James F. Driscoll

Images of the quenching of a flame by a vortex—To quantify regimes of turbulent combustion

Abstract A laminar toroidal vortex is interacted with a laminar premixed flame in order to isolate and to visualize some of the fundamental physics of turbulent combustion. Localized quenching of the flame was observed using planar laser-induced fluorescence imaging of superequilibrium OH molecules in the counterflow flamefront region near the vortex leading edge. A quenching limit curve was measured as a function of vortex size and strength. In the second part of the study, the measurements are combined with concepts proposed by Poinsot, Veynante, and Candel in order to infer the thin flame limit, namely, the onset of distributed reactions, on a classical premixed turbulent combustion regime diagram. The measured thin flame limit indicates when laminar flamelet theories become invalid, since quenching allows hot products and reactants to coexist. Results are compared with the Klimov-Williams criterion. Vortex core diameters were as small as the flame thickness in some cases. The main conclusion is that small vortices are less effective at quenching a flame than was previously believed; therefore the inferred regime within which thin flame theories are valid extends to a turbulence intensity that is more than an order of magnitude larger than that which was previously predicted. Results also indicate that micromixing models, which assume that the smallest eddies exert the largest strain on a flame, are not realistic. Measured trends are in agreement with direct numerical simulations of Poinsot et al., but absolute values differ. The measured vortex Karlovitz number that is required to quench a flame is not constant but decreases by a factor of four as vortex size increases from one to five flame thicknesses. Thin-film pyrometry was used to quantify the radiative heat losses; quenching occurs when the products cool to approximately 1300 K, which is in agreement with stretched laminar flame calculations that include detailed chemistry. The quenching Karlovitz number for propane-air flames differs from that of methane-air flames, indicating the importance of detailed chemistry and transport properties. Flame curvature was observed to cause enhancement (or reduction) of the local reaction rate, depending on the Lewis number, in a manner that is consistent with stretched flame theory.

Nitric oxide levels of turbulent jet diffusion flames: Effects of residence time and damkohler number

The global residence time and the deviations from chemical equilibrium (i.e., the Damkohler number) were varied for a number of jet diffusion flames. The resulting effects on the nitric oxide emission index were measured and were compared with existing analysis. The global residence time is defined as L//U F, where L/ is the flame length and Up is the fuel jet velocity. Flame length is varied by increasing the jet diameter, by adding either premixed air or inerts to the fuel jet, or by adding a coaxial air stream. In particular, a unique jet flame was studied that is composed of helium-diluted hydrogen fuel; this flame is free of the complicating effects of flame radiation, buoyancy, and prompt NO and provides a useful baseline comparison to theory. It is found that NO x levels for three types of fuels were consistently less than levels predicted by thermal theory, which suggests that one or both of the two mechanisms that suppress NOx, namely strain and radiative cooling, are important. The use of a Damkohler number was found to successfully correlate the NO x data for the hydrogen/helium- air flames that have simple chemistry. As the helium concentration is increased in order to reduce the Damkohler number, the measured NO x emission index exceeds that of the equilibrium theory by as much as a factor of 24, which is further indication that it is important to add the correct nonequilibrium oxygen atom chemistry to current models.

Vorticity generation and attenuation as vortices convect through a premixed flame

Abstract A sequence of PIV images shows the time history of both the vorticity field and the velocity field as vortices of different strength convect through a premixed flame. The vortices represent individual eddies in turbulent flow; the goal is to understand how each eddy wrinkles the flame and how the flame also may alter the eddy. It is found that weak vortices are completely attenuated primarily due to volume expansion. Strong vortices do survive flame passage, but only if they can weaken the flame due to stretch effects. Intense flame-generated vorticity is measured which has a magnitude that exceeds that of the incident vortex in some cases. The flame-generated vorticity in the products induces a velocity field that tends to reduce the amplitude of flame wrinkling; thus it acts as an additional flame-stabilizing mechanism. This mechanism affects the wrinkling process and should be included in models. A new nondimensional vorticity enhancement parameter ( E ) is suggested as a way to estimate the effect of vortex size, strength, Reynolds number, and Froude number on vorticity attenuation and production. Measurements are made for E approximately equal to 0, −1, and −2, corresponding to no change in vorticity, total attenuation of the vortex, and flame-generated vorticity, respectively. Buoyancy forces are important in one case that is considered, but not in other cases. The results can be used to quantify the size of the small eddies that can be neglected in large eddy simulations; the role of small eddies is estimated in one example.

Reaction zone structure in turbulent nonpremixed jet flames—from CH-OH PLIF images

Abstract It is shown that simultaneous images of the CH and OH concentration fields can be obtained throughout a high-Reynolds-number (18,600) turbulent nonpremixed, nonsooting jet flame, and that the CH-OH boundary is a useful marker of the instantaneous stoichiometric contour. Previous CH-OH imaging was confined to the flame base. The structure of the fuel-decomposition zone—identified by the CH images—includes the following regions: those with high-curvature cusps; those with low CH concentration; and those where the flame “pinches” due to oxidizer being entrained to the centerline. It is found that the reaction zone that is associated with fuel decomposition (i.e., the CH layer) remains thin and rarely exceeds 1 mm, even near the tip of the high-Reynolds-number flame. CH layers in the turbulent flame are not thicker than the CH layers in the laminar jet flame at the same x / d location. In fact, CH layer thickness is relatively insensitive to Reynolds number and the level of turbulence. This implies that turbulence does not broaden the CH reaction zone, and that flamelet concepts are justified in modeling the reaction zone associated with fuel decomposition. The CH layers become thicker in the streamwise ( x ) direction, which is expected because scalar gradients and the dissipation rate are expected to decrease in the streamwise direction. Imaging the CH layer makes it possible to measure the flame surface density (Σ), which has a typical value of 0.2 mm −1 . Surface density is shown to be related to the turbulent brush thickness and the degree of wrinkling.

A laminar vortex interacting with a premixed flame: Measured formation of pockets of reactants

A single toroidal, laminar vortex in created and a laminar premixed flame is propagated over the vortex in order to study the fundamental interaction process and to compare results with the numerical solutions of the Navier-Stokes equations obtained by Poinsot, Veynante, and Candel. Results quantify the degree of flame rollup, the flame perimeter increase due to wrinkling, the minimum vortex size that wrinkles the flame, and the boundary of the pocket formation (corrugated flame) regime. The measured trends indicate that viscous attenuation of the vortex by the hot, viscous products is important. The importance of flame stretch is verified; as Lewis number is varied it is found that in order to form pockets, diffusionally stable flames require a vortex strength that is three times larger than that required by the ditfusionally unstable flames.

Effect of Heat Release and Swirl on the Recirculation within Swirl-Stabilized Flames

Abstract In the swirl-stabilized flame studied, the heat release and the swirl were systematically varied in order to quantify their effects on the internal recirculation zone, the mixing, the velocity PDFs and the flame blowoff limits. The amount of heat release was varied by changing the overall equivalence ratio, and the swirl number was varied up to 4.0 to assess the advantages or disadvantages of very high swirl operation. The velocity fields in nine non-premixed flames and three cold flows were mapped out using laser velocimetry. It was found that increasing the heat release resulted in a number of benefits, including an increase in the recirculation, the turbulence levels, and the flame stability. Heat release helped to drive the recirculation, since in some cases the cold flow did not have recirculation but the corresponding flames did. Turbulent kinetic energy levels increased by a factor of three as heat release increased. The effect of increasing the swirl was to improve the mixing and flame st...

A comparison of bluff-body and swirl-stabilized flames

Abstract Bluff-body and swirl-stabilized flames are similar in that they represent, in simplest terms, the fundamental interaction between a fuel jet and a surrounding toroidal vortex. The vortex in this case is the recirculation vortex which affects the properties of the flames. It is found, not surprisingly, that the two most important fundamental parameters that govern both types of flames are (1) the vortex circulation (Γ), and (2) the fuel jet momentum. Comparisons are made of the properties of the two types of flames using the proper nondimensional parameters, including the fuel-to-air momentum flux ratio and the properly nondimensionalized vortex strength. Such comparisons can help to illustrate the tradeoffs between the degree of swirl and the choice of bluff-body size in devices such as industrial burners, gas turbines, and ramjets. The data also show how one can control flame properties by controlling the vortex strength Γ and fuel momentum and thus gain a degree of control that is not provided ...

Nitric oxide levels of jet diffusion flames: Effects of coaxial air and other mixing parameters

This paper quantifies the effects of certain fluid mechanical mixing perameters, such as coaxial air velocity, Reynolds number and Damkohler number, on the total nitric oxide emission index of non-premixed jet flames. It is found that forcing coaxial air into the flame can result in a significant (up to sixfold) reduction of the NOx emission index (in g/kg fuel) as compared to simple jet flames. The large NOx reduction is explained by the fact that coaxial air reduces the flame length and therefore forces the jet flame to exist closer to the nozzle where the velocities are larger, which reduces the local residence time as well as the reaction zone volume. The thermal NOx scaling proposed by Peters for jet flames with no coaxial air appears to explain the present results successfully for the case of a hydrogen-air flame but with only partial success for the case of a methane-air flame. The different scaling of the methane-air flame may be due to its greater sensitivity to the strain induced by coaxial air, or may be due to prompt NO. Coaxial air also significantly changes the fraction of NO2 within the NOx, but the NO2 scales in a manner that is different from the overall NOx. Parameters that were systematically varied include the Reynolds number, the coaxial air velocity, fuelto-air diameter ratio and fuel type. The measured trends can be used to assess recent NOx/mixing models. The NOx/emission index of simple jet flames, having no coaxial air, also are quantified. NOx levels decrease as the characteristic residence time is decreased, as expected, when the fuel jet diameter is decreased or the jet velocity is increased. However, after accounting for residence time by using the conventional scaling parameters, an additional Damkohler number effect is observed. That is, the normalized NOx emission index obtained for a wide range of jet velocities (UF) and diameters (dF) collapse to a single curve when plotted against the ratio UF/dF, rather than using the Reynolds number, as has been done previously. Some physical reasons for the observed scaling are proposed. The NOx results are compared to the limiting cases of reduced mixing (i.e., low Reynolds number laminar jet flames) and very intense mixing, which is achieved by adding swirl to the coaxial air. The trends caused by adding swirl are similar to the trends caused by adding coaxial air, since both increase the rate of fuel-air mixing and shorten the flame. Swirl can further reduce the NOx levels. The combined fluid mechanical mixing effects of coaxial air and swirl increases the percentage of NO2 in the NOx from 1.3% in a simple hydrogen jet flame to 66% in a swirl flame.

Ram-Scram Transition and Flame/Shock-Train Interactions in a Model Scramjet Experiment

The behavior of a ram-scram transition was examined using a direct-connect model scramjet experiment along with pressure measurements and high-speed laser interferometry. The work quantifies the sudden change in the wall static pressure profile and flame position that occurs as the downstream boundary condition abruptly changes when the flow becomes unchoked. Transition was studied in three ways; as a quasi-steady phenomenon due to slow decreases in fuel flow rate, or as caused by rapid variations in either fuel flow rate or test-section wall temperature. A regime diagram was constructed that plots the measured ram-scram transition boundary. Under certain conditions some periodic low-frequency oscillations of the flame position occur and they are shown to be correlated with oscillations of the upstream precombustion pseudoshock. A self-sustaining shear-layer instability, associated with the flameholding cavity, is believed to be the mechanism causing this behavior. The relevant time scales associated with...

The role of the recirculation vortex in improving fuel-air mixing within swirling flames

An important example that shows how a flame interacts with a vortex is the case of a swirlstabilized flame. Such a flame essentially is a toroidal vortex into which a jet of fuel is injected along the torus centerline. Because of the central toroidal vortex, the overall fuel-air mixing rate within a swirl-stabilized flame is found to be a factor of five times greater than that of a simple jet flame, as evidenced by a fivefold shortening of flame length. Swirl flames are unique in that the fuel-air mixing rate can be varied by controlling the amount of swirl. Flow visualization shows how internal recirculation helps to enhance the mixing. The englufment mechanisms is similar to that of a simple jet flame but is enhanced because of two factors that increase the fuel-air contact area. First, the recirculation zone acts like a large eddy with a characteristic velocity and length scale that are much larger than those associated with eddies in a simple jet. The active role of the recirculation zone contradicts a previouslyheld concept that the recirculation zone is a passive obstacle and that its internal velocity is not important. Air is entrained into the toroidal vortex primarily in the downstream region of the vortex. A second reason why mixing is more intense within swirling flames than within jet flames is that there is impingement of opposed jets at the forward stagnation point. Thus pressure gradients, which are not present in simple jets, enhance the mixing rates. Flame length, which is a measure of overall fuel-air mixing, was found to have a different scaling in swirl flames than for simple jet flames. The physical reasons are explained by a scaling which was developed and which explains four trends in the flame length data. It is proposed to use a general, nondimensional circulation parameter that allows for general comparison of mixing efficiencies of swirl, bluff-body, and dump-combustor flames.