Influence of fuel variability on the characteristics of impingingnon-premixed syngas burningCitation for published version (APA):Ranga Dinesh, K. K. J., Jiang, X., Oijen, van, J. A., Bastiaans, R. J. M., & Goey, de, L. P. H. (2013). Influence offuel variability on the characteristics of impinging non-premixed syngas burning. Proceedings of the CombustionInstitute, 34(2), 3219-3229. https://doi.org/10.1016/j.proci.2012.06.081

DOI:10.1016/j.proci.2012.06.081

Document status and date:Published: 01/01/2013

Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne

Take down policyIf you believe that this document breaches copyright please contact us at:openaccess@tue.nlproviding details and we will investigate your claim.

Download date: 16. Mar. 2020

Available online at www.sciencedirect.comProceedings

Proceedings of the Combustion Institute 34 (2013) 3219–3229

www.elsevier.com/locate/proci

of the

CombustionInstitute

Influence of fuel variability on the characteristicsof impinging non-premixed syngas burning

K.K.J. Ranga Dinesh a,⇑, X. Jiang a, J.A. van Oijen b,R.J.M. Bastiaans b, L.P.H. de Goey b

a Engineering Department, Lancaster University, Lancaster, Lancashire LA1 4YR, UKb Combustion Technology, Department of Mechanical Engineering, Eindhoven University of Technology, Eindhoven,

The Netherlands

Available online 30 June 2012

Abstract

Investigations of instantaneous flame characteristics and near-wall heat transfer of syngas mixturesincluding H2-rich and H2-lean flames have been performed in an impinging non-premixed configurationfor a Reynolds number of 2000 and a nozzle-to-plate distance of 4 jet nozzle diameters by direct numericalsimulation and flamelet generated manifold chemistry. The results presented were obtained from simula-tions using a uniform Cartesian grid with 200 � 600 � 600 points. The spatial discretisation was carriedout using a sixth-order accurate compact finite difference scheme and the discretised equations wereadvanced using a third-order accurate fully explicit compact-storage Runge–Kutta scheme. Results werediscussed for the flame characteristics, reaction progress variable, velocity field and Nusselt number distri-butions. Significant differences have been found for the flame characteristics of syngas burning depending onthe hydrogen, carbon monoxide, carbon dioxide and nitrogen percentages of the syngas mixture. High dif-fusivity in the H2-rich flame leads to form weaker vortical structures and thicker flames than those in the H2-lean flame. It has been observed that the maximum flame temperature decreases from the H2-rich to the H2-lean flames. It is also found that the maximum flame temperature occurs at the lean side of the stoichiometricmixture fraction of the syngas fuel mixtures. The composition of the syngas mixture has a significant impacton the flame characteristics of the impinging flame, including the near-wall flame structure and heat transfer.Crown copyright � 2012 Published by Elsevier Inc. on behalf of The Combustion Institute. All rightsreserved.

Keywords: Syngas combustion; Impinging jet; DNS; FGM; Near-wall heat transfer

1. Introduction

A tremendously important new climate changemitigation technology is clean combustion, whichcan significantly reduce carbon dioxide emission

1540-7489/$ - see front matter Crown copyright � 2012 PuInstitute. All rights reserved.http://dx.doi.org/10.1016/j.proci.2012.06.081

⇑ Corresponding author.E-mail address: ranga.dinesh@lancaster.ac.uk (K.K.J.

Ranga Dinesh).

while continuing to use fossil fuels, such as coal,for power generation. In moving towards highefficiency power generation systems with reducedCO2 emissions, the use of syngas (representingsynthetic gas or synthesis gas) technologiesbecomes increasingly attractive. The syngas canbe generated from a wide range of solid fuelsincluding coal, biomass and waste products.Because of the large amount of resources availableworldwide, especially coal in the U.S., Europe,

blished by Elsevier Inc. on behalf of The Combustion

3220 K.K.J. Ranga Dinesh et al. / Proceedings of the Combustion Institute 34 (2013) 3219–3229

and Asia, there is an interest in using syngas fuelsfor power generation and assessing their effects onthe operating systems. Investigation of the appli-cation of syngas is closely related to the importanttechnology of integrated gasification combinedcycle (IGCC), which turns coal and other fuelsinto syngas [1,2]. The emerging technology ofIGCC can also be integrated with the technologyof carbon capture and storage for climate mitiga-tion. Understanding the combustion of hydrogenenriched fuels could lead to better combustioncontrol and combustor design for future cleanergas turbine combustion applications, particularlyfor the design and development of IGCC basedcombustors.

Various investigations have been carried out tounderstand the flame characteristics of syngascombustion relevant to gas turbine and internalcombustion engine applications including bothnon-premixed and premixed combustion. Resultswere reported for the laminar flame speeds of leanH2/CO/CO2 premixed syngas fuel mixtures over awide range of fuel compositions including the val-idation of chemical mechanisms [3], which wassubsequently extended to applications for gas tur-bine combustors [4]. The analysis of ignition prop-erties for complete H2/CO composition range atelevated temperatures and pressures applicableto gas turbine premixed combustion was alsoreported [5]. At the same time, N2 dilution effecton flame stability of premixed syngas fuel hasbeen carried out [6]. Meanwhile investigations ofthe non-premixed combustion of syngas fuel mix-tures including blended natural gas/hydrogenhave been reported in the literature mainly onthe flame stability issues. For example, some ofthe important studies include experimental inves-tigations on the effects of H2 addition on the sta-bility limits of CH4 non-premixed flames [7],detailed scalar analysis of CO/H2/N2 jet flames[8] and the effects of H2 addition on the ignitionand combustion of high pressure CH4 combustion[9]. However, there is an insufficient amount ofinformation available in the literature regardingthe fuel variability and flame characteristics ofH2-rich and H2-lean syngas combustion in anon-premixed configuration.

There are challenges in utilising syngas in gasturbine combustion which need careful attentionand further investigations. For instance, syngascombustion could develop into undesirable flameflashback phenomenon, in which the flame propa-gates into the burner. Particularly the H2-rich syn-gas flame with high diffusivity can travel upstreamand even attach to the wall of the combustor. Thestrong heat transfer to the wall may damage thecombustor components. The consequence can bevery costly. Because of this, many existing com-bustors may need careful analysis for the burningof syngas, particularly for H2-rich fuel mixtures.A better understanding of the flame characteris-

tics is the key for achieving safe and controllablesyngas burning, which is crucial to the technologydevelopments. To overcome this bottle neck indeveloping hydrogen based clean combustiontechnology, in-depth knowledge of the flame char-acteristics of hydrogen enriched syngas is essen-tial. Furthermore, flame–wall interactions areimportant to the life span of the gas turbine com-bustor but have not been fully understood. Inorder to meet these challenges, detailed computa-tional combustion can play a significant rolewhich can be employed to investigate the struc-tures of syngas flames with high fuel flexibility.This work was aimed to investigate flame charac-teristics including the flame–wall interaction ofhydrogen enriched syngas combustion using directnumerical simulation (DNS), which provides thedetails needed to understand the flame structuresusing high-fidelity computational techniques.

Historically, most DNS flame analyses have beenbased on simple chemistry to reduce the computa-tional costs, while the continuous growth in comput-ing power has made it possible to incorporate moredetailed combustion chemistry and to perform largescale computations in recent years. Early investiga-tions reported two-dimensional simulations of non-premixed flames [10,11]. Three-dimensional DNS ofturbulent non-premixed flames including finite ratechemistry and heat release effects were also carriedout, e.g., [12]. Since then various studies have beenperformed on DNS of non-premixed combustion,for example, the effects of flow strain on a hydro-gen–air triple flame [13], scalar intermittency of CO/H2 planer jet flame [14], flame stabilisation and struc-ture of lifted hydrogen jet flame [15]. DNS studies offlame–wall interactions with one-step and multi-stepchemical kinetics and in physical domains have beenreported in the last two decades. For example, two-dimensional DNS of head-on quenching (HOQ) ina pseudo-turbulent reactive boundary layer [16] andthree-dimensional HOQ of premixed propagatingflame in constant density turbulent channel flow[17], side-wall effects on flame dynamics [18], one-dimensional simulation of hydrogen combustioninteracting with an inert wall [19], three-dimensionalDNS of sidewall quenching for v-shaped premixedflame [20] and turbulent flame–wall interaction usingthree-dimensional DNS and detailed chemical kinet-ics [21] were carried out. However, in-depth knowl-edge of the syngas flame characteristics is still notfully available while near-wall flow and combustionphenomena remain an area to be fully investigated,where the modelling challenges are also significantfor large-eddy simulation and traditional Reynolds-averaged Navier–Stokes modelling.

Fuel variability is of great importance for syn-gas combustion from perspectives of both funda-mental study and practical application. Fuelcomposition affects the combustion chemistry aswell as the molecular transport processes. Conse-quently, different combustion characteristics due

K.K.J. Ranga Dinesh et al. / Proceedings of the Combustion Institute 34 (2013) 3219–3229 3221

to fuel variability will have a significant impact onthe combustion control and the combustor design.In order to explore the effect of syngas fuelcomposition on flame structures and near-wallcombustion, a comparative DNS study was per-formed in an impinging non-premixed jet flameconfiguration. Four different syngas fuel composi-tions varying from H2-rich to H2-lean includingcarbon monoxide, carbon dioxide and nitrogenwere investigated. The flame chemistry of the syn-gas, which is a crucial element of the simulation,was represented by methods based on detailedchemistry in this study.

As an effort to understand the flame characteristicsof hydrogen enriched fuel mixtures and the near-wallreacting flow and heat transfer using detailed numericalsimulations, the research employed DNS and the flam-elet-generated manifold (FGM) approach. The detailedchemical kinetics has been employed through the FGMmethod [22], which not only uses complex chemistry,but also takes the most important transport processesinto account. The FGM data table was adopted as ameans to reduce the prohibitive computational costsof directly implementing detailed chemistry in a DNScode. The objective of the research was to analyse theimportant physics of the effects of fuel variability onthe flame characteristics of syngas combustion. In thefollowing, the methods used in this study are presentedfirst, includingthegoverningequationsandflamechem-istry, and the numerical methods. The results and dis-cussions are presented subsequently, mainly in termsof instantaneous flame characteristics. Finally conclu-sions are drawn.

2. Governing equations and flame chemistry

The three-dimensional equations of the conser-vation laws for continuity, momentum, energy,transport equations for progress variable and mix-ture fraction, and the equation of state in theirnon-dimensional form can be given as

@q@tþ@ quj

� �@xj

¼ 0; ð1Þ

@ðqujÞ@tþ@ðqujukÞ

@xkþ @p@xk� 1

Re@sjk

@xkþðqa�qÞ

gj

Fr¼ 0; ð2Þ

@qe@tþ@ðqeukÞ

@xkþ@ðpukÞ

@xk;

� 1

RePrM2ðc�1Þ@

@xkk@T@xk

� �� 1

Re@ðujsjkÞ@xk

þðqa�qÞgkuk

Fr¼ 0; ð3Þ

@ðqY Þ@tþ@ðqukY Þ

@xk� 1

ReSc@

@xkð kCp

@Y@xkÞ�xY ¼ 0; ð4Þ

@ðqZÞ@tþ@ðqukZÞ

@xk� 1

ReSc@

@xk

kCp

@Z@xk

� �¼ 0; ð5Þ

p¼ qT

cM2: ð6Þ

In Eqs. (1)–(6), t stands for time, uj is the veloc-ity components in the xj direction, e stands fortotal energy per unit mass, p for pressure, k forheat conductivity, CP for specific heat at constant

pressure, l for dynamic viscosity, c is the ratio ofspecific heats, xY is the source term of the pro-gress variable, q is the density, subscript a standsfor the ambient respectively. Symbols M, Pr, Sc,Fr and Re represent Mach number, Prandtl num-ber, Schmidt number, Froude number and Rey-nolds number respectively.

To investigate the effects of fuel variability, theflame chemistry must be realistically represented inorder to accurately predict the chemical heatrelease and the concentrations of the chemical spe-cies of the syngas combustion. In this work, theflame chemistry of syngas mixtures is representedby databases generated by using the FGM tech-nique [22], accounting for both chemical andtransport processes using the laminar flamelet con-cept [23]. The FGM databases were generatedfrom steady counter-flow diffusion flamelets usingdetailed chemistry and transport models includingdifferential diffusion effects. The detailed kineticmodel [24] incorporates the thermodynamic,kinetic, and species transport properties relevantto high temperature H2 and CO oxidation, consist-ing of 14 species and 30 reactions. In the FGMapproach, a low-dimensional manifold is createdfrom solutions of the so-called flamelet equationsand is solved by the full implicit solver CHEM1Ddeveloped at the Eindhoven University of Tech-nology [25]. In the present study, four FGM dat-abases have been generated for four differentfuels. The databases contain the variables as afunction of the mixture fraction Z and the progressvariable Y such that the resolution of the mani-folds is 201 points in the mixture fraction spaceand 201 points in the progress variable space. Itis important to note that the applicability ofFGM in numerical simulations of combustion ofhydrogen and syngas mixtures has been investi-gated and validated [26,27].

3. Computational details

Three-dimensional time dependent DNS withthe flame chemistry represented by the tabulatedFGM approach has been performed with the gov-erning equations solved in their non-dimensionalform. The set of equations listed above were solvedwith a parallel DNS code and computations werecarried out using 600 processors in the UK high-end computer HECToR. Figure 1 shows the geom-etry of the impinging configuration consideredhere and the dimensions of the computationalbox given in terms of the reference length scale –the fuel jet nozzle diameter (D). The computa-tional domain employed has a size of four jet noz-zle diameters in the streamwise direction andtwelve jet nozzle diameters in the cross-streamwisedirections. Since the main focus of the study wason the effects of fuel variability on the flamecharacteristics, a relatively small domain in the

3222 K.K.J. Ranga Dinesh et al. / Proceedings of the Combustion Institute 34 (2013) 3219–3229

streamwise direction Lx = 4.0 was selected toensure that important physical effects can be exam-ined with affordable computational costs. Theresults presented in this study were performedusing a uniform Cartesian grid with 200 � 600 �600 points resulting 72 million computationalnodes. To avoid the prohibitively high computa-tional costs of DNS of fully turbulent flows, theReynolds number used Re = 2000 was relativelylow with a Froude number of Fr = 1.0, based onthe reference quantities used. Since the results weretested as almost grid-independent, the grid pointsused are considered to be sufficient to resolve theenergy spectra. The Prandtl number Pr and thespecific heat c vary according to the FGM table.

The discretisation of the governing equationsincludes the high-order numerical schemes forboth spatial discretisation and time advancement.The spatial derivates in all three directions aresolved using a sixth-order accurate compact (Pade)finite difference scheme [28]. The scheme usessixth-order at inner points, fourth-order next tothe boundary points, and third-order at theboundary. Further details of the Pade 3/4/6scheme can be found in Ref. [28]. Solutions forthe spatial discretised equations are obtained bysolving the tri-diagonal system of algebraic equa-tions. The spatial discretised equations areadvanced in time using a fully explicit low-storagethird-order Runge–Kutta scheme [29]. The timestep was limited by the Courant number for stabil-ity and a chemical restraint.

The computational domain contains an inletand impinging wall boundaries in the streamwisedirection where the buoyancy force is acting. At

Fig. 1. Geometry of the impinging flame with thecomputational domain size of 4 jet nozzle diameters inthe streamwise direction and 12 diameters in the cross-streamwise directions.

the inflow, the flow was specified using theNavier–Stokes characteristic boundary conditionsby [30] with the temperature treated as a soft var-iable (temperature was allowed to fluctuateaccording to the characteristic waves at the bound-ary). At the inlet, the mean streamwise velocitywas specified using a hyperbolic tangent profile

u¼U fuel=2f1� tanh½ð0:5=4dÞðr=0:5�0:5=rÞ�g; ð7Þ

r¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðz�0:5LzÞ2þðy�0:5LyÞ2

q; ð8Þ

where r stands for the radial direction of the roundjet, originating from the centre of the inlet domainð0 6 x 6 Lx; 0 6 y 6 LyÞ and the initial momentumthickness d was chosen to be 10% of the jet radius.The dimensions of the impinging jet configurationused were Lx = 4.0 and Ly = Lz = 12.0. Externalunsteady disturbances were artificially added tothe mean velocity profile for all three velocity com-ponent at the inlet in a sinusoidal form such thatu0 = v0 = w0 = A sin(2pf0t). Here we assigned thevalue A = 0.05 and the non-dimensional frequencyof the unsteady disturbance f0 = 0.3. The relativelylarge disturbance was used to enhance the develop-ment of instability in the computational domain,which is rather small in the streamwise direction,while the frequency of the perturbation was chosento trigger the unstable mode of the jet. At the sideboundaries, non-reflecting characteristic boundarycondition is used. The non-slip wall boundary con-dition is applied at the downstream wall, which isassumed to be at the ambient temperature, imper-meable to mass and without surface chemical reac-tion. At the impinging wall boundary, the mixturefraction is assumed zero-gradient correspondingto the impermeability, while the progress variablefor chemistry is taken as zero at the wall boundary.The simplified wall boundary conditions wereadopted to facilitate investigations of fuel variabil-ity effects on the flame structures and near-wall heattransfer.

4. Results and discussion

DNS of flames corresponding to four differentsyngas fuel mixtures, named as flames A, B, C andD varying from H2-rich to H2-lean, have been per-formed. The compositions were taken from theBP syngas datasheet with the maximum percent-age of H2, CO, CO2 and N2 in each case. Table1 outlines the fuel compositions of these four syn-gas mixtures, where the stoichiometric mixturefraction values increase from 0.124 of flame A to0.462 of flame D, depending on the hydrogen per-centage of the syngas fuel mixture. Flame A con-tains the highest percentage of hydrogen (70.3%),flames B and C contain moderate percentages ofhydrogen (33.4% and 34.6%), while flame D con-tains the lowest percentage of hydrogen (11.4%).

K.K.J. Ranga Dinesh et al. / Proceedings of the Combustion Institute 34 (2013) 3219–3229 3223

The flame characteristics and near-wall heat trans-fer of the syngas combustion are provided here toillustrate the effects of fuel variability on the flamecharacteristics. Since the focus of the study is onthe instantaneous characteristics of flame struc-tures, only typical and representative instanta-neous results at the developed stage are showninstead of the time-averaged statistics.

Figures 2 and 3 shows the mid-plane instanta-neous flame temperature distributions of flamesA, B, C and D at non-dimensional time instantsof t = 12 and 16 respectively, when the flames havebeen developed. The temperature distributionsshow significant structural changes in the primaryand wall jet regions between the four syngasflames. In these syngas flames, combustion occursin a thin layer in the vicinity of the stoichiometricsurface of the non-premixed flames. The combus-tion reaction zone of the H2-rich syngas mixture(flame A) is considerably wider than the H2-leansyngas mixture (flame D), leading to wider temper-ature distributions in flame A. From A to D, theflame becomes increasingly thinner and morewrinkled. In a non-premixed flame, diffusion playsa major role in the localised fuel/air mixing, whichin turn controls the temperature in the reactionzone. Accordingly the flame may be thicker if thefuel has a higher diffusivity, which is indeed thecase for the H2-rich syngas flame A. When theflame is thicker, the local flow gradients becomesmaller, leading to less wrinkled structures. As aresult of inlet perturbation, an asymmetric flamestructure is observed for the temperature distribu-tion in the mid-plane of all four flames. In the pri-mary jet stream, there is a large outer vorticalstructure, which is due to the buoyancy effects[31]. In both Figs. 2 and 3, it is evident that the fuelvariability has a direct impact on the distributionsof the flame temperature and the vortex topology.The H2-lean flame D exhibits larger and morewrinkled outer vortical structures in the primaryjet stream, as well as stronger vortical structuresin the wall jet region than the other cases. Theresults in Figs. 2 and 3 reveal that the concentra-tion of hydrogen in the syngas fuel does affectthe vortex generation in both the primary andthe wall jet regions. This can be attributed to theeffect of diffusivity which depends on the hydrogen

Table 1Composition of the syngas fuels (from BP syngas datasheet).

Case Flame A

H2/CO ratio 2.36H2% 70.3CO% 29.7CO2%N2%Stoichiometric mixture fraction 0.124Maximum flame temperature 8.3 (2430 K)

amount in the syngas mixture. The effect of fuelvariability on flame structures as observed in Figs.2 and 3 is of relevance to H2-rich and H2-lean com-bustors under practical conditions.

The variation of maximum flame temperaturesfor H2-rich and H2-lean fuels is an important issuefor the design of combustors for syngas combus-tion. Figures 2 and 3 show the distributions offlame temperature from the H2-rich flame A tothe H2-lean flame D. The peak temperature of syn-gas flames A, B, C and D are shown in Table 1. Ithas been found that the peak temperature is muchhigher for the H2-rich flame A compared to the H2-lean but H2-rich flame D. This can be attributed tothe different amount of combustible H2 in the fuelmixture and the different transport properties ofthe mixture. Between Figs. 2 and 3, unsteadybehaviour of the flame can be observed, wherethe flame especially the wall jet flame has pro-gressed more at the later time instant shown inFig. 3 compared with the earlier flame shown inFig. 2, but the effects of fuel variability remainthe same. For the flame temperature distribution,experimental measurements of turbulent non-pre-mixed hydrogen enriched flames showed a maxi-mum flame temperature [32,8] in the rangeobserved in the present H2-rich flame. Althoughno directly comparable experimental data is avail-able, the temperatures of the numerically simu-lated flames seem to be in line with theexperimental data. To further analyse the temper-ature variations with respect to syngas mixtures inboth the primary jet and the near-wall regions, thescattered temperature distributions for all foursyngas flames are discussed next.

Figure 4 shows the scattered temperature dataplotted versus mixture fraction in the cross-sec-tional planes at three different axial locationsx = 0.7, 1.4 and 3.5 for flames A, B, C and D,respectively. These three axial locations have beenselected in the near-nozzle region (x = 0.7),upstream vortex region (x = 1.4) and near-wallregion (x = 3.5). In Fig. 4, the zoomed-in high-light of the variation of scattered temperaturedata near the stoichiometric mixture fraction isalso shown. Since scattered temperature resultsmay provide important insight of the flame char-acteristics, it is worthwhile to analyse these plots

Flame B Flame C Flame D

0.50 0.98 0.3733.4 34.6 11.466.6 35.3 31.1

30.157.5

0.220 0.332 0.4628.0 (2344 K) 7.2 (2109 K) 5.9 (1728K)

Fig. 2. Instantaneous flame temperature of syngas flames A, B, C and D at t = 12.

3224 K.K.J. Ranga Dinesh et al. / Proceedings of the Combustion Institute 34 (2013) 3219–3229

to reveal some details of the reacting flow. For afixed value of mixture fraction, a narrow-rangedscattered temperature data corresponds to a uni-formly burnt or un-burnt mixture while a wide-ranged scattered temperature data correspondsto a situation where fuel ignition and extinctionco-exist. The wide range of scattered temperaturedata at a fixed mixture fraction can be due to thefluid dynamic factors such as the existence of vor-tical structures. This scatter data not only revealsvariation of flame temperature with respect tosyngas fuel mixtures, but also displays the effectsof the fluid dynamic factors such as vortical struc-tures. It should be emphasised that these profilesare shown as guides for the temperature of com-bustion mixtures with implications on the rele-vance of flamelet generated manifold to the foursyngas impinging jet flames considered here.

Two important trends have been observed fromFig. 4: (1) the occurrence of peak temperature atthe lean side of the stoichiometric mixture fractionfor all the cases, indicating that the flame temper-ature does not depend on the chemistry only butalso on the transport process; and (2) the scattered

temperature distribution becomes wider from theH2-rich to the H2-lean cases. For the H2-rich flameA, the scattered temperature at x = 0.7, 1.4 and 3.5show similar profiles, which displays smooth andnarrow range of temperature at a fixed mixturefraction. This could be because of the faster fuelconsumption linked with the diffusivity factordepending on the syngas fuel mixture compositionespecially the large amount of hydrogen. It is alsoimportant to note that similar behaviour has beenobserved in previous DNS of non-premixedhydrocarbon flames [33]. The scattered data offlame B which has high CO percentage, displayswider temperature distributions at higher mixturefractions compared to flame A. This might bebecause of the changes of diffusivity level and vari-ations of transport properties. The amount ofhydrogen in the fuel and the variations of trans-port properties associated with fuel variabilitycan change the mixing rate and accordingly thechemical reaction and temperature distribution.The change of trend from flames A to B becomesmore evident for flames C and D, where flame Chas approximately equal percentage of H2, CO

Fig. 3. Instantaneous flame temperature of syngas flames A, B, C and D at t = 16.

K.K.J. Ranga Dinesh et al. / Proceedings of the Combustion Institute 34 (2013) 3219–3229 3225

and CO2 while flame D has low H2 but high N2

percentages. In both cases, the flame temperaturecontains more scatter data at the rich side of thestoichiometric mixture fraction. The proportionof data points away from the fully burnt limitsindicates the presence of unburnt or partiallyburnt mixtures. This is consistent with instanta-neous temperature distributions shown in Fig. 3where flames A, B, C and D are becoming increas-ingly more vortical. Furthermore, the scatter plotsalso reveal that the maximum flame temperaturesare significantly different for different cases. Theprevious DNS study of diluted hydrogen-air flame[34] and the present DNS results of syngas flamesindicate that further analysis on scatter plots oftemperature, heat release and scalar dissipationincluding the effect of differential diffusion isneeded to obtain more insights into the mixingand chemical processes including local flameextinction and re-ignition of the non-premixed fuelmixtures. For the flame temperature, it is alsoworth noting that heat losses due to radiationeffects are likely to reduce the maximum tempera-tures. However, inclusion of radiation in the DNS

computations still represents a computationalchallenge due to the excessive amount of calcula-tions involved.

To further analyse the vortical structures in thereacting flow fields, Fig. 5 shows the cross-sec-tional instantaneous velocity vector fields of thefour cases at t = 16, together with sample stream-line traces, where vortical structures are evident.In the reacting flow fields, complex vortical struc-tures dominate the mixing and entrainment pro-cess, which will affect the distributions ofmixture fraction, progress variable and flame tem-perature. In Fig. 5, the H2-rich flame A exhibits abuoyancy-induced vortical structure in the shearlayers (region 1) at the primary jet stream, whilethe structure becomes increasingly stronger inother cases. In the wall jet stream, all four flamesshow a head vortex ring which is the large struc-ture at the far side in wall jet region (region 2).In Fig. 5, the major difference between the fourcases seems to be in the region 3 of the wall jetregion. For the H2-rich flame A, there is no obvi-ous vortex in this region near to the stagnationpoint. However, a structure corresponding to a

Fig. 4. Scattered temperature data plotted against mixture fraction at three different axial locations: x = 0.7 (greencolour), 1.4 (blue colour), 3.5 (red colour) of flames A, B, C and D at t = 16. (For interpretation of the references tocolour in this figure legend, the reader is referred to the web version of this article.)

3226 K.K.J. Ranga Dinesh et al. / Proceedings of the Combustion Institute 34 (2013) 3219–3229

secondary vortex ring gradually appears in thecases B, C and D. The main reason for vorticitygeneration in the near-wall region is the externalskin friction that acts on the thin layer of the fluidattached to the wall, where viscosity difference dueto the fuel variability and flame temperature vari-ations can play a role in the vortex formation inthe wall jet region.

In an impinging flame, the heat flux into thesolid surface is attributed to the hot products ofcombustion formed along the impinging wall whilethe maximum wall heat flux and temperature areof great importance in practical applications. Thenear-wall heat transfer can be measured by theNusselt number, which is a dimensionless numberthat measures the enhancement of heat transferfrom a surface that occurs in a real situation, com-pared to the heat transfer that would be measuredif only conduction could occur. The Nusselt num-ber can be defined as Nu = hD/k where D is the jetnozzle diameter, k the thermal conductivity of thefluid and h the heat transfer coefficient. The Nus-selt number is used to measure the enhancementof heat transfer when convection takes place.Figure 6 shows comparison of the instantaneous

Nusselt number at t = 12, 16 for flames A, B, Cand D corresponding to the instantaneous heatfluxes near the wall. In all cases, there is no heattransfer in the stagnation point y = 6 because noflame exists here due to the unmixed fuel. The wallheat transfer takes place with increased radial dis-tance from the wall stagnation point showing fluc-tuations in the Nusselt number, which are linkedto the existence of vortical structures near the walljet regions. The instantaneous wall heat fluxes atboth t = 12 and 16 show that the distributions ofwall heat fluxes correspond to the near-wall vortic-ity distributions. In the syngas flames, the combi-nation of different fuel mixtures and buoyancylead to different heat release patterns and thus dif-ferent wall heat transfer characteristics. Corre-sponding to the temperature changes from caseto case, the maximum value of the Nusselt numberis gradually decreasing with respect to hydrogenpercentage in the syngas fuel mixture. This ismainly because near-wall temperature in H2-leanflame D is much lower than its adiabatic value,while for H2-rich flame A it is close to its adiabaticvalue. This is apparent in the head vortex ring(region 2). Figure 5 clearly shows the significance

Fig. 5. Instantaneous velocity vector fields of syngas flames A, B, C and D at t = 16.

K.K.J. Ranga Dinesh et al. / Proceedings of the Combustion Institute 34 (2013) 3219–3229 3227

of the effects of fuel variability on the near-wallheat transfer characteristics, which can be impor-tant to the design of combustors for hydrogen-enriched combustion. Although there is no directlycomparable experimental data to validate the Nus-selt number shown in Fig. 6, the values are withinthe range of experimental measurements ofimpinging flames [35,36]. It is worth noting thatthe present simulations did not take into accountheterogeneous surface chemistry, which can affectthe wall heat transfer related to the wall materialand the near-wall combustion characteristics.

Fig. 6. Instantaneous Nusselt number of syngas flames A (soliddotted-dotted) at the wall in the z = 6.0 plane at (a) t = 12 an

The significant change of flame characteristicswith fuel compositions for the syngas burning indi-cates that the combustion of hydrogen-enrichedfuels can be vastly different from that of the tradi-tional hydrocarbon fuels. This is mainly becausehydrogen has the ability to burn at very lean equiv-alence ratio which is approximately seven timesleaner than gasoline and five times leaner thanmethane [37]. Since H2-rich syngas flames havefaster flame speeds than hydrocarbon flames, H2-rich flames allow oxidation with less heat transferto the surroundings. This might improve thermal

line), B (dashed line), C (dashed-dotted) and D (dashed-d (b) t = 16.

3228 K.K.J. Ranga Dinesh et al. / Proceedings of the Combustion Institute 34 (2013) 3219–3229

efficiencies when syngas is used in practical appli-cations. In addition, efficiencies of H2-rich flamescan also be improved because of small gas quench-ing distance of H2 which allows fuel to burn morecompletely. Furthermore, since the addition of H2

tends to reduce the formation of pollutants andgreenhouse gas (GHG) emissions, syngas fuelscould play a major role in the reduction of localair pollutant and GHG emissions. However, theeffects of fuel variability as well as fuel blendingsuch as adding hydrogen to hydrocarbon fuelsneed further investigations particularly for highpressure conditions which have not been studiedextensively.

5. Conclusions

Instantaneous characteristics of flame struc-tures and flame-wall interactions of four differentsyngas impinging jet flames including the H2-richand H2-lean cases were computationally investi-gated using direct numerical simulation technique.Comparisons were discussed for the flame temper-ature, reaction progress variable, velocity andNusselt number between the four different flames.Some physical insights on the effects of fuel vari-ability of the flame characteristics have beenobtained. It has been found that diffusivity andvortical structures dominate the flame characteris-tics of hydrogen-enriched combustion. Higher dif-fusivity in the fuel leads to smoother and thickerflames that are less vortical. The flame structureand maximum flame temperature largely dependon the syngas fuel compositions. The higher diffu-sivity of the H2-rich flame leads to weaker vorticalstructure in the shear layers at the primary jetcompared to the H2-lean flame. The flame temper-ature of all syngas flames reaches their maximumvalues at the lean side of the stoichiometric mix-ture fraction, due to the specific transport proper-ties of the syngas mixture. It has been found thatthe fuel variability plays a key role in the forma-tion of the vortical structures in the flow fieldsas well as distributions of the near-wall heat flux.The unsteady vortex separation from the wallleads to variations in the Nusselt number distribu-tion due to changes in the instantaneous thermalboundary layer thickness.

Fuel variability has been identified as playing amajor role in the flame characteristics of hydro-gen-enriched combustion. Further investigationson the scatter plots and local flame structures ofH2-rich to H2-lean syngas mixtures would be ofgreat interest, especially with the influence of dif-ferential diffusion on the maximum flame temper-atures. An effort is currently underway to fullyinvestigate the influence of differential diffusionon hydrogen-enriched syngas combustion in tur-bulent flow regimes, while the flow investigatedin this study is mainly in the laminar to transitional

flow regimes. Detailed investigation of the near-wall fluid flow of syngas combustion in turbulentflow regimes would require more significant com-puting resources. Further investigations of themean wall heat flux such as the averaged heat fluxalong the wall, surface chemistry effects, radiationlosses and near-wall vorticity at the boundarylayer should provide relevant information onnear-wall heat transfer for practical applicationsof hydrogen-enriched combustion.

Acknowledgements

This research is funded by the UK EPSRCgrant EP/G062714/2. The syngas mixture compo-sitions were provided by the BP Alternative En-ergy International Ltd.

References

[1] L. Lombardi, Energy Convers. Manage. 44 (1)(2003) 93–108.

[2] A.J. Minchener, J. Fuel 84 (2005) 2222–2235.[3] J. Natarajan, T. Lieuwen, J. Seitzman, Combust.

Flame 151 (2007) 104–119.[4] J. Natarajan, Y. Kochar, T. Lieuwen, J. Seitzman,

Proc. Combust. Inst. 32 (2009) 1261–1268.[5] S.M. Walton, X. He, B.T. Zigler, M.S. Wooldridge,

Proc. Combust. Inst. 31 (2007) 3147–3154.[6] C. Prathap, A. Roy, M.R. Ravi, Combust. Flame

155 (2008) 145–160.[7] M. Karbasi, I. Wierzba, Int. J. Hydrogen Energy 23

(1998) 123–129.[8] R.S. Barlow, G.J. Fiechtner, C.D. Carter, J.Y.

Chen, Combust. Flame 120 (2000) 549–569.[9] N. Wu, G.P. McTaggart-Cowan, W.K. Bushe,

M.H. Davy, Combust. Sci. Technol. 183 (2011) 20–42.

[10] K.R. Anderson, S. Mahalingam, J. Hertzberg,Combust. Flame 118 (1999) 233–247.

[11] K.R. Anderson, S. Mahalingam, ASME J. HeatTransfer 122 (2000) 376–380.

[12] S. Mahalingam, J.H. Chen, L. Vervisch, Combust.Flame 102 (1995) 285–297.

[13] H.G. Im, J.H. Chen, Combust. Flame 126 (2001)1384–1392.

[14] E.R. Hawkes, R. Sankaran, J.C. Sutherland, J.H.Chen, Combust. Flame 31 (2007) 1633–1640.

[15] C.S. Yoo, R. Sankaran, J.H. Chen, J. Fluid Mech.640 (2009) 453–481.

[16] T. Poinsot, D.C. Haworth, G. Bruneaux, Combust.Flame 95 (1993) 118–132.

[17] G. Bruneaux, K. Akselvoll, T. Poinsot, J.H. Ferzi-ger, Combust. Flame 107 (1996) 27–44.

[18] X. Jiang, K.H. Luo, Combust. Flame 133 (2003) 29–45.

[19] F. Dabireau, B. Cuenot, O. Vermorel, T. Poinsot,Combust. Flame 135 (2003) 123–133.

[20] T.M. Alshaalan, C.J. Rutland, Combust. Sci. Tech-nol. 174 (2002) 135–165.

[21] A. Gruber, R. Sankaran, E.R. Hawkes, J.H. Chen,J. Fluid Mech. 658 (2010) 5–32.

K.K.J. Ranga Dinesh et al. / Proceedings of the Combustion Institute 34 (2013) 3219–3229 3229

[22] J.A. van Oijen, L.P.H. de Goey, Combust. Sci.Technol. 161 (2000) 113–137.

[23] N. Peters, Turbulent Combustion, Cambridge Uni-versity Press, 2000.

[24] S.G. Davis, V.J. Ameya, H. Wang, F. Egolfopou-los, Proc. Combust. Inst. 30 (2005) 1283–1292.

[25] L.M.T. Somers, Simulation of Flat Flames withDetailed and Reduced Chemical Models Ph.D. thesis,Eindhoven University of Technology, Eindhoven,The Netherlands, 1994.

[26] A.W. Vreman, J, A. van Oijen, L.P.H. de Goey,R.J.M. Bastiaans, Int. J. Hydrogen Energy 34(2009) 2778–2788.

[27] L.M. Verhoeven, W.J.S. Ramaekers, J.A. vanOijen, L.P.H. de Goey, Combust. Flame 159 (2012)230–241.

[28] S.K. Lele, J. Comput. Phys. 103 (1992) 16–42.

[29] J.H. Williamson, J. Comput. Phys. 35 (1980) 48–56.[30] T.J. Poinsot, S.K. Lele, J. Comput. Phys. 101 (1992)

104–129.[31] K.K.J. Ranga Dinesh, X. Jiang, J.A. van Oijen, Int.

J. Hydrogen Energy 37 (2012) 4502–4515.[32] R.S. Barlow, C.D. Carter, Combust. Flame 97

(1994) 261–280.[33] N. Swaminathan, R.W. Bilger, Combust. Flame 116

(1999) 519–545.[34] R. Hilbert, D. Thevenin, Combust. Flame 138

(2004) 175–187.[35] R. Viskanta, Exp. Thermal Fluid Sci. 6 (1993) 111–

134.[36] D. Lytle, B. Webb, Int. J. Heat Mass Transfer 37

(1994) 1687–1697.[37] C. Bauer, T. Forest, Int. J. Hydrogen Energy 26

(2001) 55–70.