lable at ScienceDirect

Applied Thermal Engineering 63 (2014) 559e564

Contents lists avai

Applied Thermal Engineering

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

Highly flexible burner concept for research on combustiontechnologies with recirculation of hot combustion products

Carlos E. Arrieta*, Andrés A. AmellScience and Technology of Gases and Rational Use of Energy Group, Faculty of Engineering, University of Antioquia UdeA, Calle 70 No. 52-21, Medellín,Colombia

a r t i c l e i n f o

Article history:Received 8 August 2013Accepted 24 November 2013Available online 3 December 2013

Keywords:Research burnerFlue gas recirculationMILD combustionLaminar flamesMethane

* Corresponding author.E-mail addresses: carlosarrieta@udea.edu.co, ca

E. Arrieta), anamell@udea.edu.co (A.A. Amell).

1359-4311/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.applthermaleng.2013.11.049

a b s t r a c t

This paper reports the development and testing of a research coflow burner that generates laminarflames in a hot and diluted environment, which is adequate for studying the operating conditions foundin practical combustors that use flue gas recirculation techniques. The burner has two flame zones; thefirst is an annular laminar premixed flat flame stabilized by a perforated plate, which generates a hotoxygen-rich flue gas mixture. The second is a non-premixed laminar flame, which uses the hot oxygen-rich flue gas mixture as an oxidizer. A methodology based on coflow calculations, which is the mostsignificant component, was designed to provide different temperatures and dilution levels. The burnerwas built based on this design and tested using both intrusive and non-intrusive measurement tech-niques. The device emulated several features of practical combustor systems but with a simpler geom-etry, well-defined boundary conditions and using the simplest configuration found in literature (laminarflames). In a preliminary study, the MILD combustion regime was emulated by generating methanelaminar non-premixed flames at oxygen concentrations between 3% and 9% and temperatures between973 K and 1173 K.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Combustion processes have been long-standing topics of scien-tific and technological interest because of their impact on theoptimization, efficiency and economy of energy conversion systems.Recently, the strong dependence of industry and finance on fossilfuels initiated a new phase of investigation in the field of combus-tion in which the development of advanced combustion technolo-gies has been a high priority issue. Several current studies haveshown that such developments can be achieved much faster, morecheaply and more precisely if adequate computational tools andcombustion models were available [1,2]. However, in spite of sig-nificant advances in computational tools, the modeling of advancedcombustion techniques still represents great challenges because thechemical and fluid-dynamic phenomena that occur in these devicesare highly complex. The chemistry, fluid-dynamics, heat transferand mass transport interact in a practical combustor. These phe-nomena are highly sensitive to parameters such as the temperature,gas composition, turbulence level and dilution levels [3]. Studyingadvanced combustion regimes in burners that allow the control of

rlos.e.arrieta@gmail.com (C.

All rights reserved.

these parameters while maintaining the physical phenomena ofinterest might yield valuable information. For example, in a reviewbyMasri et al. [4], the authors argue that laminarflames stabilized inthis type of burner could be suitable for studying chemical kinetics,flame extinction limits, soot formation, and thermal radiation.

The recirculation of combustion products is one of the mostsignificant combustion techniques that have been extensivelystudied over the past two decades. However, few studies hadattempted to model the chemistry of recirculation prior to the lastdecade [5]. In addition to fluid-dynamic and chemical challenges,the limited optical access found in most flue gas recirculationequipment complicates the implementation of nonintrusive diag-nostic techniques, especially those used to develop and validatecombustion models.

The emergence of technologies like MILD combustion [6,7] hasmotivated the study of the impact of the addition of diluents tolaminar non-premixed flames. Because CO2 and N2 are two of themain combustion products, many studies have been conducted toanalyze the effects of these species on the properties of non-premixed laminar flames, such extinction limits, flame length,and pollutant emission [8,9].

McEnally et al. [5] presented a coflow burner with a simplegeometry and unconfined reaction zone. The burner generatedatmospheric, axisymmetric non-premixed laminar flames. By using

C.E. Arrieta, A.A. Amell / Applied Thermal Engineering 63 (2014) 559e564560

intrusive (thermocouple techniques) and nonintrusive diagnostictechniques (Rayleigh scattering), they attempted to validate achemistry model that simulates soot formation in aircraft turbines.The fuel (ethylene diluted with nitrogen) emerged from a verticalbrass tube in the burner, and the oxidizer (air) emerged from theannular region between this tube and a 50mmdiameter concentrictube. This work was the first attempt to validate a detailed chem-istry model that incorporated amultiple section soot growthmodelin a flame with well-defined boundary conditions. However, thisexperimental setup does not emulate the operating conditions ofmany practical combustor devices in which the fuel is dischargedinto a hot mixture of oxidant and combustion products. Morerecently, Min et al. and Guo et al. [10,11] used a confined co-flowburner to experimentally and numerically study the influence ofair dilution on laminar non-premixed methane flames. The flameswere generated in a combustion chamber with square cross-sectional area, a side length of 25 cm and a height of 80 cm. Thefuel (methane) flowed into the reaction zone through a round tube,which was mounted at the center of the chamber bottom. Theoxidant emerged from the space between the fuel tube and theinner wall of the combustion chamber. They used a flexibleexperimental setup that simulates the vitiated environment of arecirculation burner by adjusting the flow rate ratio between airand diluents (CO2 and N2), but at lower temperatures than thatfound in a practical combustor.

Several recent studies have been performed in burners similarto those discussed above. Most of these studies first noted that a co-flow burner, especially a vitiated co-flow burner, allows the study ofthe recirculation process without complex recirculating flows. Itssimplified geometry provides data and boundary condition infor-mation sufficient to validate numerical simulations. Second, thereis a renewed interest in laminar non-premixed flames. Under-standing the transport and chemical reaction process in theseflames can improve the models of combustion under various con-ditions. Some authors have highlighted that laminar non-premixedflames can be considered “adequate tools” [12] or as “benchmarktest” [13] to validate the accuracy of numerical models.

A laboratory scale burner to analyze the effects of both co-flowtemperature and dilution level on the properties of non-premixedlaminar flames was developed based on a hot coflow burner. Thisburner is similar to that used by McEnally et al. [5], but it features asecondary burner in the coflow, which stabilized a lean laminarpremixed flame that provides oxygen-rich combustion productswhere the laminar fuel jet is discharged. The temperature and ox-ygen concentration of the co-flow can be modulated to provide a

Figure 1. Schematic repres

wide range of operating conditions to be investigated. In this paper,we show some details of the design and operation of this burner. Tocorroborate the great flexibility of this device, experimental resultsthat demonstrate its performance under MILD combustion condi-tions are presented.

2. The vitiated hot coflow laminar burner

The main characteristic of the burner is the possible of gener-ating atmospheric, axisymmetric non-premixed laminar flames in avitiated and hot environment where the temperature and dilutionlevel can be controlled. It consists of a central non-premixedlaminar flame (primary burner) surrounded by oxygen-rich com-bustion products that comes from a lean premix flat flame (sec-ondary burner), which is stabilized on a porous surface. Fig.1 showsa schematic representation of the burner.

From Fig. 1:

� In the primary burner the fuel flows into the reaction zone froma stainless steel tubing with inner and outer diameters of 5 mmand 6 mm, respectively. The tubing passes through the mixingchamber and the secondary burner to finally inject the fuel inthe center of the oxygen-rich combustion products by aninterchangeable nozzle.

� In the mixing chamber, the fuel, oxidizer and inert enter sepa-rately in adequate proportions to guarantee the desire temper-ature and air dilution in the coflow. Themixing chamber is madeof a stainless steel round tube. Inside the mixing chamber thereare two flame arrestors to prevent accidents. Before the zonewhere the flame arrestors are located, there is a laminarizationsystem which consists of an annular honeycomb body.

� In the coflow, to stabilize the premixed flat flame we use aperforated platewith diameter (Dcoflow) and porosity of 40.9mmand 27% respectively. This perforated plate is designed to beeasily interchangeable. Additionally, a water cooling circuit wasimplemented to refrigerate the burner plate and the walls thatprotect the flat flame from the influence of the surrounding air.

3. Methodology

3.1. Design and operation methodology

An iterative procedure was implemented to design and operatethe secondary burner, which probably is the most relevant

entation of the burner.

C.E. Arrieta, A.A. Amell / Applied Thermal Engineering 63 (2014) 559e564 561

component. First, the reactant components needed to be deter-mined according to the following criteria:

� Because a low porosity is used in the secondary area, the flamestability is an important criterion. In this work, a CH4/H2mixtureis used to better control the laminar burning velocities. UsingCH4/H2 mixtures allow working with a wide range of laminarburning velocities as it is shown by Ilbas et al. [14].

� In addition to the fuel and oxidizer, a third specie (inert) isneeded to control the temperature of the combustion products.Nitrogen (N2) and carbon dioxide (CO2) can act as this inertspecie. In this work, nitrogen is used to control the combustionproducts temperature. Although the cooling properties of CO2are better than those of N2 at high temperatures, a recent studyby Burbano et al. [15] indicated that the laminar burning ve-locity decreases considerably for mixtures diluted with CO2compared to those diluted with N2. Therefore, using CO2 doesnot meet the first criterion.

Once the reactant components have been chosen, the globalreaction equation can be derived (Equation (1)) because the reac-tion in the second burner is expected to be complete.

aCH4 þ bN2 þ uH2 þ nðgðO2 þ 3:76N2ÞÞ/tCO2 þmH2Oþ fN2 þ sO2

(1)

Where a,b and u are themoles of CH4, N2 and H2 respectively. g andn are the stoichiometric coefficient for air and the air factorrespectively. t,m, f and s are the moles of the combustion products.

The iterative model consists of supposing composition values asshown in Equation (2) andmodifying the air factor until the desiredoxygen concentration in the combustion products is achieved.

aþ bþ u ¼ 1 (2)

The adiabatic temperature and laminar premixed flame canthen be numerically calculated using the EQUIL and PREMIX(CHEMKIN-PRO package) software, respectively. In this work, thedetailed reaction mechanism Gri-Mech 3.0 is used. The concen-trations in Equation (2) must be modified until the desired dilutionlevel and temperature are found.

For example, Table 1 shows the results of the iterative processfor three dilutions levels and a temperature of 1700 K.

In order to provide laminar flows with a Reynolds number lessthan 2000 four parameters must be considered: the density andviscosity of the reactant mixtures that generated combustionproducts with an oxygen concentration of 3%e9% (typical values of

Table 1Results of the iterative process for three dilutions levels and adiabatic temperatureof 1700 K.

Percentage of O2 in the hot coflow 9% 6% 3%

Reactant molesa 0.500 0.120 0.032b 0.000 0.600 0.680u 0.500 0.280 0.288g 1.250 0.400 0.208n 1.850 1.590 1.315

Product molest 0.500 0.120 0.032m 1.500 0.520 0.352f 8.687 2.869 1.708s 1.060 0.223 0.065Adiabatic temperature [K] 1700 KLaminar burning velocity [cm/s] 10.58 14.94 25.98

MILD combustion [16,17]), flow velocities of 0.1 m/s to 0.7 m/s(typical values found in cold-coflows [5,10]) and a coflow diameterof 40.9 mm.

3.2. Experimental setup

Fig. 2 shows a schematic diagram of the experimental setupimplemented to evaluate the performance of the burner. High pu-rity certified gases were used to generate the required mixtures.The mixture composition was prepared using rotameters specif-ically calibrated for each component gas.

Several laminar non-premixed methane flames were generatedin the vitiated hot environment provided by the secondary burner.A configuration in which the O2 content of this environment varied3%, 6% and 9% was studied. The adiabatic temperature of the coflowwas set to 1400 K and the exit velocity was maintained at 0.5 m/s. Asecond configuration was studied by varying the adiabatic tem-perature of the coflow (1500 K,1700 K and 1900 K) andmaintainingthe O2 content in the hot combustion products at 3%. The exit ve-locity of the coflow was maintained at 0.5 m/s. Laminar non-premixed methane flames were generated with air flowingthrough the coflow to compare the results. For all cases, the Rey-nolds number of the coflowgases and themethane flowing throughthe center nozzle was maintained at approximately 1600 for bothflows.

To determine the flame properties, such as liftoff distance, flamecolor and size, direct flame photographs and direct chem-iluminescence photographs were captured with a digital cameraand an ICCD camera, respectively. The ICCD camera was a1024 � 1024 pixel Princeton PI-MAX equipped with a band-passfilter calibrated on the CH* wavelength (387 nme430 nm). Aswill be shown later, the exposure time was modified in both thedigital and ICCD camera because the flame luminosity decreasessignificantly with increases of the dilution level.

A SICK MAIHAK S710 analyzer was used to measure the oxygenconcentration of the coflow at a point where the combustionproducts remain free from the surrounding air. The analyzer usesthe paramagnetic principle. The flame axial temperature profileswere measured with a 35 mmwire-diameter Type R thermocouple.A rapid insertion procedure was used to avoid soot deposition ontothe thermocouple joint, especially near the flame tick.

4. Experimental results and discussions

4.1. Variation of the oxygen content

Due to the absence of adiabatic conditions, the registered tem-peratures of the hot-vitiated co-flow were 1123 K (3% O2) and1173 K (6% and 9% O2). The measurements of the oxygen concen-tration agree with the EQUIL code calculations.

Comparing the temperature profiles of the flames in Fig. 3demonstrates the benefits of MILD combustion. Specifically, therecirculation of hot combustion products, which are present inhigher proportions in flames with 3% and 6% oxygen content in theco-flow, slows the reactions and distributes the reaction zone. Asdescribed by Szegö [18], distributing the heat release to a largervolume results in a nearly uniform temperature distribution withreduced peak temperatures.

Although the temperature profile of the 9% oxygen concentra-tion flame remained mostly unchanged, it was similar to that of theconventional flame. The second peak is most likely due to the influxof air from the surroundings. The proportion of inert gases was notconducive to a constant temperature.

The direct flame photographs and direct chemiluminescencephotographs shown in Figs. 4e6 indicate that reducing the O2

Fig. 2. Schematic diagram of the experimental setup. 1 e Rotameters, 2 e Burner, 3 e Reaction zone, 4 eWater cooling circuit, 5 e Thermocouple, 6 e Rail system, 7 e CH* filter, 8 e

ICCD camera, 9 e Computer, 10 e CCD camera, 11 e Images, 12 e Fuel (primary burner), 13 e Oxidizer and inert (secondary burner), 14 e Fuel (secondary burner), 15 e Pump andtank.

C.E. Arrieta, A.A. Amell / Applied Thermal Engineering 63 (2014) 559e564562

content and maintaining a constant co-flow temperature dims theflame, produces less soot and increases the reaction zone.

The decrease in brightness is evident in the images taken withthe CCD and the ICCD cameras, which are shown in Fig. 4 and Figs. 5and 6, respectively. A longer exposure time than those used forconventional and 9% oxygen co-flow flames was necessary toobserve flames that are generatedwith an oxygen content of 6% and3% in the co-flow. This extended exposure time is necessarybecause the flame luminosity is attributed to the light emission ofsome radicals in the reaction zone, and any decrease in tempera-ture or concentration of the light emitting radical may reduce thebrightness.

Fig. 3. Axial temperature profile of the non-premixed methane flame for differentoxygen contents in the hot co-flow.

4.2. Variation of the coflow temperature

For this configuration, the oxygen content of the hot vitiated co-flow was maintained at 3%. The adiabatic temperatures of thepremixed flames in the secondary burner were set at 1515 K,1700 Kand 1900 K, but the registered temperatures of the hot vitiatedcoflow were 1053 K, 1133 K and 1256 K, respectively, due to theabsence of adiabatic conditions.

Fig. 4. CCD Flames photographs at different oxygen content in the hot coflow. a.Conventional flame with exposure time of 0.01 s, b. 9% of O2 in the hot vitiated coflowwith exposure time of 0.01 s, c. 6% of O2 in the hot vitiated coflow with exposure timeof 0.01 s, d. 6% of O2 in the hot vitiated coflowwith exposure time of 0.5 s, e. 3% of O2 inthe hot vitiated coflow with exposure time of 0.5 s.

Fig. 5. CH* chemiluminescence images for different oxygen content in the hot coflow.a. Conventional flame with exposure time of 25 ms, b. 9% of O2 in the hot vitiatedcoflow with exposure time of 25 ms, c. 6% of O2 in the hot vitiated coflow withexposure time of 25 ms, d. 3% of O2 in the hot vitiated coflow with exposure time of25 ms.

Fig. 7. CH* chemiluminescence images for different temperatures of the co-flow. a.Temperature of the hot vitiated coflow: 1053 K e with exposure time of 75 ms, b.Temperature of the hot vitiated co-flow: 1133 K e with exposure time of 75 ms, c.Temperature of the hot vitiated co-flow: 1256 K e with exposure time of 75 ms.

C.E. Arrieta, A.A. Amell / Applied Thermal Engineering 63 (2014) 559e564 563

Fig. 7 presents the CH* chemiluminescence images for the threestudied coflow temperatures. The images agree with the resultsdescribed in previous studies of laboratory-scale MILD combustionfurnaces by Rottier et al. and Ishiguro et al. [19,20]: The liftoff dis-tance of the main reaction zone increased when the preheating

Fig. 6. CH* chemiluminescence images for different oxygen content in the hot coflow.a. 6% of O2 in the hot vitiated coflow with exposure time of 75 ms, c. 3% of O2 in the hotvitiated coflow with exposure time of 75 ms.

temperature was decreased. Conversely, a decrease in the pre-heating temperature decreased the intensity of CH* chem-iluminescence signals, which indicated a decrease in the local heatrelease density.

Fig. 8 shows the influence of the preheating temperature on theaxial profile temperature of the flame. The axial temperature profileis less homogeneous for the lowest preheating temperatures.Therefore, as described by Orsino et al. [21], the preheating tem-perature inversely correlates with the flame temperature gradient.

Fig. 8. Axial temperature profile of the non-premixed methane flame for differenttemperatures of the coflow.

C.E. Arrieta, A.A. Amell / Applied Thermal Engineering 63 (2014) 559e564564

5. Conclusion

In this paper, we presented a burner with a simple geometrythat provides a controllable temperature and dilution level envi-ronment for the study of laminar non-premixed flames in a hotvitiated environment using both intrusive and non-intrusivediagnostic techniques. The experimental measurements pre-sented here indicate that the design and operation procedure inthis study generated flames in the burner with well-definedboundary conditions that emulate the low chemical reactionrates, low luminosity, uniform temperatures and large reactionzones that are observed in practical combustors that recirculate hotcombustion products. This burner is an attractive option to studythe phenomenological, theoretical and numerical aspects of newcombustion technologies, such as MILD combustion. This designcan also be used to study the effects of atmospheric conditions onthe structure of non-premixed laminar flames, especially at sub-atmospheric conditions, because the most important Latin Amer-ican cities are located at high altitudes [22].

Acknowledgements

The authors would like to acknowledge the Science and Tech-nology of Gases and Rational Use of Energy Group (GASURE) andthe program “Sostenibilidad 2013e2014” of the University ofAntioquia for the valuable economic contribution to the develop-ment of this research.

References

[1] C.K. Westbrook, Y. Mizobuchi, T.J. Poinsot, P.J. Smith, J. Warnatz, Computa-tional combustion, Proc. Combust. Inst. 30 (2005) 125e157.

[2] F. Cadavid, B. Herrera, A. Amell, Numerical simulation of the flow streamsbehavior in a self-regenerative crucible furnace, Appl. Therm. Eng. 30 (2010)826e832.

[3] R.S. Barlow, Laser diagnostics and their interplay with computations to un-derstand turbulent combustion, Proc. Combust. Inst. 31 (2007) 49e75.

[4] A.R. Masri, R.W. Dibble, R.S. Barlow, The structure of turbulent nonpremixedflames revealed by RamaneRayleigheLIF measurements, Prog. EnergyCombust. Sci. 22 (1996) 307e362.

[5] C.S. McEnally, A.M. Schaffer, M.B. Long, L.D. Pfefferle, M.D. Smooke,M.B. Colket, et al., Computational and experimental study of soot formation in

a coflow, laminar ethylene diffusion flame, Symp. (Int.) Combust. 27 (1998)1497e1505.

[6] A. Cavaliere, M. de Joannon, Mild combustion, Prog. Energy Combust. Sci. 30(2004) 329e366.

[7] J.A. Wünning, J.G. Wünning, Flameless oxidation to reduce thermal no-for-mation, Prog. Energy Combust. Sci. 23 (1997) 81e94.

[8] A. Lock, A.M. Briones, S.K. Aggarwal, I.K. Puri, U. Hegde, Liftoff and extinctioncharacteristics of fuel- and air-stream-diluted methaneeair flames, Combust.Flame 149 (2007) 340e352.

[9] H. Guo, G.J. Smallwood, A numerical study on the influence of CO2 addition onsoot formation in an ethylene/air diffusion flame, Combust. Sci. Technol. 180(2008) 1695e1708.

[10] J. Min, F. Baillot, Experimental investigation of the flame extinction processesof nonpremixed methane flames inside an air coflow diluted with CO2, N2, orAr, Combust. Flame 159 (2012) 3502e3517.

[11] H. Guo, J. Min, C. Galizzi, D. Escudié, F. Baillot, A numerical study on the effectsof CO2/N2/Ar addition to air on liftoff of a laminar CH4/air diffusion flame,Combust. Sci. Technol. 182 (2010) 1549e1563.

[12] J. Min, F. Baillot, H. Guo, E. Domingues, M. Talbaut, B. Patte-Rouland, Impact ofCO2, N2 or Ar diluted in air on the length and lifting behavior of a laminardiffusion flame, Proc. Combust. Inst. 33 (2011) 1071e1078.

[13] D. Carbonell, C.D. Perez-Segarra, P.J. Coelho, A. Oliva, Flamelet mathematicalmodels for non-premixed laminar combustion, Combust. Flame 156 (2009)334e347.

[14] M. Ilbas, A. Crayford, I. Yılmaz, P. Bowen, N. Syred, Laminar-burning velocitiesof hydrogeneair and hydrogenemethaneeair mixtures: an experimentalstudy, Int. J. Hydrogen Energy 31 (2006) 1768e1779.

[15] H.J. Burbano, J. Pareja, A.A. Amell, Laminar burning velocities and flame sta-bility analysis of H2/CO/air mixtures with dilution of N2 and CO2, Int. J.Hydrogen Energy 36 (2011) 3232e3242.

[16] M. Mancini, P. Schwöppe, R. Weber, S. Orsino, On mathematical modelling offlameless combustion, Combust. Flame 150 (2007) 54e59.

[17] P. Li, J. Mi, B.B. Dally, F. Wang, L. Wang, Z. Liu, et al., Progress and recent trendin MILD combustion, Sci. China Technol. Sci. 54 (2011) 255e269.

[18] G.G. Szegö, Experimental and Numerical Investigation of a Parallel Jet MILDCombustion Burner System in a Laboratory-scale Furnace, 2010.

[19] T. Ishiguro, S. Tsuge, T. Furuhata, K. Kitagawa, N. Arai, T. Hasegawa, et al.,Homogenization and stabilization during combustion of hydrocarbons withpreheated air, in: Symposium (International) on Combustion, Elsevier, 1998,pp. 3205e3213.

[20] C. Rottier, C. Lacour, G. Godard, B. Taupin, L. Porcheron, R. Hauguel, et al., Onthe effect of air temperature on mild flameless combustion regime of hightemperature furnace, in: Proceedings of the European Combustion MeetingVienna, Austria, 2009.

[21] S. Orsino, R. Weber, U. Bollettini, Numerical simulation of combustion ofnatural gas with high-temperature air, Combust. Sci. Technol. 170 (2001) 1e34.

[22] H.J. Burbano, J. Pareja, A.A. Amell, Laminar burning velocities and flame sta-bility analysis of syngas mixtures at sub-atmospheric pressures, Int. J.Hydrogen Energy 36 (2011) 3243e3252.