# Bluff body burner for CH4-H2 turbulent combustion

Application Challenge 2-01 © copyright ERCOFTAC 2004

## Overview of Tests

A comprehensive set of data collected in bluff-body stabilised flames and non-reacting jets is released. Measurements of flow, mixing, temperature as well as composition fields are presented. The composition field measurements include species such as CO, CO2, H2, H2O, O2, N2, Hydrocarbon, as well as OH and NO. Data are presented here for reacting and non-reacting bluff-body stabilised flows. For nonreacting flows, means and rms fluctuations of the velocity and mixture fraction fields are presented. For reacting flows, instantaneous temperature and composition data files are made available. Listings of the fluid mixtures used are given in:

Tables: CA and CB for Nonreacting Bluff-Body jets

Tables: RA and RB for Bluff-Body Stablised Flames Bluff-Body Burner

The burner is centred in a coflowing stream of air and generally consists of a circular bluff-body with an orifice at its centre for the main fuel. Bluff-body burners have been used with a range of bluff-body diameters, DB and fuel jet diameters, DJ, but in this report we consider the configuration characterised by DB = 50 mm DJ = 3.6 mm and fired with a mixture 50% CH4 and 50% H2. Stability characteristics of these flames are given in terms of the fuel jet velocity, uj and the coflow velocity, ue; and have been published elsewhere [1,10]. The length of the recirculation zone is about one bluff body diameter. It should be noted here that the addition of H2 to the CH4 fuel is intended to produce a recirculation zone which is clean from soot. In pure CH4 flames [8] the recirculation zone, generally, has soot or soot precursors which are convected downstream where they interfere with the Raman signals. Partially premixing the methane with air reduces the flame's tendency to soot and this may lead to a much cleaner recirculation zone.

### Initial and Boundary Conditions

The initial conditions (at the jet exit plane) and boundary conditions (in the coflowing stream) are described here. The mean mixture fraction, the mean velocities and the variances and covariances for the velocities are specified. The mean mixture fraction is taken to be one in the jet stream and zero in the air stream. Mixture fraction in the pilot stream is generally taken as stoichiometric unless otherwise specified. The measured initial conditions for the mean axial velocity, u and its rms fluctuations u' normalised with the mean velocity are tabulated. Initial Conditions for: Bluff-Body Stabilised Jets and Flames file: ICBBODY.dat. Modelers may chose to use the actual measured initial conditions as specified in the tables. This will ensure that the jet initial velocity and momentum are accounted for. Initial radial and circumferential velocities are taken as zero, v = w = 0. It is also assumed that the initial profiles of v' and w' are identical to those of u'.

The initial covariances have not been measured but may be taken as u'v' = v'u' = Ccv (u'2 * u'2)0.5, where Ccv is a constant equal to 0.5. All other covariances are zero.

An alternative to using the tabulated initial conditions is to use power law fits for the initial velocity profiles both in the jet (approximating fully developed turbulent pipe flow) and the coflow. Initial profiles for u' are speciefied as piece-wise linear. Sharp changes in velocity may be intentionally avoided for computational purposes. Although these specified velocities may deviate slightly from the measured ones, the momentum both in the jet and the coflow streams are within 2% of the momentum at the specified conditions. The net momentum deficit in the coflow stream due to the boundary layer is accounted for in the computations. Both u and u' are taken as zero on the bluff-body surface.

### Accuracy Considerations

There are a number of sources of error that have to be considered in evaluating the overall accuracy of laser-based, instantaneous measurements of species concentrations in flames. Only sources of error which may be influencing the data are mentioned here: (i) photon noise, (ii) interference error and (iii)spatial resolution. Photon noise is associated with the number of photons, n collected by a given detector at each laser pulse and it decreases proportionally to 1/n0.5. This noise is expected to become significant at species mole fractions less than a few percent. The interference error depends on the magnitude of the fluorescence or chemiluminescence interference with the measured species. The error due to spatial resolution is not considered to be substantial and is discussed further below. Other sources of error which are particular to the Raman system discussed here are (i) calibration drift due to the changes in laser line shape over the lifetime of the dye, (ii) shot-to-shot variation in laser line shape, and (iii)uncertainties in the temperature dependent bandwidth calibration factor f(T), especially for intermediate temperatures.

The signal to noise ratio (S/N) gives a measure of the combined shot-to-shot random errors which are primarily due to photon statistics. It should be noted that, for scalar measurements in uniform steady flows and flames, the ratio of the rms fluctuations to the mean is given by the inverse of the S/N ratio.

A measure of the S/N ratio can be obtained from the calibration data since these measurements are made in a uniform field of known temperature and species concentration. The reacting and nonreacting calibration data may be used here giving the variation of the S/N over a range of temperatures and species concentrations. Estimates of the S/N ratios obtained during experiments conducted between 1984 and 1992 are given in file STN92.dat. Since then the Raman-Rayleigh system has been improved by using better lasers and detectors.

Estimates of the S/N ratios obtained during experiments conducted in 1995 are given in file STN95.dat. S/N ratios are given for typical data samples. The results are tabulated versus temperature for the Rayleigh signal and versus species number density (molecules/cm3) for the Raman signals.

S/N ratios for data collected 1992 and before STN92.dat

S/N ratios for data collected 1995 and after STN95.dat

The general trend for the signal to noise ratios presented in stn*.dat is to increase with number density. A correlation of the form S/N = Ai * [i]0.5 produces an adequate fit for all the scalars shown in stn*.dat. Here Ai is a constant and [i] is the number density of species. This square root dependence on the number density implies that photon statistics is a major contributor to noise on all of the Raman signals.

Table 1 shows estimates of the percentage errors on various species for two typical samples collected in a CH4/H2 flame using the experimental setup. Lean and rich sample compositions are obtained from the actual data and are taken here as illustrations of typical measurement conditions. It is evident that the improvements made in 1995 have led to a significant reduction in the percentage error on all scalars. The percentage error increases with decreasing number density or mole fraction. It should be emphasised that the errors reported here do not include the effect of interferences and spatial resolution. Raman interferences affect only selected species and are believed to have a small contribution to the overall error. The fluorescence interference from soot precursors (mainly in the rich side of the flame) is very low in these flames and that improves the signal to noise ratio in all the affected Raman signals. Flames with high hydrocarbon fuels are most affected and among the Raman signals the CO line suffers the highest interference levels.

Sample Temperature Species Mass Fraction % Number Density % Error 1992 or Before % Error 1995 or After
Lean 1900 CH4 0.0 0.0 - -
O2 4.0 0.12x1018 17.0 10.0
N2 75.0 2.63x1018 5.0 0.8
CO2 8.0 0.18x1018 11.1 4.5
CO 2.0 0.07x1018 16.6 9.0
H2 0.5 0.23x1018 17.0 12.5
H2O 11.0 0.60x1018 7.1 5.0
Rich 1400 CH4 18.0 1.09x1018 10.0 2.3
O2 0.0 0.0 - -
N2 57.0 1.98x1018 6.3 1.1
CO2 5.5 0.12x1018 12.0 5.5
CO 5.5 0.19x1018 10.0 8.3
H2 2.5 1.22x1018 6.9 4.0
H2O 12.0 0.65x1018 7.2 4.0

Table 1 Estimates of percentage error on typical samples of data collected in a turbulent methane-hydrogen flame.

### Spatial Resolution

Spatial resolution issues for these data are discussed elsewhere [12]. The length of the measurement probe is 1mm and the diameter is about 0.6mm. Typical Kolmogorov length scales in the flame investigated range from 30 to 150 microns. Using estimates of the length scales, it is found that the spatial resolution error ranges from 3% to 16% depending on the axial location in the flame and on the jet velocity. More information about spatial resolution effects may be found in Mansour, M.S. et al. (Combust. Flame 82:411 (1990)).

A comprehensive set of experimental data is available for comparing velocity profiles, mixture fraction and species composition in the radial direction (r/D) for different axial positions (z). The experimental data available are listed below:

### Nonreacting Bluff-Body Jets

- Flowfield data: Mean and rms fluctuations of axial and radial velocities

- Mixing Field data (obtained from Rayleigh Imaging): Mean and rms fluctuations of mixture fraction

### Reacting Bluff-Body Stabilised Flames

Bluff-Body (50mm body, 3.6mm jet) CNG/H2(1/1)

- Flowfield data: Mean and rms fluctuations of axial and radial velocities

- Spontaneous Raman/Rayleigh/LIF data: Temperature and species mass fractions

## Velocity Measurements

### Description of Experiment

Two color LDV system with frequency-shifted beams are used to measure the horizontal and vertical velocity components. The fuel and the air are seeded in order to reduce the seeding bias. Uncertainties of the LDV technique are mainly associated with the seeding bias due to steep temperature gradients and the presence of more than one particle in the probe volume. The error due to seeding bias is very hard to quantify and is believed to be small. The error due to the presence of more than one particle in the measurement volume, however, is believed to be 4% for the mean and 7% for the RMS fluctuations. The flow field data are collected at the University of Sydney and consist of radial profiles of mean and RMS of fluctuations of the axial and radial velocity components at a range of axial locations. The flow field data are provided for selected jets and flames only.

Burner Description Fuel Mixture Data File CA. Flowfield data: Mean and RMS fluctuations of axial and radial velocities Bluff-Body(50mm body, 3.6mm jet) Air B4C1-S1.DAT Same case, Data set 2 B4C1-S2.DAT Same case, Data set 3 B4C1-S3.DAT

## Mixture Fraction Measurements

### Description of Experiment

Mixture Fraction Measurements

Measurements of mixture fraction in Nonreacting Bluff-Body jets are performed using planar imaging of Rayleigh scattering. These imaging experiments were performed at the University of Sydney. The instantaneous images had dimensions of 58x13x0.5 mm which covered the full width of the bluff-body flow. Three sets of 50 images were collected in the vicinity of the recirculation zone. The images were corrected for electronic and photon noise, background interferences and stripeness caused by the change of refractive indices which induce beam steering. The mixture fraction is estimated to have errors associated with electronic and photon noise of less than 5.7% and at best this error drops to 2.3%. These estimates do not account for any systematic errors that may be associated with these measurements. The results presented are in the form of mean and rms mixture fractions obtained at various locations within the imaged planes.

 Data File RA.Flowfield data: Mean and rms fluctuations of axial and radial velocities Bluff-Body (50mm body, 3.6mm jet) CNG/H2(1/1) Flame HM1: U-Fuel=118m/s, U-air=40m/s, 50% blowff B4F3-A-S1.dat Flame BB1: U-Fuel=108m/s, U-air=35m/s, 50% blowff B4F3-B-S1.dat Flame BB1: Data set 2 B4F3-B-S2.dat Flame BB2: U-Fuel=195m/s, U-air=35m/s, 90% blowff B4F3-C-S1.dat

## Temperature and composition Measurements

### Description of Experiment

Temperature and composition data are instantaneous measurements collected at the Combustion Research Facility, Sandia National Laboratories, Livermore CA. Measurements have been made using the Raman/Rayleigh/LIF technique to give instantaneous and simultaneous temperature and the concentration of many species at a single point in the flame. The species measured are: N2, O2, CH4 (or CH3OH), CO, CO2, H2, H2O. Other species such as OH and NO are measured for selected flames only. A range of fuel mixtures and flame velocities ranging from low to close to extinction have been studied. The following details give useful information about the processing and tabulation of the temperature and composition data:

1/ Each data file contains information about the axial and radial measurement location.

2/ The mixture fraction is obtained using the Bilger formula (Combust. Flame 80:135-149 (1990)) which is given by:

${\displaystyle z_{i}={{{2{\bigl (}Z_{c}-Z_{c,o}{\bigr )} \over W_{c}}+{2{\bigl (}Z_{h}-Z_{h,o}{\bigr )} \over {2W_{h}}}-{2{\bigl (}Z_{o}-Z_{o,o}{\bigr )} \over W_{o}}} \over {{2{\bigl (}Z_{c,f}-Z_{c,o}{\bigr )} \over W_{c}}+{2{\bigl (}Z_{h,f}-Z_{h,o}{\bigr )} \over {2W_{h}}}-{2{\bigl (}Z_{o,f}-Z_{o,o}{\bigr )} \over W_{o}}}}}$

The following values have been used in calculating the mixture fraction:

Total Mass fraction of Oxygen atoms in the air stream, Zo,o = 0.233

Total Mass fraction of Carbon atoms in the air stream, Zc,o = 0.0

Total Mass fraction of Hydrogen atoms in the air stream, Zh,o = 0.0

Total Mass fraction of Oxygen atoms in the fuel stream, Zo,f = 0.0

Total Mass fraction of Carbon atoms in the fuel stream, Zc,f = 0.6649

Total Mass fraction of Hydrogen atoms in the fuel stream, Zh,f = 0.3351

where Z(i) is a conserved scalar given by the total mass fraction of element (i), and Wi is the molecular weight of elements (carbon, c, hydrogen, h and oxygen, o). Subscripts (f) and (o) refer to the fuel and air streams, respectively.

Note: Values of Zo,o, Zh,o, Zc,o, Zo,f, Zh,f and Zc,f used to calculate zi are given for each fuel in the (*.dat) file in the relevant directory.

3/ Mixture fraction ranges from zero to one. Negative values of mixture fraction which may arise due to differential diffusion are not allowed. Users interested in differential diffusion effects can redefine and re-calculate their mixture fraction from the tabulated mass fractions.

4/ The argon contained in air is not accounted for and its mass fraction is lumped with that of nitrogen.

5/ Each file contains Favre and ensemble mean and RMS fluctuations for the data points contained in the file.

6/ Temperature is obtained either from the Rayleigh signal or from the sum of the species number densities (assuming a mixture of ideal gases).

7/ The percentage mass fraction of species (Y(i)*100) is tabulated except for NO where the percentage mass fraction*100 (Y(NO)*10000)) is given. If the mass fraction of a species is zero for the entire data set in a given directory it means that the species have not been measured. Symbol h-c refers to the parent hydrocarbon fuel (CH4, CH3OH, etc...)

8/ The factor TNDR is defined as:

TNDR = sum of species number densities measured from Raman and LIF over the total number density obtained from the Rayleigh temperature, or equivalently:

TNDR = Temp. from Rayleigh / Temp. from sum of species number densities.

TNDR = 0.0 implies that temperature is obtained from the sum of species number densities.

Otherwise temperature is obtained from Rayleigh. Generally, temperature is obtained from the Rayleigh measurements except in cases where we suspect that the Rayleigh signal is corrupted by Mie scattering. This is the case, for example, in flames of methanol where there may be scattering from fine droplets. Also, measuring in regions of the flames where there are solid particles or soot particles corrupt the Rayleigh signals.

9/ All the measured mass fractions are normalised such that the sum of the tabulated mass fractions equals one.

10/When the Rayleigh temperature is used (the factor TNDR is not equal to zero) the original measured, non-normalised species mass fractions may be recovered from the tabulated data as follows:

Y(i, original) = TNDR * Y(i, normalised, tabulated) where i corresponds to any of the tabulated species.

11/Departure of the total mass fraction of the measured species from unity is only partly due to the fact that not all species existing in the probe volume are being measured. There are random and systematic sources of error on the measured signals leading to differences between the temperature obtained from Rayleigh and that obtained from the sum of species number densities. A later section on Accuracy Considerations gives more details about these errors. In cases where the TNDR factor varies significantly from 1.0 the following guidelines are given:

TNDR > 1.0 This generally implies that the species mass fractions are affected by error which remains uncorrected for and which may be due to various kinds of interference. This leads to artificially high concentrations and hence a lower temperature from the sum of species number densities. An estimate of this error on each species is extremely difficult to quantify but a guide to the expected error is given later in the section on Accuracy Considerations.

TNDR < 1.0 This generally implies that the Rayleigh signal is subject to interference due to Mie scattering leading to artificially lower Rayleigh temperatures and hence a lower value of TNDR.

The length of the recirculation zone is about one bluff body diameter. It should be noted here that the addition of H2 or CO to the CH4 fuel is intended to produce a recirculation zone which is free of soot. In pure CH4 flames [8] the recirculation zone, generally, has soot or soot precursors which are convected downstream where they interfere with the Raman signals. Partially premixing the methane with air reduces the flame's tendency to soot and this may lead to a much cleaner recirculation zone. Methanol fuel is extremely useful in this regard since it produces a clean recirculation zone without the need of premixing with another fuel.

### Axisymmetric Bluff Body Turbulent Flame

Code name: B4F3

Wind Tunnel Dimension: 0.305m x 0.305m

Jet Diameter: 0.0036m

Bluff Body Diameter: 0.05m

Fuel mixture: H2-CH4 = 50-50 (% by vol.)

Stoichiometric Mixture Fraction: 0.050

Blowoff velocity: 235m/s

Wind tunnel air velocity: 40m/s

Open flame, No enclosure

Burner Description Fuel Mixture Input Data File RB. Spontaneous Raman/Rayleigh/LIF data: Temperature and species mass fractions Bluff-Body(50mm body, 3.6mm jet) CH4/H2(1/1) B4F3.dat (1995)

Data files provided for this fuel mixture:

x/D r(mm) Filename Velocity 118 m/s 0.3 -27.2 a013.txt 0.6 -26.3 a030.txt 0.9 -27.3 a045.txt 1.3 -26.4 a065.txt 1.8 -27.4 a090.txt 2.4 -27.4 a120.txt

Table 3 Position measurements and data files for species composition measurements

The following table summarizes the experimental data

NAME GNDPs PDPs MPs
Nonreacting Bluff-Body Jets Re Airflow velocity m/s Bluff-Body Geometry Detailed Data DOAPs
EXP 1 (Mean and rms fluctuations of axial and radial velocities) 1.4x105 40.0 Dout=50 mm Djet=3.6 mm Uz,Vr
Reacting Bluff-Body Stabilised Flames Re Airflow Velocity m/s Bluff-Body Geometry Detailed Data DOAPs
EXP 3 (Temperature and species mass fractions) 1.4x105 40.0 Dout=50 mm Djet=3.6 mm Z, N2, O2, CH4 (or CH3OH), CO, CO2, H2, H2O

Table EXP - A Summary description of all test cases
MP 1

U V W (ms-1)

MP 2

k (m2s-2)

MP 3

Z (Mass fraction)

DOAPs, or other miscellaneous data
EXP 1 X X X -
EXP 2 ü ü ü -
EXP 3 ü ü ü B4F3.dat

Table EXP-B Summary description of all measured parameters and available datafiles

The complete experimental database is reported at the web site: http://www.aeromech.usyd.edu.au/thermofluids/main_frame.htm

## References

Masri, A.R. and Bilger, R.W., 1985, Turbulent Diffusion Flames of Hydrocarbon Fuels Stabilised on a Bluff Body', Twentieth Symposium (International) on Combustion. The Combustion Institute, Pittsburgh, pp. 319-326.

Masri, A.R., Dally, B.B., Barlow, R.S. and Carter, C.D., 1994, The Structure of The Recirculation Zone of a Bluff-Body Combustor', Twenty-fifth Symposium on Combustion, The Combustion Institute, Pittsburgh, pp.1301-1308.

Masri, A.R., Dibble, R.W. and Barlow, R.S., Raman-Rayleigh Measurements in Bluff Body Stabilised Flames of Hydrocarbon Fuels', Twenty-fourth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1992, pp. 317-324.

Masri, A.R., Dally, B.B., Barlow, R.S. and Carter, C.D., The Structure of The Recirculation Zone of a Bluff-Body Combustor', Twenty-fifth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1994, pp.1301-1308.

Masri, A.R., Dibble, R.W., and Barlow, R.S., The Structure of Turbulent Nonpremixed Flames Revealed by Raman-Rayleigh-LIF Measurements', Prog. Energy Combust. Sci., 22:307-362 (1997).

Dally, B.B, Masri, A.R., Barlow, R.S., Fiechtner, G.J., and Fletcher, D.F., Measurements of NO in Turbulent Nonpremixed Flames Stabilised on a Bluff Body', Twenty-sixth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1996, Vol. 2, pp.2191-2197.

Adami, P. and Martelli, F., “Computation of 3D Turbulent Not-premixed Reacting Flows Using an Implicit Unstructured Solver” IGTI ASME TURBO EXPO '2000, Munich, Germany.

PROCEEDINGS from:

4th TNF Workshop - Darmstadt 1999

3rd TNF Workshop - Boulder 1998

2nd TNF Workshop - Heppenheim 1997

1st TNF Workshop - Naples 1996

2.3 Test case EXP2

(as per EXP1)