Advanced Methods for the Prediction of Lean-burn Combustor Unsteady Phenomena

Final Report Summary - AMEL (Advanced Methods for the Prediction of Lean-burn Combustor Unsteady Phenomena)

Executive Summary:

Aviation gas turbines now reach a very high level of combustion efficiency. However, there are significant advances that must be made from the perspective of reducing emissions, such as NOx, soot and other aerosols and particulates, CO, and organic compounds. Attention is placed currently on a new concept, called “lean burn” or “lean direct injection” (LDI), where the aim is to create a flame that does not contain large regions of stoichiometric or close-to-stoichiometric mixtures. This implies significant premixing before the flame and large amounts of air through the injector, which has serious implications in terms of flame oscillations (“thermo-acoustics”) and flame stability (“lean blow-off”). The prediction of both these phenomena with computer simulation based on Computational Fluid Dynamics (CFD) would result in a quicker and cheaper design and development process in industry, but current simulation methods are not accurate enough. In addition, most current work focuses on single flames, while in practice gas turbine flames are located in an annulus allowing hence flame-flame and acoustic interactions. Therefore, research with annular flames is needed.

The overall aim of this project is to improve our simulation capability for thermoacoustics and blow-off for gas turbine flames. In particular, the objectives are to develop CFD methods for the acoustic damping efficiency of advanced cooling systems, to measure and develop modes for the Flame Describing Functions (FDF) that are used in network models for thermoacoustics, to develop multi-dimensional CFD models for the FDF and for flame blow-off, and to experimentally determine swirl spray flame structure and blow-off behaviour.

The project met its stated aims completely and all Technical Deliverables have been accomplished. In summary: (i) new methods to predict the cooling behaviour of impingement-effusion tiles have been developed and tested; (ii) new low-order models for more accurate boundary conditions (non-choked compressor), flame behaviour (steady and unsteady), and flame-acoustics interaction have been developed to quantify thermoacoustic oscillations; (iii) a series of experiments on single and annular configurations, aimed at quantifying oscillating behaviour and lean blow-off have provided new intuition and data for model validation; and (iv) CFD models for thermoacoustics, flame structure, and blow-off were developed. All models and computational codes have been delivered to industry, while they open new avenues for further research in gas turbine combustion and other fuel technologies.

Project Context and Objectives:

Aviation gas turbines now reach a very high level of combustion efficiency, in the sense of burning all the fuel used. In contrast, significant advances must be made from the perspective of reducing dangerous emissions, such as NOx, soot and other aerosols and particulates, CO, and organic compounds. In order to meet this objective, manufacturers of aero-engines currently focus attention on a new concept called “lean burn” or “lean direct injection” (LDI). In these technologies, the aim is to create a flame that contains small regions of stoichiometric or close-to-stoichiometric mixtures. This implies significant premixing before the flame and large amounts of air through the injector. This has some important implications for the behaviour of the flame. First, the flame is more susceptible to self-excited oscillations (“thermo-acoustics”), which can produce large-amplitude pressure oscillations and hence be detrimental to the engine performance and longevity. Second, there is an increased danger of flame extinction (“lean blow-off”), which needs to be mitigated without sacrificing the advantages of lean combustion. The theoretical treatment of both these phenomena is currently lacking. However, being able to predict flame and combustor behaviour in terms of thermoacoustics and blow-off at the design stage would be very effective and would result in a quicker and cheaper design and development process in industry. Hence, advances are needed in the accuracy of current-generation Computational Fluid Dynamics (CFD) codes that are used for describing flames in gas turbine combustors. Third, most current research work focuses on single flames, while in practice gas turbine flames are located in an annulus allowing hence flame-flame and acoustic interactions. These interactions can affect significantly both the thermoacoustic and the blow-off behaviour of the combustor. Therefore, research with annular flames is needed. Finally, most research in lab-scale swirl flames concerning lean extinction focuses on simple geometries where there is a single fuel and air entry; however, in practice, the air is delivered from multiple inlets. In this project, the stability of swirl flames with air delivered from two co-axial swirling flows will also be examined to provide data on lean blow-off that are of higher relevance to realistic gas turbine combustors.

The overall aim of this project is to improve our simulation capability for thermoacoustics and blow-off for gas turbine flames, with particular focus on thermoacoustics, blow-off, and annular interactions. In more detail, the specific objectives are: (i) to undertake measurements and develop CFD methods for the acoustic damping efficiency of advanced cooling systems; (ii) to measure Flame Describing Functions (FDF) and develop new low-order theoretical models for the FDF, in single-flame and annular configurations, which are used in network models for thermoacoustics; (iii) to develop multi-dimensional CFD models for the FDF and for flame blow-off; and (iv) to experimentally determine swirl spray flame structure and blow-off behaviour for flames with two co-axial swirling flows, which is the configuration used in industrial injectors. Meeting these objectives will result in higher accuracy in low-order and CFD modelling strategies.

Project Results:

The project was divided in four main work packages. The management (WP1) was successful in mitigating project risks and ensuring effective communications. The achievements of the technical work packages are described below.

Work Package 2: Low-order acoustics modelling

T2.1: Modelling of cooling systems

An optimised experimental methodology has been developed for determining the acoustic resistance and reactance characteristics of various geometries that simulate Impingement-Effusion (IE) tiles for use in future, low emission, gas turbine combustion systems. The experimental set-up includes an impedance style tube and a multi-microphone analysis technique to provide the acoustic resistance and reactance of various IE style geometries. A parallel computational effort has also provided knowledge on how to model such configurations.

The IE tile designs, heat transfer and impedance characteristics, together with the methodologies of simulating those, have proven successful and transferred to industry.

T2.2: Compressor boundary effects

The focus of the present research was on the formulation of the boundary conditions and the acoustic model of the upstream compressor for the case when the compressor's exit is not choked. An acoustic network approach was chosen. The primary variables are decomposed into a mean and a fluctuating contribution. The propagation of the fluctuations can be treated as one-dimensional wave transmission, i.e. two acoustic waves and one entropy wave, being governed by the cross-sectional area, radius, and length of the segment. The method has been implemented in LINEAR-B (a network code developed previously at the University of Cambridge), and this new capability has been delivered to the industrial partner.

Work Package 3: Validated methods for thermoacoustics of full annular combustion systems

T3.1: Improvements to low-order modelling

To model thermoacoustics, various flame transfer function modelling approaches have been considered and the one based on G-equation and a distribution of time delays has been selected and the work shows it is successful in capturing experimentally-obtained flame transfer functions. The tests against the annular combustor experimental results of Task 3.2 shows good agreement. The code has been delivered to industry.

T3.2: Experiments with forced flames

Experiments with the transverse forcing in the annular combustor have been performed for a range of forcing amplitudes and frequencies and for a range of operating conditions. For some of these conditions, the flame was self-excited and hence the experiments provided interesting insights on the response of already fluctuating flames to forcing at a different frequency. For the conditions where the natural self-excited oscillation was small, the forced experiments provided a flame transfer function that was evaluated based on the global response but also based on the response of individual burners of varying distance from the burner closest to the forcing. A comparison between the FDF of the whole combustor with the FDF of a single flame shows that under some conditions there are large differences, while at other conditions the single flame FDF describes reasonably well the FDF of the whole combustor. The results are useful for industry to understand when costly annular tests are needed and when it is sufficient to test at the single-sector level.

A single spray kerosene flame has also been studied to provide the reference of the physical processes occurring in forced spray flames. In these experiments, the flame motion as a function of kerosene type, forcing frequency, and flow rate has been measured and analysed, suggesting that the part of the flame in closest proximity to the annular pulsed air is the one that responds more, while the spray cone responds less.

T3.3: Prediction of the FDF with CFD for annular combustors

The CFD capabilities for predicting the flame transfer function of single-sector flames have been fully demonstrated for single sector flames from previously published flames and for the present annular combustor tests of Task 3.2. A detailed evaluation of the FDF is very expensive computationally, but the physics of flame motion captured in the CFD agrees with the experimental observations. For forced flames in annular configurations, which is a very novel configuration never studied before either experimentally or computationally, the CFD also produces good results, showing hence the usefulness of computational modelling for this very important combustion phenomenon.

Work Package 4: Validated methods for predicting lean extinction limits

T4.1: Experiments on lean extinction

A new dual-swirl burner flame has been built and used for premixed, non-premixed, and spray flames and the influence of the second, outer swirling stream on the overall flame shape has been quantified. The results show a significant alteration of the flame location, blow-off limits, and local flame extinction as a function of flow parameters, and in particular as a function of flame mode, and the data set can be used for further CFD model validation.

T4.2: Prediction of lean extinction

The Large-Eddy Simulation / Conditional Moment Closure (LES/CMC) prediction of blow-off of the multiple non-premixed flames has been completed and improved relative to the previous version of the model. The LES / Scalar Dissipation Rate (SDR) model for the prediction of blow-off of the premixed flames showed very good results for the flame structure as extinction is approached. Finally, the global blow-off behaviour of swirling spray flames has been successfully simulated with LES/CMC. The development not only show theoretical advances in the combustion modelling, but also in terms of scope of application, since this was the first CFD study of extinction of annular combustors. All computer codes have been delivered to industry.

Potential Impact:

IMPACT

The environmental benefits of lean-burn aviation jet engines are numerous. Lean-burn leads to very low NOx and soot levels. Lower emissions of particulates and NOx mean better air quality at ground level, and also an improved capability to meet future regulations, but also less radiative forcing and ozone depletion at high altitudes. The project’s outputs are assisting jet engine designers produce cleaner engines. In addition, a more effective design process leads to greater innovation, lower cost, and a larger number of jobs in industry. Although it is not easy to directly quantify the number of these jobs long-term, in the short-term, it is interesting to note that four of the project's six post-doctoral researchers are now employed by a European gas turbine industry, while the other two have found academic jobs, one of which is in Europe.

EXPLOITATION

1. Through continuing interactions with the ITD Member who received the outputs of the project, further development of the codes and penetration of the results in the industry's R&D activities is possible and planned through continuing presentations and collaborations.
2. The academic partners now have some new theoretical, experimental, and modelling capabilities that can be used for further research projects. Two Marie-Curie proposals based on the annular rig and the blow-off work at Cambridge have already been submitted in January 2017.
3. The new CFD capabilities at Cambridge can be applied to other combustion problems, and hence licensing discussions with European nuclear industry in terms of fire extinction and with a European diesel engine manufacturer in terms of pollutant prediction have begun.