Search

The effects of inert diluents in the fuel mixture of a model swirl stabilized gas turbine combustor, under thermoacoustically unstable limit cycle operation, were studied experimentally. The measurements included particle image velocimetry (PIV), high-speed CH* chemiluminescent imaging, and dynamic pressure signals. The paper focuses on the dynamic phenomenon of the period doubling bifurcation, which came about when the equivalence ratio ϕ of undiluted flames was enriched from 0.55 to 0.60 under constant Reynolds number (Re=18,000). The bifurcation featured an emergence of an aerodynamically related timescale in addition to the fundamental timescale that was induced from an unstable acoustic eigenmode. The aerodynamic timescale is introduced by azimuthal convection of a high-heat-release-rate region and is linked in the literature with a precessing motion of the recirculation zone. It was found that on increasing the diluent molar fraction up to 50%, the amplitude and the frequency of the limit cycle fundamental acoustic mode decreased. A mechanism to interpret this suppression is that increasing the diluent molar fraction of the fuel makes the flame more susceptible to quenching because the extinction strain rate of the mixture is decreased. The paper argues that the existence of inert diluents in the fuel can significantly alter the dynamic state of the combustor, because the anchoring locations of the flame greatly depend on the composition-sensitive extinction strain rate of the mixture.

The effects of inert diluents in the fuel mixture of a model swirl stabilized gas turbine combustor, under thermoacoustically unstable limit cycle operation, were studied experimentally. The measurements included Particle Image Velocimetry (PIV), high speed CH* chemiluminescent imaging and dynamic pressure signals. The paper focuses on the dynamic phenomenon of the period doubling bifurcation, which came about when the equivalence ratio (Ø) of undiluted flames was enriched from 0.55 to 0.60 under constant Reynolds number (Re). The bifurcation featured an emergence of an aerodynamically related timescale in addition to the fundamental timescale which was induced from an unstable acoustic eigenmode. The aerodynamic timescale is introduced by azimuthal convection of a high heat release rate region, and is linked in the literature with a precessing motion of the recirculation zone. Prior to the bifurcation, the flame anchored between the wall and the outer shear layers of the recirculation zone assuming a V-shape. The dynamics were attracted to a Period-1 limit cycle. Post the bifurcation the flame expanded in the inner shear layers of the recirculation zone assuming an M-shape, while the dynamics were attracted to a Period-2 limit cycle. The latter operational condition was subject to nitrogen dilution in order to parametrically increase the molar fraction of inert diluent in the fuel stream. It was found that on increasing the diluent molar fraction the amplitude and the frequency of the limit cycle fundamental acoustic mode decreased. Also, inert dilution suppressed the aerodynamic timescale and the flame assumed a V-shape again. A mechanism to interpret this mechanism is suggested. Increasing the diluent molar fraction of the fuel makes he flame susceptible to quenching because the extinction strain rate of the mixture decreased. Thus, the flow imposed strain rates quenched the flame when it attempted to anchor on the inner shear layers of the vortex breakdown induced recirculation zone. The inability of the flame to anchor on the shear layers suppressed the aerodynamic mode. The paper argues that the existence of inert diluents in the fuel can significantly alter the dynamic state of the combustor, since the anchoring locations of the flame greatly depend on the composition-sensitive extinction strain rate of the mixture.

In the current paper, time resolved high speed optical Particle Image Velocimetry and CH* chemiluminescence measurements were performed, to study self-excited limit cycle combustion instabilities in a swirl stabilized model gas turbine combustor operating at atmospheric pressure with choked propane and air flow supplies. The combustor was operated under a constant Reynolds number (Re=22,000) and four equivalence ratios, namely ϕ = 0.50 for operation susceptible to extinction, and ϕ=0.55, ϕ=0.60 and ϕ=0.65 for operation under a thermoacoustically unstable combustion regime, to encounter two limit cycle dynamic states. The period-1 limit cycle was driven by thermoacoustic coupling between the acoustic and the thermal field at a fundamental timescale dictated by an acoustic eigenmode of the combustor. The period-2 limit cycle, further to the fundamental acoustic timescale featured a subharmonic aerodynamic signature in the heat release rate and dynamic pressure spectra caused by the helical coherent structure of a Precessing Vortex Core (PVC). Previous studies have shown that the PVC in the limit cycle regime may be suppressed by the temperature stratification at the inlet of the combustor. A mechanism is suggested to interpret the flame anchoring locations which effectively regulated whether PVC was excited or suppressed. It is showed that the conditions under which the flame attached to the centerbody and suppressed the PVC can be explained by the spatial distribution of the relative ratio of the flow imposed to the mixture extinction strain rate. The PVC was excited due to local extinction by aerodynamic straining at the inlet of the combustor, at the phase angle of maximum dynamic pressure. On increasing the equivalence ratio, the flame became robust to aerodynamic straining and flashed back at the phase angle of maximum dynamic pressure. The PVC was then suppressed due to the relative ratio of the flow imposed to the extinction strain rate, which allowed the establishment of swirl-damping temperature gradients at the combustor inlet. The paper underlines the importance of quantifying the relation between the flow imposed and extinction strain rate, as it largely dictates the eventual combustor limit cycle dynamic state and its resonant frequencies.

The current paper studies the thermoacoustically unstable combustion, under elevated mean pressure, of a commercial swirl stabilized gas turbine burner fitted with optically accessible windows. The considered measurements include particle image velocimetry (PIV), OH∗ chemiluminescence imaging, high speed broadband flame imaging and dynamic pressure signals. We study cases A and B, wherein natural gas flames at mean pressures equal to 3 bar and 6 bar delivered thermal loads equal to 335 kW and 685 kW respectively. The flow field demonstrated a typical vortex breakdown induced inner recirculation zone and a sudden step expansion induced outer recirculation zone. In case A, high amplitude dynamic pressure bursts were observed amidst a quiescent acoustic background. The flame was conical, it anchored on the shear layers of the recirculation zone and it periodically expanded in the outer recirculation zone (ORZ). In case B, the flame was consistently thermoacoustically unstable with seldom requiescent events, while at the same time expansion to the ORZ was suppressed. By applying Dynamic Mode Decomposition on high speed images of case A, it was showed that this expansion introduced an additional time scale, further to the fundamental acoustically related timescale. The superposition of two timescales over a turbulent background established an intermittent regime of thermoacoustic instabilities, wherein the dynamics transitioned between quiescent and fully oscillatory. A physical mechanism is suggested to explain the differences between the flame shapes on adjusting mean pressure. The mechanism considers that the premixture is characterized by a Lewis number lower than unity, the laminar flame speed increases on decreasing mean pressure and the flow imposed on the flame strain rate oscillated over a period of thermoacoustic instability. This combination resulted in oscillatory heat release rate, in the region of the outer shear layers. The phenomenon was more pronounced in case A than in case B, because flow dilatation imposed-strain rates are higher for the former than for the latter flame. The paper argues that for a given fuel, elevated mean pressure introduces time scales that can significantly affect the dynamic regime the combustor operates in.

A novel velocimetry method is proposed for point velocity measurement, which is based on tracking a laser-induced plasma in a flow. The plasma’s behaviour is first analysed spatially, temporally and spectrally in quiescent air. The dependence of this technique on the delay time between subsequent plasma images and the processing methods are described. It is found that, for optimized operation of the technique in a turbulent air jet (exit diameter 10.0 mm from a 480 mm long pipe; with an averaged velocity of 50 m/s at the jet exit resulting in Reynolds number of 34,000) with 100 µs time delay between plasma images, the systematic and random components of the velocity uncertainty are − 0.51 m/s and ± 3.6 m/s along the laser beam direction, and 1.25 m/s and ± 0.86 m/s along other directions perpendicular to the laser beam. These uncertainties are mainly caused by the asymmetric laser energy deposition during the formation of plasma, and the associated spatial resolution (in this realisation of the instrument) of 5 mm. The mean velocity measurements in the turbulent air jet flow are consistent with the reported flow behaviour in the literature for mean velocity: the turbulent intensity of axial velocity fluctuations is comparable to those in the literature but difference arises due to the limited spatial resolution. This velocimetry method is an alternative to traditional tracer-based velocimetry methods, because it does not require ‘seeding’ of particles or other substances in the flow. It also has the ability to measure local gas mixture composition, using laser-induced breakdown spectroscopy approach, simultaneously with flow velocity, but this aspect is not explored in the current study. Concept of Laser-Induced Plasma Image Velocimetry (LIPIV) applied to a turbulent jet flow. The LIPIV technique measures the flow velocity vector by using the temporal displacement of the laser-induced plasma in a flowing fluid. For this reason, two sequential images of the plasma with time delay Δt are used.

The present work investigates the effect of the injection dwell time and injection pressure on soot reduction potential in an optical single-cylinder light-duty diesel engine. The engine operated under a double-injection scheme under low load and low engine speed conditions. The conducted experiments considered two different dwell times for three different injection pressures. The fuel quantity of the main injection was adjusted to maintain the same indicated mean effective pressure (IMEP) value among all cases considered. Findings were analyzed via means of pressure trace and apparent heat release rate (AHRR) analyses, as well as a series of optical diagnostics techniques, namely high-speed imaging and planar laser-induced incandescence (pLII). The combination of dwell time and injection pressure substantially affects charge reactivity and soot oxidation potential. The analysis suggests that a shorter dwell time combined with a higher injection pressure can lead to an enhanced potential for engine-out particulate reduction by creating an in-cylinder environment that promotes soot oxidation. Overall, results indicate that a close-coupled pilot and main injection scheme can reduce soot levels, albeit while increasing specific fuel consumption by up to 12% to maintain the same engine power output levels.

The interaction between droplet dispersion and evaporation in an acetone spray evaporating under ambient conditions is experimentally studied with an aim to understand the physics behind the spatial correlation between the local vapour mass fraction and droplets. The influence of gas-phase turbulence and droplet–gas slip velocity of such correlations is examined, while the focus is on the consequence of droplet clustering on collective evaporation of droplet clouds. Simultaneous and planar measurements of droplet size, velocity and number density, and vapour mass fraction around the droplets, were obtained by combining the interferometric laser imaging for droplet sizing and planar laser induced fluorescence techniques (Sahu et al., Exp. Fluids, vol. 55, 1673, 2014b, pp. 1–21). Comparison with droplet measurements in a non-evaporating water spray under the same flow conditions showed that droplet evaporation leads to higher fluctuations of droplet number density and velocity relative to the respective mean values. While the mean droplet–gas slip velocity was found to be negligibly small, the vaporization Damköhler number ($Da_{v}$) was approximately ‘one’, which means the droplet evaporation time and the characteristic time scale of large eddies are of the same order. Thus, the influence of the convective effect on droplet evaporation is not expected to be significant in comparison to the instantaneous fluctuations of slip velocity, which refers to the direct effect of turbulence. An overall linearly increasing trend was observed in the scatter plot of the instantaneous values of droplet number density ($N$) and vapour mass fraction ($Y_{F}$). Accordingly, the correlation coefficient of fluctuations of vapour mass fraction and droplet number density ($R_{n\ast y}$) was relatively high (${\approx}0.5$) implying moderately high correlation. However, considerable spread of the $N$ versus $Y_{F}$ scatter plot along both coordinates demonstrated the influence on droplet evaporation due to turbulent droplet dispersion, which leads to droplet clustering. The presence of droplet clustering was confirmed by the measurement of spatial correlation coefficient of the fluctuations of droplet number density for different size classes ($R_{n\ast n}$) and the radial distribution function (RDF) of the droplets. Also, the tendency of the droplets to form clusters was higher for the acetone spray than the water spray, indicating that droplet evaporation promoted droplet grouping in the spray. The instantaneous group evaporation number ($G$) was evaluated from the measured length scale of droplet clusters (by the RDF) and the average droplet size and spacing in instantaneous clusters. The mean value of $G$ suggests an internal group evaporation mode of the droplet clouds near the spray centre, while single droplet evaporation prevails near the spray boundary. However, the large fluctuations in the magnitude of instantaneous values of $G$ at all measurement locations implied temporal variations in the mode of droplet cloud evaporation.

[EN] This paper studies the effect of liquid viscosity on the atomisation regimes of initially spherical and non-spherical droplets and also kinematic characteristics of non-spherical droplets. The droplets consisted of water-glycerol solutions with viscosities ranging from 6.3 to 697 mPa s, and the initial aspect ratio was 1, The authors would like to acknowledge the financial support of Volvo Car Corporation (VCC), Gothenburg, Sweden and EPSRC grant EP/K019732/1.

This paper discusses the interaction between droplets and entrained turbulent air flow in the far-downstream locations of a confined polydispersed isothermal spray. Simultaneous and planar measurements of droplet and gas velocities in the spray along with droplet size are obtained with the application of a novel experimental technique, developed by Hardalupaset al. (Exp. Fluids, vol. 49, 2010, pp. 417–434), which combines interferometric laser imaging for droplet sizing (ILIDS) with particle image velocimetry (PIV). These measurements quantified the spatial correlation coefficients of droplet–gas velocity fluctuations ($R_{dg}$) and droplet–droplet velocity fluctuations ($R_{dd}$) conditional on droplet size classes, for various separation distances, and for axial and cross-stream velocity components. At the measurement location close to the spray edge, with increasing droplet size,$R_{dg}$was found to increase in axial direction and decrease in cross-stream direction. This suggests that as the gas-phase turbulence becomes more anisotropic away from the spray axis, the gravitational influence on droplet–gas correlated motion tends to increase. The effective length scales of the correlated droplet–gas motion were evaluated and compared with that for gas and droplet motion. The role of different turbulent eddies of the gas flow on the droplet–gas interaction was examined. The flow structures were extracted using proper orthogonal decomposition (POD) of the instantaneous gas velocity data, and their contribution on the spatial droplet–gas velocity correlation was evaluated, which quantified the momentum transfer between the two phases at different length scales of the gas flow. The droplets were observed to augment turbulence for the first three POD modes (larger scales) and attenuate it for the rest of the modes (smaller scales). It has been realized that apart from droplet Stokes number and mass loading, the dynamic range of length scales of the gas flow and the relative turbulent kinetic energy content of the flow structures (POD modes) must be considered in order to conclude if the droplets enhance or reduce the carrier-phase turbulence especially at the lower wavenumbers.

Spontaneous chemiluminescence emission has been suggested as one of the most promising options for flame stoichiometry monitoring, with particularly interesting application for lean premixed combustors. Most published works are focused on global and time-averaged flame stoichiometry; however, the results reported in previous studies it can be concluded that chemiluminescence measurement could be also valid for local or instantaneous equivalence ratio monitoring. This could enable the development of novel diagnostic techniques with interesting applications in many combustion situations, like the study of thermo-acoustic instabilities. This was the main motivation for this study, focused on analyzing the feasibility of chemiluminescence for space- and/or time-resolved stoichiometry measurements. The results revealed unexpected results, not reported in previous studies, with a non-negligible variation of the relation between chemiluminescence and equivalence ratio that seems to be related to changes in local flame properties such as temperature or composition. This finding has important practical consequences since the chemiluminescence vs. equivalence ratio curves should be carefully verified for each application.

Thermoacoustic instabilities under lean operation in gas turbine burners hinder the development of lean premixed combustion mode of operation, thus limiting the potential to decrease NO x emissions. A method to improve stability of lean combustion while maintaining low thermal NO x formation is to add hydrogen in typical gas turbine fuels such as natural gas. The present work examines the thermoacoustic dynamic characteristics of hydrogen-enriched methane blends in a swirl stabilized model gas turbine combustor. We blend CH4 with increasing H2 molar content (from 0% to 40%) at a global equivalence ratio of ϕ = 0.55 on a constant Reynolds number Re = 19000. The reference case of pure methane is susceptible to blow off at the same equivalence ratio. On increasing the hydrogen content at 10% H2, the flammability limits are extended. However, further increase of the H2 content leads to manifestation of random short bursts of dynamic pressure and heat release. Thermoacoustic dynamics are intermittently injected between a quiescent state and a regime of high amplitude oscillations. On further increasing H2 content, the dynamics are attracted towards a limit cycle; a fully established high amplitude regime, where no requiescence is observed. In this regime, heat release rate and dynamic pressure oscillate in phase and at the same frequency. Based on the observed dynamics, we seek a mixture property to characterize the dynamic state, the combustor operates in. We show that the extinction strain rate associated with each mixture can collapse the dynamic transitions from quiescent to intermittent instabilities and finally to fully established thermoacoustic oscillations. In each dynamic state, we examine the mechanism that affiliates coherent structures of the underlying thermodynamic flow field with the flame through the relation of the spatial distribution of the flow imposed strain rate over the extinction strain rate of each mixture. It is shown that the flame anchoring locations for a given mixture are dictated by the flow imposed strain rate and then that increased extinction resistance couples the flame with naturally excited to swirling flows helical instabilities, that may instigate mechanisms responsible for intermittent heat release rate bursts. Finally, the combustor demonstrates a limit cycle behaviour where the flow-imposed strain rate oscillates about the extinction strain rate along the inner shear layers causing local extinction of the flame at the root close to the centerbody.

Non-intrusive chemiluminescence measurements have been used as heat release rate and equivalence ratio indicators for gas turbine combustor active control. In the present study, measurements and modelling of OH*, CH(A)*, C 2 *, and CO 2 * chemiluminescence are used to examine chemiluminescence sensing of heat release rate and equivalence ratio in premixed counterflow methane – air flames with equivalence ratio from 0.6 to 1.3 and strain rate from 80 to 400 s -1 . Two spectrally resolved detecting optical systems were used to detect spatially-averaged (global) and spatially resolved (local) chemiluminescence characteristics in the reaction zone. A recently published reaction mechanism 1 for the chemiluminescence of the OH*, CH*, and C 2 * species is incorporated to GRI-Mech 3.0. The augmented mechanism is further validated against the experimental results of the present study and is used to predict the chemiluminescence characteristics of premixed counterflow methane – air flames. The mechanism includes OH* chemiluminescence formation paths from hydrogen reaction, which have not been evaluated before in premixed counterflow flames. The CHEMKIN based counterflow flame code, OPPDIF is employed to simulate the experiments. The calculated OH* and CH(A)* chemiluminescence agrees well with the experimental results measured by both optical methods. Both the experimental and numerical results demonstrate the ability of OH* and CH(A)* intensities to mark heat release rate in methane – air flames. Overall, CH* may be preferable for heat release rate sensing applications at elevated equivalence ratio and strain rate. For equivalence ratio sensing in methane combustion, the measured and simulated OH*/CH(A)* chemiluminescent intensity ratio is highly dependent on equivalence ratio and nearly independent of strain rate. Thus, this ratio can be used to monitor equivalence ratio. However, a non-monotonic behavior of the OH*/CH* ratio for very lean combustion (ER < 0.7) is observed, in agreement with previous studies. This behavior can be reproduced by the reaction mechanisms. The behavior of OH*/CH(A)* chemiluminescent intensity ratio for flames of methanepropane blends are also calculated with the detailed chemistry model. The addition of propane in methane modifies the behaviour of OH*/CH(A)* chemiluminescent intensity ratio dramatically. However, the numerical results suggest that the OH*/CH(A)* chemiluminescent intensity ratio is an indicator of equivalence ratio in lean methanepropane fuel blended flames.

A swirl-stabilised, lean, partially premixed combustor operating at atmospheric conditions has been used to investigate the local curvature distributions in lifted, stable and thermoacoustically oscillating CH4-air partially premixed flames for bulk cold-flow Reynolds numbers of 15,000 and 23,000. Single-shot OH planar laser-induced fluorescence has been used to capture instantaneous images of these three different flame types. Use of binary thresholding to identify the reactant and product regions in the OH planar laser-induced fluorescence images, in order to extract accurate flame-front locations, is shown to be unsatisfactory for the examined flames. The Canny-Deriche edge detection filter has also been examined and is seen to still leave an unacceptable quantity of artificial flame-fronts. A novel approach has been developed for image analysis where a combination of a non-linear diffusion filter, Sobel gradient and threshold-based curve elimination routines have been used to extract traces of the flame-front to obtain local curvature distributions. A visual comparison of the effectiveness of flame-front identification is made between the novel approach, the threshold binarisation filter and the Canny-Deriche filter. The novel approach appears to most accurately identify the flame-fronts. Example histograms of the curvature for six flame conditions and of the total image area are presented and are found to have a broader range of local flame curvatures for increasing bulk Reynolds numbers. Significantly positive values of mean curvature and marginally positive values of skewness of the histogram have been measured for one lifted flame case, but this is generally accounted for by the effect of flame brush curvature. The mean local flame-front curvature reduces with increasing axial distance from the burner exit plane for all flame types. These changes are more pronounced in the lifted flames but are marginal for the thermoacoustically oscillating flames. It is concluded that additional fuel mixture fraction and velocimetry studies are required to examine whether processes such as the degree of partial-premixedness close to the burner exit plane, the velocity field and the turbulence field have a strong correlation with the curvature characteristics of the investigated flames.

This paper investigates the structure of the continuous liquid jet of a coaxial air-blast atomiser over a range of Weber numbers 60-1040, Reynolds numbers of liquid jet 5400-21700 and air to liquid momentum ratios of the two streams of 1.7–335. A novel optical technique, based on internal illumination of the liquid jet through the jet nozzle by a laser pulse, which excites a fluorescing dye introduced in the atomizing liquid, was used to obtain instantaneous measurements of the breakup length and the three dimensional location of the liquid core of the continuous liquid jet. The latter was achieved by simultaneously imaging the liquid jet from two directions normal to each other. Such measurements are usually prevented by droplets surrounding the liquid jet at the dense spray near the nozzle exit. The measurements showed that the break-up length of the liquid jet scaled well with the air to liquid momentum ratio. The standard deviation of the temporal fluctuations of the break-up length was around 10% of the mean breakup length for each considered flow condition. The instantaneous jet surface does not develop axi-symmetric wave structures but the time-averaged liquid jet is axi-symmetric around the nozzle axis, while the maximum deflection of the liquid jet occurs close to the breaking point.

A novel optical method is proposed for the measurement of the length of the continuous liquid jet core in atomizers. Laser light is directed through the injecting nozzle to illuminate internally the liquid jet, which acts as an optical fiber transmitting the laser beam. The continuous core of the liquid jet is visualized by means of laser-induced fluorescence, generated by the addition of dye in the liquid. The instantaneous continuous length of the liquid jet is measured as the distance from the nozzle exit, where the emitted laser-induced fluorescence intensity from the liquid jet becomes negligibly small due to interruption of laser light transmission through the liquid jet following the liquid jet breakup. The method provides a nonintrusive measurement approach, promising improved measurements in dense sprays, where droplets obstruct the illumination and the imaging path of shadowgraphic techniques. Measurements in an air-blast atomizer showed that the continuous liquid length measured by the novel approach is systematically shorter than that measured by shadowgraphy. As such, existing empirical correlations for the breakup length, which have been mainly measured by shadowgraphy, may need to be revisited.

The measurement of the rate of fuel injection using a constant volume, fluid filled chamber and measuring the pressure change as a function of time due to the injected fluid (the so called “Zeuch” method) is an industry standard due to its simple theoretical underpinnings. Such a measurement device is useful to determine key timing and quantity parameters for injection system improvements to meet the evolving requirements of emissions, power and economy. This study aims to further the understanding of the nature of cavitation which could occur in the near nozzle region under these specific conditions of liquid into liquid injection using high pressure diesel injectors for heavy duty engines. The motivation for this work is to better understand the temporal signature of the pressure signals that arise in a typical injection cycle. A preliminary CFD study was performed, using OpenFOAM, with a transient (Large Eddy Simulation -LES), multiphase solver using the homogenous equilibrium model for the compressibility of the liquid/vapour. The nozzle body was modelled for simplicity without the nozzle needle using a nozzle hole of 200μm diameter and the body pressurised to values typical for common rail engines. Temperature effects were neglected and the wall condition assumed to be adiabatic. The chamber initial static pressure was varied between 10 and 50 bar to reflect typical testing conditions. Results indicate that vapour formation could occur in areas 10-30mm distant from the nozzle itself. The cavitation was initiated around 100 μs after the jet had started for low ΔP cases and followed the development period required for the formation of vortices associated with the vortex roll up of this jet. These vortices had localised sites, in their core region, below the vapour pressure and were convected downstream of their initial formation location. It was also found that vapour formation could occur at chamber static pressures up to 50 bar (the highest tested) due to cavitation in the shear layer and this vortex effect. The pressure signal received at the chamber would therefore be more difficult to interpret with additional error components.

Schlieren [Schlieren and Shadowgraphy Techniques (McGraw-Hill, 2001); Optics of Flames (Butterworths, 1963)] is a non-intrusive technique that can be used to detect density variations in a medium, and thus, under constant pressure and mixture concentration conditions, measure whole-field temperature distributions. The objective of the current work was to design a schlieren system to measure line-of-sight (LOS)-averaged temperature distribution with the final aim to determine the temperature distribution inside the cylinder of internal combustion (IC) engines. In a preliminary step, we assess theoretically the errors arising from the data reduction used to determine temperature from a schlieren measurement and find that the total error, random and systematic, is less than 3% for typical conditions encountered in the present experiments. A Z-type, curved-mirror schlieren system was used to measure the temperature distribution from a hot air jet in an open air environment in order to evaluate the method. Using the Abel transform, the radial distribution of the temperature was reconstructed from the LOS measurements. There was good agreement in the peak temperature between the reconstructed schlieren and thermocouple measurements. Experiments were then conducted in a four-stroke, single-cylinder, optical spark ignition engine with a four-valve, pentroof-type cylinder head to measure the temperature distribution of the reaction zone of an iso-octane–air mixture. The engine optical windows were designed to produce parallel rays and allow accurate application of the technique. The feasibility of the method to measure temperature distributions in IC engines was evaluated with simulations of the deflection angle combined with equilibrium chemistry calculations that estimated the temperature of the reaction zone at the position of maximum ray deflection as recorded in a schlieren image. Further simulations showed that the effects of exhaust gas recirculation and air-to-fuel ratio on the schlieren images were minimal under engine conditions compared to the temperature effect. At 20 crank angle degrees before top dead center (i.e., 20 crank angle degrees after ignition timing), the measured temperature of the flame front was in agreement with the simulations (730–1320 K depending on the shape of the flame front). Furthermore, the schlieren images identified the presence of hot gases ahead of the reaction zone due to diffusion and showed that there were no hot spots in the unburned mixture.

Lean-burn spark-ignition engines exhibit higher efficiency and lower specific emissions in comparison with stoichiometrically charged engines. However, as the air-to-fuel (A/F) ratio of the mixture is made leaner than stoichiometric, cycle-by-cycle variations in the early stages of in-cylinder combustion, and subsequent indicated mean effective pressure (IMEP), become more pronounced and limit the range of lean-burn operation. Viable lean-burn engines promote charge stratification, the mixture near the spark plug being richer than the cylinder volume averaged value. Recent work has shown that cycle-by-cycle variations in the early stages of combustion in a stratified-charge engine can be associated with variations in both the local value of A/F ratio near the spark plug around ignition timing, as well as in the volume averaged value of the A/F ratio. The objective of the current work was to identify possible sources of such variability in A/F ratio by studying the in-cylinder field of fuel-droplet distribution during the early intake stroke. This field was visualised in an optical single-cylinder 4-valve pentroof-type spark-ignition engine by means of laser-sheet illumination in planes parallel to the cylinder head gasket 6 and 10 mm below the spark plug. The engine was run with port-injected isooctane at 1500 rpm with 30% volumetric efficiency and air-to-fuel ratio corresponding to both stoichiometric firing (A/F=15, Φ =1.0) and mixture strength close to the lean limit of stable operation (A/F=22, Φ =0.68). Images of Mie intensity scattered by the cloud of fuel droplets were acquired on a cycle-by-cycle basis. These were studied in order to establish possible correlations between the cyclic variations in size, location and scattered-light intensity of the cloud of droplets with the respective variations in IMEP. Because of the low level of Mie intensity scattered by the droplets and because of problems related to elastic scattering on the walls of the combustion chamber, as well as problems related to engine “rocking” at the operating conditions close to the misfire limit, the acquired images were processed for background subtraction by using a PIV-based data correction algorithm. After this processing, the arrival and leaving timings of fuel droplets into the illuminated plane were found not to vary significantly on a cycle-by-cycle basis but the recorded cycle-by-cycle variations in Mie intensity suggested that the amount of fuel in the cylinder could have been 6–26% greater for the “strong” cycles with IMEP 115% higher than the average IMEP, than the ones imaged for “weak” cycles at less than 85% the average IMEP. This would correspond to a maximum cyclic variability in the in-cylinder equivalence ratio Φ of the order of 0.17.

An infrared sensor was used for measurements of spatial distribution of hydrocarbon concentration in a model of a gas turbine combustor, using absorption tomography along multiple lines of sight (LOS). This sensor has the potential for monitoring the degree of premixedness of reacting fuel and air in stationary gas turbine combustors, where operation with lean premixed mixtures is important for reduction of NOx emissions. A numerical simulation, using a stochastic representation of a turbulent hydrocarbon distribution in the cross section of a duct was generated, in order to evaluate the ability of computed tomography (CT) to reconstruct the initial distribution. The simulation considered the experimental execution of the measurements, the optimal number of parallel LOS and projection angles and the optimal reconstructed grid size that would be required to minimize the error of reconstruction of the distribution with the algebraic reconstruction tomography algorithm for planar parallel geometry of LOS. We conclude that the quality of reconstruction is increasing asymptotically as both parallel and angular projections increase. As the reconstructed object grid is refined and given a maximum acceptable error of reconstruction, there is a minimum threshold for both the amount of view angles and parallel projections. For instance, the spatially integrated error of reconstruction can remain below 30% in a 11 × 11 object grid for at least 10 parallel LOS and at least 6 view angles. For similar quality of reconstruction and an increasing number of projection angles, the required amount of parallel LOS is slightly decreasing. An experimental assessment of the spatial resolution of the sensor is presented by comparing infrared absorption (INFRA) measurements with those from optical planar visualisation of fuel concentration, based on a Mie scattering technique. The flow investigated was a confined coaxial flow of a 8 mm-diameter central premixed air/methane jet (initial equivalence ratio of 2; Reynolds number of 3800), surrounded by a co-flow of air (0.5 m/s) in a 144 mm diameter duct. The INFRA sensor combined with CT used in this case an 11 × 11 reconstruction grid with 25 parallel LOS and 6 projection angles, because of limitations particular to this experiment, and assumed an axisymmetric flow to provide information to improve the accuracy of the reconstruction. The numerical simulation predicted for this geometry an error of about 30%, which is lowered by the axisymmetric assumption. The CT concentration measurements for low radial gradients (i.e. for Z≫2.75D) had an error in the reconstructed hydrocarbon concentration profile (here calculated by integration over the full width of the duct), of 25% of the corresponding Mie scattering measurements, which were previously calibrated with a flame ionisation detector. The error was found to increase with the local radial gradient of the concentration, which was confirmed by a supplementary numerical simulation that investigated the influence of the spatial gradient of concentration and laser beam diameter. This computational work showed that INFRA combined with CT should have a maximum laser beam width of 20% of the reconstruction grid spatial resolution for optimal accuracy and that the technique can resolve gradient scales of half a pixel of the reconstruction grid size.

This paper describes a Cassegrain optics-based chemiluminescence sensor (CS) for measurements in gas turbine combustors. The chemiluminescence sensor measures the equivalence ratio of reacting fuel and air mixtures, and can identify the flame location, in partially premixed flames. It has the potential for monitoring the degree of premixedness of reacting fuel and air in industrial gas turbine combustors, where operation with lean premixed mixtures is important for reduction of NO, emissions. The spatial resolution of the sensor is evaluated by comparing OH* chemiluminescence measurement from the CS with laser induced OH fluorescence, in the cone-shaped premixed flame of a Bunsen burner. The ability of the sensor to measure in a modified micro-gas turbine environment burning a methane/air, as well as, a methane/ water/air flame (humidified flame) is also demonstrated. (Less)

The potential of lean combustion for the reduction in exhaust emissions and fuel consumption in spark ignition engines has long been established. However, the operating range of lean-burn spark ignition engines is limited by the level of cyclic variability in the early-flame development stage that typically corresponds to the 0-5 per cent mass fraction burned duration. In the current study, the cyclic variations in early flame development were investigated in an optical stratified-charge spark ignition engine at conditions close to stoichiometry [air-to-fuel ratio (A = 15] and to the lean limit of stable operation (A/F = 22). Flame images were acquired through either a pentroof window (‘tumble plane’ of view) or the piston crown (‘swirl plane’ of view) and these were processed to calculate the intra-cycle flame-kernel radius evolution. In order to quantify the relative effects of local fuel concentration, gas motion, spark-energy release and heat losses to the electrodes on the flame-kernel growth rate, a zero-dimensional flame-kernel growth model, in conjunction with a one-dimensional spark ignition model, was employed. Comparison of the calculated flame-radius evolutions with the experimental data suggested that a variation in A/F around the spark plug of Φ(A/F) ≈ 4 or, in terms of equivalence ratio Φ, a variation in Φ ≈ 0.15 at most was large enough to account for 100 per cent of the observed cyclic variability in flame-kernel radius. A variation in the residual-gas fraction of about 20 per cent around the mean was found to account for up to 30 per cent of the variability in flame-kernel radius at the timing of 5 per cent mass fraction burned. The individual effect of 20 per cent variations in the ‘mean’ in-cylinder velocity at the spark plug at ignition timing was found to account for no more than 20 per cent of the measured cyclic variability in flame kernel radius. An individual effect of similar level was attributed to the heat losses to the electrodes, and a much lower effect than that, namely 5-10 per cent at most, could be attributed to cyclic variations in spark-energy release traces. Even when the variations in mean velocity were coupled to both the effects of heat losses to the electrodes and spark-energy release variations, the combined effect was found to account for no more than 40-50 per cent of the observed cyclic variability in flame-kernel radius at the 5 per cent mass-fraction burned timing of 40 ° crank angle after ignition timing.

The acquired images of interferometric particle sizing techniques are characterized by intense fringe pattern overlapping in dense droplet and bubble areas, which hinders the image processing process and subsequent information extraction. Methods employed, such as thresholding and the Hough transform and template cross-correlation, exhibit weaknesses when processing such dense areas of interest. We investigate the viability of applying the wavelet transform (WT) for the detection of the fringe pattern centers and the evaluation of the particle size. We present the basics of the WT using the Mexican hat, which exhibits excellent localization properties and present two different alternatives routes in detecting the fringe patterns in the compressed and uncompressed fringe pattern cases. We found that in comparison to the most reported methods for image evaluation, such as intensity thresholding and plain cross-correlation, the WT is a very efficient tool for detecting the patterns, even in images with high-number fringe pattern areas. The usage of the WT for the sizing of the imaged droplets and bubbles is also examined, in comparison to the Fast Fourier Transform (FFT).

This study reports an experimental investigation of the phenomena which occur when discrete, monodisperse droplets of a water–ethanol–glycerol solution in the size and velocity ranges of 160

This paper reports size measurement of droplets with optically inhomogeneous media by shadow Doppler velocimetry (SDV), which can provide spatially and temporally precise in situ readings of the size and velocity of a single particle with irregular shape and with arbitrary optical properties of the particle medium. In this work, water, instant coffee solution and water-based paint with various solid contents were measured to evaluate the capability and limitations of the measurement. The experiment with instant coffee solutions of 2 and 5% (wt:wt), which contained m discrete particles, atomized by a standard paint spray gun, demonstrated that the accuracy of sizing was not affected by the optical properties of the medium. Insensitivity to the optical properties is one of the primary advantages of SDV over the other optical, single-particle sizing methods. As a further demonstration, paint samples atomized by the same gun containing solid flakes of nominal diameter m were also measured. The results revealed a spatially uniform arithmetic mean diameter of m and suggested that the atomization characteristics were influenced in the highest flake volume fraction case (red paint containing mica, 1.5%) with the result that the mean diameter was 20% larger than that of a similar paint with a smaller flake volume fraction (green paint containing aluminium, 0.4%). It was also found that the measurable number density is limited to no more than 1000 droplets in the case of droplets with an average size of m.

This paper presents a novel optical technique for simultaneous planar measurement of droplet and vapour phases in an evaporative spray. The technique combines the out-of-focus imaging technique 'Interferometric Laser Imaging for Droplet Sizing' (ILIDS) for planar simultaneous measurements of droplet size, velocity, and number density with the in-focus technique 'planar laser-induced fluorescence' (PLIF) for measurements of vapour concentration around individual droplets. As with the combined ILIDS and PIV technique (Hardalupas et al. in Exp Fluids 49:417-434, 2010), a discrepancy occurs in the location of the same droplet while imaging through both techniques. The centre discrepancy is experimentally estimated, and so, the position of the droplet centres in the ILIDS image can be corrected, which is essential for the measurement of the correlation of the local vapour concentration with its corresponding droplet size and velocity. The combined technique is first applied to mono-sized acetone droplet stream for two droplet sizes, which have different inter-droplet distances. A new method is proposed for the correction of the 'halation' effect around droplets on vapour fluorescence intensity around the mono-sized droplets in the PLIF image. A phase discrimination method to separate droplet and vapour fluorescence in PLIF images of evaporative sprays is explained. It is shown that the combined ILIDS and PLIF technique is an effective tool for studying group evaporation of droplets and interaction between the two phases in an evaporative spray � Springer-Verlag Berlin Heidelberg 2014.

The four-valve head of a VTEC engine was mounted on an open cylinder and the valves and fuel injection system operated as in the engine with a rotational speed of 1200 rpm. Local measurements of droplet characteristics were obtained with a phase-Doppler velocimeter and iso-octane injected over 5 ms intervals, corresponding to 36 crank angle degrees, with manifold depression of 20 mbar. The results show that most of the fuel droplets were located close to the liner and on the side of the cylinder adjacent to the exhaust valves. In the plane of the measurement, 10 mm below TDC, the liquid flux diminished as the initiation of injection was advanced before opening of the inlet valves. With injection with the inlet valves closed, there were two waves of droplets, one from each of the two valves and separated by 60 deg CA and both with the Sauter mean diameter of about 120 μm. With injection with the inlet valves open, most of the droplets emerged from the main inlet valve and with Sauter mean diameters of about 50 μm, smaller than those of the unconfined spray.

Phase Doppler anemometry and a two-colour pyrometer have been combined to provide simultaneous, local measurements of kerosene droplet velocity and size and temperature of the burning soot mantle. Preliminary measurements within the flame brush that develops in the shear layer of a swirl-stabilized, gas-supported kerosene flame with a swirl number of about 0.19 and potential heat releases of 10.6 and 15.5 kW, respectively, are described. The results show that the maximum burning fraction of the droplets occurs adjacent to the region denoted as gas flame but the value ranges between 20+or-5 and 40+or-5% depending on the axial station, and decreases sharply across the shear layer. The flame mantle temperature was found to be independent of droplet diameter, which agrees with previous results in the literature. The likely experimental errors are assessed in the context of the measurements in the swirl burner.

Mass flux, mass fraction and concentration of kerosene droplets have been measured by a phase-Doppler instrument in a swirl-stabilized burner during inert and reacting flow. In the reacting flow, the flame was supported by natural gas added through the fuel stalk, while the kerosene was added to and atomized by the combusting air. The bulk mean velocity of the combustion air was 30.4 m/s, corresponding to a Reynolds number of 30,000, and the swirl numbers were 0.48 and 0.29, respectively, upstream and downstream of the kerosene injection. The equivalence ratios were 0.45 and 2.11 for the natural gas and the kerosene, respectively. The results show that the minimum swirl necessary for flame stability caused fuel to centrifuge from the region of combustion, so that only 8 kW of energy could be released out of the available 37.4 kW of the kerosene fuel. An important conclusion is that an optimum swirl number will exist with every atomization and burner arrangement of a liquid-fuelled flame and will be different from that associated with the corresponding gas-fuelled flame. The measurement techniques appropriate to regions where the flow instantaneously reverses are described and the existence of large droplets, which moved towards the injector inside the recirculation zone and supplied fuel to the base of the flame, are explained in terms of a “fountain effect” based on the mean drag between the gas phase and the droplets. Sources of uncertainties of the mass flux and concentration measurements with the phase Doppler instrument are considered in detail.

A new approach for simultaneous planar measurement of droplet velocity and size with gas phase velocities is reported, which combines the out-of-focus imaging technique ‘Interferometric Laser Imaging Droplet Sizing’ (ILIDS) for planar simultaneous droplet size and velocity measurements with the in-focus technique ‘Particle Image Velocimetry’ (PIV) for gas velocity measurements in the vicinity of individual droplets. Discrimination between the gas phase seeding and the droplets is achieved in the PIV images by removing the glare points of focused droplet images, using the droplet position obtained through ILIDS processing. Combination of the two optical arrangements can result in a discrepancy in the location of the centre of a droplet, when imaging through ILIDS and PIV techniques, of up to about 1 mm, which may lead to erroneous identification of the glare points from droplets on the PIV images. The magnitude of the discrepancy is a function of position of the droplet’s image on the CCD array and the degree of defocus, but almost independent of droplet size. Specifically, it varies approximately linearly across the image along the direction corresponding to the direction of propagation of the laser sheet for a given defocus setting in ILIDS. The experimental finding is supported by a theoretical analysis, which was based on geometrical optics for a simple optical configuration that replicates the essential features of the optical system. The discrepancy in the location was measured using a monodisperse droplet generator, and this was subtracted from the droplet centres identified in the ILIDS images of a polydisperse spray without ‘seeding’ particles. This reduced the discrepancy between PIV and ILIDS droplet centres from about 1 mm to about 0.1 mm and hence increased the probability of finding the corresponding fringe patterns on the ILIDS image and glare points on the PIV image. In conclusion, it is shown that the proposed combined method can discriminate between droplets and ‘seeding’ particles and is capable of two-phase measurements in polydisperse sprays.

The dispersion of glass beads in an air flow through sudden step expansions in the direction of gravity has been investigated with phase-Doppler anemometry, which provided measurements of the velocity, flux and concentration characteristics. The purpose was to quantify the effect of increasing the air velocity over a factor of 5 for bead diameters of 40 µm and 80 µm and two expansion diameter ratios, 3.33 and 5, and at mass loadings of beads up to 90% of the air mass flow-rate. The results showed that the beads dispersed into the recirculation zone in the lee of the step by interaction with eddies characterized by length and velocity scales of the order of the expansion step height and the downstream area-averaged velocity. Particle dispersion into the recirculation zone was reduced when the bead mean transit time across the recirculation zone was shorter than the bead relaxation time, defined as the time required for a motionless bead suddenly exposed to a constant velocity fluid stream to reach 63% of its surrounding fluid velocity. Also the centrifuging effect, caused by the mean stream line curvature of the recirculation zone, could reduce particle dispersion into the recirculation zone, when its characteristic dimensionless group was less than unity. Beads, because of their mass, left the recirculation zone by sliding down the wall and past the air reattachment point or near the step, giving rise to bimodal probability distributions of velocity.

Spatial and temporal profiles of the velocity of the entrained air and 60 and 30 μm droplets, together with the associated fluxes, from a 5-hole diesel spray exhausting into atmosphere at a repetition rate of 10 Hz have been measured with a phase-Doppler anemometer. The nozzle diameters, fuel charge per hole, injection duration and the area-averaged spray velocity during this duration were 0.18 mm, 2.35 mm3, 0.7 ms and U0 = 132 m/s, respectively. The Sauter mean diameter of the fuel droplets decreased from a maximum centreline value of around 80 μm at 100 diameters from the nozzle to 38 μm at 780 diameters, and a similar decrease was observed between 1 and 2 ms after the start of injection at the upstream location. The flux carried by the 30 μm droplets was up to twice that associated with the 60 μm droplets, 2 ms after injection, although the velocities of the larger droplets were consistently higher than those of the smaller droplets. The maximum measured ensemble-averaged relative velocity was 0.45 U0 for 60 μm droplets just after the arrival of the spray at 550 diameters from the nozzle. The magnitudes of the Weber number imply that droplet breakup was always confined to the leading edge of the spray and was limited to, at most, the initial 1 2 ms of the passage of the spray past a given point. Breakup was mostly complete by 550 diameters from the nozzle. Thus, the measured decrease in the mean diameter was due to small droplets, generated by breakup at the leading edge of the spray, losing velocity due to aerodynamic drag and falling behind the leading edge. Droplets generated late in the injection schedule were likely to overtake those generated earlier and together with the fan-spreading effect, which arises from the combination of the root mean square (RMS) droplet radial velocity and the radial profile of the ensemble-averaged droplet axial velocity, led to RMS velocities in the axial component of the droplets that were not associated with the transfer of turbulent motion from the air.

The relationship between size and intensity of laser light scattered from coal particles, glass beads and calibration pinholes in the size ranges 10–140 mm, 17–240 mm and 5–200 mm respectively has been measured. The purposes were to determine the effect of shape of non-spherical particles and beads by varying three parameters in the collection optics. These were: the angle, θc, between the axis of the collection optics and that of the incident laser beam; the aperture of the f/3.66 collection lens; and the diameter of the spatial filter on the photodetector. A spatial filter diameter of 0.5 mm was used on the detector as a compromise between image vignetting and reducing the size of the effective measuring volume of the instrument. Due to refraction, and particularly for diameters larger than about 80 mm, the beads scattered more light than the pinholes or the coal particles. A mask, introduced to limit the collection lens aperture to angles less than 5.3°, reduced the intensity collected from the beads, while the response curves for the pinholes and coal particles were unaffected. Non-spherical beads larger than about 60 mm with aspect ratios up to 1.8 resulted in departure from the response curve, established from the calibration pinholes, by up to a factor of two. The aspect ratio was defined by the maximum and minimum projected lengths of the image of the scatterer. The response curves for beads and coal particles approached that of the pinholes as θc was changed from 2.5° to 1.4° and as the angular position of the mask limiting the aperture of the collection lens nearest to the incident beam axis, θi, was reduced from 2.9° to 0.9°. The best results were obtained for θi and θc equal to 0.9° and 1.4° respectively, for which the precision and accuracy were 9 mm and + 5 mm for nominally spherical glass beads and 5 mm and −1 mm for the coal particles for diameters up to 60 mm and aspect ratio up to 2.5.

A high-swirl low-compression-ratio, optically accessed engine that was able to produce a stratified charge was used to investigate the differences in homogeneous charge compression ignition (HCCI) combustion and in the propagation of the autoignition front between a non-stratified and a stratified charge. Natural-light images were acquired using a fast camera to visualize HCCI combustion and to quantify the location of autoignition, the apparent ‘propagation speed’ of the autoignition front, and its variations between closed-valve injection timing (leading to a nearly homogeneous charge) and open-valve injection timing (leading to a strongly axially stratified charge), owing to temperature inhomogeneities that were introduced by utilizing a camshaft which allowed 40 per cent internal exhaust gas recirculation (iEGR). Experimental results show that, in the case without exhaust gas recirculation (EGR) and with closed-valve injection timing, autoignition started under the primary intake valve near the cylinder wall, while, in the case without EGR and with open-valve injection timing, autoignition started between the exhaust valve and the secondary intake valve, closer to the centre of the piston. With 40 per cent iEGR and closed-valve injection timing, autoignition started between the exhaust valve and the primary intake valve near the cylinder wall. These differences can be explained by the difference in the location of hot gases due to the injection timing or due to iEGR. Finally, without EGR, a ‘uniform’ autoignition front of HCCI combustion from the original sites of autoignition was observed compared with a more ‘random development’ of the autoignition front with 40 per cent iEGR. Strong local inhomogeneities (possibly a very rich mixture at a low temperature) could be present with 40 per cent iEGR.

Flow turbidity, when introduced between the transmitting and receiving optics and the measuring volume of a laser-Doppler anemometer, changes the pedestal amplitude and visibility of the signal. The purpose of this work is to assess the effect on the accuracy of particle sizing, based on measurements of these two quantities, for depths of field of 5 and 10 cm, interrupting particle diameters between 14 to 212 μm in three discrete ranges and void fractions up to 0.1 percent. The turbidity introduces random fluctuations in visibility which increase with void fraction and the resulting rms errors in particle diameter for turbidity introduced on the receiving side of the optics are smaller than 10 percent at void fractions below 0.1 percent. For particles larger than about one third of the beam diameter, the influence of turbidity is largely due to the interruption of the incident beams over the 5 cm nearest to the measuring volume.

A space-filling, low blockage fractal grid is used as a novel turbulence generator in a premixed turbulent flame stabilized by a rod. The study compares the flame behaviour with a fractal grid to the behaviour when a standard square mesh grid with the same effective mesh size and solidity as the fractal grid is used. The isothermal gas flow turbulence characteristics, including mean flow velocity and rms of velocity fluctuations and Taylor length, were evaluated from hot-wire measurements. The behaviour of the flames was assessed with direct chemiluminescence emission from the flame and high-speed OH-laser-induced fluorescence. The characteristics of the two flames are considered in terms of turbulent flame thickness, local flame curvature and turbulent flame speed. It is found that, for the same flow rate and stoichiometry and at the same distance downstream of the location of the grid, fractal-grid-generated turbulence leads to a more turbulent flame with enhanced burning rate and increased flame surface area.

Laminar counterflow flames have been used to study a variety of combustion problems, for example extinction conditions and chemical kinetic mechanisms with a review in Ref. [1], and to provide information for flamelet models of turbulent flames [2]. Recently, this flow configuration for turbulent premixed and non—premixed flames has also attracted interest [3, 4] because the velocity field provides a uniform strain rate, a quantity important for flame extinction, and because the simple geometry facilitates theoretical analyses [5, 6]. Mastorakos et al. [4] found that partial premixing improved flame stability and that a spectrum of flames from “pure” diffusion to premixed flames could be established in the counterflow geometry. The present paper is concerned with the appearance of the range of flames that can be stabilised in turbulent counterflowing streams and gives additional information on the consequences of stretching on laminar and turbulent counterflow flames with partial premixing and on the mixing between two turbulent isothermal opposed jets, information relevant to opposed jet combustion. In the next Section, the geometies and the methods used to visualise the flames are described, with the results presented and discussed in Section 3. Finally, Section 4 summarises the more important conclusions.

Two orthogonal components of velocity and associated Reynolds stresses have been measured in a square-sectioned, 90 degree bend of 2.3 radius ratio using laser-Doppler velocimetry for Reynolds numbers of 790 and 40,000. The boundary layers at the bend inlet were 0.25 and 0.15 of the hydraulic diameter and resulted in secondary velocity maxima of 0.6 and 0.4 of the bulk flow velocity respectively. Comparison with fully-developed inlet flow shows that the boundary layer thickness is important to the flow development (mainly in the first half of the bend), particularly so when it is reduced to 0.15 of the hydraulic diameter. Turbulent flow in an identical duct with a radius ratio of 7.0 gives rise to smaller secondary velocities than in the strongly curved bend, although their effect is more important to the streamwise flow development because of the smaller pressure gradients. The detail and accuracy of the measurements make them suitable for evaluation of numerical techniques and turbulence models. Partially-parabolic techniques are applicable to the flows studied and their reduced storage requirements seem essential if satisfactory numerical accuracy is to be achieved.

Measurements have been made of the effect of flow turbidity on the visibility and pedestal amplitude of an anemometer signal when incident laser beams are interrupted by particulate flow. The purpose is to assess the likely accuracy of particle sizing and the reliability of discrimination between continuous and particulate phase velocities. Optical depths of field were varied between 2.5 × 10−2 and 14 × 10−2 mm the diameter of the interrupting particles ranged between 14 and 800 μm in six discrete ranges and the corresponding void fractions lay between 0.003% and 0.378%. The incident beam diameter was approximately 400 μm. The measured size is subject to both systematic and random errors when inferred from measurements of pedestal amplitude: the random error increases as the ratio of the incident beam diameter to that of the particulate phase decreases. Systematic errors corresponding to a 10% underestimation of diameter occur at void fractions of 0.003%, 0.01% and 0.018% for particles below 40 μm 75 μm and 105 μm respectively over a 5 cm depth of field. The r.m.s. error is smaller than 7% for particles below 40 μm for all conditions studied but increases with increasing diameter and exceeds 10% at void fractions greater than 0.1% for particles above about 100 μm. The random error in measured diameter derived from measurements of visibility is influenced mostly by the flow turbidity over the 5 cm of the incident beams closest to the measuring volume. For interrupting particles smaller than about 100 μm the r.m.s. error is similar to that for measurements based on the pedestal amplitude. Discrimination of the velocity signal between the particulate and dispersed phase, based on the separation of pedestal amplitudes, is likely to be unreliable if the particle diameter is comparable to the diameter of the incident beams and if the probability of two particles simultaneously present in each beam is not negligible. A method for estimating the level of turbidity at which discrimination is no longer possible is described.

The streamwise components of velocity in the flow through a ninety degree bend of circular cross section for which the ratio of radius of curvature to diameter is 2.8 were measured. The development of strong pressure driven secondary flow in the form of a pair of counter rotating vortices in the steamwise direction is shown. Refractive index matching at the fluid wall interface was not employed; the displacement of the measurement volume due to refraction is allowed for in simple geometrical calculations.