Experimental and Numerical Studies of Diesel Spray Evaporation and Combustion in a Gas Turbine Matrix Burner

Lean gas turbine combustion instability and control is currently a subject of interest for many researchers. The motivation for running gas turbines lean is to reduce NOx emissions. For this reason gas turbine combustors are being design using the Lean Premixed Prevaporized (LPP) concept. In this concept, the liquid fuel must first be atomized, vaporized and thoroughly premixed with the oxidizer before it enters the combustion chamber. One problem that is associated with running gas turbines lean and premixed is that they are prone to combustion instability. The matrix burner test rig at the Institute of Steam and Gas Turbines at the RWTH Aachen University is no exception. This matrix burner is suitable for simulating the conditions prevailing in stationary gas turbines. Till now this burner could handle only gaseous fuel injection. It is important for gas turbines in operation to be able to handle both gaseous and liquid fuels though. This paper reports the modification of this test rig in order for it to be able to handle both gaseous and liquid primary fuels. Many design issues like the number and position of injectors, the spray angle, nozzle type, droplet size distribution, etc. were considered. Starting with the determination of the spray cone angle from measurements, CFD was used in the initial design to determine the optimum position and number of injectors from cold flow simulations. This was followed by hot flow simulations to determine the dynamic behavior of the flame first without any forcing at the air inlet and with forcing at the air inlet. The effect of the forcing on the atomization is determined and discussed.

Numerical Simulation of Broadband Combustion Noise With the RPM-CN Approach

A new numerical approach called RPM-CN approach is applied to predict broadband combustion noise. This highly efficient hybrid CFD/CAA approach can rely on a reactive RANS simulation. The RPM method is used to reconstruct stochastic broadband combustion noise sources in the time domain based on statistical turbulence quantities. Subsequently, the propagation of the combustion noise is computed by solving the acoustic perturbation equations (APE-4). The accuracy of the RPM-CN approach will be demonstrated by a good agreement of the simulation results with acoustic measurements of the DLR-A flame. The high efficiency and therefore low computational costs enable the usage of this numerical approach in the design process.

Simulation of Methanol-Air Two-Phase Flames Using Various Turbulent Combustion Models

Though fossil fuel is running out, liquid fuels nowadays still provide the most energy used by industrial furnaces, automotive and aero engines. How to predict a two-phase turbulent combustion flame is still a big problem to designers. Generally, the liquid fuel is sprayed and mixed with oxygen, and the flame characteristics depends on the fuel atomization, the fuel droplet spatial distribution, and its interaction with the turbulent oxidizer flow field: turbulent heat, mass and momentum transfer, complicated chemical kinetics, and turbulent-chemistry interaction. Turbulent combustion model is a key point for the two phase combustion simulation. For its short time consuming, Reynolds Averaged Navier Stokes (RANS) method nowadays still is the major tool for gas turbine chamber (GTC) designers, but there is not a universal method in RANS GTC spray combustion simulation at present especially for the two-phase turbulent combustion. The Eddy-Break-Up turbulent combustion model (EBU), Eddy Dissipation Concept turbulent combustion model (EDC), steady Laminar Flame-let turbulent combustion Model (LFM) and the Composition PDF transport turbulent combustion model (CPDF) are all widely used models. In this paper, these four turbulent models are used to simulate a methane-air turbulent jet flame measured by Sandia Lab first, then three methanol-air two-phase turbulent flames, in order to know the ability of these turbulent models. In the gas turbulent jet flame simulation, the result of LFM model and CPDF model are in better agreement with the experimental data than those of the EBU and the EDC models’ results. The reason is that the EBU model and EDC model are overestimated the effect of turbulent. In the three different cases of the two phase combustion simulation, CPDF is the best. The prediction ability of the other three models is different in different cases. The EDC predictions are closer to the experimental data when the air flow rate value is lower, whereas the LFM predictions are better when the air flow rate value is higher.

Simulations of Unsteady Oscillations in Turbulent Non-Premixed Swirling Flames

Simulations of turbulent non-premixed swirling flames based on the Sydney swirl burner experiments under different flame characteristics are conducted using large eddy simulations (LES). The simulations attempt to capture the unsteady flame oscillations and explore the underlying instability modes responsible for a centre jet precession and the large scale recirculation zone oscillation. The selected flame series known as SMH flames have a fuel mixture of methane-hydrogen (50:50 by volume). The LES program solved the governing equations on a structured Cartesian grid using finite volume method and the subgrid turbulence and combustion models used the localized dynamic form of Smagorinsky eddy viscosity model and the steady laminar flamelet model respectively. The results show that the LES predicts two types of instability modes near fuel jet region and the bluff body stabilized recirculation zone region. The Mode I instability defined as cyclic precession of a centre jet is identified using time periodicity of the centre jet in flames SMH1 and SMH2. The Mode II instability defined as cyclic expansion and collapse of the recirculation zone is identified using time periodicity of the recirculation zone in flame SMH3. The calculated frequency spectrums found reasonably good agreement with the experimental precession frequencies. Overall, the LES yield a good qualitative and quantitative agreement with the experimental observations, although some discrepancies are apparent.

Effects of Pilot Fuel and Liner Cooling on the Flame Structure in a Full Scale Swirl-Stabilized Combustion Setup

The flame structure and the limits of operation of a lean premixed swirl flame were experimentally investigated under piloted and non-piloted conditions. Flame stabilization and blow out limits are discussed with respect to pilot fuel injection and combustor liner cooling for lean operating conditions. Two distinctly different flow patterns are found to develop depending on piloting and liner cooling parameters. These flow patterns are characterized with respect to flame stability, blow out limits, combustion noise and emissions. The combustion system explored consists of a single burner similar to the burners used in Siemens annular combustion systems. The burner feeds a distinctively non-adiabatic combustion chamber operated with natural gas under atmospheric pressure. Liner cooling is mimicked by purely convective cooling and an additional flow of ‘leakage air’ injected into the combustion chamber. Both, this additional air flow and the pilot fuel ratio were found to have a strong influence on the flow structure and stability of the flame close to the lean blow off limit (LBO). It is shown by Laser Doppler Velocimetry (LDV) that the angle of the swirl cone is strongly affected by pilot fuel injection. Two distinct types of flow patterns are observed close to LBO in this large scale setup: While non-piloted flames exhibit tight cone angles and small inner recirculation zones (IRZ), sufficient piloting results in a wide cone angle and a large IRZ. Only in the latter case, the main flow becomes attached to the combustor liner. Flame structures deduced from flow fields and CH-Chemiluminescence images depend on both the pilot fuel injection and liner cooling.

Emissions Characteristics of a Legacy Military Aircraft

Emissions from aircraft and associated ground equipment are major sources of local pollution at airports and military bases. These pollutant emissions, especially particulate matter (PM), have been receiving significant attention lately due to their proven harmful health and environmental effects. As the U.S. Environmental Protection Agency (EPA) tightens environmental standards, it is likely that military operations, including the basing of advanced and legacy aircraft, will be impacted. Accurate determination of emission indices from aircraft is necessary to properly assess their environmental burden. As such, the gaseous and PM emissions of a B-52 Stratofortress aircraft were characterized in this effort. This emissions study supports the Strategic Environmental Research and Development Program (SERDP) project WP-1401 to determine emissions factors from military aircraft. The main purpose of the project is to develop a comprehensive emissions measurement program using both conventional and advanced techniques to determine emissions factors for pollutants of fixed and rotating wing military aircraft. Standard practices for the measurement of gaseous emissions from aircraft have been well established; however, there is no certified methodology for the measurement of aircraft PM emissions. In this study, several conventional aerosol instruments were employed to physically characterize the PM emissions from two of the aircraft’s TF33 turbofan engines. Exit plane pollutant emissions were extracted via probes and transported through heated lines to the analytical instruments. Particle concentrations, size distributions and mass emissions, as well as engine smoke numbers (SN), soot volatile fraction and total hydrocarbon emissions were measured. The engines were tested at four power settings, from idle to 75% normal rated thrust (NRT) (95% N2 – turbine speed). Test results show relatively consistent PM and gaseous emissions between the two engines for most conditions tested. The measured TF33 PM mass emission indices (EI), including estimated sampling line losses, were in the range of 1.0–3.0 g/kg-fuel and the particle number (PN) EI were between 4.0–10.0E+15 particles/kg-fuel. The particle size data followed a single mode lognormal distribution for all power settings with particle geometric mean diameters ranging from 52 to 85 nm. In general, the aerosol instrumentation provided consistent and reliable measurements throughout the test campaign, therefore increasing confidence on their use for turbine engine PM emissions measurements.

Effects of Combustion Chamber Geometry Upon Exit Temperature Profiles

The purpose of this research was to investigate the effects of combustion chamber geometry on exit temperature fields using an ambient pressure test rig. The apparatus contained a 120° sector of a combustion section of a Rolls Royce (previously Allison) T56-A-15 gas turbine engine. A thermocouple rake acquired high-resolution temperature measurements in the combustion chamber exit plane. Rig test conditions were set to simulate an engine operating condition of 463 km/h (250 knots) at 7620 m (25000ft) by matching the Mach number, the equivalence ratio and the Sauter mean diameter of the fuel spray. To quantify the geometric deviations of the combustion chamber specimens, which varied in service conditions, a three-dimensional laser scanning system was used. Combustion chamber geometric deviations were extracted through comparison of the scanned data to a reference model using the selected software. The relationship between combustion chamber exit temperature profile and geometric deviation was then compared. The main conclusion of this research was that small deviations from nominal dimensions in the dilution zone of the combustion chamber correlated to an increase in pattern factor. A decrease in the mixing of the products of combustion and dilution air was observed as damage in the dilution zone increased. This reduction in mixing created a more compact, higher temperature core flow. The results obtained from this research were compared to past studies.

Assessment of the Operational Performance of Fischer-Tropsch Synthetic-Paraffinic Kerosene in a T63 Gas Turbine Compared to Conventional Jet A-1 Fuel

The performance implications of operating on Synthetic-Paraffinic Kerosene (SPK) were investigated using a RR-Allison T63-A-700 Model 250-C18 B gas turbine and compared to conventional Jet A-1. The SPK was aromatic–free and possessed a greater hydrogen/carbon ratio than petroleum derived Jet A-1. The variation in aromatic content had several implications with respect to soot and NOx emissions. Reduced aromatics also implied a reduction in the radiative heat transfer to the combustor liner. A simple model was used to explore the effect of H/C ratio on the adiabatic flame temperature, the combustor exit temperature and the engine efficiency via the impact on the gas properties and these were compared to the experimental data. It was found that operation with SPK changed directionally toward improving energy extraction via a turbine and an overall efficiency gain of about 1.2% was attained with operation on SPK through increased combustion efficiency, a reduction in liner pressure loss and an improvement in the combustion products properties. A modified combustion liner was fitted to enable the thermal loading on the combustor liner to be investigated and the expected trend with the SPK fuel was confirmed and quantified.

Flow Field Derived Equivalent Reactor Networks for Accurate Chemistry Simulation in Gas Turbine Combustors

A method has been developed in which the flow field predicted by Computational Fluid Dynamics (CFD) is automatically condensed into an Equivalent Reactor Network (ERN), composed of well stirred reactors, allowing rapid and accurate analysis of emissions. This paper presents the effectiveness of utilizing an ERN that is a direct abstraction of the computational flow field for combustion analysis. The CFD results are divided into reactors using various filters on flow-field variables to construct an ERN that represents the 3-D combustor flow field and flame structure. Detailed kinetics can then be used in ERN simulations to analyze effects of fuel composition and operating condition on emissions. The technique is applied to a commercial industrial gas turbine combustor fuel injector and compared against experimental emissions results. Sensitivity of emissions predictions to different parameters in the network extraction is also presented. Parameter variations in fuel flow rate are applied to the ERN to obtain relative impacts of fuel-air ratio on the emissions of NOx without requiring new CFD solutions. This automatic approach has been found to reduce the time required to construct and analyze flow field derived ERNs with detailed chemistry by 90%. A local calculation of Damko¨hler number, important for stability analysis, is also presented. This calculation also uses abstracted information from the CFD flow field and detailed-kinetics simulations for more accurate, cost-effective analysis.

Numerical Characterisation of a Gas Turbine Model Combustor Applying Scale-Adaptive Simulation

In this contribution the three-dimensional reacting turbulent flow field of a swirl-stabilized gas turbine model combustor is analyzed numerically. The investigated partially premixed and lifted CH4/air flame has a thermal power load of Pth = 35kW and a global equivalence ratio of φ = 0.65. To study the reacting flow field the Scale Adaptive Simulation (SAS) turbulence model in combination with the Eddy Dissipation/Finite Rate Chemistry combustion model was applied. The simulations were performed using the commercial CFD software package ANSYS CFX-11.0. The numerically achieved time-averaged values of the velocity components and their appropriate turbulent fluctuations (RMS) are in very good agreement with the experimental values (LDA). The same excellent results were found for other flow quantities like temperature and mixture fraction. Here, the corresponding time-averaged and the appropriate RMS profiles are compared to Raman measurements. Furthermore the instantaneous flow features are discussed. In accordance with the experiment the numerical simulation evidences the existence of a precessing vortex core (PVC). The PVC rotates with a frequency of 1596Hz. Moreover it is shown that in the upper part of the combustion chamber a tornado-like vortical structure is established.