Dmitriy Shcherbik

Active Control of Combustion Oscillations by Non-Coherent Fuel Flow Modulation

This paper describes the application of active, open loop, control in effective damping of severe (i.e., peak to peak pressure amplitudes of up to 35 psi relative to a mean combustor pressure of 155 psi) combustion instabilities in a gas turbine combustor simulator. Active control was applied by harmonic modulation of the fuel injection rate into the combustor using a fast response actuating valve. To determine the dependence of the performance of the active control system upon the frequency, the fuel injection modulation frequency was varied between 50 and 500 Hz while the frequency of instability was around 386 Hz. These tests showed that the amplitude of the combustor pressure oscillations varied ten fold over the range of investigated frequencies, indicating that applying the investigated open loop control approach at the appropriate frequency could effectively damp detrimental combustion instabilities. When the same active control system was operated in closed loop with an optimized control signal phase at the frequency of instability, the amplitude of the unstable oscillations was reduced by a factor of two, thus causing the combustor to operate with lower amplitude instability.

Dynamics of Non-Premixed Bluff Body-Stabilized Flames in Heated Air Flow

This paper describes a study of the fundamental flame dynamic processes that control bluff body-stabilized combustion of liquid fuel with low dilatation. Specifically, flame oscillations due to asymmetric vortex shedding downstream of a bluff body (i.e., the Benard/von-Karman vortex street) were characterized in an effort to identify the fundamental processes that most affect the intensity of these oscillations. For this purpose, the spatial and temporal distributions of the combustion process heat release were characterized over a range of inlet velocities, temperatures, and overall fuel-air ratios in a single flame holder combustion channel with full optical access to the flame. A stream of hot preheated air was supplied to the bluff body using a preburner, and Jet-A fuel was injected across the heated gas stream from discrete fuel injectors integrated within the bluff body. The relative amplitudes, frequencies, and phase of the sinusoidal flame oscillations were characterized by Fourier analysis of high-speed movies of the flame. The amplitudes of the flame oscillations were generally found to increase with global equivalence ratio, reaching a maximum just before rich blowout. Comparison of the flame dynamics to the time-averaged spatial heat release distribution revealed that the intensity of the vortex shedding decreased as a larger fraction of the combustion process heat release occurred in the shear layers surrounding the recirculation zone of the bluff body. Furthermore, a complete transition of the vortex shedding and consequent flame stabilization from asymmetric to symmetric modes was clearly observed when the inlet temperature was reduced from 850°C to 400°C (and hence, significantly increasing the flame dilatation ratio from Tb /Tu ∼ 2.3 to 3.7).Copyright

Suppression of Instabilities in Gaseous Fuel High-Pressure Combustor Using Non-Coherent Oscillatory Fuel Injection

This paper describes the application of active, open loop, control in effective damping of severe combustion instabilities in a high pressure (i.e., around 520 psi) gas turbine combustor simulator. Active control was applied by harmonic modulation of the fuel injection rate into the combustor. The open-loop active control system consisted of a pressure sensor and a fast response actuating valve. To determine the dependence of the performance of the active control system upon the frequency, the fuel injection modulation frequency was varied between 300 and 420 Hz while the frequency of instability was around 375 Hz. These tests showed that the amplitude of the combustor pressure oscillations strongly depended upon the frequency of the open loop control. In fact, the amplitude of the combustor pressure oscillations varied ten fold over the range of investigated frequencies, indicating that applying the investigated open loop control approach at the appropriate frequency could effectively damp detrimental combustion instabilities. This was confirmed in subsequent tests in which initiation of open loop modulation of the fuel injection rate at a non resonant frequency of 300Hz during unstable operation with peak to peak instability amplitude of 114 psi and a frequency of 375Hz suppressed the instability to a level of 12 psi within approximately 0.2 sec (i.e., 75 periods). Analysis of the time dependence of the spectra of the pressure oscillations during suppression of the instability strongly suggested that the open loop fuel injection rate modulation effectively damped the instability by “breaking up” (or preventing the establishment of) the feedback loop between the reaction rate and combustor oscillations that drove the instability.Copyright

ACTIVE CONTROL OF INSTABILITIES IN HIGH-PRESSURE COMBUSTOR BY NON-COHERENT OSCILLATORY FUEL INJECTION

This paper describes the application of active, open loop, control resulted in effective damping of severe (i.e., peak to peak pressure amplitudes of up to 60 psi relative to a mean combustor pressure of 485 psi) combustion instabilities in a high pressure combustor. Active control was applied by harmonic modulation of the fuel injection rate into the combustor using a fast response actuating valve, which modulated the entire fuel flow rate. To determine the dependence of the performance of the active control system upon the amplitude of fuel flow modulations generated by the actuating valve the latter was varied between zero and about 16% (±8%) of the mean value of the fuel flow rate. Successful suppression of the 352Hz fundamental acoustic mode in the combustor was attained at about 12% fuel flow rate modulation at the frequency fCS=293Hz. These tests showed that the amplitude of the combustor pressure oscillations varied six fold over the range of investigated oscillatory outputs of the control valve, indicating that applying the investigated open loop control approach at the appropriate frequency could effectively damp detrimental combustion instabilities. It was shown that fundamental acoustic mode of the combustor collapsed at a certain level of the control output when it was gradually increased during control application. Modulation frequency at about 290Hz was determined to be optimal in our earlier study at relatively low power of the combustor (mean combustor pressure of 160 psi). In the current study this value has also proved to be optimal in the control of 352Hz instability at full power operating conditions by conducting frequency sweep controllability test in the range 285-330Hz. In an effort to gain better understanding of the control system operation its characteristics were investigated in the cold flow simulation tests using extensive instrumentation. These included dynamic pressure sensors in the actuating valve and in the fuel line near the injection orifice as well as hot film anemometer (which measured mass flow rate oscillations at the point of injection). This additional instrumentation provided data for the monitoring of the control input propagation through the system in the combustor.

Controllable Injection for Supercritical Combustion

*† ‡ § ** †† This paper describes experimental investigation of subcritical and supercritical liquid fuel injection and combustion in a high-pressure combustor. The back-scattered laser light diagnostics was implemented for online monitoring and characterization of heat release oscillations and spray quality. The change in spray quality in the subcritical regime affects combustor dynamics causing various regimes of stable and unstable combustor operation. Thus, spray quality optimization can be used to suppress combustor instabilities in subcritical regime, with the particular mode of instability depending on the incoming air temperature. Supercritical injection of heptane resulted in gradual decrease in combustor oscillatory pressure amplitude. The back-scattered laser light intensity gradually decreased with decreasing droplet size in subcritical regime and disappeared altogether when fuel became supercritical. The CO emission was reduced to an insignificant level for supercritical injection indicating complete combustion.

OPEN LOOP CONTROL OF SEVERE COMBUSTION INSTABILITIES BY FUEL FLOW MODULATION AT NON RESONANT FREQUENCIES

This paper describes a study of open loop control of combustion instability employing harmonic fuel injection rate modulation at a non-resonant frequency. It’s shown that this approach can effectively damp severe instabilities whose amplitude equal up to 30% of the mean combustor pressure in high pressure (i.e., Pc,max=580psi). The investigated control approach was also studied using a low order electronic simulator (LOES) in an effort to understand the effect of nonlinearities upon the open loop control approach. Initially, open loop control was applied to an unstable combustor operating at 20% of full power output to determine the dependence of the combustor’s response upon the control frequency. The fuel injection modulation frequency was varied between 50 and 500 Hz while the frequency of instability was around 400 Hz. These tests revealed that when the control frequency was in the 330-250Hz range, the controller completely damped the instability, apparently by “destroying” the feedback-like mechanism(s) that drives the instability. This study also revealed that when the ratio of the amplitude of the pressure oscillations and mean pressure was high, nonlinear effects severely distorted the instability waveform and hindered the effectiveness of the open loop control approach. Some of the initial tests were repeated with the combustor operating at full power. These also showed that modulating the fuel injection rate at specific, nonresonant, frequencies can completely damp severe instabilities. In fact, modulating the fuel injection rate at 336Hz damped 100psi peak to peak 386Hz oscillations within 0.15 seconds (~50 periods of the fundamental mode). Finally, we performed computer simulations of open loop control of the LOES to study the processes that control the effectiveness of the investigated open loop control approach. The developed LOES consisted a resonator with “damping” boundary conditions and an oscillating energy source. The resonator was modeled by two arrays of L-C elements and the boundary conditions simulated the fuel-air pre-mixer and nozzle. The energy source simulated the feedback interactions between the combustion process heat release and resonator flow oscillations. With proper choice of parameters, the LOES was able to reproduce the instability waveforms and relative amplitudes in the uncontrolled combustor. This study also revealed that severe waveform distortion, occurring when the ratio of the amplitude of the oscillations and the mean pressure is high, severely hinders the effectiveness of the open loop control approach.

Application of Planar Laser-Induced Phosphorescence to Investigate Jet-A Injection Into a Cross-Flow of Hot Air

This paper describes the development of the Planar Laser-Induced Phosphorescence (PLIP) technique for mapping the fuel temperature and concentration distributions in a jet-in-cross-flow (JICF) spray study. The spray was produced by injecting cold liquid Jet-A into hot cross-flowing air. The application of PLIP required the seeding of liquid fuel with micron-size thermographic phosphor particles before injection. The resulting spray produced phosphorescence and droplets Mie-scattering signals when illuminated by a 355nm planar UV laser sheet of 0.054J/pulse energy. The technique was investigated as a potential alternative to the use of Jet-A Planar Laser-Induced Fluorescence (PLIF) for the mapping of fuel concentration in sprays, because the low signal intensity of Jet-A’s fluorescence at high T prevents the use of the PLIF approach. In contrast, PLIP provides a strong signal at high T, and allows the simultaneous determination of local T and fuel concentration when two spectral bands of the phosphorescence emission are imaged and their ratio-of-intensities (RI) determined. In addition, the locations where liquid fuel droplets exist were imaged from the UV Mie-scattering of the laser-sheet (which can also be done in PLIF).In the present investigation, an optical system that imaged two spectral bands of phosphorescence and one wavelength of Mie-scattering was developed. It consisted of three CCD cameras with dichroic beam-splitters and interference narrow bandpass filters. The spray-pattern within a span of ∼80×30 orifice diameters was captured, with spatial resolution of about 0.1mm/px. The investigated jet-in-cross-flow spray was produced by injecting Jet-A fuel from a 0.671mm diameter orifice located on the wall of a rectangular channel (25.4×31.75mm cross-section). The cross-flow air was preheated to temperatures encountered in modern gas turbines (up to 480°C), while the temperature of the injected Jet-A fuel was in the T = 27–80°C range. YVO4:Eu phosphor particles with a median size of 1.8 microns were used to seed the fuel.Since the emissions of the commonly used Dy:YAG thermographic phosphor were found to be too weak and had wavelengths that overlapped with Jet-A fluorescence signals, YVO4:Eu was used for the JICF studies instead. It was observed that while the emissions of YVO4:Eu were stronger than Dy:YAG, the range of T where it can be applied in the PLIP technique was more limited — just sufficient for the investigated JICF. Preliminary results from the study showed rapid changes in fuel concentration and T from the injector up to z/dinj∼30 for momentum ratios of J = 5, 10 and 20, followed by a more gradual mixing/heat-up downstream. It was also found that deposition of phosphor particles on channel-walls interfered with the spray characterization, reducing the accuracy of the measurements.Copyright

Determination of Equivalence Ratio and Oscillatory Heat Release Distributions in Non-Premixed Bluff Body-Stabilized Flames Using Chemiluminescence Imaging

In an effort to elucidate the fundamental processes controlling bluff body flame stabilization, the dependence of the spatial distribution of the local equivalence ratio and the heat release dynamics upon the mode of fuel injection was studied. Experiments were performed in a single flame holder combustion channel which was supplied with a high-temperature air stream. Jet-A fuel was injected across the incoming air stream from one of two locations: a cylindrical fuel bar installed 0.25 m upstream of the bluff body, or from fuel injectors integrated within the bluff body 2.5 cm upstream of the trailing edge (i.e., close-coupled injection). The time-averaged spatial distributions of the combustion heat release were characterized by CH* and C2 * chemiluminescence imaging of the flame, and ratios of the C2 * to CH* light emission were used to characterize the local equivalence ratio. The spatial average of the C2 */CH* value in the flame was found to increase linearly with increasing global equivalence ratio for fuel injection upstream of the bluff body, whereas this value was relatively constant for close-coupled injection. This constant value equaled the same average C2 */CH* value obtained for upstream fuel injection at globally stoichiometric conditions, suggesting that combustion resulting from close-coupled fuel injection took place, on average, in stoichiometric flamelets throughout the combustor. The heat release dynamics due to asymmetric (von Karman) vortex shedding were also investigated for each operating condition by recording high-speed movies of the flame at 24 kHz. Upon processing of these movies, the amplitudes of heat release fluctuations due to von Karman vortex shedding were found to be significantly higher for close-coupled injection than for injection well upstream of the flame holder for all operating conditions. This is attributed to an increase in span-wise fuel-air mixing and near-wake heat release for upstream fuel injection, resulting in a hotter recirculation zone which suppressed the von Karman instability more than the close-coupled case.Copyright

Onset and Suppression of Instabilities in High Pressure Air- Breathing Combustor

Combustion instabilities were investigated in experiments where the fuel was rapidly heated close to critical temperatures, but the pressure in the combustion chamber was kept below the critical value for the injected fuel, n-heptane (C7H16). Two different unstable modes (~90Hz and ~400Hz) were excited depending on whether the intake air was preheated or not. Where the 90Hz mode was dominant, higher fuel preheat temperatures led to a lower level of registered instabilities. When the 400Hz mode was dominant, the fuel heater was set to operate at the highest productivity but severe combustion instabilities remained essentially unchanged. The observed qualitative trends agreed with the known data obtained in the high-power tests, where pressure in the combustor exceeded critical value.

Autoignition of a Jet-A Fuel Spray in a High Temperature Vitiated Air Flow

This study investigates the autoignition of a Jet-A fuel spray in a preheated air flow in a 75mm inner diameter quartz tube at atmospheric pressure. Fuel was injected via a pressure-swirl atomizing injector enclosed in an aerodynamically-shaped outer body (7mm diameter) that is installed coaxially with the flow using 3 water-cooled pylons. The air temperature and oxygen content were in the range of 1000–1400 K and 9–12%, respectively by controlling the equivalence ratio in the primary zone of the pre-burner/vitiator and by adding dilution air downstream. The co-flow air velocity ranged from 25–30 m/s for this study (Re = 10,000–12,000), with a trapezoidal profile. Fuel spray was characterized using PDPA. Shape of the spray was transitional between hollow and solid cone with the corner angle of 35° to the axis near injector which reduced to 22° downstream. Spray density varied significantly over cross-section of the tube with the minimum on the axes. Droplets produced have average diameters (D10) of 15–70μm on the axes and periphery, respectively, at 6 cm downstream from the injector. Character of the droplet size distribution was polydisperse. Auto-ignition time delays were captured using a time-averaged camera with CH* (432nm) filter. The measured values agree with delay times previously reported in literature. Two synchronized high-speed cameras with 432nm and 307nm filters were used to investigate dynamics of auto-ignition kernel initiation and convection by capturing of CH* and OH* chemiluminescence at 5000 f ps. This methodology allowed qualitative characterization of the equivalence ratio of kernels in process of their convection and growth. It was shown that kernels are always initiated on the axes of the spray where the average droplet size is minimum. Kernels were formed leaner and become richer as they grow down-stream as indicated by the increase of CH*/OH*intensity ratio. Additionally, kernel behavior depends greatly on air temperature with kernels transitioning from randomly appearing (i.e. single kernel), to periodic, to a constantly auto-igniting flame with the spatial scatter of ignition kernels decreasing with temperature.Copyright