12
15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010 - 1 - Experimental Study of Combustion and Flow Dynamics in a Meso-Scale Whirl Combustor Siqian Liu 1 , Bruno Renou 1 , Marianne Sjöstrand 1 , Yves D’Angelo 1 , Fabio Cozzi 2 1: UMR 6614 CORIA, INSA de Rouen, FRANCE, [email protected] 2: Department of Energy, Politecnico di Milano, Milan, Italy, [email protected] Abstract Combustion may constitute an interesting solution for small-scale power generation. Using hydrocarbons high energy density at small scales is still a scientific and technological challenge. Driven by industrial applications such as the replacement of existing batteries by a lighter system or the design of miniature propulsion device, successful development of such small scale systems requires an extensive understanding of combustion behavior while scaling down conventional combustors to a miniaturized size. For this purpose, a meso-scale whirl combustor is designed and experiments are performed. The internal flow fields have been studied by Particle Image Velocimetry (PIV) in non-reacting conditions. Combustion stability diagrams have been established which evidenced quite wide flame stability limits. Pressure fluctuation data have been analyzed for the purpose of exploring the combustion instabilities. Pollutant emissions have been measured simultaneously to the pressure fluctuations not only in order to detect the exhaust species, but also to estimate the combustion efficiency. At the same time, CH* chemiluminescence flame imaging has been carried out by an intensified CCD camera to observe the flame location inside the combustor and instantaneous flame structures corresponding to different total flow rates and different equivalence ratios. High-speed imaging has also been implemented and synchronized with pressure acquisition to study the thermo-acoustic coupling. In order to improve the performance of the combustor, hydrogen is added to the fuel mixture. The beneficial effects of hydrogen enrichment on the flame stabilization and the reduction of pollutant emissions have been pointed out, which provides a potential alternative to fuel utilization in meso-scale combustion. 1. Introduction In the recent years, one can easily notice the trends for the miniaturization of electronic and mechanic systems. However, the battery autonomy still remains a challenge. At present time, what we use as power supply are mainly the primary batteries and the rechargeable ones. Considered as one of the currently available top chemical batteries with an energy density of 1.2MJ/kg, lithium is widely used in typical portable consumer electronics. In comparison, the hydrogen and hydrocarbon fuels demonstrate an extremely high energy density (~45MJ/kg). As a result, meso- and micro-scale combustion is supposed to be a potential candidate in small-volume power generation systems. The successful design of such systems requires an extensive understanding of the combustion behavior while scaling down conventional combustors to a miniaturized size. Scaling issues related to fluid and thermal transport phenomena as well as chemical reaction have been theoretically reviewed in the literature [1]. In comparison to macro scale combustors, meso-combustors operate at much lower Reynolds numbers due to the miniaturized size. As a result, mixing process can hardly be achieved by turbulent effects. Instead, it is mainly controlled by diffusion, which may cause poor mixing. Moreover, combustion efficiency can be negatively affected by the reduction in the residence time. Finally, the high surface to volume ratio resulted from the small characteristic length of the combustor leads to significant heat losses. When heat release from chemical reaction cannot take pace with heat losses by wall, combustion can no longer be sustained, thus occurs the flame quenching. A possible way to overcome the short residence time effect is to reduce the kinetic reaction

1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 1 -

Experimental Study of Combustion and Flow Dynamics in a Meso-Scale Whirl

Combustor

Siqian Liu1, Bruno Renou

1, Marianne Sjöstrand

1, Yves D’Angelo

1, Fabio Cozzi

2

1: UMR 6614 CORIA, INSA de Rouen, FRANCE, [email protected]

2: Department of Energy, Politecnico di Milano, Milan, Italy, [email protected]

Abstract Combustion may constitute an interesting solution for small-scale power generation. Using hydrocarbons high energy density at small scales is still a scientific and technological challenge. Driven by industrial applications such as the replacement of existing batteries by a lighter system or the design of miniature propulsion device, successful development of such small scale systems requires an extensive understanding of combustion behavior while scaling down conventional combustors to a miniaturized size. For this purpose, a meso-scale whirl combustor is designed and experiments are performed. The internal flow fields have been studied by Particle Image Velocimetry (PIV) in non-reacting conditions. Combustion stability diagrams have been established which evidenced quite wide flame stability limits. Pressure fluctuation data have been analyzed for the purpose of exploring the combustion instabilities. Pollutant emissions have been measured simultaneously to the pressure fluctuations not only in order to detect the exhaust species, but also to estimate the combustion efficiency. At the same time, CH* chemiluminescence flame imaging has been carried out by an intensified CCD camera to observe the flame location inside the combustor and instantaneous flame structures corresponding to different total flow rates and different equivalence ratios. High-speed imaging has also been implemented and synchronized with pressure acquisition to study the thermo-acoustic coupling. In order to improve the performance of the combustor, hydrogen is added to the fuel mixture. The beneficial effects of hydrogen enrichment on the flame stabilization and the reduction of pollutant emissions have been pointed out, which provides a potential alternative to fuel utilization in meso-scale combustion.

1. Introduction

In the recent years, one can easily notice the trends for the miniaturization of electronic and

mechanic systems. However, the battery autonomy still remains a challenge. At present time, what

we use as power supply are mainly the primary batteries and the rechargeable ones. Considered as

one of the currently available top chemical batteries with an energy density of 1.2MJ/kg, lithium is

widely used in typical portable consumer electronics. In comparison, the hydrogen and hydrocarbon

fuels demonstrate an extremely high energy density (~45MJ/kg). As a result, meso- and micro-scale

combustion is supposed to be a potential candidate in small-volume power generation systems.

The successful design of such systems requires an extensive understanding of the combustion

behavior while scaling down conventional combustors to a miniaturized size. Scaling issues related

to fluid and thermal transport phenomena as well as chemical reaction have been theoretically

reviewed in the literature [1]. In comparison to macro scale combustors, meso-combustors operate

at much lower Reynolds numbers due to the miniaturized size. As a result, mixing process can

hardly be achieved by turbulent effects. Instead, it is mainly controlled by diffusion, which may

cause poor mixing. Moreover, combustion efficiency can be negatively affected by the reduction in

the residence time. Finally, the high surface to volume ratio resulted from the small characteristic

length of the combustor leads to significant heat losses. When heat release from chemical reaction

cannot take pace with heat losses by wall, combustion can no longer be sustained, thus occurs the

flame quenching.

A possible way to overcome the short residence time effect is to reduce the kinetic reaction

Page 2: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 2 -

time by increasing the reaction temperature. Excess-enthalpy burners are thus inspired by this idea,

which recirculate the thermal energy stored in the exhaust port to preheat the reactants [2].

Concerning the poor mixing level, asymmetric whirl combustors are regarded as a promising

strategy to promote turbulent mixing [3,4,5].

Although the sources and mechanisms of flame instabilities in small scale combustion are hard

to be identified at the moment, chemical quenching and thermal quenching are assumed to be two

important contributions [6]. As the surface-to-volume ratio is increased, the possibility of radical

termination by wall is enhanced. Therefore, surface-induced catalytic combustors become an

attractive solution to positively make use of the wall in presence [7]. The general concept of

catalytic design lies in catalyst deposition such as Platinum coating on the interior wall.

Based on a preliminary study on the feasibility of different strategies found in literature, we

chose the whirl flow concept. The experimental needs, especially the optical access, are taken into

account as well. For these reasons, a quasi-cubic whirl combustor is designed. The combustor

volume is less than one centimeter cube, usually referred to as meso-scale.

As far as we know, most papers related to micro-/meso-scale combustion deal with the designs

and performances of the combustors with the aim of tackling the quenching problem. Very few

discussions concerning the sources of instabilities are available so that we are not able to explain

the quenching phenomenon. For this reason, CH* and OH* chemiluminescence measurements have

been performed which provides us with the information on flame structure and local heat release.

By coupling this information with instantaneous pressure fluctuation, thermo-acoustic instabilities

will be discussed.

Moreover, the lack of velocity field data can be very inconvenient to properly understand the

internal flow structure, especially for a whirl combustor inside which flow and flame structures are

more complex. Hence, Particle Image Velocimetry (PIV) is chosen on account of its efficiency as a

planar measurement method.

Most of the existing results involving the determination of the combustion stability only gave

us a rough idea on the flame self-sustainability conditions, ignoring the occurrence of the acoustic

phenomena [4,5,8]. In order to classify the different combustion stability regimes, acoustic analysis

has been performed, leading to the establishment of a detailed stability diagram.

In meso-scale combustors, flame stabilization mainly involves competition between the rates of

the chemical reactions and the rate of turbulent diffusion of species and energy. From this point of

view, hydrogen has better flame holding characteristics than methane in terms of both adiabatic

flame temperatures and flammability limits. Thus, hydrogen enrichment in fuel mixture is supposed

to enhance meso-scale combustion. Nevertheless, injection of hydrogen has negative effects in

some aspects such as higher NOx emissions due to higher adiabatic flame temperature. In

consequence, research towards determination of beneficial ranges of hydrogen enrichment is

crucial. The impact of hydrogen addition on both gaseous and liquid fuel/air mixture (e.g.

methane/air, propane/air, kerosene air) in macro-scale combustion system (e.g. gas turbine) has

been widely investigated [7,8,9]. But little information can be found with respect to meso-scale

combustion.

In the present work, we aim to demonstrate the interest of hydrogen-enriched methane/air

combustion in a meso-scale whirl combustor inside which the flow configurations appear to be

much more complex. At first, the internal flow field under isothermal conditions will be studied

aiming to investigate the flow motions inside the combustor. Then, emphases will be put on the

analysis of the global performance of the meso-combustor, including establishment of stability

diagrams coupled with flame structure observation and estimation on the combustion efficiency.

Finally, the effect of hydrogen addition on flame stability and pollutant emissions will be discussed

in detail.

Page 3: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 3 -

2. Experimental approaches

2.1) Meso-combustor

A quasi-cubic combustor having a volume of about 640mm3

(8mm×8mm×10mm) has been

designed and manufactured (Fig.1). The meso-combustor operates in a non-premixed configuration,

fed by methane and hydrogen mixture as fuel and air as oxidizer. More precisely, air is injected

tangentially in order to generate a whirling flow, while gaseous fuel injection is achieved in the

radial direction. Two different pipes can be used for fuel injection in order to investigate its effect

on the mixing process. The flow rates are controlled by three thermal mass flow meters (Bronkhorst

EL-FLOW Select) calibrated for Hydrogen (max: 0.5L/min), Methane (max: 0.55L/min), and Air

(max: 5L/min) respectively. Therefore, various concentration of hydrogen can be studied so as to

point out its effect on flame stability and pollutant emissions. A maximum input thermal power of

about 220W can be reached.

Fig. 1 Meso-combustor (Left: meso-combustor compared with a one-euro coin; Middle: 3D schematic; Right: plan X-

Z)

In addition to the methane/air combustion, measurements are carried out for 3 fuel mixtures

with different hydrogen enrichment rates: CH4 (20%) / H2 (80%), CH4 (50%) / H2 (50%), CH4

(80%) / H2 (20%), where the percentage of hydrogen is defined as the ratio of the mass flow rate of

H2 over the mass flow rate of the fuel mixture:

Stability diagrams are established for all the 3 cases, each of which requires at least 40

measurement points. In this study, we choose CH4 (50%) / H2 (50%) / air as a representative case.

The equivalence ratio of the mixture is determined by the following expression:

with (stoichiometric fuel air ratio for methane) and (stoichiometric fuel air ratio for hydrogen).

2.2) Cold flow velocity measurement

PIV measurements are processed initially to the non-reacting flows, allowing us to obtain

quantitative information on flow motions inside the combustor. For this purpose, double-pulsed Nd-

YAG laser (λ = 532nm, E=120mJ/pulse) is employed. A cylindrical lens (f=30cm) is used to create

a laser sheet with a constant thickness of 200μm (FWHM of the intensity profile) in the

measurement volume. A 3-axe moving system is used to adjust with high precision (sensibility: 1

μm, resolution: 10μm) the position of the combustor so that we are able to measure the velocity

field of different planes in two orthogonal directions. Flow is seeded with DEHS droplets with a

Exhaust pipe

Fuel injection

Air injection

Page 4: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 4 -

typical diameter of 1μm. The acquisition system is composed of a Lavision Flowmaster CCD

camera (1024×1280 pixels, 3Hz) with a macro-lens (105mm, f/2.8). The PIV software Davis 7.2 is

used to register and post-process the acquired images. The field of view is set to be 6mm×8mm. By

applying the multi-pass cross-correlation procedures (64 pixels → 32 pixels) we can obtain a spatial

resolution close to 0.18mm. This typical resolved length scale ensures the mean flow velocity to be

fully characterized.

2.3) Flame imaging

Both CH* and OH* chemiluminescence flame imaging has been carried out by an intensified

CCD camera (Roper Scientific, 512×512 pixels) coupled with narrow band pass filters (BG12) in

order to observe the flame position in the combustor and instantaneous flame structures

corresponding to different total flow rates and different equivalence ratios. In each operating

condition, 500 instantaneous flame images have been taken consecutively (frequency = 10Hz).

Exposure time is set to be 2ms while the gain is adjusted to 175 based on the case corresponding to

maximum level of collected CH* signal. These two parameters are kept constant during all the

measurements so that the results from different experimental conditions are comparable. All the

flame images presented in this study correspond exactly to the plan X-Z (Fig.1) where the effects of

the tangential injection of air flow and radial injection of fuel flow can be observed clearly.

2.4) Pressure fluctuation acquisition

Pressure fluctuation data are collected using a piezoelectric dynamic pressure sensor (Kistler

6053CC60) coupled with a charge amplifier. Spectral analysis is performed for the purpose of

exploring combustion instabilities. Sampling frequency is fixed to 10 kHz. The temporal averaged

power spectrum for the pressure fluctuation signal is determined from the ensemble averaged of 40,

temporally Hamm-windowed, acquisition.

2.5) Thermo-acoustic coupling

In order to analyze thermo-acoustic instabilities, a high-speed ICCD camera (Photron,

FastcamSA1, 512×512 pixels) is also implemented. The frame rate is set as 20000images/second.

Flame imaging is synchronized with the pressure signal acquisition with the help of a Multi-

Channel-Data-Link unit. As a result, pressure fluctuation signals are sampled at 20 kHz and 20000

corresponding images are obtained in 1 second.

2.6) Pollutant emissions

Burned gases are sampled for pollutant emissions analysis using a chemiluminescence analyzer

to measure the concentrations of NO and NOx, a non-dispersive infrared (NDIR) analyzer to

measure CO and CO2 and paramagnetic technique for O2 concentrations. A flame ionization

detector (FID) is applied for the continuous trace analysis of total hydrocarbon (THC). The gas

analyzer probe is located close to the gas exit and operates at 3L/min sampling flow rate which is

higher than that of the exhaust gas. Therefore, some extra air in the surroundings could have been

sucked up by the probe. This will cause a decrease in the measured concentrations of CO2, CO,

NOx and THC since the amount of air is slightly overestimated. The detected exhaust species are

regarded as a good indicator of the combustion completeness level. They are also served to estimate

the combustion efficiencies. Accordingly, conclusions on the scaling impact on the residence time

can be reached. The combustion efficiency is calculated by taking into account the loss caused by incomplete

Page 5: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 5 -

combustion which is calculated on the basis of the measured CO concentration in the combustion gases according to the following formula:

3. Results and discussions

3.1) Internal flow fields

At first, the internal flow fields have been studied by Particle Image Velocimetry (PIV) in non-

reacting conditions, thus providing the information on the structure of whirling flows.

We are restricted to 2D PIV measurements which are performed in consecutive planes and in

two orthogonal directions. An illustration of different planes of measurements can be found in

Fig.2. The distance between two adjacent planes is 500μm. Since the laser sheet thickness is about

200μm, we can thus consider the planes of measurements to be almost side by side.

Fig. 2 Consecutive planes of PIV measurements in 2 orthogonal directions

Fig. 3 Algorithm for velocity component reconstruction in plane y-z

13 Planes (X, Y) 13 Planes (Y, Z)

Fuel

Air

6mm

8mm

z

y

Plane Y-Z

Mesh size = 60 x 80

Plane: Vy and Vz

13 lines: Vx and Vz’

If error (Vz, Vz’) < 8%,

Then Vz’’ = average (Vz, Vz’)

Linear Interpolation:

Vx (lines) Vx (plan)

Page 6: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 6 -

Based on this procedure, a time-averaged 3D and 3C velocity field can be reconstructed from a

series of 2D measurements. We assume implicitly the flow field as fully stationary. Fig.3 is the

schematic of velocity component reconstruction algorithm in plane y-z. Totally we have 13 planes

of measurements in this direction. Since in each plane we repeat the same procedure, we use one

plane as an example to explain how to reconstruct the velocity component.

In plane y-z (Fig.3), as mesh size is 60 x 80, Vy and Vz are measured for these points. From the

measurements in the orthogonal direction, we also have information on the velocity components

(Vx and Vz’) for the 13 interlines. The first step is to compare Vz with Vz’ because for 60 x 13

points the velocity components in axis-z have been measured twice. The criterion we applied to this

comparison is the repeatability of the measurements which is estimated at about 8% after repeating

several times the measurements under the same operating conditions. If the estimated error

overpasses 8%, we consider that the measurements are not precise enough. Once the error is less

than 8%, we calculate the average of Vz and Vz’ and replace the measured Vz and Vz’ with the

new values Vz’’. The next step is to interpolate measured Vx (in the 13 interlines) to the entire plan.

Consequently, we know 3 velocity components for all the points in this plane.

The same procedure has been applied to the other 12 planes in this direction as well as the 13

planes in the orthogonal direction by a program written in Matlab. Once the 3D velocity field is

reconstructed, post-processing and visualization such as streamlines can be done by Paraview and

Tecplot.

Fig.4 represents the streamlines of the mean flow. We clearly observe the injection zone which

impact to the wall before to create two distinct flow motions. The main zone travels directly

downstream along the side-wall and exits. A portion of the injected flow recirculates azimuthally in

the upstream part of the chamber before convecting downstream. Similar to swirling flows, whirling

flows were originally used to improve and control the mixing rate between fuel and oxidant

streams. But unlike swirling flows, as air is injected with no axial velocity component, the

structures of whirling flows appear to be very different, with a relatively quiescent center. In our

case, the mixing process will be take place near the upper wall of the combustor at relatively low

Reynolds numbers.

Fig. 4 Streamlines coloured by velocity magnitude (Air Velocity = 40m/s, Fuel Velocity = 5m/s), reconstructed from

2D PIV results

3.2) Stability diagram

We concentrate our attention on the operating range of the combustor. In order to choose the

operating conditions for further analysis, the stability diagram is established by the following

procedure: at a fixed total flow rate, we change the equivalence ratio gradually by changing both H2

and CH4 flow rates, and then we repeat the operation for several different total flow rates.

Mixing controlled mainly

by diffusion, near the

upper wall of the

combustor

Outlet

Fuel

Air

Quiescent center

Page 7: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 7 -

Fig. 5 Stability diagram for CH4 (50%) / H2 (50%) / air Flame (Left); Criterion for classification of different stability

regimes (Right)

Fig.5 is the stability diagram for CH4 (50%) / H2 (50%) / air flames, in which axis-x represents

the total flow rate in Nml/min while axis-y corresponds to the equivalence ratio. As shown in Fig.5,

three principal regimes are differentiated by calculating the fluctuation energy of dynamic pressure

under each operating condition (Fig.5, right). Between the green line and violet line, the flames are

relatively more stable. In the other regions, although the flames are classified as less stable, they are

observed to possess good self-sustainability even near the extinction limits. Obviously, the region in

which the flames exhibit the most stable behavior corresponds to the orange zone indicated in this

figure.

Besides the chosen criteria (RMS), other evidence can be provided so as to validate the

established stability diagram. Power spectral density (PSD) is another useful statistic signal

processing tool based on dynamic pressure analysis. It not only indicates the instability level but

also capture the frequency contents.

When the total flow rate is fixed at 1000Nml/min, equivalence ratio can be changed gradually.

The representation of the results is inspired by Campbell Diagram, initially defined as a system’s

response spectrum as a function of its oscillation regimes. In this case, axis-x represents the

combustion regimes such as equivalence ratio (Fig.6) and total flow rate (Fig.7); axis-y shows the

frequency level; the color bar corresponds to PSD of pressure fluctuations in logarithmic scale.

With this kind of interpretation, we can easily compare the instability levels under various operating

conditions and identify at which frequency comes out the peak.

In Fig.6, two main acoustic modes are observed respectively at 500 Hz and 1100 Hz. When the

fuel mixture burns in lean conditions, the low frequency instability is a dominant factor. On the

contrary, for fuel-rich flames the high frequency instability becomes a more important contribution

to global combustion instabilities together with the total disappearance of low frequency instability.

Similar proof can be provided by the flame visualization. The large-scale flame motions inside the

combustor are indicated by the RMS of the image sets which qualitatively implies the instability

level. The results are quite consistent with the PSD analysis in the present case. The most stable

flame in this case (φ=1.01) corresponds to stoichiometric condition. From Fig.7, we can conclude

that instabilities occur at about 1100 Hz for all the cases. When the total flow rate increases, the

tendency to enlarge the resonance frequency range is more and more obvious, which implies

different acoustic modes appear at the same time due to the fact that a diversity of dominant

resources of instabilities coexist when the total flow rate is large enough.

According to the instantaneous information on pressure fluctuation and flame images, the most

stable flame states can be found at stoichiometry while the total equivalence ratio is around

1000Nml/min. If we trace back to the stability diagram, these flames states are well situated inside

the orange zone.

Less Stable

Less Stable

Stable Very Stable

Page 8: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 8 -

Fig. 6 Campbell diagram and corresponding flame images in Plan X-Z obtained by CH* chemiluminescence (CH4 (50%)/H2 (50%)/Air, Total Flow Rate = 1000Nml/min)

Fig. 7 Campbell diagram (CH4 (50%)/H2 (50%)/Air, Equivalence Ratio = 1.01)

3.3) Impact of H2 addition on flame stability

In general, methane flames can be stabilized with lower equivalence ratio in fuel-lean

conditions as the H2 concentration increases. Results displayed in Fig.8 proved this widely accepted

conclusion. We notice that the addition of small amounts of H2 shows little impact upon blow-out

limits for most levels of total flow rates. When the H2 molar fraction overpasses 20%, the

sensitivity of the blow-out equivalence ratio to H2 level variations remains essentially constant

across the rest of the entire range of H2 enrichment level. An exception is found in the case of

800Nml/min, where no significant discontinuous or steep drop in Lean Blow-Out (LBO)

equivalence ratio is observed from the beginning. Also in this case, the response of LBO

equivalence ratio to hydrogen addition is the strongest among all the five operating conditions

covering the whole operating range of the combustor in terms of total flow rates. In conclusion,

φ = 1.01

φ = 0.62

Average RMS

PSD (dB)

PSD (dB)

Page 9: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 9 -

hydrogen shows beneficial effects on combustion stabilization by extending the LBO limits to

lower equivalence ratio which allow the combustion to occur in leaner conditions. This can be

explained by the fact that hydrogen has better flame holding characteristics than methane thanks to

its low flammability limits and small gap quenching distance.

Fig. 8 Dependence of Lean Blow-Out (LBO) limit equivalence ratio upon H2 molar fraction at different total flow rates

3.4) Thermo-acoustic instability analysis

In whirl combustors, the flow is too complicated to have only one source of instability that

takes precedence over all others. As the air and fuel injection velocities are at very different levels,

the hydrodynamic instabilities such as Kelvin-Helmholtz instability can occur when there is

sufficient velocity difference across the interface between air and fuel flows. The imbalance

between thermal diffusion and mass diffusion can also induce thermo-diffusive instabilities.

Furthermore, we observe that in meso-combustor flame tends to oscillate between extinction and re-

ignition phases. The flame stabilization is achieved with the help of the inner recirculation zone

located in the center of the combustor while the flame itself extinguishes or re-ignites locally. These

extinction and re-ignition processes may be caused by the instabilities induced by local heat release.

However, no direct proof has been provided concerning these different possibilities. Since

experimentally we can imaging some possible ways to achieve thermo-acoustic coupling, we

focused our attention on the thermo-acoustic analysis as a first step to study the instabilities inside

the meso-combustor.

As mentioned before, flame imaging by high-speed ICCD camera is synchronized with

pressure fluctuation signal collection. By associating the pressure signals to flame images, thermo-

acoustic instabilities can also be explored. In order to evaluate the relevance between local heat

release and pressure fluctuations, Rayleigh index has been introduced. Rayleigh index is regarded

as an important indicator of thermo-acoustic coupling. The definition of normalized Rayleigh index

in a simplified form can be found in the study of Yilmaz et al. [10]. In our study, the pressure

fluctuation is uniform over the entire image, which leads to a discretized formulation:

where is the pressure oscillation for each OH* image, is the root mean square of the

pressure oscillations for each set of images, is the fluctuation of the OH* intensity indicating the

local heat release and is the averaged total OH* intensity, N is the number of images in each image set. The normalization by the number of images is achieved so that the computed Rayleigh

Page 10: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 10 -

index is comprised between -1 and 1. If Rf is positive, pressure oscillation and heat release oscillation are in phase so the instabilities tend to grow; if it is negative, the instabilities tend to decay. Experimentally, a set of 1000 consecutive images is recorded corresponding to duration of 50ms.

Fig. 9 Impact of H2 addition on flame location and thermo-acoustic instability

As illustrated in Fig.9, when burning 100% CH4, the flame is located in the centre of the

combustor with a relatively high instability level. When the hydrogen is added, we observe that the

flame is much closer to the combustor wall. This is mainly because of the small quenching distance

that hydrogen possesses. Concerning the computed Rayleigh index, two major conclusions can be

drawn. First of all, it is evident that the instabilities occurred in 100% CH4 combustion is more

relevant to thermo-acoustic effects as Rayleigh index is at a relatively higher value. Secondly, when

the concentration of hydrogen increases, the thermo-acoustic instability level is decreased but it is

also spread out, covering the entire combustor. This is probably the result of dispersivity of

hydrogen.

3.5) Impact of H2 addition on pollutant emissions

In addition to the ability to burn at extremely lean conditions, hydrogen has also been shown to

decrease the formation of CO and unburned hydrocarbons. Theoretically, hydrogen addition has

three major effects on the pollutant emissions. First of all, the flame velocity of hydrogen at 20°C

and stochiometric conditions (237cm/s) is much higher than that of methane (42cm/s), allowing

oxidation with less heat transfer to the surroundings. Likewise, higher burning rate implies lower

burning duration, leading to a higher completeness level of combustion. Secondly, the effect of

hydrogen enrichment can also be summarized as a rise of the heat release rate and the mixture

reactivity [11]. In fact, when methane is H2-enriched a higher OH concentration can be obtained via

HO2 in low temperature regimes, which is generally the case of meso-combustor. Finally, compared

to methane, hydrogen has a very small gap quenching distance (0.06cm VS 0.2cm for methane).

The results shown in Fig.10 are clearly consistent with this analysis. As the residence time

increases, more time is available for chemical reactions to be completed. Subsequently, both CO

and unburned hydrocarbon emissions decrease. In order to compare the pollutant emissions at

various levels of H2 addition, the measurement data are normalized by corresponding CH4 flow

rates with the aim of avoiding the substitution effect. Evidently, hydrogen addition plays a very

CH4 (100%), Equivalence Ratio=1.03, Total Flow Rate = 1500Nml/min

CH4 (50%), Equivalence Ratio=1.03, Total Flow Rate = 1500Nml/min

Average RMS Rayleigh Index

Page 11: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 11 -

important role in the reduction of CO and THC. On the basis of pollutant emissions measurements

we can deduce that H2 enrichment improves the combustion efficiency, which increase from 0.37 to

0.76.

Fig. 10 Impact of H2 addition on CO and THC emissions at stochiometric conditions (Combustion efficiencies are calculated at residence time = 15.5ms)

Carbon monoxide emissions are the results of incomplete hydrocarbon combustion and more

likely to occur during rich mixture conditions. We notice that under stochiometric condition, a great

amount of CO emissions are detected. This can be explained as the fact that CO2 can easily

dissociate and be converted back to CO as the combustion temperature increases. In addition, flow

residence time is much smaller than chemical time and mixing time as the combustor size is

enormously reduced in comparison to conventional sizes.

Unburned hydrocarbons are produced due to the thermal quenching phenomenon. In the near-

wall regions inside the combustor, flame tends to extinguish because the heat loss through the

chamber wall is greater than the heat needed to sustain the flame. The area of flame quenching is

probably where hydrocarbon is left unburned. In meso-combustors, the high surface-to-volume

ratio indicates a much higher possibility for the flames to be quenched. That’s why in our

measurements, unburned hydrocarbon (THC) emissions are also at a high level.

NOx emissions are also measured but we didn’t present the results in this study because their

amounts are negligible. The principal mechanism of NOx formation consists of thermal NO which

appears only when the combustion temperature becomes high enough (about 1800K). In our cases,

the magnitudes of combustion temperatures are estimated from measured exhaust gas temperatures

and vary from 600K to 1000K approximately, which is far from the critical temperature for NOx

formation.

4. Conclusion

A meso-scale whirl combustor has been designed and experiments have been carried out in

order to investigate the global performance of the combustor as well as to propose a possible way to

Combustion efficiency

= 0.37

Combustion efficiency

= 0.76

Page 12: 1, Marianne Sjöstrand1, Yves D’Angelo1, Fabio Cozzi2ltces.dem.ist.utl.pt/lxlaser/lxlaser2010/upload/1684_pktpkr_1.7.4.Full... · 15th Int Symp on Applications of Laser Techniques

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 12 -

improve the combustion stability and efficiency.

Firstly, PIV measurements have been performed under non-reacting conditions. Whirl flow

structure has thus been studied. Under combustion conditions, the combustor demonstrated its

ability to operate with different gaseous hydrocarbon fuel mixtures. A first step to thermo-acoustic

analysis has been achieved to evaluate the relevance between local heat release and pressure

fluctuations.

Considered as a possibility to improve the combustor performance, the injection of hydrogen

into fuel mixture has been proved to largely extend the lean blow-out limits. In addition, both CO

and unburned hydrocarbon emissions have been significantly reduced by adding hydrogen. The

thermo-acoustic analysis showed the ability of hydrogen to reduce thermo-acoustic instabilities.

Further studies can be focused on the combustor performance under constant combustion

power conditions which is a common approach in industry. This will also help us to achieve a more

reliable estimation and comparison of the combustion efficiencies.

So far we have no clue about the frequency of the thermo-acoustic instability. For this reason,

the post-processing such as spectral analysis of a much larger amount of flame images recorded by

high-speed ICCD camera should be implemented.

Last but not least, the feasibility study on PIV implementation when burning CH4 and H2 is

also in progress. Quantitative results can hopefully be obtained afterwards. In addition, we envisage

the further studies to characterize the flame structures by OH-PLIF, so that we will be able to

identify the quenching zones. Consequently, quenching phenomena will be discussed in detail.

References

1) C. Fernandez-Pello, Proceedings of the Combustion Institute 2002, 31.

2) Paul D.Ronney, “Analysis of non-adiabatic heat recirculating combustors,” Combustion and

Flame 135 (2003) 421-439

3) R. A. Yetter; I. Glassman; H. C. Gabler, Proceedings of the Combustion Institut,2000, 28,

1265-1271

4) M.-h. Wu, Y. Wang, V. Yang, R. A. Yetter, Proceedings of the Combustion Institute 2007,

31, (2), 3235-3242

5) F. Cozzi, A. Coghe, Y. D'Angelo, B. Renou, A. Boukhalfa, in: 4th European Combustion

Meeting, Vienna, Austria, 2009.

6) S.Karagiannidis, J.Mantzaras, G.Jackson, K.Boulouchos, Proceedings of the Combustion

Institute 2007, 31, 3309-3317

7) C.M.Miesse, R.I.Masel, C.D.Jensen, M.A.Shannon, M.Short, American Institut of Chemical

Engineers, 50: 3206-3214 (2004)

8) F. Cozzi, A. Coghe, A. Olivani, M. Rogora, in: 30th meeting of combustion, Ischia, Napoli,

Italy, 2007.

9) J.P.Frenillot, G.Cabot, M.Cazalens, B.Renou, M.A.Boukhalfa, International Journal of

Hydrogen Energy, 2009, 3930-3944

10) Yilmaz, A.Ratner, M.Ilbas, Y.Huang, International Journal of Hydrogen Energy, 2010,

329-336

11) Y.Lafay, B.Renou, G.Cabot, M.Boukhalfa, Combustion and Flame, 153 (2008) 540-561