Numerical Modeling of Soot Production in Aero-Engine Combustors Using Large Eddy Simulations

Numerical simulations are regarded as an essential tool for improving the design of combustion systems since they can provide information that is complementary to experiments. However, although numerical simulations have already been successfully applied to the prediction of temperature and species concentration in turbulent flames, the production of soot is far from being conclusive due to the complexity of the processes involved in soot production. In this context, first Large Eddy Simulations (LES) of soot production in turbulent flames are reported in the literature in laboratory-scale configurations, thereby confirming the feasibility of the approach. However numerous modeling and numerical issues have not been completely solved. Moreover, validation of the models through comparisons with measurements in realistic complex flows typical of aero-engines is still rare. This work therefore proposes to evaluate the LES approach for the prediction of soot production in an experimental swirl-stabilized non-premixed ethylene/air aero-engine combustor, for which soot and flame data are available. Two simulations are carried out using a two-equation soot model to compare the performance of a hybrid chemical description (reduced chemistry for the flame structure/tabulated chemistry for soot precursor chemistry) to a classical full tabulation method. Discrepancies of soot concentration between the two LES calculations will be analyzed and the sensitivity to the chemical models will be investigated.

Uncertainty Quantification and Stochastic Variations of Renewable Fuels

Renewable fuels have been successfully used in gas turbine combustion chambers and the layout of the chamber does not require major interventions if the composition is known. However, the variation in the composition in renewable fuels is higher than in fossil ones and it is stochastic. In principle, this variation affects the stability of the combustion, the emissions and the temperature distribution. The combustion chamber tested in this work has been designed to reproduce the temperature distribution of MT1 test case and modelled using reactive CFD simulations. The fuel is an ideal natural gas with a random mix of methane and hydrogen. In order to account the stochastic variation of the fuel composition, a probabilistic analysis is carried out with two sampling methods: a Monte Carlo simulation with meta-models and a Probabilistic Collocation Method. The two methodologies show similar results in terms of mean value and standard deviation. The paper proves that is possible to predict the mean value of temperature and emissions in a modern chamber and their associated standard deviation by applying an uncertainty quantification methodology. One of the major drawbacks of the composition change is the maximum temperature variation at the exit that can reduce the life of the downstream turbine. The variation in the emissions seems less important and all the major differences in the composition are mixed out before the combustion chamber exit.

Design Improvement Survey for NOx Emissions Reduction of a Heavy-Duty Gas Turbine Partially Premixed Fuel Nozzle Operating With Natural Gas: Experimental Campaign

As government regulations become increasingly strict with regards to combustion pollutant emissions, new gas turbine combustor designs must produce lower NOx while also maintaining acceptable combustor operability. The design and implementation of an efficient fuel/air premixer is paramount to achieving low emissions. Options for improving the design of a natural gas fired heavy-duty gas turbine partially premixed fuel nozzle have been considered in the current study. In particular, the study focused on fuel injection and pilot/main interaction at high pressure and high inlet temperature. NOx emissions results have been reported and analyzed for a baseline nozzle first. Available experience is shared in this paper in the form of a NOx correlative model, giving evidence of the consistency of current results with past campaigns. Subsequently, new fuel nozzle premixer designs have been investigated and compared, mainly in terms of NOx emissions performance. The operating range of investigation has been preliminarily checked by means of a flame stability assessment. Adequate margin to lean blow out and thermo-acoustic instabilities onset has been found while also maintaining acceptable CO emissions. NOx emission data were collected over a variety of fuel/air ratios and pilot/main splits for all the fuel nozzle configurations. Results clearly indicated the most effective design option in reducing NOx. In addition, the impact of each design modification has been quantified and the baseline correlative NOx emissions model calibrated to describe the new fuel nozzles behavior. Effect of inlet air pressure has been evaluated and included in the models, allowing the extensive use of less costly reduced pressure test campaigns hereafter. Although the observed effect of combustor pressure drop on NOx is not dominant for this particular fuel nozzle, sensitivity has been performed to consolidate gathered experience and to make the model able to evaluate even small design changes affecting pressure drop.

Effects of Inlet Air Distortion on Gas Turbine Combustion Chamber Exit Temperature Profiles

Localized damage to turbine inlet nozzles is typically caused by non-uniform temperature distributions at the combustion chamber exit. This damage results in decreased turbine performance and can lead to expensive repair or replacement. A test rig was designed and constructed for the Rolls-Royce Allison 250-C20B dual-entry combustion chamber to investigate the effects of inlet air distortion on the combustion chamber’s exit temperature fields. The rig includes a purposely built water cooled thermocouple rake to sweep the exit plane of the combustion chamber. Test rig operating conditions simulated normal engine cruise conditions by matching the quasi-non-dimensional Mach number, equivalence ratio and Sauter mean diameter. The combustion chamber was tested with an even distribution of inlet air and a 4% difference in airflow at either combustion chamber inlet. An even distribution of inlet air to the combustion chamber did not produce a uniform temperature profile and varying the inlet distribution of air exacerbated the profile’s non-uniformity. The design of the combustion chamber promoted the formation of an oval-shaped toroidal vortex inside the combustion liner, causing localized hot and cool sections separated by 90° that were apparent in the exhaust. Uneven inlet air distributions skewed the oval vortex, increasing the temperature of the hot section nearest the side with the most airflow and decreasing the temperature of the hot section on the opposite side.

Prediction of Thermoacoustic Oscillations With a Linearised Euler Methodology

Combustion instabilities in the aviation, aerospace and power generation industries have been a matter of concern for engineers since the 1950s, but with the increase in computer processing speed and the development of CFD it is now possible to attempt to predict frequencies and stability of a combustion system by numerical means, or by combining numerical, analytical and experimental approaches. Currently available analytical methods for the prediction of the frequency and stability of thermoacoustic oscillations make use of one-dimensional models where the frequency of oscillation is assumed to be low enough that only plane waves propagate in the burner, with higher order modes decaying quickly. While accurate and well-suited for longitudinal oscillations, these methods are unable to predict the frequency of instabilities where the unsteady heat release couples with the higher frequency transverse acoustic modes. Therefore a method is needed for applications where high frequency transverse oscillations are important. A method in which the linearised Euler equations are employed to calculate the propagation of acoustic waves is then suitable for solving this thermoacoustic problem. When a flame model that appropriately represents the frequency-dependent dynamics of the flame front is included, this method can predict the frequency of the oscillation resulting from the coupling between acoustics and combustion in an arbitrarily complex geometry. In this paper, a linearised Euler solver called INSTANT is introduced and validated against a well known theoretical model for the calculation of thermoacoustic oscillations in a one dimensional cylindrical duct with rigid walls and a radially uniform mean flow. The frequencies of oscillation and the modeshapes for this stable configuration match the theoretical ones well. An example calculation of transverse acoustic resonant mode is then presented. The ability of the code to predict the production of an entropy mode as a result of the interaction between an acoustic wave and a heat source region and its ability to predict frequencies of oscillation and modeshapes in a one dimensional configuration give confidence it can serve as a predictive tool for high frequency, transverse thermoacoustic oscillations in the more complex geometries of practical combustion systems once a suitable model for the frequency dependent flame response is included. The development of such a flame model is left for future work.

Unsteady Simulations of a Low NOx LDI Combustor for Environmentally Responsible Aviation Engines

A key objective of NASA’s Environmentally Responsible Aviation (ERA) research program is to develop advanced technologies that enable 75% reduction of LTO NOx emissions of N+2 aviation gas turbine engines relative to the CAEP 6 standard. To meet this objective, a new advanced multi-point fuel injector was proposed and tested under the NASA ERA program. The new injector, called the three-zone injector, or 3ZI, uses fifteen spray cups arranged in three zones. Swirling air flows into each cup and fuel is introduced via pressure swirl atomizers within the cup. Multiple design parameters impact the performance of the injector, such as the location of the atomizer within the spray cup, the spray angle and cup-to-cup spacing. To fully understand the benefits and trade-offs of various injector design parameters and to optimize the performance of the injector, detailed CFD simulations are an essential tool. Furthermore, the CFD methodology must allow easy changes in design parameters and guarantee consistent and comparable accuracy from one design iteration to the next. This paper investigates the use of LES in reacting and non-reacting flows and compares against the NOx experimental data for the multi-point atomization strategy of the injector. The CFD simulations employ an automatically generated Cartesian cut-cell meshing approach with mesh refinement applied near complex geometry and spray regions. Adaptive Mesh Refinement (AMR) is used to refine mesh in regions of high gradients in velocity and temperature. The CFD simulations use boundary and operating conditions based on experimental data for air flow and spray atomization obtained from LDV and PDPA characterizations of the spray respectively. The results are extended to reacting flow using a detailed reaction mechanism and predictions of NOx emissions are compared to experimental data. Overall NOx predictions were consistently less than experimental values. However, the NOx prediction trends showed excellent agreement with experimental data across the wide range of equivalence ratios investigated.

Determination of Criteria for Flame Stability in an Annular Trapped Vortex Combustor

Development of lean premixed (LP) combustion is still a challenge as it results in considerable constraints for the combustor design. Indeed, new combustors using LP combustion are more prone to flashback, blow-off, or even thermo-acoustic instabilities. A detailed understanding of mechanisms leading to such extreme conditions is then crucial to reduce pollutant emissions, widen the range of operating conditions, and reduce design time. This paper reports the experimental study of an innovative LP trapped vortex combustor (TVC). The TVC concept uses a recirculating rich flow trapped in a cavity to create a stable flame that continuously ignites a main lean mixture passing above the cavity. This concept gave promising performances but some workers highlighted the existence of combustion instabilities for some operating conditions. Detailed studies have therefore been carried out in order to understand the occurrence of these drastic operating conditions. Results showed that the cavity flow dynamics in conjunction with the location of the interfacial mixing zone (between the cavity and the mainstream) were the driving forces to obtain stable combustion regimes. The goal of this work has been to take advantage of these detailed recommendations to determine stability maps, trends, and dimensionless parameters which could be easily used as early-design rules. For this reason, the study introduced a simple and robust criterion, based on the global pressure fluctuation energy. The latter was used to distinguish stable and unstable modes. An aerodynamic momentum flux ratio and a chemical stratification ratio (taken between the cavity and the mainstream) were defined to scale all measurements. Results indicated that the mainstream velocity was critically important to confine the cavity and to prevent combustion instabilities. Remarkably, this trend was verified and even more pronounced for larger cavity powers. In addition, flame stabilization above the cavity resulted in the existence of specific stratification ratios, in order to obtain a soft gradient of gas composition between the rich and lean regions. Finally, a linear relation between the mainstream and cavity velocities became apparent, thereby making possible to simply predict the combustor stability.

Generation, Characterization and Ignition of Lube Oil Mists

In this work, the generation conditions of lube oil mists by spray, their droplet size distributions and their flammability characteristics were studied. At first, tests were carried out on brand new lube oil having a viscosity of 32 mm2.s−1 at 40°C (ISO VG32) and a flashpoint higher than 200°C. Experiments were performed in a vertical semi-open tube with a 0.07 m-square cross section connected to ignition systems supplying energies ranging from 1 mJ to 5 kJ. The average droplet diameter, determined using a laser diffraction sensor, ranged from 4 to 60 μm depending on operating parameters. In the case of brand new mineral oil mists, the probability of ignition by a spark discharge can be considered as low, the minimum ignition energy (MIE) being 2 kJ for 5 μm droplets. The minimum explosive concentration is fairly high, in the order of 250 g.m−3. For mists with droplet diameters around 60 μm, MIE slightly increases to reach 2.5 kJ. The flammability of oil mists generated in case of maloperation (accidental fuel/lube mixing) was also studied. The ignition properties of used lube oil were not notably modified with regard to the brand new oil. However, the addition of organic volatiles compounds leads to a modification of mists flammability, nevertheless it is only perceptible for fuel contents greater than 20%v.

Auto-Ignition of In-Line Injected Hydrogen/Nitrogen Fuel Mixtures at Reheat Combustor Operating Conditions

De-carbonization of the power generation sector becomes increasingly important in order to achieve the European climate targets. Coal or biomass gasification together with a pre-combustion carbon capture process might be a solution resulting in hydrogen-rich gas turbine (GT) fuels. However, the high reactivity of these fuels poses challenges to the operability of lean premixed gas turbine combustion systems because of a higher auto-ignition and flashback risk. Investigation of these phenomena at GT relevant operating conditions is needed to gain knowledge and to derive design guidelines for a safe and reliable operation. The present investigation focusses on the influence of the fuel injector configuration on auto-ignition and kernel development at reheat combustor relevant operating conditions. Auto-ignition of H2-rich fuels was investigated in the optically accessible mixing section of a generic reheat combustor. Two different geometrical in-line configurations were investigated. In the premixed configuration, the fuel mixture (H2 / N2) and the carrier medium nitrogen (N2) were homogeneously premixed before injection, whereas in the co-flow configuration the fuel (H2 / N2) jet was embedded in a carrier medium (N2 or air) co-flow. High-speed imaging was used to detect auto-ignition and to record the temporal and spatial development of auto-ignition kernels in the mixing section. A high temperature sensitivity of the auto-ignition limits were observed for all configurations investigated. The lowest auto-ignition limits are measured for the premixed in-line injection. Significantly higher auto-ignition limits were determined in the co-flow in-line configuration. The analysis of auto-ignition kernels clearly showed the inhibiting influence of fuel dilution for all configurations.

Validation of a Novel LES Approach Using Tabulated Chemistry for Thermoacoustic Instability Prediction in Gas Turbines

For the prediction of thermoacoustic instabilities in gas turbines, a compressible, unsteady LES approach was validated for 1D, lab-scale and technical-scale test cases. The simulations were performed with a novel combustion model, which relies on a combination of tabulated chemistry and flame thickening. A 1D case was used for fundamental verification and to quantify the mesh resolution dependent error. The verification demonstrated sufficiently accurate predictions. In a second step an elevated pressure lab-scale flame was considered to calibrate the model parameter of the turbulence chemistry interaction at relevant conditions. For this case CO2 field data measured with the Raman scattering method is available. Finally, a test rig configuration of a can type combustor was investigated. The comparison between experiment and simulation was performed in terms of thermoacoustic pressure amplitudes at stable and unstable operating conditions. The LES combustion model relies on an artificial thickened flame approach to ensure that the flame propagation speed is reproduced with tolerable error. Tabulated chemistry provides the source term for CO2 (governing species) as a function of the mixture fraction and the CO2 concentration based on premixed, laminar flamelets. The model distinguishes between inherent thickening due to sub grid scale turbulence and explicit laminar thickening. This novel thickening approach is presented for the first time. The presented approach was able to predict the thermoacoustic stability behavior of a gas turbine combustion system correctly.