Comparison of Numerical Combustion Models for Hydrogen and Hydrogen-Rich Syngas Applied for Dry-Low-NOx-Micromix-Combustion

The Dry-Low-NOx (DLN) Micromix combustion technology has been developed as low emission combustion principle for industrial gas turbines fueled with hydrogen or syngas. The combustion process is based on the phenomenon of jet-in-crossflow-mixing. Fuel is injected perpendicular into the air-cross-flow and burned in a multitude of miniaturized, diffusion-like flames. The miniaturization of the flames leads to a significant reduction of NOx emissions due to the very short residence time of reactants in the flame. In the Micromix research approach, CFD analyses are validated towards experimental results. The combination of numerical and experimental methods allows an efficient design and optimization of DLN Micromix combustors concerning combustion stability and low NOx emissions. The paper presents a comparison of several numerical combustion models for hydrogen and hydrogen-rich syngas. They differ in the complexity of the underlying reaction mechanism and the associated computational effort. For pure hydrogen combustion a one-step global reaction is applied using a hybrid Eddy-Break-up model that incorporates finite rate kinetics. The model is evaluated and compared to a detailed hydrogen combustion mechanism derived by Li et al. including 9 species and 19 reversible elementary reactions. Based on this mechanism, reduction of the computational effort is achieved by applying the Flamelet Generated Manifolds (FGM) method while the accuracy of the detailed reaction scheme is maintained. For hydrogen-rich syngas combustion (H2-CO) numerical analyses based on a skeletal H2/CO reaction mechanism derived by Hawkes et al. and a detailed reaction mechanism provided by Ranzi et al. are performed. The comparison between combustion models and the validation of numerical results is based on exhaust gas compositions available from experimental investigation on DLN Micromix combustors. The conducted evaluation confirms that the applied detailed combustion mechanisms are able to predict the general physics of the DLN-Micromix combustion process accurately. The Flamelet Generated Manifolds method proved to be generally suitable to reduce the computational effort while maintaining the accuracy of detailed chemistry. Especially for reaction mechanisms with a high number of species accuracy and computational effort can be balanced using the FGM model.

Evaluation of Kinetic Mechanisms for Direct Fired Supercritical Oxy-Combustion of Natural Gas

Existing kinetic mechanisms for natural gas combustion are not validated under supercritical oxy-fuel conditions because of the lack of experimental validation data. Our studies show that different mechanisms have different predictions under supercritical oxy-fuel conditions. Therefore, preliminary designers may experience difficulties when selecting a mechanism for a numerical model. This paper evaluates the performance of existing chemical kinetic mechanisms and produces a reduced mechanism for preliminary designers based on the results of the evaluation. Specifically, the mechanisms considered were GRI-Mech 3.0, USC-II, San Diego 204-10-04, NUIG-I, and NUIG-III. The set of mechanisms was modeled in Cantera and compared against the literature data closest to the application range. The high pressure data set included autoignition delay time in nitrogen and argon diluents up to 85 atm and laminar flame speed in helium diluent up to 60 atm. The high carbon dioxide data set included laminar flame speed with 70% carbon dioxide diluent and the carbon monoxide species profile in an isothermal reactor with up to 95% carbon dioxide diluent. All mechanisms performed adequately against at least one dataset. Among the evaluated mechanisms, USC-II has the best overall performance and is preferred over the other mechanisms for use in the preliminary design of supercritical oxy-combustors. This is a significant distinction; USC-II predicts slower kinetics than GRI-Mech 3.0 and San Diego 2014 at the combustor conditions expected in a recompression cycle. Finally, the global pathway selection method was used to reduce the USC-II model from 111 species, 784 reactions to a 27 species, 150 reactions mechanism. Performance of the reduced mechanism was verified against USC-II over the range relevant for high inlet temperature supercritical oxy-combustion.

Assessment of Scale Resolved CFD Methods for the Investigation of Lean Burn Spray Flames

Incoming standards on NOx emissions are motivating many aero-engines manufacturers to adopt the lean burn combustion concept. However, several technological issues have to be faced in this transition, among which limited availability of air for cooling purpose and thermoacoustics phenomena that should be managed to safely implement this burning mode. In this scenario, standard numerical design tools are not often capable of characterizing such devices. Thus, considering also the difficulties of experimental investigations in a highly pressurized and reactive environment, unsteady scale resolved CFD methods are required to correctly understand the combustor performances. In the last years Large Eddy (LES) and hybrid RANS-LES models such as Scale Adaptive Simulations (SAS) have undergone considerable developments. Such approaches have been already applied for gaseous flames, leading to a strong enhancement in phenomena prediction with respect to standard steady-state simulations. However, huge research efforts are still required when spray flames are considered, since all the numerical models chosen to describe spray dynamics and the related reactive processes can have a strong impact on the accuracy of the whole simulation. In this work a set of scale resolved simulations have been carried out on the DLR Generic Single Sector Combustor spray flame for which measurements both in non-reactive and reactive test conditions are available. Exploiting a two-phase Eulerian-Lagrangian approach combined with a Flamelet Generated Manifold (FGM) combustion model, LES simulations have been performed in order to assess the potential improvements with respect to steady state solutions. Additional comparisons have also been accomplished with SAS calculations based on Eddy Dissipation combustion model (EDM). The comparison with experimental results shows that the chosen unsteady strategies lead to a more physical description of reactive processes with respect to RANS simulations. FGM model showed some limitations in reproducing the partially-premixed nature of the flame, whereas SAS-EDM proved to be a robust modelling strategy within an industrial perspective. A new set of spray boundary conditions for liquid injection is also proposed whose realiability is proved through a detailed comparison against experimental data.

Online Monitoring of Thermoacoustic Eigenmodes in Annular Combustion Systems Based on a State Space Model

Thermoacoustic instabilities have the potential to restrict the operability window of annular combustion systems, primarily as a result of azimuthal modes. Azimuthal acoustic modes are composed of counter-rotating wave pairs, which form traveling modes, standing modes, or combinations thereof. In this work, a monitoring strategy is proposed for annular combustors that accounts for azimuthal mode shapes. Output-only modal identification has been adapted to retrieve azimuthal eigenmodes from surrogate data, resembling acoustic measurements on an industrial gas turbine. Online monitoring of decay rate estimates can serve as a thermoacoustic stability margin, while the recovered mode shapes contain information that can be useful for control strategies. A low-order thermoacoustic model is described, requiring multiple sensors around the circumference of the combustor annulus to assess the dynamics. This model leads to a second order state space representation with stochastic forcing, which is used as the model structure for the identification process. Four different identification approaches are evaluated under different assumptions, concerning noise characteristics and preprocessing of the signals. Additionally, recursive algorithms for online parameter identification are tested.

Liquid Fuel Composition Effects on Forced, Non-Premixed Ignition

The effects of jet fuel composition on ignition probability have been studied in a flowfield that is relevant to turbine engine combustors, but also fundamental and conducive to modeling. In the experiments, a spark kernel is ejected from a wall and propagates transversely into a crossflow. The kernel first encounters an air-only stream before transiting into a second, flammable (premixed) stream. The two streams have matched velocities, as verified by hot-wire measurements. The liquid fuels span a range of physical and chemical kinetic properties. To focus on their chemical differences, the fuels are prevaporized in a carrier air flow before being injected into the experimental facility. Ignition probabilities at atmospheric pressure and elevated crossflow temperature were determined from optical measurements of a large number of spark events, and high speed imaging was used to characterize the kernel evolution. Eight fuel blends were tested experimentally; all exhibited increasing ignition probability as equivalence ratio increased, at least up to 1.5. Statistically significant differences between fuels were measured that have some correlation with fuel properties. To elucidate these trends, the forced ignition process was also studied with a reduced order numerical model of an entraining kernel. The simulations suggest ignition is successful if sufficient heat release occurs before entrainment of colder crossflow fluid quenches the exothermic oxidation reactions. As the kernel is initialized in air, it remains lean during the initial entrainment of the fuel-air mixture; thus richer crossflows lead to quicker and higher exothermicity.

Impact of Swirling Flow Structure on Shear Layer Vorticity Fluctuation Mechanisms

Vorticity fluctuations have been identified as an important coupling mechanism during velocity-coupled combustion instability in swirl-stabilized flames. Acoustic oscillations in the combustor can cause all components of vorticity to oscillate, particularly the cross-stream, or azimuthal, vorticity that is excited in shear layer roll-up, and streamwise, or axial, vorticity that is excited during swirl fluctuations. These fluctuations can be induced by longitudinal acoustic fluctuations that oscillate across the swirler and dump plane upstream of the flame. While these fluctuations have been identified in a number of configurations, the sensitivity of this mechanism to flow configuration and boundary conditions has not been studied parametrically. In this study, we investigate the impact of time-averaged swirl level, confinement, and forcing frequency and amplitude on vorticity fluctuation dynamics in the azimuthal direction of a non-reacting swirling jet. The goal of this work is to better understand the dependence of vorticity fluctuations on these parameters as well as the vorticity conversion processes that occur in the flow. We have shown that vorticity fluctuation levels vary with time-averaged swirl number, particularly in the presence of a self-excited precessing vortex core, which dampens most acoustically-driven motion. Additionally, variations in forcing frequency excite flow response in different portions of the flow, particularly for different swirl numbers. Finally, confinement drastically changes the flow topology and unforced dynamics, resulting in significantly different response to forcing and generation of vortical fluctuations.

Spatio-Temporal Distribution of Soot Temperature for Flames Using Optical Pyrometry Under Unsteady Inlet Airflow Conditions

Two pyrometric tools for measuring soot temperature response in fuel-rich flames under unsteady inlet airflow conditions are developed. High-speed pyrometry using a high-speed color camera is used in producing soot temperature distributions, with its results compared with those of global soot temperature response measured using a multi-wavelength pyrometer. For the former, the pixel RGB values pertaining to respective bandwidths of red, green and blue filters are used to calculate temperature and for the latter, the emission from whole flame at 660 nm, 730 nm and 800 nm is used to measure temperature. The combustor, running on Jet-A fuel, achieves unsteady inlet airflow using a siren running at frequencies of 150 and 250 Hz and with modulation levels (RMS) 20–50% of mean velocity. Spatiotemporal response of flame temperature measured by the high speed camera is presented by phase-averaged with average subtracted images and by fast Fourier transform at the modulation frequencies of inlet velocity. Simultaneous measurement of combustor inlet air velocity and flame soot temperature using the multi-wavelength pyrometer is used in calculating the flame transfer function of flame temperature response to unsteady inlet airflow. The results of global temperature and temperature fluctuation from the 3-color pyrometer show qualitative agreement with the local temperature response measured by the high speed camera. Over the range of operating conditions employed, the overall flame temperature fluctuation increases linearly with respect to the inlet velocity fluctuation. The two-dimensional map of flame temperature under unsteady combustion determined using a high-speed digital color camera shows that the local temperature fluctuation during unsteady combustion occurs over relatively small region of flame and its level is greater (∼10–20%) than that of overall temperature fluctuation (∼1%).

An Acoustic Time-of-Flight Approach for Unsteady Temperature Measurements: Characterization of Entropy Waves in a Model Gas Turbine Combustor

Lean premixed combustion promotes the occurrence of thermoacoustic phenomena in gas turbine combustors. One mechanism that contributes to the flame-acoustic interaction is entropy noise. Fluctuations of the equivalence ratio in the mixing section cause the generation of hot spots in the flame. These so called entropy waves are convectively transported to the first stage of the turbine and generate acoustic waves that travel back to the flame; a thermoacoustic loop is closed. However, due to the lack of experimental tools, a detailed investigation of entropy waves in gas turbine combustion systems has not been possible up to now. This work presents an acoustic time-of-flight based temperature measurement method which allows the detection of temperature fluctuations in the relevant frequency range. A narrow acoustic pulse is generated with an electric spark discharge close to the combustor wall. The acoustic response is measured at the same axial location with an array of microphones circumferentially distributed around the combustion chamber. The delay in the pulse arrival times corresponds to the line-integrated inverse speed of sound. For validation of this new method an experimental setup was developed capable of generating well defined entropy waves. As a reference temperature measurement technique a hot-wire anemometer is employed. For the measurement of entropy waves in an atmospheric combustion test rig, fuel is periodically injected into the mixing tube of a premixed combustor. The subsequently generated entropy waves are detected for different forcing frequencies of the fuel injection and for different mean flow velocities in the combustor. The amplitude decay and phase lag of the entropy waves adheres well to a Strouhal number scaling for different mean flow velocities. In addition, simultaneously to the entropy wave measurement, the equivalence ratio fluctuations in the mixing tube are detected using the Tunable Diode Laser Absorption Spectroscopy (TDLAS) technique.

Flame Holding in the Premixing Zone of a Gas Turbine Model Combustor After Forced Ignition of H2-NG-Air-Mixtures

Flashback and self-ignition in the premixing zone of typical gas turbine swirl combustors in lean premixed operation are immanent risks and can lead to damage and failure of components. Thus, steady combustion in the premixing zone must be avoided under all circumstances. This study experimentally investigates the flame holding propensity of fuel injectors in the swirler of a gas turbine model combustor with premixing of H2-NG-air-mixtures under atmospheric pressure and proposes a model to predict the limit for safe operation. The A2EV swirler concept exhibits a hollow, thick walled conical structure with four tangential slots. Four fuel injector geometries were tested. One of them injects the fuel orthogonal to the air flow in the slots (jet-in-crossflow-injector, JICI). Three injector types introduce the fuel almost isokinetic to the air flow at the trailing edge of the swirler slots (trailing edge injector, TEI). A cylindrical duct and a window in the swirler made of quartz glass allow the application of optical diagnostics (OH* chemiluminescence and Planar Laser Induced Fluorescence of the OH radical (OH-PLIF)) inside the swirler. The fuel-air-mixture was ignited with a focused single laser pulse during steady operation. The position of ignition was located inside the swirler in proximity to a fuel injection hole. If the flame was washed out of the premixing zone not later than four seconds after the ignition the operation point was defined as safe. Operation points were investigated at three air mass flows, three air ratios, two air preheat temperatures (573 K, 673 K) and 40 to 100 percent per volume hydrogen in the fuel composed of hydrogen and natural gas. The determined safety limit for atmospheric pressure yields a similarity rule based on a critical Damköhler number. Application of the proposed rule at conditions typical for gas turbines leads to these safety limits for the A2EV burner: With the TEIs the swirler can safely operate with up to 80 percent per volume hydrogen content in the fuel at an air ratio of two. With the JIC injector safe operation at stoichiometric conditions and 95 percent per volume hydrogen is possible.

Prediction of Confined Flame Flashback Limits Using Boundary Layer Separation Theory

Premixed combustion is a common technology applied in modern gas turbine combustors to minimize nitrogen oxide emissions. However, early mixing of fuel and oxidizer opens up the possibility of flame flashback into the premixing section upstream of the combustion chamber. Especially for highly reactive fuels boundary layer flashback is a serious challenge. For high preheating and burner surface temperatures, boundary layer flashback limits for burner stabilized flames converge to those of so-called confined flames, where the flame is stabilized inside the burner duct. Hence, the prediction of confined flashback limits is a highly technically relevant task. In this study, a predictive model for flashback limits of confined flames is developed for premixed hydrogen-air mixtures. As shown in earlier studies, confined flashback is initiated by boundary layer separation upstream of the flame tip. Hence, the flashback limit can be predicted identifying the minimum pressure rise upstream of a confined flame causing boundary layer separation. For this purpose, the criterion of Stratford is chosen which was originally developed for boundary layer separation in mere aerodynamic phenomena. It is shown in this paper that it can also be applied to near wall combustion processes if the pressure rise upstream of the flame tip is modeled correctly. In order to determine the pressure rise, an expression for the turbulent burning velocity is derived including the effects of flame stretch and turbulence. A comparison of the predicted flashback limits and experimental data shows high prediction accuracy and wide applicability of the developed model.