This article was published in an Elsevier journal. The attached copyis furnished to the author for non-commercial research and

education use, including for instruction at the author’s institution,sharing with colleagues and providing to institution administration.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copy

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/ijhydene

Laminar burning behaviour of biomassgasification-derived producer gas

C. Serranoa,�, J.J. Hernandeza, C. Mandilasb, C.G.W. Sheppardb, R. Woolleyb,1

aE.T.S. Industriales, Universidad de Castilla-La Mancha, Camilo Jose Cela s/n, 13071 Ciudad Real, SpainbSchool of Mechanical Engineering, University of Leeds, Leeds LS2 9JT, UK

a r t i c l e i n f o

Article history:

Received 28 October 2007

Accepted 30 October 2007

Available online 4 December 2007

Keywords:

Producer gas

Burning velocity

Stretch

Cellularity

a b s t r a c t

In the currently reported work, a mixture of H2, CO and N2 (21:24:55 vol%) has been

considered as representative of the producer gas coming from gasification of lignocellulosic

biomass. Laminar burning velocities have been determined, with simultaneous study of

the effects of flame stretch rate and instabilities. Experimentally determined laminar

burning velocities derived from schlieren flame images, over a range of equivalence ratios,

have been compared with those determined using the CHEMKIN code. Good agreement

obtained for 1 bar flames, but significant differences were observed for high pressure

cellular flames. Markstein numbers were also derived from the experimental data and

corresponding Lewis numbers were calculated. Hydrogen thermo-diffusive effects tended

to destabilise lean flames, while the CO content resulted in laminar burning velocity

peaking at very high equivalence ratios. The peak burning rate of producer gas proved

faster than those of conventional fuels, such as isooctane and methane.

& 2007 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights

reserved.

1. Introduction

Recent European energy policies [1] strongly encourage the

use of biomass in order to address three targets: diversifica-

tion of energy supply, reduction of CO2 emissions and

contribution to rural development. One of the main advan-

tages of gasification is the possibility of installing small, low-

cost and efficient gasifier-engine plants. These enable use of

the biomass close to source and so elimination of most of the

biomass waste storage and transportation costs. In this

manner, biomass gasification constitutes an attractive option

and an alternative to direct combustion. A low-energy-

content gas is generated in the gasification process through

an oxygen deficient reaction; the so-called ‘producer gas’ is an

H2 and CO rich fuel, which also comprises N2, CO2 and small

quantities of CH4 and H2O. Previous studies have shown that

the autoignition delay time of producer gas is longer than that

of isooctane at temperatures below 1100 K [2]; in fact the

knock tendency of a spark ignition engine fuelled with

producer gas is quite low, yielding good engine performance

at very high compression ratios [3,4]. Although the calorific

value of producer gas is ten times lower than that of natural

gas, its stoichiometric fuel/air ratio is ten times higher (it

needs less air to burn). Thus the energy density of the fuel–air

mixture is similar for the two fuels [5] and the loss of power

(cf. natural gas) is low, especially under lean conditions. With

regard to pollutant emissions: the low adiabatic flame

temperature of producer gas helps in limiting NOx produc-

tion, whilst its hydrogen content assists reduction in parti-

culate and unburned hydrocarbon emissions [6,7]. These

properties render this alternative fuel attractive for mechan-

ical or electrical energy production in internal combustion

ARTICLE IN PRESS

0360-3199/$ - see front matter & 2007 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.doi:10.1016/j.ijhydene.2007.10.050

�Corresponding author. Tel.: +34 676753120.E-mail address: Clara.Serrano@uclm.es (C. Serrano).1 Current address: Department of Mechanical Engineering, University of Sheffield, UK.

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2

Author's personal copy

engines, or in external combustion systems such as Stirling

engines, gas burners and micro-turbines [8]. However, engine/

burner design and the operating parameters must be

optimised to suit the producer gas composition and its

thermo-chemical properties. Laminar burning velocity (ul) is

one of the most important properties governing the combus-

tion behaviour of a fuel. Measurement of laminar burning

velocities is also important to the development and validation

of chemical kinetic mechanisms of unconventional fuels, as

well as for development of predictive models to estimate the

performance and emissions of combustion equipment, and to

prevent and control possible explosion hazards.

Laminar burning velocity is classically defined in relation to

one-dimensional, steady and unstretched flames. These are

also the assumptions typically adopted in most theoretical

combustion models, such as the CHEMKIN code [9] adopted in

the currently reported study, for the calculation of theoretical

laminar burning velocities. However, for experimental spheri-

cally expanding flames similar to those in the combustion

chamber of an SI engine or in the combustion bomb described

below, flame curvature and aerodynamic strain cause the

premixed flame front to become stretched. Differences

between experimentally observed laminar burning velocities

and ideal planar and theoretical unstretched values may be

attributed to the stretch rate experienced by ‘real’ laminar

flames. Additionally, instability resulting from hydrodynamic

disturbances associated with thermal expansion of the burnt

gas may wrinkle and break the flame front into small

combustion cells, causing an increase in effective flame front

area, and associated acceleration of the combustion process

[10]. This phenomenon, very prevalent at high pressures and

particularly for fuels containing significant concentrations of

hydrogen, is usually known as cellularity, which might be

thought of as an intermediate regime between laminar and

turbulent combustion. Much of the variation in values of

laminar burning velocity reported in the literature, generated

using different experimental and theoretical methods, can be

associated with neglect of flame stretch rate and cellularity

effects.

The non-dimensional Markstein number (Ma) is another

important parameter for a burning mixture, characterising its

flame stability and the response of its laminar burning

velocity to stretch rate effects. A fuel’s Markstein number

decreases with pressure and is strongly related to its Lewis

number (Le), Prandtl number and the activation energy of the

combustion process [11]. In non-equidiffusive mixtures,

flame stability is also affected by thermo-diffusive effects,

characterised by Lewis number (defined as the mixture’s

thermal diffusivity divided its mass diffusivity). An effective

Le for a reacting mixture has been defined by Matalon et al.

[12], although for non-stoichiometric mixtures it approaches

the Lewis number based on the deficient species. A flame

having Leo1 may show thermo-diffusive instability, whereas

for Le41 thermo-diffusive effects tend to stabilise the flame

front [13].

Although laminar burning velocities have been experimen-

tally determined for a wide range of pure conventional fuels

(e.g. isooctane [14], propane [15] and methane [16]), there is a

dearth of reliable corresponding data for alternative and

renewable fuel/air mixtures; this is particularly so for H2/CO

rich fuels such as the producer gas coming from biomass

gasification. Huang et al. [17] have determined the burning

velocity of mixtures of primary reference fuels and H2/CO/N2

mixtures for a counterflow burner at room pressure and

temperature (1 bar and 298 K), using digital particle image

velocimetry. They showed that the laminar burning velocity

of isooctane increased with the addition of H2/CO/N2. Han

et al. [18] studied the effect of adding simulated reformer gas

(22.1% H2–7.4% CO, by volume) to a methane/diluent mixture.

They employed a cylindrical combustion vessel and pro-

cessed combustion event pressure records to show an

increase in burning velocity and reduction in equilibrium

NO concentration with increasing reformer gas concentra-

tion. Hassan et al. [19] analysed shadowgraph flame images

for a range of H2/CO/air mixtures, with H2 concentration in

the fuel mixture up to 50 vol%. They suggested that, for low H2

concentration, the effect of stretch rate was insignificant and

that laminar burning velocity peaked at an equivalence ratio

close to two. A number of other workers have provided

theoretically derived values of laminar burning velocity for

such fuels on the basis of chemical kinetic modelling of the

flame front. Sung et al. [20] used the CHEMKIN code in

conjunction with their own reaction mechanism to show an

increase in laminar burning velocity of n-butane and iso-

butane with the addition of reformer gas (30% H2–25% CO–45

%N2, by volume). Calculations were made for combustion

pressures of up to 20 atm. and the separate effects of H2 and

CO were investigated; these showed that H2 led to both an

ARTICLE IN PRESS

Nomenclature

A flame front area

Dij mass diffusivity

Lb Markstein length

Le Lewis non-dimensional number

Ma Markstein number

p pressure

r flame radius

Sn stretched laminar flame speed

Ss unstretched laminar flame speed

t time

T temperature

ul laminar burning velocity

Greek symbols

a stretch rate

at thermal diffusivity

dl laminar flame thickness

f equivalence ratio

rb burned gas density

ru unburned gas density

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2852

Author's personal copy

increase in radical concentrations and mixture flame tem-

perature, whilst laminar burning velocity enhancement

associated with CO was mainly due just to the second effect.

Sridhar et al. [21] and Hernandez et al. [22] have provided

theoretically derived laminar burning velocity data for

producer gas, correlated in a relationship of the form

advanced by Metghalchi and Keck [23]. Both studies were

carried out for a wide range of producer gas compositions and

have shown that the CO and H2 contents in producer gas

greatly affect the magnitude of the laminar burning velocity

(ulÞ. In Ref. [22], computed values of ul at 1 bar were compared

to experimental results derived from instantaneous pressure

records generated by deflagrations in a non-optically accessed

200 mm-diameter combustion vessel. Differences between

experimental and theoretical results were possibly associated

with neglect of flame stretch and cellularity. Further insight to

the intrinsic burning properties of producer gas flames is still

needed, as a contribution to the development of more

efficient technologies based on this alternative fuel. The

currently reported work contributes to this. The laminar

combustion behaviour of a mixture comprising 21% H2, 24%

CO, 55% N2 (vol%), representative of the producer gas

generated during the gasification of biomass wastes [24,25],

has been examined with the following objectives:

� To obtain experimental ul values from schlieren images of

the flame over a range of pressures and equivalence ratios,

and to compare those values with those coming from

theoretical modelling.

� To study the effect of stretch rate and cellularity on the

burning rate of producer gas, both at atmospheric and

elevated pressure.

� To compare laminar burning velocities of the producer gas

with those previously found for other fuels (methane,

hydrogen, isooctane) and with those reported for other

H2/CO/N2 mixture compositions.

2. Theoretical and experimental tools

2.1. The CHEMKIN software

The CHEMKIN 4.0 software consists on a set of applications

for solving a range of chemical kinetic problems. The Flame

Speed Calculator application was used to compute laminar

burning velocities for a steady, adiabatic, one-dimensional

and unstretched premixed flame. The code used a hybrid

time-integration and steady-state algorithm to solve compre-

hensive mass species and energy conservation equations. To

enhance the convergence properties, a good starting estimate

of the temperature profile through the flame thickness was

required. Following an earlier study [22], this was accom-

plished by adopting a sigmoid function; with upper limit

above the adiabatic flame temperature, to minimise conver-

gence failures. The problem was solved in an iterative way,

with convenient coarse-to-fine grid refinement and mesh

adaptation parameters at every iterative step, until the

converged solution was found.

The CHEMKIN code has been programmed to evaluate the

thermodynamic and diffusive properties of a multi-compo-

nent mixture and to process chemical reaction mechanisms.

In the currently reported work, the GRI-Mech Version 3 [26]

detailed chemical kinetic mechanism, comprising 325 ele-

mentary reactions and 53 species, has been employed. This

well-known mechanism for methane oxidation has been

verified against more than 60 experimental studies, including

unstretched laminar burning velocity data for CH4/air and

H2/CO/air mixtures.

Covered in the currently reported study are experiments for

a range of lean (f ¼ 0:8) to rich (f ¼ 1:8) equivalence ratios for

three different initial pressure values (1, 2.5 and 5 bar). Since

temperature is unimportant to the hydrodynamics and

morphology of the flame front [14,27], all tests were carried

out for unburned gas at room temperature (295 K). These

same conditions were adopted for the theoretical calculations

in order to assess the effects of flame stretch (well defined at

low pressure) and cellularity (appearing at high pressure) on

the experimentally determined values of laminar burning

velocity.

2.2. Experimental set-up

The experiments were conducted using the Leeds Mk II

combustion vessel, with central spark ignition (Fig. 1). This

high optical-access, spherical, 380 mm-diameter, stainless

steel bomb had three pairs of orthogonally opposed windows

of 150 mm-diameter and was equipped with four fans driven

by electrical motors. These fans were designed for the study

of turbulent flames [14], but for the current laminar experi-

ments were just used to homogenise the producer gas/air

mixture, being switched off 2 min prior to ignition to allow

any mixture motion to decay. The gases comprising the fuel

mixture were injected separately into the bomb to the partial

pressures appropriate to the required equivalence ratio, using

a Druck PDC 081-0499 absolute pressure transducer. Dry air

ARTICLE IN PRESS

Fig. 1 – Optically accessed spherical combustion bomb.

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2 853

Author's personal copy

from a cylinder was then added until the required total

pressure was reached.

During the combustion, pressure was measured with a

Kistler 701 transducer, resulting in pressure graphs similar to

that shown in Fig. 2. The propagating flames were recorded by

schlieren cine photography using a Photosonics Phantom 9

digital camera, which ran at a framing rate of 2500 frames/s

with a resolution of 768� 768 in all the experiments. The

leading edge of the flame front visualised by the schlieren

technique corresponded to that of the first significant change

in density gradient/refractive index associated with rise in

temperature from that of the unburned gas [28]. The ‘flame

radius’ was calculated as that of a circle encompassing the

same area as that enclosed by the schlieren imaged flame.

Several experiments were performed at each condition in

order to check the repeatability of the results. Following an

experiment, the vessel was evacuated and flushed twice with

dry air to remove any residual products before a subsequent

mixture was prepared. All other aspects of the equipment and

experimental technique are fully described elsewhere [14,29].

3. Experimental data analysis methods

Schlieren images of the spherical flame propagating within

the combustion vessel, from the spark plug to a radius

corresponding to that of the bomb viewing window, are

presented in Appendix A at the end of this paper (Figs. A.1

and A.2). Since no significant rise in pressure occurred during

the flame image recording period (see Fig. 2), pressure and

temperature could be considered constant and equal to the

initial values during that period. In each test, successive

flame radii (r) were measured and the stretched laminar

flame speed (Sn) was obtained at any instant as

Sn ¼drdt

. (1)

The stretch rate (a) in a spherical flame of area A is given by

[29,30]

a ¼1A

dAdt¼ 2

Sn

r(2)

and its effect on the flame speed is such that

Ss � Sn ¼ Lba, (3)

where Ss is the unstretched laminar flame speed and Lb the

Markstein length.

Shown in Fig. 3 are plots of measured flame speed against

mean flame radius and stretch rate for two different

conditions, representative of (a) laminar and (b) cellular

flames. Laminar flames (Appendix A, Fig. A.1) obtained at

low pressure, while cellular flames (Appendix A, Fig. A.2) were

observed at higher pressure. In their earliest stages of

development (for very small radii and high stretch rate) the

propagation of all flames was affected by the excess energy

from the spark; flame speed data for this period were

therefore disregarded. It would have been possible to adjust

the spark ignition energy to the minimum for each experi-

mental condition to minimise this effect; however, previous

work had shown that ignition energy does not affect the

flame speed for radii greater than 5 mm [29,31] and so a

constant ignition energy was adopted in all experiments.

Since pressure and temperature remained constant during

the analysed period of flame development, variation in the

flame speed with radius/time for smooth laminar flames (e.g.

those at 1 bar) could be ascribed solely to changes in stretch

rate. Thus, from Eq. (3), the best straight line fit to the

experimentally derived Sn vs. a data (e.g. Fig. 3) yielded the

Markstein length as the slope of that line and the unstretched

laminar flame speed as its extrapolation to zero stretch rate.

For the case shown in Fig. 3(a), positive Lb, flame speed can be

seen to reduce with increasing stretch rate and vice versa (i.e.

stretch is detrimental to burning rate); conversely, for the

negative Lb case shown in Fig. 3(b), flame speed can be seen to

increase with higher stretch rates (i.e. stretch is beneficial to

burning rate). However, as also indicated in that figure

(illustrating behaviour typical of that noted for producer gas

at 2.5 and 5 bar), the flame rapidly became cellular with

increasing flame radius (and reducing stretch rate). This onset

of cellularity was accompanied with enhanced magnitude of

Sn, Fig. 3(b). For such flames, as discussed in Section 4 (below),

if the non-cellular region became too restricted for reliable

determination of Ss by the extrapolation technique then Ss

was put equal to the minimum recorded Sn value.

Finally, laminar burning velocity (ulÞ data were obtained

from values of Ss on the basis of the following equation:

ul ¼ Ssrb

ru, (4)

where rb and ru represent the densities of the burned (b) and

the unburned (u) gases, respectively.

The experimentally determined values of Markstein length

were similarly expressed as non-dimensional Markstein

number via the following equation [29]:

Ma ¼Lb

dl, (5)

ARTICLE IN PRESS

Fig. 2 – Evolution of the flame at the beginning of the

combustion.

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2854

Author's personal copy

where dl is the laminar flame thickness, calculated as the

ratio of burned product kinematic viscosity to laminar

burning velocity.

The set composition and equivalence ratio of the mixture

were functions of the accuracy of pressure measurement

during the process of filling the bomb using the partial

pressure method. The maximum absolute errors on the initial

pressure and temperature were 0.005 bar and 5 K, respectively.

In the worst case (1 bar and f ¼ 0:8), this would result in a

relative error of 6% in the molar fraction of hydrogen.

4. Results and discussion

4.1. Response to stretch rate

Set out in Fig. 4 are plots of flame speed (SnÞ against

flame radius for the ranges of initial pressure and equivalence

fuel/air ratio examined.

As shown in Table 1, several experimental tests were

performed at each condition in order to check the consistency

of the results. At 1 bar, the repeatability of ul values proved

good, with event to event variations of less than 2%, except

for mixtures at f ¼ 1:8, for which the difference between tests

could be as much as 6%. The repeatability in Markstein

number values was not as good, even for these stable low

pressure flames. This was associated with uncertainty in

determination of the values of Lb and dl. As seen in Fig. 3, with

increasing propensity to cellularity, rather less experimental

points were available for the linear fit required for the

extrapolation of flame speed to zero stretch and correspond-

ing evaluation of laminar burning velocity and Markstein

length/number. The number of data points used was subject

to a degree of judgement, particularly as flames became more

cellular. Changes in the number of points used in linear fits

generally led to changes in ul of only 1.5%; however the

corresponding values of Lb was found to change as much as

20%. Similar levels of uncertainty in the determination of

Markstein lengths have been reported previously [29]. Since

the flame thickness is definition dependent, there exist

considerable uncertainties in its value; an accurate determi-

nation of dl would require determination of at least the

thermal profile of the flame for which the laminar burning

velocity is required for validation.

At higher pressures, flames became cellular soon after

inception, with the range of stretch rates available for the

processing method described above becoming very limited. In

consequence, no meaningful Lb and Ma values could be

presented in Table 1 for the experiments at 2.5 or 5 bar;

similarly, since extrapolation to zero stretch was impossible,

the given values of ul might be considered less reliable at

those pressures.

At 1 bar, the effect of the stretch rate can be determined

from the flame speed evolutions, shown in Fig. 4(a) for the

range of equivalence ratios explored. For rich producer gas/air

mixtures, the flame speed (SnÞ can be seen slowly to increase

as the flame propagates (and stretch rate decreases). For

lean and stoichiometric mixtures the effect of the stretch

rate can be seen to be opposite, with decreasing flame

speed as flames propagate and become larger (although,

arguably, for all flames at 1 bar Sn might be considered to be

almost constant at the lower stretch rates consistent with

flame radii 430 mm). This behaviour with increasing flame

radius and reducing stretch rate effect can also be observed at

higher pressures, Figs. 4(b) and (c), prior to the onset of

cellularity—at which point flames accelerated. It should

be noted that while Sn changes associated with stretch rate

are of the order of 0.2 m/s, Fig. 4(a), those associated with

cellular flame development can be as much as 1 m/s, Figs. 4(b)

and (c).

ARTICLE IN PRESS

Fig. 3 – Flame speed vs. flame radius and stretch rate at (a) 1 bar, / ¼ 1:8 and (b) 5 bar, / ¼ 0:8.

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2 855

Author's personal copyARTICLE IN PRESS

Fig. 4 – Flame speed vs. flame radius at (a) 1 bar, (b) 2.5 bar and (c) 5 bar.

Table 1 – Flame speed, burning velocity and Markstein numbers

p0;T0 f Ss (m/s) ru=rb ul (m/s) Lb (mm) dl (mm) Ma

1 bar, 295 K 0.8 1.99 5.67 0.35 �0.43 0.05 �8.88

1.96 5.67 0.35 �0.53 0.05 �10.88

1.0 3.02 6.08 0.49 �0.06 0.03 �1.63

2.94 6.08 0.48 �0.23 0.03 �6.67

2.92 6.08 0.48 �0.18 0.04 �5.21

1.2 3.48 5.79 0.61 0.08 0.03 2.93

3.42 5.79 0.60 0.12 0.03 4.22

1.4 3.60 5.47 0.65 0.11 0.03 4.05

3.58 5.47 0.65 0.23 0.03 8.83

1.6 3.49 5.19 0.68 0.31 0.03 12.03

3.48 5.19 0.67 0.28 0.03 10.65

1.8 3.27 4.94 0.66 0.43 0.03 16.31

3.27 4.94 0.66 0.35 0.03 13.28

3.09 4.94 0.63 0.83 0.03 29.78

3.29 4.94 0.67 0.49 0.03 18.60

3.27 4.94 0.66 0.02 0.03 0.66

2.5 bar, 295 K 1 2.62 6.10 0.40 – 0.02 –

2.48 6.10 0.40 – 0.02 –

5 bar, 295 K 0.8 1.26 5.68 0.26 – 0.01 –

1.53 5.68 0.28 – 0.01 –

1.0 2.02 6.11 0.36 – 0.01 –

2.03 6.11 0.36 – 0.01 –

1.4 2.52 5.47 0.46 – 0.01 –

2.46 5.47 0.47 – 0.01 –

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2856

Author's personal copy

The values of Markstein number shown in Table 1

characterise the effects of the stretch rate, as outlined above.

The polarity of the value of Ma for each mixture indicates

whether burn rate Sn will increase or decrease with increasing

stretch rate (Eqs. (3) and (5)); the magnitude of Ma is indicative

of the sensitivity of the burn rate to stretching.

A general relationship between the value of Ma at 1 bar and

flame surface stability was also noted. At 1 bar, the values of

Ma were mostly positive and no flame became fully cellular

during the observation period, Fig. A.1 in Appendix A.

However, in that figure (particularly with the lean and

stoichiometric mixtures) it can be seen that what appear to

be ‘cracks’ do develop in the flame surface. These distur-

bances seem to originate in the vicinity of the spark plug and

develop over the surface as the radius increases. Bradley et al.

[31] suggested that such cracks are associated with regions of

negative curvature that propagate when thermo-diffusive

effects are not able to stabilise them. Increasing values of Ma

with f coincided with more stable (smooth) laminar flames,

with rich producer gas/air flames apparently more stable than

lean ones.

The general direction of stretch rate response with increas-

ing equivalence ratio (i.e. increasing Lb and Ma, more

detrimental burn rate response to stretch and reduced

tendency to cellular instability) of producer gas is similar to

that observed for methane [16] and hydrogen [27,32] but

opposite to that of isooctane [29] and other heavy hydro-

carbons (where Lb and Ma decrease with increasing equiva-

lence ratio). This may have practical implications where

biomass derived fuels are substituted for conventional

hydrocarbons. To help interpret the effects of stoichiometry

on flame stability, Lewis numbers (Le) for H2/CO/N2–air

mixtures at various equivalence ratios were also estimated:

Le ¼at

Dij, (6)

where Dij is the mass diffusivity of the deficient species in the

excess reactant [12] and at is the thermal diffusivity of the

unburned mixture. The deficient reactant, denoted by A, was

taken as O2 for rich mixtures and H2 and CO for the lean

mixtures. The diffusion coefficient of A in the dominant

species (N2) was obtained using the empirical correlation of

Fuller, Schettler and Giddings [33]; the corresponding thermal

diffusion coefficient was determined via the HYSYS program

[34]. Data in Table 2 show the resultant calculated Lewis

numbers, based upon the deficient reactant at each f value.

The so determined value of the diffusion coefficient DH22N2

was much higher than those of DCO2N2 and DO22N2 , and also

much higher than the thermal diffusivity of the mixture.

Inspection of the Lewis numbers obtained at 1 and 5 bar

suggests that the influence of pressure is very small; this is

since both at and DAB are roughly proportional to T1:75p�1.

The high diffusivity of H2 renders lean flames more

unstable; this effect being even more significant for lean

H2/air flames, where cellularity was noted even at 1 bar [35].

On the contrary, the mass diffusivity of CO (very similar to

that of O2) is nearly equal to the thermal diffusivity of the

mixture, leading to LeCO values slightly higher than 1. Thus,

the destabilising effect of H2 in the producer gas (LeH251)

might be thought to counter the stabilising effect of CO

(LeCO41). However, the study carried out by Law and Kwon

[36] suggests that CO would have no significant effect on the

effective Lewis number of the mixture. These authors studied

the effect of adding ethylene (same molar mass as CO) to H2/

air mixtures, finding no influence on thermo-diffusive effects;

addition of heavier hydrocarbons, such as propane, did make

H2 flames more stable.

The images shown in Fig. A.1 of Appendix A suggest that

flames become more stable with increasing equivalence ratio

(and so Lewis number). Thus, thermo-diffusive effects make

lean flames (LeH2o1) more unstable than rich flames (LeO2

41),

with completely smooth flames resultant at f ¼ 1:8. For rich

mixtures at 2.5 and 5 bar, the stabilizing effects of thermo-

diffusive phenomena are insufficient to compensate the

very strong hydrodynamic instability associated with the

elevated pressure. From the relationship between Ma and Le

[13], at 1 bar, Leo1 contributes to negative Ma for lean

mixtures, while Le41 corresponds to positive Ma for rich

mixtures.

4.2. Comparison of experimental and theoretical data

The experimentally and theoretically determined values of

laminar burning velocity for the H2/CO/N2–air mixtures are

shown in Fig. 5.

At 1 bar, agreement can be seen to be good; with differ-

ences, arguably, as much due to uncertainties in fitting the

experimental data to Eq. (3) as to any deficiency in the model.

The theoretical model incorporated no mechanism for

prediction of burning rate enhancement associated with any

kind of instability. Nevertheless, agreement with experiment

proved no worse for the cracked f ¼ 0:8 flames than for richer

completely smooth laminar flames. The differences between

experimental and theoretical values of burning velocity were

much more noticeable for the 5 bar flames. At this pressure,

as seen in Fig. A.2 of Appendix A, flames became cellular

almost from inception. In consequence, as discussed pre-

viously, in image processing experimental Sn data to yield Ss

(and hence ulÞ it was necessary to restrict consideration to a

very narrow range of stretch rates (a) or to put Ss equal to the

ARTICLE IN PRESS

Table 2 – Lewis number at 1 and 5 bar, for differentequivalence ratios

Conditions f Aa at ðcm2=sÞ DA2N2ðcm2=SÞ Le

1 bar, 295 K 0.8 H2 0.264 0.777 0.34

CO 0.205 1.29

1.2 O2 0.276 0.205 1.35

1.4 0.282 1.37

1.6 0.286 1.40

1.8 0.289 1.41

5 bar, 295 K 0.8 H2 0.053 0.155 0.34

CO 0.049 1.08

1.4 O2 0.056 0.041 1.37

a Denotes the deficient reactant.

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2 857

Author's personal copy

minimum observed value of Sn. The difficulty of extrapolating

to zero stretch is likely to have lead to a degree of over-

prediction of ul at the higher pressure, since in Fig. 4(c) the

stretch was seen to increase the burn rate.

Data shown in Fig. 5 suggest that there may be a need to

incorporate some mechanism to account for cellularity in

modelling real flames at high pressure. Nevertheless, there is

agreement between experiment and theory in that laminar

burning velocity decreases with pressure and peaks at very

rich mixtures (f ¼ 1:6). The reason for the latter behaviour is

that, unlike the adiabatic flame temperature of conventional

fuels (which peak at f � 1), the net reaction rate of the

oxidation process for H2/air and CO/air mixtures peaks at

f � 2 [37].

Avoidance of cellularity in high pressure flame experiments

for mixtures containing significant hydrogen content is

problematic. The Princeton group [38] has been able to obtain

laminar flames at high pressure by increasing the Lewis

number of the mixture (e.g. substituting N2 by He). However

this generates a different burning velocity to that obtained

with nitrogen and one-dimension kinetic computations

(tuned using the He data) were required to correct mea-

sured burning velocity; introducing new uncertainties. An

alternative solution might be to quantify the burning

enhancement of cellularity by means of fractal considera-

tions of the flame surface area [39,40]; the ratio of burning

velocities with and without instabilities is taken equal to

that of the surface areas, provided flame stretch effects can

be neglected. To date this technique remains in the develop-

ment stage and requires further validation with well-char-

acterised fuels.

The possible influence of cellular instabilities on premixed

turbulent flames is currently unknown. It is possible that the

high pressure laminar burning velocities presented here (with

some cellularity effects included) might be more appropriate

for use in simplified turbulent burning velocity correlations

than theoretical one-dimensional values. This is the subject

of ongoing research.

4.3. Comparisons with other fuels

The experimental ul values of the producer gas are compared

in Fig. 6 with those for other fuels, also obtained in the Leeds

MkII combustion vessel at room temperature, for 1 and 5 bar.

Under lean and stoichiometric conditions, producer gas

laminar burning velocity is not dissimilar to that of methane

[16] or isooctane [29], albeit somewhat slower than that of

pure hydrogen [35]. In Fig. 6, laminar burning velocity data for

a 50:50 H2/CH4 mixture (vol%) at 1 bar have also been included

[41]. The laminar burning velocity of the hydrogen containing

mixtures is notably higher than those of the other fuels at

f41, and the upper flammability limit of the producer gas

seems to be even higher than that of the H2/CH4 mixture. This

ARTICLE IN PRESS

Fig. 6 – Comparison of laminar burning velocities for

different fuels at 1 and 5 bar: hydrogen [35], CH4–H2 50:50

[41], isooctane [29], methane [16] and producer gas.

Fig. 5 – Experimental and theoretical laminar burning

velocity vs. equivalence ratio (/) at 295 K for pressures of 1

and 5 bar.

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2858

Author's personal copy

is because the value of ul for carbon monoxide/air mixtures

peaks at higher equivalence ratio than that of hydrogen/air

mixtures.

A number of other workers have published laminar burning

velocity data over a range of equivalence ratios at 1 bar and

room temperature for various other H2/CO fuels (with and

without excess nitrogen), Table 3.

The results obtained in these various studies are compared

in Fig. 7. The values of laminar burning velocity reported by

Huang et al. [17] for reformer gas (triangular symbols in Fig. 7)

can be seen to be higher than those obtained for the producer

gas in the current study (circular symbols in Fig. 7); this is

associated with the lower H2 content and greater N2 content

of the producer gas. Similarly, since the H2/CO ratio is smaller

for the producer gas, its laminar burning velocity (ulÞ peaks at

a higher equivalence ratio.

The dotted lines in Fig. 7 show the laminar burning velocity

values for a range of H2/CO mixtures obtained by Hassan et al.

[19]; these data clearly show that ul increases with H2/CO ratio

at any given equivalence ratio. It can be observed that the

magnitude of laminar burning velocity for the simulated

producer gas is similar to that of a mixture comprising 10%

H2/90% CO, although the value of ul of the former peaks at a

lower equivalence ratio than that of the latter. Of course, it is

the nitrogen content of the producer gas, coming from the air

used in the gasification process of the biomass, which results

in a ul value lower than one would expect for a hydrogen rich

fuel.

5. Summary and conclusions

In the present work, data for unstretched laminar burning

velocity at 1 and 5 bar have been determined, experimen-

tally and theoretically, for a H2/CO/N2 mixture represen-

tative of biomass gasification-derived producer gas. In

determining the experimental laminar burning velocities

from schlieren images of developing spherical flames, it

was necessary to correct for the effects of flame stretch

rate and cellularity. The flame stretch rate effects have

been characterised in terms of Markstein length and

noted flame instabilities discussed in terms of Lewis

number. The burning velocity data have been compared

with those for some reported by other workers, including

those of other H2/CO containing mixtures. The principal

findings are

� At 1 bar, 295 K, the laminar flames of simulated producer

gas/air mixtures remained essentially ‘smooth’, with

flame speed enhanced by increasing stretch rate for

stoichiometric and lean mixtures, and reduced in the case

of rich mixtures. This response has been characterised by

negative values of Markstein length/number for lean

mixtures and increasingly positive values for rich mix-

tures.

� At higher pressure (2.5 and 5 bar), transition to cellular

flame structure occurred very soon after ignition, with

laminar flame speed (SnÞ values rapidly increasing by as

much as 1 m/s over the range of radii observed (up to

60 mm). As a result, Markstein length/number determina-

tion proved problematic and hence it was not possible

similarly to characterise the effect of flame stretch on

burning velocity in terms of these parameters at elevated

pressure.

� Lewis numbers, based on the deficient species of the

reactant mixture, have been calculated to characterise

thermo-diffusive effects on flame stability. For rich flames

at 1 bar, the mass diffusivity of oxygen proved much

smaller than the thermal diffusivity of the mixture; in

consequence, flames remained perfectly smooth and

laminar. Conversely, the higher mass diffusivity of the

deficient reactant (H2) in lean producer gas flames

rendered them more unstable, with the appearance of

cracks over the flame front observed in the schlieren

images. For rich flames at higher pressure values (2.5 and

5 bar), the stabilising effect of the thermo-diffusive phe-

nomena proved insufficient to compensate the strong

hydrodynamic instabilities associated with elevated pres-

sure, with resultant cellular flame structure in all the

cases.

� Experimental and theoretical values of laminar burning

velocity proved in good agreement at 1 bar. At this

condition, flames were smooth and stable, allowing

extrapolation of experimental data to the unstretched

state assumed in the planar flame propagation considered

ARTICLE IN PRESS

Fig. 7 – Comparison of ul data for the H2/CO/N2 mixture

listed in Table 3 (� present work, m [17] and dotted lines, four

H2/CO mixtures [19]).

Table 3 – Composition of H2/CO fuels (vol%)

H2 CO N2

Present work 21 24 55

Huang et al. [17] 28 25 47

Hassan et al. [19] Various ratios –

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2 859

Author's personal copy

by the theoretical model. The peak laminar burning

velocity was 0.67 m/s (experimental), and 0.65 (theoretical),

at an equivalence ratio of 1.6.

� Agreement between experiment and theory was much

poorer at 5 bar; this was associated with the onset of

cellularity. There was no mechanism for the enhancement

of laminar burning velocity by cellularity in the computer

model and the phenomenon precluded reliable determi-

nation of corresponding (stretch and instability free)

experimental data. Increased pressure resulted in sub-

stantially reduced laminar burning velocity (both theore-

tical and experimental).

� For stoichiometric and lean mixtures, the experimentally

determined laminar burning velocities of producer gas–air

mixtures (at both 1 and 5 bar) were about 20% and 45%

faster than those of isooctane–air and methane–air

mixtures, respectively; and approximately 40% and 140%

slower than those of 50% methane/50% hydrogen–air and

pure hydrogen–air mixtures, respectively. However, at

1 bar, producer gas–air laminar burning velocity peaked

at a somewhat richer equivalence ratio (f ¼ 1:6) than did

those of the other fuels (hydrogen was only studied

to f ¼ 1:0). This also appeared to be the case at 5 bar.

This behaviour was associated with the CO content of

the producer gas; the addition of hydrogen to methane,

while increasing burn rate at any given equivalence ratio,

did not appear to shift the peak to a significantly richer

mixture—this seemed to be confirmed by examination of

data for other hydrogen/carbon monoxide/nitrogen mix-

tures.

Acknowledgements

The authors would like to acknowledge Dr. M. Lawes for

useful discussions regarding flame instabilities, and also to

thank Dr. M. Ormsby and M. Harker for assistance with some

experiments. This study was partially supported by a post-

doctoral fellowship grant from Junta de Comunidades de

Castilla-La Mancha.

Appendix A

Schlieren photographs for different equivalence ratios are

shown in Figs. A.1 and A.2.

Appendix B. Supplementary materials

Supplementary data associated with this article can be found

in the online version at doi:10.1016/j.ijhydene.2007.10.050.

ARTICLE IN PRESS

p0 = 1 bar � = 0.8 p0 =1 bar � = 1 p0 =1 bar � =1.2 p0 = 1 bar � = 1.4 p0 =1 bar � =1.6 p0 =1 bar � =1.8

t = 8 ms

t = 16 ms

t = 24 ms

t = 32 ms

Fig. A.1 – Schlieren photographs at p0 ¼ 1 bar (T0 ¼ 295 K) for different equivalence ratios.

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2860

Author's personal copy

R E F E R E N C E S

[1] European Commission. Biomass Action Plan. Communica-tion COM(2005) 628, hhttp://ec.europa.eu/energy/res/biomass_action_plan/index_en.htmi; 2005.

[2] Hernandez JJ, Serrano C, Perez J. Prediction of the autoigni-tion delay time of producer gas from biomass gasification.Energy Fuels 2006;20(2):532–9.

[3] Sridhar G, Sridhar HV, Dasappa S, Paul PJ, Rajan NKS,Mukunda HS. Development of producer gas engines. ProcInst Mech Eng Part D: J Automob Eng 2005;219:423–68.

[4] Ahrenfeldt J, Jensen TK, Henriksen U, Schramm J.Experiments with wood gas engines. SAE Paper 2001; 2001-01-3681.

[5] Lapuerta M, Hernandez JJ, Tinaut FV, Horrillo A. Thermo-chemical behaviour of producer gas from gasification oflignocellulosic biomass in SI engines. SAE Paper 2001; 2001-01-3586.

[6] Ahrenfeldt J, Jensen TK, Henriksen U, Schramm J. Investiga-tion of continuous gas engine CHP operation on biomassproducer gas. SAE Paper 2005; 2005-01-3778.

[7] Tsolakis A, Hernandez JJ, Megaritis A, Crampton M. Dual fueldiesel engine operation using H2. Effects on particulateemissions. Energy Fuels 2005;19(2):418–25.

[8] Biomass Engineering Ltd. Development of a micro-turbine

plant to run on gasifier producer gas. Technical Report URN

06/675, hhttp://www.dti.gov.uk/i; 2004.

[9] Kee RJ, Rupley FM, Miller JA, Coltrin ME, Grcar JF, Meeks E,et al. CHEMKIN Software, Release 4.0. San Diego, CA: Reaction

Design, Inc.; 2004.

[10] Bechtold JK, Matalon M. Hydrodynamic and diffusion effects

on the instability of spherically expanding flames. Combust

Flame 1987;6:77–90.

[11] Bradley D, Cresswell TM, Puttock JS. Flame acceleration due

to flame-induced instabilities in large-scale explosions.Combust Flame 2001;124:551–9.

[12] Matalon M, Cui C, Bechtold JK. Hydrodynamic theory of

premixed flames: effects of stoichiometry, variable transport

coefficients and arbitrary reaction orders. J Fluid Mech

2003;487:179–210.

[13] Law CK, Sung CJ. Structure, aerodynamics, and geometry of

premixed flamelets. Prog Energy Combust Sci

2000;26:459–505.[14] Gillespie L, Lawes M, Sheppard CGW, Woolley R. Aspects of

laminar and turbulent burning velocity relevant to SI

engines. SAE Paper 2000; 2000-01-0192.

[15] Kitagawa T. Effects of pressure on burning velocity and

instabilities of propane–air premixed flames. JSME Int J Ser B

2005;48(1).

ARTICLE IN PRESS

p0 = 5 bar � = 0.8

t = 8 ms

t = 16 ms

t = 24 ms

t = 32 ms

p0 = 2.5 bar � = 1 p0 = 5 bar � = 1 p0 = 5 bar � = 1.4

Fig. A.2 – Schlieren photographs at p0 ¼ 2:5 and 5 bar (T0 ¼ 295 K) for different equivalence ratios.

I N T E R N AT I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2 861

Author's personal copy

[16] Gu XJ, Haq MZ, Lawes M, Woolley R. Laminar burning velocityand Markstein lengths of methane–air mixtures. CombustFlame 2000;121:41–58.

[17] Huang Y, Sung CJ, Eng JA. Laminar flame speeds of primaryreference fuels and reformer gas mixtures. Combust Flame2004;139:239–51.

[18] Han P, Checkel MD, Fleck BA, Nowicki NL. Burning velocity ofmethane/diluent mixture with reformer gas addition. Fuel2007;86:585–96.

[19] Hassan MI, Aung KT, Faeth GM. Properties of laminarpremixed CO/H2/air flames at various pressures. J PropulPower 1997;13(2):239–45.

[20] Sung CJ, Huang Y, Eng JA. Effects of reformer gas addition onthe laminar flame speeds and flammability limits ofn-butane and iso-butane flames. Combust Flame2001;126:1699–713.

[21] Sridhar G, Paul PJ, Mukunda HS. Computational studies of thelaminar burning velocity of a producer gas and air mixtureunder typical engine conditions. Proc Inst Mech Eng Part A JPower Energy 2005;219:195–201.

[22] Hernandez JJ, Lapuerta M, Serrano C, Melgar A. Estimation ofthe laminar flame speed of producer gas from biomassgasification. Energy Fuels 2005;19:2172–8.

[23] Metghalchi M, Keck JC. Laminar burning velocities ofpropane mixtures at high temperature and pressure. Com-bust Flame 1980;38:43–154.

[24] Melgar A, Perez JF, Laget H, Horrillo A. Thermochemicalequilibrium modelling of a gasifying process. Energy ConversManage 2007;48:59–67.

[25] FAO. Wood gas as engine fuel. Food and AgriculturalOrganization of the United Nations (FAO), Rome, 1993.

[26] Smith GP, Golden DM, Frenklach M, Moriarty NW, Eiteneer B,Goldenberg M, et al. GRI-Mech 3.0. Gas Research Institute,hhttp://www.me.berkeley.edu/gri_mech/i; 1995.

[27] Verhelst S, Woolley R, Lawes M, Sierens R. Laminar andunstable burning velocities and Markstein lengths of hydro-gen–air mixtures at engine-like conditions. Proc. Combust.Inst. 2005;30:209–16.

[28] Weinberg FJ. Optics of flames. London: Butterworth; 1963.[29] Bradley D, Hicks RA, Lawes M, Sheppard CGW, Woolley R. The

measurement of laminar burning velocities and Marksteinlengths for iso-octane–air and iso-octane–n–heptane–air

mixtures at elevated temperatures and pressures in anexplosion bomb. Combust Flame 1998;115:126–44.

[30] Bradley D, Gaskell PH, Gu XJ. Burning velocities, Marksteinlengths, and flame quenching for spherical methane–airflames: a computational study. Combust Flame1996;104:176–98.

[31] Bradley D, Sheppard CGW, Woolley R, Greenhalgh DA, LockettRD. The development and structure of flame instabilities andcellularity at low Markstein numbers in explosions. CombustFlame 2000;122:195–209.

[32] Aung KT, Hassan MI, Faeth GM. Flame stretch interactions oflaminar hydrogen/air flames at normal temperature andpressure. Combust Flame 1997;109:1–24.

[33] Poling BE, Prausnitz JM, O’Connell JP. The properties of gasesand liquids. New York: McGraw-Hill; 2001.

[34] Hyprotech. HYSYS 3.01, Documentation. Hyprotech Ltd.Calgary, Alberta, 2002.

[35] Verhelst S. A study of the combustion in hydrogen-fuelledinternal combustion engines. PhD thesis, University ofGhent, Belgium, hhttp://www.floheacom.ugent.be/H2/h2_research_en.htmi; 2005.

[36] Law CK, Kwon OC. Effects of hydrocarbon substitution onatmospheric hydrogen–air flame propagation. Int J HydrogenEnergy 2004;29:867–79.

[37] Griffiths JF, Barnard JA. Flame and combustion. London:Chapman & Hall; 1995.

[38] Burke MP, Qin X, Yu Y, Dryer FL. Measurements ofhydrogen syngas flame speeds at elevated pressures. In:Fifth US combustion meeting. California: San Diego; March2007.

[39] Smallbone A, Kitagawa T, Oonishi T. Unstretched laminarburning velocity estimations from unstable flames at ele-vated pressures. JSME Int J 2006;91-06:15–20.

[40] Al-Shahrany AS, Bradley D, Lawes M, Woolley R. Measure-ment of unstable burning velocities of iso-octane–air mix-tures at high pressure and the derivation of laminar burningvelocities. Proc. Combust. Inst. 2005;30:225–32.

[41] Burluka AA, Fairweather M, Ormsby MP, Sheppard CGW,Woolley R. The laminar burning properties of premixedmethane–hydrogen flames determined using a novel analy-sis method. In: Third European combustion meeting. Crete:Chania; 2007.

ARTICLE IN PRESS

I N T E R N A T I O N A L J O U R N A L O F H Y D R O G E N E N E R G Y 3 3 ( 2 0 0 8 ) 8 5 1 – 8 6 2862