Stability Analysis and Flow Characterization of Multi-Perforated Plate Premixed Burners

Transcription

1 Stability Analysis and Flow Characterization of Multi-Perforated Plate Premixed Burners José M. N. Rodrigues 1, Edgar C. Fernandes 1 1: Dept. Mech Eng. Center IN+, Instituto Superior Técnico, University of Lisbon, Portugal * Corresponding author: jose.neves.rodrigues@tecnico.ulisboa.pt Abstract: NO x reduction on multi-perforated plate burners, often used in domestic water-heater appliances, constitutes a great design challenge for manufacturers. Reducing the flame temperature by using lean premixed mixtures is one method used to attain low-no x emissions but this results in instability issues on most existing designs. A study of the stabilization of methane and propane flames on several configurations of burner plates was carried out, concluding that the stabilization degrades as the distance between holes on the hole pattern of the perforated plate increases, whilst the diameter of holes and number of holes are not as relevant. Direct visualization of the flow and PIV measurements suggest that the stabilization of the flame is affected by entrained cold air that flows into the gaps between the main flow jets of air-fuel mixture, diluting the premixed reacting mixture and decreasing the reacting zone temperature and chemical reaction rates, accordingly. The results also show that as the gap, or the distance between holes is higher, the more cold air is entrained and the less stable is the flame. Introduction The control of pollutant emissions produced by combustion has gained a predominant role on the design constrains of domestic appliances using combustion, due to the more and more strict regulations imposed, especially on NO x radicals emissions. Great effort has been applied on the investigation of technologies and burning regimes that lead to low NO x emissions, without compromising CO emissions. A popular method to reduce the production of thermal NO x during combustion is by reducing the flame temperature, which in most cases means burning lean and ultra-lean mixtures. However, most of the burner configurations installed on domestic appliances present combustion instability issues and are not ready to sustain stable combustion at ultra-lean or lean conditions, unless they are modified. These burners consist of a perforated stainless-steel plate, with hole and slits matrixes of 1 to 3 mm in diameter and 1 to mm of space between holes. A study [1] comparing the stability of a commonly used stainless steel hole matrix plate and a porous ceramic matrix with the same hole pattern showed that the ceramic plate was able to attain stable combustion at equivalence ratios close to the flammability limit for a wide range of power output, due to the material porosity, which produces smooth velocity gradients at the burner exit, whereas the stainless-steel plate, with sharp velocity gradients and high heat loss to the plate did not withstand stable combustion for mixtures as low as two times the flammability limit of the fuel and was greatly affected by the increase in power output, showing the importance of the velocity profile at the burner exit to flame stability []. Manufacturers do not possess a great understanding on the dynamics of the combustion governing this type of burners and, consequently, the design process is based on empirical observations, trial and error and on the experience of the designer. Additionally, the designer is often constrained by budget, time and manufacturing cost factors, which leave the designer only a few design variables to account on the optimization process. The combustion instabilities observed on perforated-plate burners are of two types, thermo-acoustic resonance and blow-off, and this work will focus on the latter, namely on the role of the flow velocity distribution on flame stabilization. The mechanisms of flame stabilization and blow-off on Bunsen flames and inverted flames has been extensively studied and analyzed in the past by several researchers. These physical mechanisms are governed by complex coupling between aerodynamic stretching, diffusion effects, heat loss to the burner plate by conduction and radiation. Lewis and von Elbe [3,] proposed the first theory on flame blowoff, which advocates that blowoff occurs when a critical velocity gradient in the nozzle near

2 the burner plate is reached in the unburnt mixture. This theory introduces the concept of flame stretch, which characterizes non-uniformities in the aerodynamic flow motion and curvature of the flame front [] and has been used extensively to study and characterize flame stabilization and blowoff [, ]. However, the consequent theories formulated based on the flame stretch and curvature take only into account the aerodynamic effects and neglect the impact of heat loss, species preferential diffusion and secondary air dilution [, 7, 8]. There is still some debate about the leading mechanisms of flame stabilization, specially on the role of heat loss effect against the predominant effect of flame stretch and several investigations on this matter show contradictory results [9, 10]. A numerical characterization of the blow-off and stabilization mechanisms on perforated-plate burners was investigated by other authors [], concluding that stabilization of a flame is a compromise between heat loss to the plate and the velocity gradients. However, these results were obtained for a perfectly premixed flow, on a -dimensional numerical analysis, where air dilution was not considered. In the present work, a set of stainless steel burning plates is analyzed and their stability is investigated according to three geometric parameters of the burner plate for two different fuels: methane and propane. These parameters are hole diameter, number of holes and distance between holes. First the stability limits of each burner plate are measured and then follows the flow characterization using visualization methods and PIV measurements, focusing on the particular stability mechanisms and on the corresponding effects on such y x Figure 1 Hole pattern on the perforate plates mechanisms of the geometry of the burner plates. Experimental setup description The combustor is constituted by an aluminum mixing chamber where the air and fuel flows are mixed and developed until it reaches the top exit. At the top exit, a stainless-steel 0,mm thick perforated plate distributes the reacting mixture in several small jets, preparing the mixture for combustion. The distribution plate, which is the main subject of this study, will be referred to as burner-plate throughout the remainder of Burner Number of holes Hole Diamater [mm] Table 1 - Burner plate configurations Distance between holes [mm] A 7, 1 B 7 1, C 7 1, 1 D 7, E 19, 1 F 19 1, G 19 1, 3 H 19 1, 1 I 19, J 37 1, 1

3 ϕ U d [s-1] A - BlowOff B - BlowOff C - BlowOff D - BlowOff E - BlowOff F - BlowOff G - BlowOff H - BlowOff I - BlowOff J - BlowOff Cerâmico (LII) Figure Blow-off limits for the plates described on Table 1. this document. A number of burner plates, with different geometric configurations was analyzed during this study. The hole pattern on the perforated plates follows a hexagonal centered distribution, illustrated in Figure 1 and differ from each other on hole diameter, spacing between holes and number of holes. Methane-air and Propane-air mixtures were used for this study. The air and fuel flow were controlled by Alicat Flow controllers with a capacity of 0 SLPM and SLPM, accordingly, and both with an accuracy of ± (0.8% of reading + 0.% of full scale). PIV setup For the velocity measurements of the flow just downstream of the burner-plate, a Dantec D-PIV system was used. The laser-sheet was produced by a Dantec Dual Power -1 Yag laser with two laser cavities, each with 1Hz of maximum laser pulse frequency, and a Kodac MegaPlus ES1.0 camera with a CCD resolution of 1008x1008 along with a Nikkor 0mm lenses was used to record the image samples. The synchronization between the laser and the camera was assured by a FLOWHUB 100 hub connected to a workstation with the Dantec FlowMap software to control the data acquisition and data process. A purpose made device was built to suspend fine alumina Al O 3 particles on inlet air flow in order to seed the flow, assuring proper particle density for the PIV image samples. The average size of the alumina particles was of 0.3µm. Stability and blowoff limits For each burner, the blowoff points were measured and plotted in a stability diagram, ranging from Re=0 to 00. Each blowoff point was measured by decreasing the equivalence ratio (phi) of the mixture, while fixing the Re of the flow from a stable combustion condition until total blowoff occurs. For every iteration during the process, a proper time interval was given to stabilize the flow. Figure shows the plot of the blowoff points in terms of U/d and equivalence-ratio for methane-air mixtures, obtained for all the different configurations of the burner-plates.

4 Even though most of the points converge for low flow velocities (and power), the blowoff regions diverge from plate to plate as the flow velocity is higher (higher power). In order to find a correlation between all the measured data, a non-linear model with the form of equation 1 was built to fit the results according to the parameters f, g and h which are the exponents for the hole diameter, space between holes and number of holes, accordingly, on the structure of the non-linear fitting model (equation 1). = a + b!! d! e! U!!"!!"#"$% (Eq. 1) The numerical process was performed using the software Mathematica [11] and Figure 3 shows the resulting model fit, along with the correlated data transformed according to equation. Factor Estimate Standard Error f g h Table Resulting parameters of the non-linear model fit.!!!!!!!"!"#"$%! = = a + b!! (Eq. ) Table summarizes the resulting parameters of the non-linear model fit. Parameter g, the exponent on equation 1 translating the weight of the distance between holes on the model, is much larger than parameters f and h, suggesting that the distance between holes has a much stronger effect on the flame stability limit of the burners of this study. On the other hand, parameters f and h are close to zero which also indicate that the influence of hole diameter and the number of holes of the burner is almost negligible. Figure shows the blow-off limits solely for burners with 1.mm diameter holes and 19 holes. Figure emphasizes the two trends of blow-off limits, one being for distances between holes of mm and 3mm, for which the blow-off limit is greatly affected as the flow velocity increases, whilst, for distance between holes of 1mm, the blowoff limit increases slightly at the low power regime, but remains almost constant as the flow velocity is increased. These trends are also verified for propane mixtures. Moreover, the plot of the fitted data of Figure 3, evidences a sudden divergence of the data from the correlation model fit line for U/d higher than 00s -1, which might be related to a sudden change in the flame morphology. The following section will focus on the flame shape and flow characteristics of plates F, G and H, observed and measured using direct visualization and PIV techniques. Flow Characterization around burners F, G and H Figure 3 - Data fitting, Ø* vs. U/d, of the blowoff limits according to equation.

5 ϕ U/d This section covers the characterization of the reacting and isothermal flow with plates H, F and G. Several visualization and measuring methods were used to analyze the isothermal and the reacting flow around the burner. Direct photography of the luminous contour of the flame is a suitable method to observe the position and shape of the flame but it provides no information about the flow itself. By tracking the position of small seed particles injected in the flow, it was possible to follow the flow paths and picture an overview of its structure as it crosses the flame front. Then, a PIV algorithm was used to post-process the position of the particles and get the actual velocity field of the reacting flows resulting from each burner plate. Direct Flame Visualization picture. F - CH G - CH H - CH Figure - Blowoff limits for burner plates with 1,mm diameter and 19 holes. Figure shows a series of direct photography of the resulting propane flame for different operating conditions on the plates F, G and H. These photographs were taken from a side perspective, so that all the plate holes were aligned perpendicularly to the plane of the Considering that the light intensity captured by the CCD of the camera is related to the rate of chemical reactions occurring at the flame front, it is possible to guess the flame front shape and position by observing the light contours of the photograms taken of the flame and presented on Figure. Thus, each picture set, corresponding to the flame on plates H, F and G, from left to right, shows the flame front evolution as the premixed mixture equivalence ratio decreases towards blow-off, for a Re number of 300. Overall, the resulting flame on the three burners is composed by small conical flames aligned with each hole of the burner. As the equivalence ratio decreases from a rich premixed mixture towards blow-off, it is notable that the evolution of the flame on plates F and G are similar, but different from plate H. The peripheral flames are detached from the burner rim for all the three plates, however, plates F and G appear to be more sensible to the reduction of equivalence ratio. As the equivalence ratio decreases, not only the peripheral flames are displaced from the burner rim but also the interior flames start to detach on burners F and G and eventually the whole flame detaches completely, stabilizing at a finite distance from the burner before blow-off. On the other hand, despite

6 a) Plate H b) Plate F c) Plate G Ø=1, Ø=,1 Ø=, Ø=1, Ø=,0 Ø=1,9 Ø=,1 Ø=1,3 Ø=1,8 Ø=,0 Ø=1, Ø=1, Ø=1, Ø=1,1 Ø=1,0 Figure Direct photography of the flame on burner plates H, F and G, from left to right. Flame evolution from a rich to lean burning conditions at Re=300.

7 a) Plate H, ϕ = 1,10 b) Plate F, ϕ = 1,80 c) Plate G, ϕ = 1,90 Figure PIV samples showing the particles scatter of the reacting flow on plates H, F and G. The seed particles were injected on the main stream of reacting mixture. the peripheral flames on burner H being detached from the burner, the interior flames remain stable and attached to the burner rim. Just before blow-off occurs on burners G and F, the detached flame starts oscillating slowly, showing clear signs that it is in the imminence of being blown away from the burner. On burner H, though, the blow-off happens suddenly showing no warning behavior of instability. Another distinct characteristic between the flames of the studied burners, that is inferred from the pictures on Figure, is the fact that on burners F and G the flame front is discontinuous, broken at the base of each individual flame and forming a chalice shape, whereas, the small flames forming the flame front on plate H are all connected at their base, establishing a single-continuous flame surface. In fact, the flame shape of burner H resembles a saw -shape flame, ending with two inverted-flame wings at each end. Moreover, while the small flames on burner H are conical and short, on burners F and G the flames formed downstream of each jet are severely elongated along the mixture jet direction, with roughly twice the length of the flames of burner H. Flow visualization In this section, the flow pattern of the reacting flow of air-propane mixture is highlighted and analyzed. Figures a-c show the scatter of the seed particles, injected on the air-fuel stream and illuminated by a laser sheet as they exit the mixing chamber, through the burner plate and feed the reacting zone. The pictures were taken having the camera focused on a normal plane on the horizontal axis of the burner, represented on Figure 1. Only the three rightmost jets (holes) of the burner were framed on these recorded PIV samples, as the flow is symmetric. All the scatter pictures were recorded for Re = 300 and an equivalence ratio of 1.10, 1.80 and 1.90, for plates H, F and G, accordingly. The chosen equivalence ratio corresponds to a stable burning condition, away from the blow off point. Figures a-c show a sudden decrease on particle density accompanied by a slight dispersion of the particles as they cross the flame front, which is an evidence of the thermal expansion of the gases due to the heat release at the flame front. On Figure a, corresponding to burner H, three triangular regions with higher density standout at the bottom. These regions correspond to the stream of unburnt fresh gases emerging from the holes of the burner and are bounded by the conical flame front, for the case of the inner flames, whereas the outer flame is bounded by the wing of the inverted flame, at the left side, and by surrounding fresh air on the right side. Because the flame is lifted on burners F and G, the fresh mixture region, having the higher particle density, extends longer, as seen on Figure b and c. On the other hand, the regions between the jets are free of particles, meaning the flow carrying the particles is not able to inject them into those regions. Further downstream, on burner H, the particles scatter uniformly

8 immediately after crossing the flame front, merging the jets and leaving no empty spaces between them whilst on burners F and G it is clearly seen a blank region between the jets. Assuming that particles are carried solely by the fresh unburnt fuel-air mixture exiting the burner and that the surrounding freestream of fresh air has no particles in suspension to transport, the blank areas are probably occupied by entrained air from the surrounding environment, dragged and flowing inside the gaps between the jets. Presumably, this secondary flow of entrained air plays an important role on the stability of the flame. Supporting this hypothesis, Figures 8a-c show the scatter of seed particles injected on the free air stream surrounding the burner. This was attained by injecting a reduced air flow into a container, pushing a particle cloud previously generated and stabilized inside through an annular exit centered on the burner axis and coplanar with the burner plate face, as represented on The scheme on Picture 7 shows the experimental apparatus set up to produce the particle cloud and seed the freestream of air surrounding the burner. The scatter of Figure 8a, evidencing air entrainment induced by the reacting flame on Plate H, shows that the free air stream transporting the particles is not able to penetrate into de flow and is promptly pulled up by the main flow boundary layer as it reaches the lateral sides of the attached flame. For the plates with higher space between holes, Plates F and G, the scatter pictures show that the free air stream reaches the inner gaps between jets, being more evident on Plate G, with the highest gap. Across the jet regions of Figures 8b and c, there were no particles recorded in the picture, meaning that the secondary air stream reaches the inner gaps by crossing through the open gaps on the peripheral crown of holes of the plate pattern, but not crossing the jets directly. The previous observations support the thesis that a compact flame with small gaps between holes blocks the entrained fresh air from breaching into the inner gaps between air-fuel jets, whereas, for greater gaps, a secondary flow of cold air is developed and penetrates into the inner gaps of the flames, as seen on Figure. Seeded freestream Container Mixing Chamber Seed inlet Flame Plate Figure 7 Apparatus for seeding the freestream of air surrounding the burner. a) Plate H, ϕ = 1,10 b) Plate F, ϕ = 1,80 c) Plate G, ϕ = 1,90 Figure 8 PIV samples showing the particles scatter of the reacting flow on plates H, F and G. The seed particles are transported by the entrained air secondary flow.

9 The breaching air dilutes the reacting mixture at the base of the flame by mass diffusion mechanisms, affecting only the jets boundary layer and decreasing the equivalence ratio at these interface regions between the two flows. Additionally, the same secondary cold flow exchanges heat with the reacting mixture, inducing heterogeneities in the temperature and chemical rates distribution. These phenomena can explain the less stable flames of plates with greater gaps between holes, namely the high equivalence ratio of the blow-off point, the detachment of the flame and the elongated flame cone. To further investigate this, a spectroscopic analysis can be carried out to track the concentration of C-radicals, CO and NO which are expected to increase due to the slower chemical kinetics. However, the latter analysis is beyond the scope of this preliminary analysis. Velocity distribution In this section are presented the PIV measurements of a stabilized propane flame on Plates H, F and G with Re=300. The measurements were taken in half of the flame, taking advantage of the symmetry of the flame, and on a vertical plane intersecting the face of the plate at the horizontal axis represented on Figure 1. A 3x3 pixels interrogation window with 0% overlap on both directions was used to process the particle samples with an acceptable signal-to-noise ratio over the seed areas. Figures 9 show the vector plot for the velocity distribution across the flames on the three plates, with the contours of the respective flame front on the background. On plate H, the flow accelerates as it crosses the flame front and slightly deflects outwards at the flame front, whilst on plate F and G the velocity increase is not observed, despite the deflection is. The plot of Figure 10, where the axial velocity component of the reacting and non-reacting flow along the central axis of the central hole for plates H and G is plotted, show that the axial velocity for plate H quickly raises about 30% at the flame front, whereas for plate G the velocity profile along the centerline for the reacting flow is almost similar to the non-reacting flow, decreasing slightly along the axis. For plate H, the flow acceleration and streamline deflection is due to the thermal expansion of 0 8 x-pos[mm] the mixture as it reacts at the flame front. As seen on the vector plot of Figure 9a and the axial velocity plot along a horizontal line downstream of the flame at 9,mm of the burner plate, shown on Figure 11a, the velocity profile of the reacting flow is maximum at the center of the flame (x=0mm), slowly decreasing in the outward direction until x=mm. From that point, the velocity profile rapidly decays as the free stream of air surrounding the burner starts mixing with the reacting mixture and both flows exchange momentum. Note that for x > 7mm on the reacting flow of Plate H, there is no valid PIV information due to the lack of seeding and, thus, the velocity profile on the plot of Figure 11a is neglected accordingly. The reacting flow develops y-pos[mm] y-pos[mm] y-pos[mm] a) Plate H, ϕ = 1, x-pos[mm] b) Plate F, ϕ = 1, x-pos[mm] c) Plate G, ϕ = 1,90 Length Length Length Figure 9 PIV results. Velocity vector plots with the contour of the flame as background of propane-air flames at Re=300. x vectors

10 line, Re=300.. v [ms-1]. Plate G PHI=1.90 isothermic Plate G PHI=1.90 w/ reaction Plate H PHI=1.10 isothermic Plate H PHI=1.10 w/ reaction y [mm] Figure 10 - axial velocity component of the reacting and non-reacting flow of a propane-air mixture along the central axis of the central hole for plates H and G. a single jet of combustion products downstream the flame with higher axial velocity and wider than the three-jet structure present on the isothermal flow, for the same inlet conditions. This may be explained again by the thermal expansion of the gases caused by the sharp increase of the flow temperature provided by the highly exothermal combustion reactions, accompanied by the increase of specific volume of the combustion products gases. Additionally, due to their proximity, the incoming jets leaving the burner plate end up merging and interacting with each other, developing a single jet of combustion products downstream of the flame front.

11 The flow acceleration across the flame front due to the thermal expansion, as a consequence of the combustion heat release, is not evident on the reacting flow with plate G, with larger space between holes and for the same conditions, although the deflection of the streamlines is still present. From the horizontal velocity profile downstream of the flame and shown on Figure 11b, the axial velocity of the reacting flow and the isothermal flow match at the corresponding position of the holes of the plate (x=0mm, x=,mm and x=9mm). However, while the velocity profile for the isothermal flow evidences the three isothermal jets developed, with the corresponding velocity peak at the center of the jet and the typical developed parabolic velocity profile, the reacting flow shows a quasi-plug velocity profile across the horizontal axis, decaying only after the outmost jet interacting with the surrounding still air. The local minima at x=mm of the velocity profile corresponds to an area without seed particles and, thus is neglected. The plug velocity profile of the flow downstream the flame front is explained by the deflection of the streamlines, induced by the local thermal expansion at the flame front, as seen of Figure 9c. Nevertheless, the local thermal expansion at the flame front is not sufficient to accelerate the flow, as it was observed for the more compact Plate H, and an explanation to this fact may lay on the thermal expansion that is being inhibited by the entrained fresh air, dragged by the main reacting flow and flowing inside the large gaps between the jets, as evidenced before. Due to the entrainment of cold air, the hot combustion products will be diluted immediately after reaction by that cold flow, hence, losing heat to the cold flow and the thermal expansion is less evident. If this hypothesis is confirmed, the velocity profile of Figure 11b is an evidence that the large gaps between the 7 V@Y=9,mm, Re=300 Plate H PHI=1.10 isothermic Plate H PHI=1.10 w/ reaction v [ms-1] x [mm] V@Y=1mm, Re=300 7 Plate G PHI=1.90 isothermic Plate G PHI=1.90 w/ reaction v [ms-1] x [mm] Figure 11 axial velocity component of the reacting and non-reacting flow of a propane-air mixture, downstream de flame front, for plates H and G at y=9,mm and y=1mm respectively.

12 burner holes induce an excessive mass of entrained air into the inner gaps of the flame, causing a secondary flow of cold fresh air between the main reacting flow jets. Consequently, this observation sustain the thesis proposed in the previous section, stating that this secondary flow of entrained fresh air dilute the premixed reacting flow, decreasing the equivalence ratio locally and the reaction zone temperature, which will undermine combustion reactions and cause flame instabilities. Summary and Conclusions The stabilization of methane and propane flames on a matrix-hole plate burner was analyzed, describing on the influence of hole diameter, distance between holes and number of holes of the flow distribution plate as function of the velocity gradient U/d. This analysis shown that the flames are less stable as large is the distance between holes on the distribution plate and that the flame stability is almost insensible to the other number of holes of the plate and the hole diameter. The second part focused on characterization the reacting propane-air flow around the burner plates to understand the influence of the distance between holes on flame stability. From direct visualization of the flow, it was observed that for a flow of Re=300, the flame stabilized on the plate with smaller gaps between holes had a continuous saw -shape flame front, attached to the burner and with the outer flames behaving has V-shape flames, whereas, the plates with larger gaps produced individual conical flame fronts stabilized downstream of each jet, highly elongated on the jet direction and detached from the burner in a chalice shape. The scatter pictures of seed particles in suspension on the surrounding air evidenced the air entrainment on the flame and showed that the flow penetrates deeper into the gaps between main premixed flow jets as the gaps are larger. The latter observation suggests that the entrained air affects the stability of the flame, which exchanges mass by diffusion mechanisms with the main flow jets, decreasing the equivalence ratio at the interface between the reacting mixture flow and the entrained air flow, and reduces temperature of the reacting zone and combustion products. Finally, the PIV results presented lastly show plausible evidences of the entrained air and the secondary flow of cold air running in between the main flow jets leaving the burner plate. References [1] Leitão, I.. Experimental and Analytical Flame Transfer Functions of Multi-Perforated Plate Burners, MSc Thesis dissertation, 009. [] K. S. Kedia e A. F. Ghoniem, Mechanisms of stabilization and blowoff of a premixed flame downstream of a heat-conducting perforated plate, Combustion and Flame, 011. [3] B. Lewis, von Elbe, J. Chem. Phys. 11 (193) [] B. Lewis, von Elbe, Combustion, Flames and Explosions of Gases, Academic Press, New York, 191. [] Law, C. K., Twenty-Second Symposium (international) on Combustion, The Combustion Institute, 1988, pp [] Choi, W. C., Puri, I. K., Flame Stretch Effects on Partially Premixed Flames, Combustion and Flame, The Combustion Institute, 000, pp [7] C. Trevino, S. Donnerhack, N. Peters, Combust. Flame 8 (1991) [8] C.W. Choi, I.K. Puri, Combust. Flame 1 (001) [9] R.M.M. Mallens, L.P.H. de Goey, C.K. Law, Combust. Sci. Technol. 19 (000) [10] T. Kawamura, K. Asato, T. Mazaki, Combust. Flame (198) 33. [11] Wolfram Research, Inc., Mathematica, Version 9.0, Champaign, IL (01).