Laminar Burning Velocities Of Hydrogen Oxygen Nitrogen Mixtures

Published Date: 02 Nov 2017

R.T.E. Hermanns, R.J.M. Bastiaans, and L.P.H. de Goey

Department of Mechanical Engineering, Eindhoven University of Technology,

Eindhoven, The Netherlands

Published in : Proceedings of the European Combustion Meeting 2003

Editors: C. Chauveau and C. Vovelle

Place: Orleans, France

Pages: paper 121, (2003)

Corresponding Author:

R.T.E. Hermanns

WH 3.142, Den Dolech 2, PO BOX 513

5600 MB Eindhoven, The Netherlands

Tel : +31 40 2475995

Fax : +31 40 2433445

Email: [email protected]

1

Laminar Burning Velocities of Hydrogen-Oxygen-Nitrogen Mixtures

R.T.E. Hermanns_, R.J.M. Bastiaans and L.P.H. de Goey

Section Combustion Technology, Department of Mechanical Engineering

Eindhoven University of Technology

P.O. Box 513, 5600 MB Eindhoven, The Netherlands

Abstract

Experimental results of the adiabatic burning velocities in hydrogen-oxygen-nitrogen mixtures are presented. Non-stretched

flat flames were stabilized on a perforated burner at atmospheric pressure. The oxygen content in the oxidizer varies between

0.07 to 0.10 molar percent. The heat flux method was used to determine the burning velocities under conditions at which the

net heat loss of the flame to the burner is zero. Comparing the presented results with detailed combustion reaction mechanisms,

the numerical data show a lower laminar burning velocity over the entire range. Also a large variation is found when comparing

our results with experimental data of other authors.

Introduction

In a future sustainable society, hydrogen will play

an important role as an energy carrier. Contrary to the

oxidation of fossil fuels, at favorable burning conditions,

water is the only reaction product and no emissions

like CO2 will occur. The development of a hydrogen

economy is an important step forward in the worldwide

effort to reduce CO2 emissions and, with it, the

greenhouse effect. In a Dutch research project ”Greening

of Gas”, a gradual transition is studied of employing

the Dutch natural gas infrastructure in which an additional

hydrogen supply will be introduced. One of

the crucial parameters to the safety of burner devices is

the laminar burning velocity. This property determines

for instance the limits of flash-back and lift-off. In the

present research we measured the adiabatic burning velocity

of hydrogen-oxygen-nitrogen mixtures. To measure

this property the heat flux method [1] has been developed

in the combustion group of the Technische Universiteit

Eindhoven in recent years. This method has

been proven to be very accurate. Additionally, the laminar

burning velocity is an important parameter needed

for the understanding of combustion, because it contains

fundamental information of reactive and diffusive

properties of a mixture.

In this paper results are presented of measurements

of the burning velocity of mixtures of hydrogenoxygen-

nitrogen. The oxygen content, O2=(O2 + N2),

was varied between 0.07 and 0.10 molar percent. The

equivalence ratio was varied between 0.7 and 1.0. The

results are compared to other experimental data and to

numerical calculations using several detailed chemical

kinetics schemes. In this paper, first the experimental

method is discussed with an analysis of the errors. Then

the numerical method is specified shortly. In the next

section the results are presented and discussed and the

paper ends with some conclusions.

_Corresponding author: [email protected]

associated website: http://www.combustion.tue.nl

Proceedings of the European Combustion Meeting 2003

Heat Flux Method

Originally, Botha and Spalding [3] had the idea for

determining the burning velocity using the heat loss

necessary to stabilize the flame as a measure. They

measured the temperature increase of the water for

cooling the burner. In 1993, de Goey et al. [6] introduced

a perforated plate burner, figure 1, where a flame

is stabilized on a brass plate of 2 mm thickness, and

30 mm in diameter. The burner plate is perforated with

a hexagonal pattern of small holes. With an appropriately

chosen perforation pattern, it can be shown that

flames stabilized on this burner remain flat [5]. The hole

diameter is determined by the flow velocity for which

the burner will be used. A larger flow requires smaller

holes for the flow to become uniform before reaching

the flame. The burner plate used in the current research

is perforated with holes of 0.5 mm and a pitch of 0.7

mm as is shown in figure 1b.

Eight copper-constantan thermocouples of 0.1 mm

in diameter are glued into holes of the perforated plate.

The thermocouples are positioned at different radii

and different angles to measure the temperature profile

across the burner plate. These thermocouples are positioned

at the following pares of radius-angle combinations

0 (0°), 2.8 (330°), 4.9 (150°), 7.7 (270°), 9.1 (30°),

10.5 (90°), 12.6 (210°), 14.7 (330°). Radiation correction

for the thermocouple readings was neglected because

the measured temperature of the burner plate was

always below 400 K. The improved perforated plate

burner which is used in the present research has been

extensively tested in recent years, [1, 2, 11]. It is shown

in figure 1.

The burner head has a heating jacket supplied with

thermostatic water to keep the temperature of the burner

plate constant. During experiments this temperature

was kept at _360 K. The plenum mixing chamber has a

separate cooling system supplied with water of 298 K.

Thus the heating jacket keeps the burner plate edges at

a certain temperature higher than the initial gas temperature,

which causes the unburned gas mixture to heat

up when flowing through the burner plate. By doing so,

the heat loss necessary for stabilizing the flame can be

compensated by the heat gain of the unburnt mixture,

2

burner plate

mixing chamber

heating

jacket

thermocouples

r

x

cooling

jacket

(a) The heat flux burner

perforation pattern :

d = 0.5 mm

l = 0.7 mm

l

d

(b) Top view of the burner showing the perforation pattern of the

burner plate

Figure 1: The heat flux setup

leading to a stabilized adiabatic flame.

In [5] it is shown that the axial temperature profile

of the stabilized adiabatic flame only deviates around

the burner plate and is unaltered elsewhere. Because

of the low burner plate temperature, lower than 400 K,

the flame structure above this value will not be significantly

influenced by the insertion of the burner plate.

Also due to the low burner plate temperature it is believed

that radicals don’t react with the burner plate.

Furthermore Van Maaren et al. [12] showed that the

stretch rate is negligibly small _1 sô€€€1.With this burner

it is therefore possible to create a flat, almost stretchless,

adiabatic laminar flame. It should be emphasized

that the flame is not perfectly one-dimensional but flat,

which is the best experimental approximation of a onedimensional

flame. To find the laminar adiabatic burning

velocity with the heat flux method there is no need

to extrapolate experimental results. In fact the adiabatic

burning velocity is found by interpolation, since flames

where the unburnt gas is slightly higher (typically a few

cm/s) than the adiabatic burning velocity still remain

flat and can stabilize on this burner. The flow velocity

and the gas mixture were regulated by carefully recalibrated

mass flow controllers (MFC) to correct for any

nonlinear effect in the devices and electronic circuits.

In this way, the uncertainty will be _1% per MFC, provided

the MFC is used in a range above 10% of its maximum

flow rate.

Typical Measurement

Typical examples of temperature profiles (markers)

in the burner plate are given in figure 2, where

H2-O2-N2 flames with _ = 0:95 and O 2=(O2 + O2) =

0:107 are stabilized, having different velocities around

the adiabatic velocity of 52.3 cm/s. The gas velocities

are in the range of 51 to 54 cm/s.

Each of the temperature profiles in figure 2 can be

fitted to a parabolic function T = _2r2 + Tc, where r

is a radial position of each thermocouple. The fit yields

two parameters, the temperature in the center of the

burner plate Tc and the parabolic coefficient _2. This

parabolic fit can be used due to the assumption of rotational

symmetry of the temperature distribution in the

burner plate, by measuring the temperature profile with

different gas velocities. The adiabatic burning velocity

can be interpolated to a uniform radial temperature distribution

in the burner plate. In this case the coefficient

_2 is equal to zero.

Error Estimation

In [2] Bosschaart and de Goey showed a thorough

analysis of the error estimation when using the heat flux

method to measure the adiabatic burning velocity SL.

The resulting errors in _ and SL are summarized here.

To estimate the resulting error in the equivalence ratio,

a straightforward analysis using partial derivates is performed,

and leads to:

__

_

=

_qH2

qH2

+

_qO2

qO2

+

_qN2

qN2

: (1)

The uncertainties for qH2 , qO2and qN2 were estimated

from deviations during calibration. Typically an un-

3

0 5 10 15

72

74

76

78

80

82

84

86

88

Radial position (mm)

Temperature (ºC)

54.0 (cm/s)

53.0

52.0

51.5

51.0

Figure 2: Typical temperature profiles in the burner

plate, when a H2-O2-N2 flame with _ = 0:95 and

O2=(O2 + N2) = 0:107 is stabilized. The gas velocities

range from 51.0 to 54.0 cm/s. Markers are measurements

and dotted lines are fitted parabolae.

certainty of _q = 5 _ 10ô€€€3 _ Qmax was observed. This

leads to typically uncertainties of _1% for equivalence

ratios around 1.0. For other equivalence ratios the error

becomes larger because the relative error of the MFCs

increases when producing smaller flows. Therefore it

is important to choose appropriate dimensions for the

MFCs to keep the error small.

According to Bosschaart and de Goey the error of

the burning velocity due to the scatter in the thermocouples

can be estimated by:

_SL =

1

s

2_tc

r2

b

: (2)

where rb equals the position of the thermocouple on the

edge of the burner, e.g. 14.7 mm, _tc is the remaining

thermocouple scatter and is typically 0.5 K. The sensitivity

s is defined as the variation of the parabolic parameter

_2 in the case of a change in the gas velocity

Ugas,

s =

d_2

dSL

____

_2=0

: (3)

This quantity is determined during the interpolation

procedure. In figure 3 the parabolic coefficients of _2

are plotted against the gas velocity. The sensitivity is

calculated with equation 3 at a gas velocity equal to the

laminar burning velocity.

Numerical specification

The experiments performed are very important to validate

numerical results. Therefore numerically obtained

data with state-of-the-art methods are included as a reference.

To that end, the burning velocity of freely propagating,

steady, adiabatic premixed flames are computed

by using the CHEM1D code [4]. This code has

the EGlib libraries [8] included in it to have detailed

51 51.5 52 52.5 53 53.5 54

−0.02

−0.015

−0.01

−0.005

0

0.005

0.01

0.015

0.02

0.025

0.03

Ugas (cm/s)

2 (K/mm2)

Measurements

Fit

SL = 52.338 (cm/s)

Figure 3: Interpolation to the adiabatic burning velocity

by plotting the parabolic parameter _2 as a function

of the gas velocity. This mixture of H2-O2-N2 and

_ = 0:95 and O2=(O2 + N2) = 0:107 gives a laminar

burning velocity of 52.3 cm/s.

multicomponent diffusion transport modeling available.

Enthalpy fluctuations due to species diffusion (Sorret

effect) and species diffusion due to a temperature gradient

(Dufour effect) are taken into account. The calculations

are performed with an unburnt mixture temperature

of 298 K at ambient pressure. Two detailed reaction

mechanisms were used for modeling. They include the

commonly used GRI-mech 3.0 [9] which is optimized

for natural gas combustion and the hydrogen-oxygennitrogen

mechanism of Marinov et al. [10].

Results and discussion

Experimental and numerical results of the laminar

burning velocities with varying oxygen content in the

oxidizer stream are presented in figure 4. When comparing

the experimental results of Egolfopoulos and

Law [7] with the heat flux method, the latter one gives

lower laminar burning velocities for an oxygen fraction

up to 0.09, whereas above an oxygen content of 0.09

the results of Egolfopoulos and Law give lower results

than the presented heat flux measurements. The largest

difference between the experimental results in the measurement

range is around an oxygen content of 0.08

where it is approximately 6 cm/s. It should be emphasized

that the discrepancy between the measurements of

Egolfopoulos and Law and the heat flux method are outside

the error estimate. A significant deviation in laminar

burning velocities can also be seen when comparing

the numerical results of the GRI-mech 3.0 and Marinov

et al. combustion reaction mechanisms with the measurements

performed with the heat flux method. The

trend of the combustion reaction mechanisms compared

with the presented heat flux measurements is comparable,

which is not the case for the measurements of

Egolfopoulos and Law in the measured range. Higher

laminar burning velocities are found with the heat flux

4

0.07 0.08 0.09 0.1

0

10

20

30

40

50

60

O2/(O2+N2) (−)

SL (cm/s)

Heat flux method

Egolfopoulos (1990)

Marinov

GRI−mech 3.0

Figure 4: Heat flux laminar burning velocity results with error bars. The presented results are at an equivalence

ratio of 1.058 with different oxygen fractions in the oxidizer stream. The measurements are performed with a

gasflow temperature of 298 K and at ambient pressure.

method, typically 6 to 10 cm/s, over the entire measured

range when compared to the numerical calculations performed

with the GRI-mech 3.0 and the Marinov reaction

mechanisms.

The results of the laminar burning velocities with

varying equivalence ratio are presented in figure 5. In

this case also the experimental results of Egolfopoulos

and Law [7] are compared with the experimental results

of the heat flux method. The heat flux method gives

lower laminar burning velocities compared to the measurements

of Egolfopoulos and Law for equivalence ratios

lower than 0.9, whereas above an equivalence ratio

of 0.9 the heat flux method gives higher velocities than

the presented results of Egolfopoulos and Law. At an

equivalence ratio of 0.7 the difference in laminar burning

velocity is 10 cm/s. At higher equivalence ratios

the disagreement becomes smaller, but above an equivalence

ratio of 0.9 the disagreement increases again.

Also, in this case with varying equivalence ratio the inconsistency

between the measurements performed with

the heat flux method and the measurements of Egolfopoulos

and Law is outside the error estimate performed

for the heat flux data. When comparing the experimental

results with numerical data of the GRI-mech

3.0 and Marinov combustion reaction mechanisms, the

heat flux method gives over the whole measured range

a larger laminar burning velocity, approximately 6 cm/s

at _ = 0:7 and increasing to 11 cm/s at _ = 0:95.

The average error estimate in the laminar burning

velocity with the heat flux method is less than 1 cm/s in

the case of varying equivalence ratios, 95% confidence

interval. Only at very lean equivalence ratios the error

becomes larger, up to 4 cm/s, due to fact that all of the

MFCs where used in a low working range _ 10%. The

error in the equivalence ratio is less than 0.025 over the

whole measured range. The average error in the laminar

burning velocity with varying oxygen content is

less than 1 cm/s over the whole measured range. In

this case the error estimate in oxygen content is less

than 0.0025 over the entire measured range. A note

concerning the reaction mechanisms is that a dedicated

hydrogen-oxygen-nitrogen mechanism as presented by

Marinov [10], give larger differences with the measurements

than a low hydrocarbon mechanism, GRI-mech

3.0 [9], as used in present study. Both mechanisms

predict lower laminar burning velocities than the experimental

results which are presented here with the

heat flux method and results of other authors, like Egolfopoulos

and Law [7].

Conclusions

In the present investigation the laminar burning velocity

of hydrogen-oxygen-nitrogen mixtures has been

measured. A number of assumptions behind this

5

0.6 0.7 0.8 0.9 1 1.1

0

10

20

30

40

50

60

(−)

SL (cm/s)

Heat flux method

Egolfopoulos (1990)

Marinov

GRI−mech 3.0

Figure 5: Heat flux laminar burning velocity results with error bars. The presented results are at O 2=(O2 +N2) =

0:107 with different equivalence ratios. The measurements are performed with a gasflow temperature of 298 K

and at ambient pressure.

method of nonstretched adiabatic flame stabilization

have been validated in recent years at the Technische

Universiteit Eindhoven. Furthermore a thorough error

estimation has been presented by Bosschaart et al.

[2] recently. The measurements presented here give an

overall error in the laminar burning velocity which can

be estimated to be smaller than _ 1 cm/s, with a 95%

confidence interval, in the case of varying equivalence

ratios and smaller than _ 4 cm/s with varying oxygen

content, due to fact that MFCs were used in a very low

working range _10%. The error estimate in equivalence

ratio is less than 0.025 over the entire measured

range.

Comparing the heat flux measurements with experimental

results show a large deviation over almost

the entire measured range, only at _ = 0:9

and at O2=(O2 + N2) = 0:09 the differences are negligible.

Subsequently we have compared the experimental

results with detailed computations, using

GRI-mech 3.0 [9] and Marinov [10] combustion reaction

mechanisms. These mechanisms show a lower

laminar burning velocity, typically 5 to 10 cm/s over

the entire measured range.

6