LEAN-BLOWOUT DYNAMICS OF A LIFTED-STABILIZED, NON-PREMIXED SWIRL FLAME Paris A. Fokaides∗1, Nikolaos Zarzalis,1

1 Chair of Combustion Technology, Engler-Bunte-Institute, University of Karlsruhe (TH)

Abstract We report on the lean blowout dynamics of a lifted-stabilized confined non-premixed swirl flame, by apply-ing a novel Airblast atomiser [1]. Within the scope of this work, useful information regarding the flow pat-tern, the mixing evolution and the temperature distribution of a lifted-stabilized swirl flame near its lean blowout limits is obtained. Since flame instabilities in the blowout process are very sensitive to external disturbance, nonintrusive 3D-Laser Doppler Anemometry was applied for the determination of all three mean velocity components as well as the six Reynolds stress components. Measurements for the tem-perature and mixture field were accomplished by adapting in-flame measurement techniques. Carbon monoxide in the post flame regime, formed as a result of local extinction phenomena, served as indicator for the onset of combustion instabilities, and thus of the blowout process. The lean blowout limits of the lifted-stabilized flame are compared with those obtained by an attached swirl flame, yielding that under specific circumstances lifted swirl flames may retain stability in a greater range of equivalence ratios (Φ).

∗ Corresponding author: Paris.Fokaides@vbt.uni-karlsruhe.de Associated Web site: http://www.vbt.uni-karlsruhe.de Engler-Bunte-Ring 1, 76131, Karlsruhe, Germany

1. INTRODUCTION Scope of this study is the examination of the lean blowout behaviour of a lifted-stabilized confined non-premixed swirl flame. The character of the investigated flame is lifted-stabilized within the whole range of its ignitable equivalence ratios (Φ), hence its properties should not be mistaken with those of the blowout intermediate stage of lift-off. Lifted-stabilized flames are regarded as the major application of the partially premixed combustion, through which all the advantageous features of the premixed and non-premixed combustion, concern-ing operational safety, lower pollutant emissions and stability, are exploited. The beneficial aspects of this kind of combustion in lean area concerning the NOx emissions have already been demon-strated in several works [2,3]. This work is moti-vated by these findings and aims to expand the knowledge concerning the fundamental factors governing the stability of lifted swirled flames near their extinction limits. Lean combustion, already explored in several studies [4], is considered to be a breakthrough step which should allow the reduc-tion of NOx emissions. Emission legislations have motivated current lean, premixed combustor de-signs, enhancing however the risk of blowout. NOMENCLATURE

a Heat transfer coefficient [W/m²K] D Diameter [m] K Stability factor [-] L Length [m] •

M Mass flux [kg/s] Pe Peclet-Number [-] Re Reynolds-Number [-]

LS Laminar burning velocity [m/s]

T Temperature [K] u Axial velocity component [m/s] V Mean velocity magnitude [m/s] p Pressure [bar] x Radial coordinate [m] z Axial coordinate [m] Greek Symbols ∆ Finite Difference [-] δ Flame front thickness [m]

Cτ Heat release time scale [s]

Uτ Residence time of reactants [s] Φ Equivalence ratio [-] Subscripts A Air O Referencing nozzle throat PR Preheating R Recirculated Operators

Time averaged

Abbreviation CO Carbon Monoxide EICO Emission Index Carbon Monoxide NOx Nitric Oxides LDA Laser Doppler Anemometry UHC Unburned Hydrocarbons

2. FUNDAMENTALS In practical combustors, sustained combustion can be retained near the lean blowout limits through the fulfilment of the following conditions: • The ignitable mixture should be premixed in

molecular level • The reactants should reach the ignition tem-

perature

THIRD EUROPEAN COMBUSTION MEETING ECM 2007

• The turbulent burning velocity and the flow velocity should be in equilibrium

Under these assumptions, the maintenance of the reaction should be possible if sufficient energy from hot reaction products is released to the un-burned gas mixture. This procedure can be ex-pressed physically in terms of a balance between the time scale for the reaction kinetics ( )Cτ and the residence time of the reaction gases within the stabilization area ( )Uτ . In case the residence time exceeds the necessary time for the complement of the reaction, then combustion is sustained. The stability criterion in this case can be expressed as follows:

KU

C ≈ττ

(1)

According to numerous models [5], ( )Cτ can be expressed by means of a quotient between the flame front thickness ( )δ and the laminar burning velocity ( )LS , whereas ( )δ can be substituted ac-cording to Zeldovich’s analysis of flame propaga-tion with ( )LSa . Hoffmann [6] assumed a perfect stirred reactor, and thus adopted for the determina-tion of ( )Uτ the stability zone length ( )L and the

mean flow velocity ( )u . According to this approach stability is assured if

KuL

Sa 2L

U

C ≈=ττ

( )( )( )a

LSKu

2L≈ (3)

Equation (3) can be expressed in terms of ( )Pe numbers as follows

( )( )( )

( )( )( )

2L

aLSK

aLu

≈ 2

SU LKPePe ≈ (4)

The main purpose of studies concerning the lean blowout limits of combustors is the determination of the factor ( )K for specific combustors, in order to resume its properties from laboratory scale to in-dustrial applications. However, in terms of this work the efforts were focused on the structure and the dynamics of the lifted stabilized flame near its lean blowout limits, and not on the determination of a similarity factor. This methodology was found appropriate, due to the fact that the above theory refers to swirl non-confined diffusion flames, whereas the lifted stabilized swirl flame of this study is categorized under the double swirl partially premixed confined flames. Nevertheless much of the features of the investigated system can be explained through the described time scales bal-ance.

3. EXPERIMENTAL SETUP The investigated atomiser is an assembly based on the Airblast concept (Fig. 1.) [1]. It consists of a modular arrangement of two radial swirl genera-tors, an atomizer lip which separates the two air-streams from each other within the nozzle, and an air diffuser. Methane is delivered to the atomizer through a central pipe along the axis of the nozzle holder. The primary airflow is fed at the lower half of the nozzle holder and is then surrounded by secondary air, which is supplied through a con-necting section at the upper half of the nozzle as-sembly. In case of the lifted-stabilized flame, the high swirl number of the primary swirl body results in the break down of the vortex, whereas the con-struction of the secondary radial swirler ensures the lift-off of the flame. A test facility operating under atmospheric condi-tions was used in terms of this study. This test stand consists of a nozzle holder, in which the mounting and positioning of the investigated noz-zle occurs, and an optic accessible combustion chamber for the implementation of the measure-ment techniques. The nozzle holder can be moved axially inside the combustion chamber, in order to enable the performance of field measurements along the axial direction. Its concept allows the independent adjustment of the mass flow rate and the preheat temperature of the primary and the secondary flow, whereas in terms of this study this option was not utilized. The combustion chamber is isolated and water cooled in order to ensure the performance of reproducible measurements under nearly adiabatic conditions. Its diameter is four times the throat diameter of the nozzle diffuser, and its length was limited by an annular ring outlet at four and a half times its diameter. This outlet geometry was chosen in order to ensure that no back-flow through the exit section occurs. Optical access to the cylindrical burning chamber is en-abled through flat silica-glass-windows which are applied appropriate to the chosen LDA arrange-ment. Moreover, four connecting sections are

Figure 1. : Investigated Airblast atomiser

THIRD EUROPEAN COMBUSTION MEETING ECM 2007

Atmospheric Pressure Experiments (Methane)

Atmospheric Pressure Experiments (Methane and JetA1)

Lifted-Stabilized – Field Measurement Lifted-Stabilized; Attached Lean Blowout Measuerement

T PR,A [K] 373 T PR,A [K] 373 – 673 pA [bar] 1 pA [bar] 1

AM•

[kg/s] 0.00185 ∆Ρ [%] 2

uo [m/s] 40 Re [-] 45000 ∆Ρ [%] 2 Φ [-] 0.65 ; 0.45

Table 1. Operating conditions for selected test case

attached to the combustion chamber to provide access for suction probe measurements. The methodology of this work is as follows: two similar Airblast-nozzles producing a lifted-stabilized flame and an attached flame were employed for the determination of their lean blowout limits. A more detailed investigation of the flow field proper-ties was then performed for the lifted-stabilized flame for two operating points; a stable one by Φ=0.625 – identified in this work as stable, and an equivalence ratio (Φ) located at the lean-blowout regime of the flame, at Φ=0.45 – identified as non-stable. Operation conditions are summarized in table 1. A measurement campaign employing a commer-cial 3D-DANTEC LDA was conducted for the time-resolved determination of all three velocity compo-nents. Multi-species concentrations in the field have been conducted by means of conventional gas analysis, based on molecular excitation proc-esses, due to modulated infrared absorption. Two commercial analysers (Hartman and Braun Model URAS 14 and Magnos 14) and a suction probe of 1 mm were applied. High frequency temperature measurements were performed by means of an S type (Pt/Pt -10 percent Rh) compensated micro thermocouple probe, with wire diameter (D) of 100 µm. 4. RESULTS AND DISCUSSION 4.1. Lean blowout limits The blowout phenomenon is considered to be a transient process, with a series of events occurring consecutively. According to the dynamics of swirl flames, blowout process can be divided into three regions:

• The instability onset • The flame lift-off • The extinction

If the laminar burning velocity and thus the turbu-lent burning velocity in combustion regimes be-come lower than the flow velocity the onset of the instability in terms of pulsation is observed. Further reduction of the turbulent burning velocity due to temperature drop leads to the lift-off of the flame and finally to its extinction. The problem becomes more complicated when the lean blowout of a lifted-stabilized flame is examined. The flame pul-sation is followed by its quenching and thus no intermediate phase is observed. The intense stretching of the flame which is caused due to the disruption of the balance be-tween the burning and the flow velocities, leads to local extinction areas within the flame. In conse-quence the finite rate of the chemical kinetics drops in levels where the generation of products is lower than the rate with which reactants are deliv-ered. The combustion temperature gradually de-creases and the reaction rate becomes low

Figure 2. : CO emissions of lifted-stabilized flame for stable (right) and non-stable (left)

combustion

THIRD EUROPEAN COMBUSTION MEETING ECM 2007

enough to terminate the combustion. Under these conditions the residence time of the hot reactants is not long enough for the complement of the ele-mentary reactions and thus intermediate products, such as UHC and CO in post combustion area are detected. Figure 2 demonstrates a comparison between the CO distribution in the lifted-stabilized flame for the two examined operation conditions. These results are in a very good agreement with the measured temperature fields, as well as with the spontane-ous chemiluminescence of the flame, locating the main reaction zone. In case of the stable combus-tion, the reaction zone is located at 100mm stream upwards from the nozzle and has a compact an-chor shape form, whereas in the post-combustion region, CO is fully oxidized. The widespread re-gion in which carbon monoxide is measured in case of the non-stable measurement is caused by the effective restriction of its oxidation due to the decay of the reaction rate. The highest concentra-tion of carbon monoxide in the vicinity of the com-bustion chamber walls can be attributed to the flame-wall interaction phenomena. A confined flame is constantly exposed to wall-induced heat losses. When the magnitude of the wall heat flux exceeds the heat generated by combustion, then the flame is locally quenched, resulting in large amounts of unburned hydrocarbons and interme-diate products. Comparison of the stable (Fig. 2) and the quenching field yields that by the lean blowout limits of a swirl confined flame, regions near the walls are ominous in quenching, even if combustion is still sustained in the flame core.

Figure 3. : CO formation near lean blowout lim-

its and instability regimes Since this study concerns confined combustion and optical observation of the flame status is not possible, the blowout behaviour was determined by means of CO measurements in a reference point in the post combustion area. The form of CO was related to the instability onset, as well as with the pulsation of the flame and finally its extinc-tion.(Fig. 3) The experiments were performed in such a way that the air flow was exactly adjusted

and the fuel flow was gradually reduced until the instability limits were reached. In this way the pres-sure drop and thus the combustor entry flow veloc-ity were held constant. Figure 4 shows a comparison of the results ob-tained for a lifted and an attached swirl flame against the preheating temperature. The attached swirl flame becomes unstable at 0.53<Φ<0.44, and extinction occurs at 0.5<Φ<0.41. The lean blowout limits of the lifted- stabilized flame is lower than hat of the attached flame (0.44<Φ<0.41) and seems relatively insensitive to preheating tempera-ture. These findings seem to be consistent with previous works studying the stability of an attached and a lifted swirl flame [2].

Figure 4. : Pulsation Onset and Extinction of

attached and lifted-stabilized swirl flame 4.2. Field Measurements The main features of the lifted-stabilized swirl flame for the investigated conditions are

Figure 5. : Stream lines and mean velocity con-

tour (V) of lifted-stabilized flame for stable (right) and non-stable (left) combustion

THIRD EUROPEAN COMBUSTION MEETING ECM 2007

Figure 6. : Axial velocity (u) at nozzle vicinity for stable (right) and non-stable (left) combustion

demonstrated in Fig. 5, in terms of the stream lines and the magnitude of the mean velocity contour of the non-stable operating point. As it is shown by the stream lines, the mean-structure of the lifted-stabilized flame remains remarkably unaffected near its lean blowout limits. This confirms the lean blowout behaviour of the lifted-stabilized flame, and thus the absence of an intermediate status and its limited unstable boundary. The structure of the lifted stabilized non-premixed swirl flame was detailed described in our previous work [3]. This flame is devoid of an inner recirculation zone within the combustion area, and its stability is maintained mainly due to a considerably large outer recircula-tion zone. The swirl enables the macromixing in the pre-combustion area, whereas the hot outer recirculation zone provides the mixture the re-quired energy, by means of heat transfer, so that ignitable levels, regarding molecular mixing and temperature, can be accomplished. Under these conditions, the lifted swirl flame was found to be homogeneous and devoid of temperature peaks, resulting in combustion with considerably lower NOx Emissions.

Figure 7. : Temperature field (T) for stable (right) and non-stable (left) combustion

The lift-off height of the lifted-stabilized flame seems to be affected near the blowout limits. Ac-cording to the theory of the lifted jet turbulent flames, the lift-off height of a lifted-stabilized flame should depend linear on its jet exit velocity[7]. However, in this case, although the jet velocity is kept constant (Fig.6), the flame is displaced stream upwards in case of the non-stable operation, and thus of the leaner combustion (Fig.7.). Although the mixing procedure (Fig.8) as well is unaffected in case of the non-stable combustion, local extinc-tion effects, as well as the drop of the temperature of the recirculated gases, may postpone the igni-tion and thus elevate the lift-off height. This em-phasizes the role of the heat transfer from the hot gases around the jet, to the jet core, regarding the ignition and thus the stabilization of the lifted-stabilized flame.

Figure 8. : Local equivalence ratio (Φ) for stable

(right) and non-stable (left) combustion

The confinement of the flow and the size of the outer recirculation zone also explain the insensi-tiveness of the lean blowout limits of the lifted sta-bilized flame for elevated air preheated tempera-

THIRD EUROPEAN COMBUSTION MEETING ECM 2007

ture. In case of attached swirl flames, the increase of the preheated temperature expands the lean blow out limits, as this has an impact on the turbu-lent burning velocity and thus the balance between this velocity and the flow velocity is achieved for lower equivalence ratios (Φ) (Eq.3). However, in case of the lifted stabilized flame, the reduction of

Figure 9. : Recirculated mass (Mr) for lifted

stabilized swirl flame the density ratio between the downstream propa-gating unburnt mixture to the surrounding burnout fluid results in the increased entrainment of hot products in the flame core (Fig.9). This effect obvi-ously counterbalances the turbulent velocity im-pact, and thus for elevated preheating tempera-tures the lean blowout limits seem more or less indifferent (Fig.4). CONCLUSIONS The lean blowout behaviour of a lifted-stabilized confined non-premixed swirl flame has been ex-perimentally investigated. Two different swirl flames, an attached and a lifted-stabilized, where examined by means of conventional lean limit measurements. The lifted-stabilized flame was found to be more insensitive in intermediate insta-bilities and thus lower equivalence ratio (Φ) were achieved. The results also provide some very useful information regarding the velocity field of a lifted-stabilized swirl flame near its extinction, its turbulence, as well as inside look concerning the mixing procedure and the temperature distribution. While the mixing field and the stream pattern retain their properties, the drop of temperature in the recirculated gases plays a crucial role in the elimi-nation of the reaction rate, and thus to the lift-off height of the lifted-stabilized flame. The interaction of the chamber wall with the flame proved to have a crucial role on the flame stability, as local extinc-tion phenomena occurred.

Acknowledgments We kindly acknowledge financial support by the European Commission through project TLC (Con-tract No. AST4-CT-2005-012326 Literature [1] European Patent Office, Application No./ Pat-ent. No. 06009563.5-, 2006 [2] Johnson, M.R. et al., Proceedings of 30th Sym-posium (International) on Combustion, pp. 2867-2874, 2004 [3] Fokaides, P.A. et al.: Proceedings of Turbo Expo 2007, GT 2007-27126, in Press [4] Zarzalis, N. et al.: Proceedings of RTO Meeting 14, pp. 7.1-7.12, 1999 [5] Bradley, D.: Proceedings of 24th Symposium (International) on Combustion, pp. 247–262, 1992 [6] Hoffmann, S.: Untersuchungen der Stabilisie-rungsverhaltens und der Stabilitätsgrenzen von Drallflammen mit innerer Rückströmzone, Disserta-tion University of Karlsruhe, 1994 [7] Peters, N.: Turbulent Combustion. Cambridge University Press, 1999

THIRD EUROPEAN COMBUSTION MEETING ECM 2007