzle and expands along the top and bottom walls. A choked wall port is used for both injection configurations. Different sets of operating conditions have been simulated. It was found that injecting fuel obliquely results in higher efficiency as well as effectiveness. Unlike the traverse configuration, oblique injection makes use of the beneficial interaction of the injection-induced shock waves with the air/fuel shear layer. This interaction was proven in previous research to be effective for mixing enhancement in supersonic flows. However, in contrast with the results of previous research, normal or oblique injection at large angles (30° or 60°) is not necessary for the achievement of sufficient mixing in supersonic flows. Substantial mixing improvement was found at angles as small as 5°. Fuel injection at such small angles improves the fuel-air mixing while minimizing the injection-induced pressure losses, which leads to increased thrust. Our results on mixing under non-reacting conditions provide good preliminary insights on a more favorable fuel injection configuration that provides better mixing with lower losses and higher thrust.

A thin wedge-shaped blade, sometimes known as pylon, was used to enhance transverse fuel mixing in Mach 2 airflow. Supersonic mixing experiments were conducted using 45° and 90° fuel injections from the wall in the immediate wake of the pylon blade, and the results were compared with a baseline case using 90° transverse fuel injection without any pylon assistance. All the injectors had a same-size diameter, and the injector diameter matched the maximum blade thickness at the pylon base. For both qualitative and quantitative comparison, wall-pressure measurements, planar Mie-scattering of smokeseeded fuel streams, and instantaneous and time-averaged schlieren visualization were employed. Fuel penetration height was measured as a function of axial distance from the schlieren images, while flow losses associated with pylon and jet-induced shocks were assessed from the wall pressure measurements. Also, the wake flow extending downstream from the pylon base was characterized by planar Mie-scattering images. Substantial improvement in mixing performance was observed with the use of pylon as the fuel penetration height was increased by 100~120% and the flow losses associated with jetinduced shock were reduced by 13~30%. The 90° injection achieved greater fuel penetration heights, while the 45° injection incurred greater savings in flow losses. The results open up the possibility of further increasing the performance by optimizing the fuel injection angle behind the pylon.

The mixing of a gaseous fuel jet injected into a shock-wave-rich supersonic airflow has been examined. Different airflow Mach numbers and injection configurations (parallel, oblique, and normal) have been investigated numerically using a validated code. It was found that the shock-wave-rich environment, i.e. the shock train of the airflow, aids to a great extent in mixing enhancement for all injection configurations, especially at small oblique injection angles close to the parallel configuration. Unlike what was shown in previous research on fuel injection in shock-free supersonic airflows that normal or oblique injection at angles as large as 30° or 60° is inevitable to achieve sufficient mixing, substantial mixing improvement is provided by shock-wave-rich airflows. Fuel injection at small oblique angles of only about 5° improves mixing while minimizing the pressure losses accompanying the fuel injection, which leads to increased thrust. Due to the lack of experimental data for validation, the code was validated by using it to numerically simulate part of the results of a past experimental investigation from the literature. Good agreement of the simulation results and the actual flowfield data was observed. The global features of the flow under high-speed conditions are similar to those reported previously, so that our results of mixing in shock-rich supersonic flows provide good insights on a more favorable fuel injection configuration that provides better mixing with lower losses and higher thrust.

Nonreacting supersonic mixing experiments were conducted to characterize and compare two different fuel-injection schemes applied to a high-aspect ratio supersonic combustor. The combustor, which has a constant-width rectangular cross-section, is attached to a Mach 2 nozzle and expands along the top and bottom walls. Helium was injected through a choked orifice using transverse wall injection and parallel ramp injection. Pressure traces along the top wall of the combustor were used to determine flow field characteristics as well as to evaluate the static pressure rise caused by each system. Instantaneous and time-averaged Schlieren images were taken to visualize the resulting flow structures of each injection and to measure fuel penetration depth. Planar Mie scattering using ethanol droplets was also conducted to visualize the lateral and vertical spreading of the fuel in the near and far field. Results show that the ramp can be used with parallel injection providing just as effective fuel penetration as in the normal injection, but it also caused higher static pressure rise indicating increased flow loss in the main airflow. While combustion experiments are necessary to assess the impact on actual combustion performance and thermal efficiency, the present nonreacting mixing study provides more insights into the limitations of each injector configuration.

Compact liquid-fueled swirl-stabilized combustors using advanced fuel atomization techniques provide a potential solution to challenges associated with propulsion of submerged naval vehicles. The submerged pressurized operation of a combustor may alter the operational characteristics and performance of the combustor and also involves two-phase phenomena when the combustor exhaust gases interact with surrounding water. This work describes an investigation of swirl-stabilized flames created in a combustor featuring coannular swirling airflows under submerged conditions. A central atomization-air jet was used to atomize methanol fuel, providing great flexibility and control over fuel spray properties in a compact geometry. Direct photography was used to examine the global features of the flame. High-speed imaging was used to examine the two-phase shear-layer behavior of the exhaust jet from the combustor. Sound spectra associated with exhaust jets under different operational conditions were compared. Three nozzle geometries were examined, and the impact of nozzle geometry on the two-phase jet interaction was assessed. The two-phase interaction of the exhaust jet was found to depend heavily on the pressure drop over the exhaust nozzle. The dynamic behavior of the exhaust jet was buoyancy-driven at low-pressure drops and was affected by complex instability mechanisms at high-pressure drops. Nozzle features that have been shown to affect the behavior of single-phase jet shear-layer instability were found to have little effect in a two-phase situation. The instability mechanisms involved in the two-phase cases investigated here are shown to be significantly different from jets in single-phase cases. Evidence is presented that a pressure-wave interaction mechanism, similar to Richtmyer―Meshkov instability, may play an important role in the evolution of unstable structures at the two-phase interface when high combustor pressures are involved.

The effects of flame enclosure and combustor pressure on the combustor flowfield and structure of turbulent spray flames have been investigated. The exhaust jet from the combustor was directed into water to simulate underwater propulsion applications. Twophase interactions between the exhaust jet from the pressurized combustor and liquid water in an attached mixing chamber have been examined to address issues associated with the operation of a submerged combustor for underwater propulsion applications. Enclosure of the flame, in the absence of pressurization, was found to affect the flame structure and dynamic behavior significantly. At normal pressure, the combustor flow attached to one side of the exit port to result in large-scale, low-frequency flame oscillations. In contrast the pressurized flame, obtained by constricting the exhaust flow from the combustor, was not found to display the same large-scale distortions associated with the enclosed, unpressurized case. The global structure of the pressurized and unenclosed flame cases was similar; however, the flame in pressurized case was found to be shorter. Axial and radial velocities of the air flowfield in the combustor were examined using particle imaging velocimetry. The interaction of the gaseous combustor exhaust jet with water in a downstream mixing chamber was also examined. Unchoked and choked isothermal exhaust jets at two combustor pressures were examined using two exhaust nozzle geometries. The primary parameter influencing the behavior of the isothermal exhaust jet was found to be the pressure drop across the exhaust port. At relatively low combustor pressures the exhaust jet was found to emerge as a series of distinct bubbles. The inertia of the liquid phase and buoyant forces acting on the bubbles appear to dominate the dynamic behavior of the jet. Swirl imparted to the combustion air did not affect the behavior of the jet significantly, since the momentum of the jet is insignificant in comparison to other factors associated with the two-phase jet interaction. The exhaust nozzle geometry affects the shape of the bubble during emergence and detachment. However, the frequency of the bubble formation and detachment was not influenced strongly by exhaust nozzle geometry. The Strouhal number associated with the cycle in the two-phase flow case was found to be two orders of magnitude smaller than that associated with large-scale jet mixing instabilities in single-phase mixing. The maximum diameter of the large-scale bubble structure was found to be effectively independent of combustor chamber pressure. The structure and dynamics of the exhaust jet from a reacting flow case, issuing from the pressurized combustor, were also examined. The structure of the heated jet was found to be more chaotic than the isothermal flow, displaying events occurring across a wide range of time scales. A distinct, repeatable bubble-formation mode was not observed. The exhaust jet associated with the reacting flow displayed a greater tendency to produce extremely small as well as large bubbles, with diameters ranging from microns to a few millimeters.