Jens Klingmann

Large eddy simulation and experimental studies of a confined turbulent swirling flow

Laser Doppler velocimetry (LDV) measurement and large eddy simulation (LES) were used to study confined isothermal turbulent swirling flows in a model dump combustor. The aim was to gain deeper understanding of the flow and turbulence structures in dump combustors and to examine the capability of LES for prediction of turbulent swirling flows. A refractive index matching technique is used in the LDV measurement to improve the near-wall data. A high-order finite difference scheme on Cartesian grids with a scale-similarity subfilter scale model is used in the LES. Turbulent inflow boundary conditions with different energy spectra, different outflow boundary conditions, and grid resolutions are tested in the LES. Three test cases with different swirl numbers and Reynolds numbers are studied in the measurements and the simulations. The Reynolds numbers range from 10 000 to 20 000, and the swirl number is varied from 0 to 0.43. With appropriate inflow, outflow boundary conditions, and fine grid resolution, the LES results are in fairly good agreement with the LDV data. The experimental and numerical results show that turbulence in the dump combustor is highly anisotropic behind the sudden expansion and in the internal recirculation zone near the axis of the combustor. Turbulence decays rapidly along the streamwise direction downstream, and the structure of turbulence depends highly on the level of inlet swirl. At low swirl numbers, turbulence is primarily generated in the shear layer behind the sudden expansion; at high swirl numbers the near axis flow becomes very unstable and vortex breakdown occurs. The shear layer near the axis of the combustor caused by vortex breakdown generates most of the turbulent kinetic energy. Large-scale motions (coherent structures) are found in the near axis vortex breakdown region. A helical flow in the guiding pipe breaks down near the sudden expansion to form a large bubble-like recirculation zone whose center moves slowly around the axis. Downstream of the bubble the core of the rotational large scale azimuthal flow motion is off the combustor axis and rotates around the axis at a frequency about 18-25 Hz (Strouhal number about 0.17-0.4). As the swirl number increases the coherent structure becomes more evident, and the internal recirculation zone moves upstream. LES successfully simulated the vortex breakdown, the internal recirculation zones and the anisotropic turbulence structures for all the swirl numbers considered

Reacting Boundary Layers in a Homogeneous Charge Compression Ignition (HCCI) Engine

An experimental and computational study of the nearwall combustion in a Homogeneous Charge Compression Ignition (HCCI) engine has been conducted by applying laser based diagnostic techniques in combination with numerical modeling. Our major intent was to characterize the combustion in the velocity- and thermal boundary layers. The progress of the combustion was studied by using fuel tracer LIF, the result of which was compared with LDA measurements of the velocity boundary layer along with numerical simulations of the reacting boundary layer. Time resolved images of the PLIF signal were taken and ensemble averaged images were calculated. In the fuel tracer LIF experiments, acetone was seeded into the fuel as a tracer. It is clear from the experiments that a proper set of backgrounds and laser profiles are necessary to resolve the near-wall concentration profiles, even at a qualitative level. Partial resolution of the velocity boundary layer was enabled by using a slightly inclined LDA probe operated in back-scatter mode. During these conditions, it was possible to acquire velocity data within 0.2 mm from the wall. A one-dimensional model of the flow field was devised to make the connection between the thermal and the velocity boundary layer. The investigations suggest that wall interaction is not the responsible mechanism for the rather high emissions of unburned hydrocarbons from HCCI engines. It is believed that the delayed oxidation, indicated by the fuel tracer LIF experiments and numerical simulations, is due to the thermal boundary layer. From the data at hand, it is concluded that the thermal boundary layer is on the order of 1 mm thick. In this boundary layer the reactions are delayed but not quenched. (Less)

Investigation of two-equation turbulence models applied to a confined axis-symmetric swirling flow

The modeling of industrial combustion applications today is almost exclusively based on two-equation turbulence models. Despite its known limitations, the most the widely used model is still the standard k-e model. The objective of this paper is to investigate the performance of two-equation turbulence models applied to a confined swirling flow. Numerical modeling of an axis-symmetric confined sudden expansion, followed by a contraction with the assumption of steady flow and an incompressible fluid, has been conducted. The flow field is what can be expected in simplified dump gas turbine combustor geometry. In this investigation, three different swirl cases were considered: no swirl, moderate swirl (no central re-circulation zone) and strong swift (a central recirculation zone occurring). The models investigated were: the standard k-e model, a curvature-modified k-e model, Chens k-e model, a cubic non-linear k-e model, the standard k-ω model and the Shear Stress Transport (SST) k-ω model. The results show that almost all models were able to predict the major impact of the moderate swirl: reduced outer re-circulation lengths and retardation of the axial velocity on the center-line. However, the Chen k-e model and the SST k-ω model were found to better reproduce the mean velocity field and the turbulent kinetic energy field from the measurements. For a strong swift, a large re-circulation zone is formed along the center-line, which the standard k-e model and the modified k-e model fail to predict. However, the shape and size of the re-circulation zone differ strongly between the models. At this swirl number, the performances of all models were, without exception, worse than for the lower swift numbers. The SST k-ω model achieved the best agreement between computations and experimental data. (Less)

Visualization of Different Flashback Mechanisms for H2/CH4 Mixtures in a Variable-Swirl Burner

Flame flashback from the combustion chamber to the premixing section is a major operability issue when using high H-2 content fuels in lean premixed combustors. Depending on the flow-field in the combustor, flashback can be triggered by different mechanisms. In this work, three flashback mechanisms of H-2/CH4 mixtures were visualized in an atmospheric variable-swirl burner using high speed OH* chemiluminescence imaging. The H-2 mole fraction of the tested fuel mixtures varied between 0.1 and 0.9. The flow-field in the combustor was varied by changing the swirl number from 0.0 to 0.66 and the total air mass-flow rate from 75 to 200 SLPM (standard liters per minute). The following three types of flashback mechanism were observed: Flashback caused by combustion induced vortex breakdown (CIVB) occurred at swirl numbers >= 0.53 for all of the tested fuel mixtures. Flashback in the boundary layer (BL) and flame propagation in the premixing tube caused by auto-ignition were observed at low swirl numbers and low total air mass-flow rates. The temporal and spatial propagation of the flame in the optical section of the premixing tube during flashback was studied and flashback speed for different mechanisms was estimated. The flame propagation speed during flashback was significantly different for the different mechanisms. (Less)

Computational and Experimental Investigation of Emissions in a Highly Humidified Premixed Flame

Emission formation and flame stability were investigated, both experimentally and computationally, for premixed combustion with varying amounts of water vapor in the mixture. Emission measurements were made in a gas turbine combustor at atmospheric conditions, using Danish Natural Gas (NG) as fuel. The emissions were mapped as a function of humidity, inlet air temperature, equivalence ratio and aerodynamic load. Operating conditions were chosen to match what can be expected from e.g. an EvGT cycle for power generation. The inlet air temperature was slightly lower than the inlet temperatures that would be found in a recuperated cycle. The degree of humidity was varied from 0w% to 33w% of the airflow in the experiment, while the air inlet temperature was varied from 500K to 800K. Computations were made using a single Perfectly Stirred Reactor (PSR) model and a reaction scheme with 821 reactions and 69 species. It was found that the NOX emissions were strongly reduced by the addition of water. Most of this decrease vanishes in practical combustion since richer combustion is required to keep CO emissions (combustion efficiency) at a tolerable level. The maximum humidity was found to be dependent on inlet air temperature and aerodynamic load. In this experiment, the maximum humidity achieved was 33%. (Less)

A comparison between the combustion of natural gas and partially reformed natural gas in an atmospheric lean premixed turbine-type combustor

A small-scale combustor was set up to analyze the combustion of natural gas and two mixtures of partially reformed natural gas. The partially reformed mixtures can be formed using biomass to feed the endothermic reforming reactions. Before combusting these mixtures in a gas turbine, experimental work was done on a primary zone combustion chamber to examine the combustor behavior when switching from natural gas to the wet and dry hydrogen-rich mixtures. Temperature profiles, flame location and ignition limits have been investigated for a variety of stoichiometries and several air temperatures. Possible problems concerning blow-off, flashback, increased pollutant products and excessive liner wall temperatures were analyzed. It was concluded that the switch in operation from natural gas to these wet and/or dry partially reformed natural gas mixtures lowers the blow-off limits while maintaining similar liner wall temperature profiles. Furthermore, no significant changes in pollutant production were observed. Flame area, shape and position display considerable differences in combustion regime for the three tested fuel types.

Investigation of Turbulence Models Applied to Premixed Combustion Using a Level-Set Flamelet Library Approach

Most of the common modeling approaches to premixed combustion in engineering applications are either based on the assumption of infinitely fast chemistry or the flamelet assumption with simple chemistry. The level-set flamelet library approach (FLA) has shown great potential in predicting major species and heat release, as well as intermediate and minor species, where more simple models often fail. In this approach, the mean flame surface is tracked by a level-set equation. The flamelet libraries are generated by all external code, which employs a detailed chemical mechanism. However a model for the turbulent flame speed is required, which, among other considerations, depends on the turbulence intensity, i.e., these models may show sensitivity to turbulence modeling. In this paper, the FLA model was implemented in the commercial CFD program Star-Cd, and applied to a lean premixed flame stabilized by a triangular prism (bluff body). The objective of this paper has been to investigate the impact on the mean flame position, and hence on the temperature and species distribution, using three different turbulent flame speed models in combination with four different turbulence models. The turbulence models investigated are: the standard k-epsilon model, a cubic nonlinear k-e model, the standard k-omega model and the shear stress transport (SST) k-omega model. In general, the computed results agree well with experimental data for all computed cases, although the turbulence intensity is strongly underestimated at the downstream position. The use of the nonlinear k-epsilon model offers no advantage over the standard model, regardless of flame speed model. The k-omega based turbulence models predict the highest turbulence intensity with the shortest flame lengths as a consequence. The Muller flame speed model shows the least sensitivity to the choice of turbulence model. (Less)

Experimental Investigation of the Effect of Steam Dilution on the Combustion of Methane for Humidified Micro Gas Turbine Applications

ABSTRACT Water introduction in the micro gas turbine (mGT) cycle is considered the optimal route for waste heat recovery and flexibility increase of such a small-scale combined heat and power (CHP) unit. However, humidification of the combustion air in a mGT affects combustion stability, efficiency, and exhaust gas emissions. This can lead to a non-stable, incomplete combustion, which will affect the global efficiency negatively. Additionally, CO emissions will increase. The non-stable, incomplete combustion might result in an engine shutdown due to a flameout. To study the impact of humidification on the combustion of methane in a humidified mGT, we performed combustion experiments in an atmospheric, variable-swirl, premixed combustion chamber. The results of these experiments are summarized in this article. The effect of the humidification of the combustion air was simulated by adding steam to the combustion air. The impact of the steam injection on methane combustion has been studied at variable swirl number and steam fraction. Experimental results showed a linearly increasing lean blowout (LBO) equivalence ratio for methane combustion with increasing steam fraction. In addition, CO emission levels started to rise at higher equivalence ratio for higher steam fractions compared to combustion under dry conditions. The CO emission levels at stable combustion were however still the same order of magnitude as for the dry combustion. The swirl number has little effect on the LBO limit. Final results indicated the possibility to maintain complete and stable combustion under humidified conditions with low CO emissions at higher equivalence ratio compared to the dry combustion.

Parametric Study of Emissions from Low Calorific Value Syngas Combustion, with Variation of Fuel Distribution, in a Prototype Three Sector Burner

The emission composition is measured for a prototype burner while varying the equivalence ratio in discrete portions of the burner. The burner is a three sector system, consisting of a separate igniter, pilot/stabilizer and main burner. The design allows for discrete control of equivalence ratio in each of the three sectors. The ignition sector, designated RPL (Rich- Pilot-Lean), operates from rich to lean equivalence values, and serves to ignite the pilot sector, which, in turn, stabilizes the main combustion sector. All three burner sections are premixed. The burner is operated at atmospheric pressure with inlet flows heated to 650 K (±8 K). Tests were performed for three gases: methane, a model syngas (10% CH4, 22.5% CO, 67.5% H2), and dilute syngas. The dilute gas includes sufficient nitrogen to lower the heating value to 15 MJ/m3. The model syngas and diluted syngas are representative of fuels produced by gasification process. The burner emissions, specifically, CO, CO2, O2 and NOx, are measured while holding the RPL equivalence value constant and varying the equivalence ratio of the pilot and main sectors. The equivalence ratios for pilot and main sectors are chosen such that the total burner equivalence ratios remain constant during a test sequence. The target total equivalence ratio for each gas is chosen such that all experiments should have the same flame temperature. (Less)

Chemical Analysis of Combustion Products From a High-Pressure Gas Turbine Combustor Rig Fueled by Jet A1 Fuel and a Fischer-Tropsch-Based Fuel

A comparative experimental investigation has been performed, comparing the emissions from a synthetic jet fuel and from Jet AI. In the investigation, the unburned hydrocarbons were analyzed chemically and the regulated emissions of NOX, CO and HC were measured. All combustion tests were performed under elevated pressures in a gas turbine combustor rig. A Swedish company, Oroboros AB, has developed a novel clean synthetic jet fuel, LeanJet [registered trademark] . The fuel is produced synthetically from synthesis gas by a Fischer-Tropsch process. Except for the density, the fuel conforms to the Standard Specification for Aviation Turbine Fuels. The low density is due to the lack of aromatics and polyaromatics. Organic emissions from the gas turbine combustor rig were collected by adsorption sampling and analyzed chemically. Both the fuels and the organic emissions were analyzed by gas chromatography/flame ionization (GC/FlD) complemented with gas chromatography/mass spectrometry (GC/MS). Under the operating conditions investigated, no significant differences were found for the regulated emissions, except for emission of CO from the synthetic fuel, which, at leaner conditions, was one-quarter of that measured for Jet Al. Detailed analysis of the organic compounds showed that the emissions from both fuels were dominated by fuel alkanes and a significant amount of naphthalene. It was also found that Jet Al produced a much higher amount of benzene than the synthetic fuel. Copyright (Less)