Modeling of turbulent opposed-jet mixing flows with – model and second-order closure

Velocity measurements in the isothermal flows created by an opposed nozzle configuration are reported with emphasis on the axis, stagnation plane and the distributions of mean and instantaneous strain rates. The instrumentation comprised particle image velocimetry (PIV) with silicon oil droplets added to the flows upstream of both nozzles with the laser sheet passing through the axis between the nozzles. The results identify the regions of high strain rates and quantify the development of the mean and turbulent components of the flow from the nozzle exits as a function of bulk velocities from 3 to 8.2 m/s and nozzle separations from 0.4 to 1.0 diameters. Results show, for example, the rise in the values of axial and radial normal stress towards the stagnation plane with values increasing by up to 300% and 160% respectively. The maximum mean strain rate occurred just over one nozzle radius from the axis at the smallest separation and with values that increased from 450 to 950 s −1 with decreasing separation at a bulk velocity of 3.0 m/s. Probability density functions were near Gaussian and hence much larger instantaneous strain rates were observed. The PIV image size had the advantage that it allowed the entire flow field to be viewed in terms of velocity vectors and derived quantities include mean strain rates. Small asymmetry of the flow and the higher strain rates at finite distances from the nominal impingement plane were observed. The experimental results permitted the domain of applicability of different modelling approaches to be defined more accurately and calculations were performed with different turbulence models. The results showed that two-equation turbulence models did not represent turbulence intensities close to impingement and that Reynolds stress closures produce superior agreement. It was further shown that ad hoc modifications to the dissipation equation, such as those based on the ratio of the turbulent to mean strain time scale, can improve results at the expense of generality. It is also shown that mean flows are well reproduced by a Reynolds stress closure for all nozzle separations. Comments are included on the implications of the results for investigations of reacting flows and extinction. NomenclatureD = Nozzle diameter (mm) H = Nozzle separation (mm) k = Turbulent kinetic energy (m 2 /s 2 ) L t = Integral length scale (m) P = Production (m 2 /s 3 ) r = Radius (mm) 170 R.P. LINDSTEDT ET AL. R = Nozzle radius (D/2) S b = Bulk strain rate (s −1 ) S rad = Radial component of strain rate (s −1 ) S ax = Axial component of strain rate (s −1 ) t = time (s) u 2 = Axial component of Reynolds stress (m 2 /s 2 ) U = Axial component of velocity (m/s) U b = Bulk nozzle exit velocity (m/s) uv = Shear component of Reynolds stress (m 2 /s 2 ) v 2 = Radial component of Reynolds stress (m 2 /s 2 ) V = Radial component of velocity (m/s) x = Axial distance from top nozzle (mm) √ U 2 + V 2 = Velocity magnitude (m/s) δ i j = Kronecker delta (equal to 1 when i = j) µ t = Turbulent viscosity (m 2 /s) ρ = Densit...

Section: Numerical Simulationsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Velocity and Strain-Rate Characteristics of Opposed Isothermal Flows

Lindstedt

Luff

Whitelaw

2005

“…A number of numerical studies, mostly based on one-and two-dimensional simplifications of turbulent stagnation point flows in the RANS framework, have been published by Craft et al [7], Dianat et al [8], Lindstedt and Vaos [26,27], Eckstein et al [9], Chou et al [5], Korusoy and Whitelaw [22] and others. Generally, mean quantities were predicted more accurately than fluctuating components and Reynolds stress turbulence models yielded better results than eddy viscosity approaches.…”

Section: Introductionmentioning

confidence: 99%

Highly-resolved LES and PIV Analysis of Isothermal Turbulent Opposed Jets for Combustion Applications

Stein

Böhm

Dreizler

et al. 2010

Turbulent opposed jet (TOJ) burners are an interesting test case for fundamental combustion research and a good benchmark for the available modelling approaches. However, these opposed jet flames strongly depend on the turbulence generation inside the nozzle, which is usually achieved through a perforated plate upstream of the nozzle exit. The present work investigates the flow from these perforated plates and the subsequent turbulence generation in great detail. We present results from highly-resolved large eddy simulations (LES) of the in-nozzle flow in turbulent opposed jets alongside state-of-the-art particle image velocimetry (PIV) at standard and high repetition rates taken inside a glass nozzle. The in-nozzle PIV data provides the LES inflow conditions with unprecedented detail, which are used to follow the initial jet development behaviour known from PIV, before jet coalescence, turbulence production and decay further downstream in the nozzles are successfully predicted. In regions where the PIV experiment suffers from inherent limitations like reflections and the velocity bias, the LES data is available to still obtain a detailed picture of the flow. The sensitivity of the simulations to various physical and numerical parameters is discussed in detail. Results from LES and PIV are compared Flow Turbulence Combust (2011) 87:425-447 qualitatively and quantitatively in terms of first and second moments of velocity, temporal autocorrelations, and energy density spectra. Significant deviations are found in the frequency (20%) and strength of vortex shedding from the inlet plane only, whereas the qualitative and quantitative agreement between simulation and experiment is otherwise excellent throughout, implying that a successful large eddy simulation of a turbulent opposed jet can be performed in a domain that includes the perforated plates.

“…The mean velocities, the mean and variance of the mixture fraction, the mean scalar dissipation rate of the mixture fraction, the turbulent diffusivity, the turbulent kinetic energy and its dissipation come from a CFD solution with FLUENT [38] by a Reynolds stress model [13], as suggested previously for counterflow simulations [12,14], and with FLUENT's flamelet combustion model.…”

Section: Cfd Predictionsmentioning

confidence: 99%

“…Eckstein et al [11] calculated turbulent inert mixing in the opposed jet flow with a Monte Carlo PDF simulation that used a high value for the timescale ratio constant, C D , used in the conventional model for the mean scalar dissipation rateχ = C Dε k ξ 2 : C D as high as 10 was needed to produce a reasonable agreements with the data [2]. Chou et al [12] presented a comparison of turbulence models (k −ε and Reynolds stress) for the inert mixing flow. The Reynolds stress model by Launder et al [13] provided the most accurate result, a conclusion that has also been reached by the earlier modelling efforts for the opposed-jet flow [14], although thek −ε model has also provided good results [15].…”

Section: Introductionmentioning

confidence: 99%

Simulations of Turbulent Non-Premixed Counterflow Flames with First-Order Conditional Moment Closure

Kim

Mastorakos

2006

Self Cite

Simulations of turbulent CH 4 -air counterflow flames are presented, obtained in terms of zero and two-dimensional first-order Conditional Moment Closure (CMC) to study the flame structure and extinction limits. The CMC equation with detailed chemistry is solved without the need for operator splitting, while the accompanying flow field is determined using a commercial CFD software employing a Reynolds stress turbulence model and additional transport equations for the turbulent scalar flux and for the mean scalar dissipation rate. Two detailed chemical mechanisms and different conditional scalar dissipation rate models have been examined and small differences were found.The first-order CMC captures the overall structure of the counterflow flame accurately for the unconditional averages. The calculated conditional averages behave as if the scalar dissipation rate were under-predicted, although a comparison with measurement of the conditional scalar dissipation rate is reasonable. The calculated extinction velocity is found to be much higher than the experimental value, but the trend of increasing extinction velocity with air dilution of the fuel stream is captured well. The discrepancies with the data are mostly attributed to the neglect of conditional fluctuations.