Combustion and Flame 158 (2011) 15421552

Contents lists available at ScienceDirect

Combustion and Flame

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

A combustion concept for oxyfuel processes with low recirculation rate Experimental validation
Valentin Becher , Jan-Peter Bohn, Adrian Goanta, Hartmut Spliethoff
Technische Universitt Mnchen, Boltzmannstr. 15, 85748 Garching b. Mnchen, Germany

a r t i c l e

i n f o

a b s t r a c t
Oxyfuel combustion is a technology for Carbon Capture & Storage from coal red power plants. One drawback is the large necessary amount of recirculation of cold ue gases into the combustion chamber to avoid inadmissible high ame temperatures. The new concept of Controlled Staging with Non-stoichiometric Burners (CSNB) makes a reduction of the recirculation rate possible without inadmissible high ame temperatures. This reduction promises more compact boiler designs. We present in this paper experiments with the new combustion concept in a 3 70 kW natural gas combustion test rig with dry ue gas recirculation of 50% of the cold ue gases. The new concept was compared to a reference air combustion case and a reference oxyfuel combustion case with recirculation of 70% of the cold ue gases. FTIR emission spectroscopy measurements allowed the estimation of spectral radiative heat uxes in the 25.5 lm range. The mixing of the gases in the furnace was good as the burnout and the emissions were comparable to the reference cases. The ame temperatures of the CSNB case could be controlled by the burner operation stoichiometry and were also similar to the reference cases. The heat ux in the furnace through radiation to the wall was higher compared to the oxyfuel reference case. This is an effect of the lowered recirculation rate as the mass ow out of the furnace and therefore the sensible heat leaving the furnace decreases. The higher oxygen consumption with lower recirculation rate could be compensated by a lower furnace stoichiometry. This was possible due to better burnout with increased oxygen concentrations in the burner. The results prove that a reduction of the ue gas recirculation rate in oxyfuel natural gas combustion from 70% down to 50% is possible while avoiding inadmissible high ame temperatures with the concept of Controlled Staging with Non-stoichiometric Burners. 2011 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Article history: Received 14 June 2010 Received in revised form 12 August 2010 Accepted 27 December 2010 Available online 27 January 2011 Keywords: Oxyfuel Recirculation rate CSNB Experiments Natural gas

1. Introduction The man made climate change has its main reason in the burning of fossil fuels for production of power and heat and the resulting release of carbon dioxide into the atmosphere. Coal is the most carbon intense fossil fuel. But it will be necessary for future power generation due to its widely distributed resources and its large proven reserves. One way of mitigating the release of carbon dioxide is the separation of the carbon dioxide with the oxyfuel combustion process and the following storage of the captured gases into underground reservoirs. The high reaction temperatures are the main problem in oxyfuel combustion as the fuel is burned with pure oxygen without the presence of the temperature moderating nitrogen in the combustion air. Several options exist to lower these temperatures:

1. External ue gas recirculation (commonly used option). 2. Internal ue gas recirculation. 3. Burner operation under non-stoichiometric conditions. Almost all proposed oxyfuel combustion concepts rely only on the rst option for controlling the ame temperature. A number of review articles are available on the topic [17]. The concept of Controlled Staging with Non-stoichiometric Burners (CSNB) uses additionally to the rst option the third option to control the ame temperature (Fig. 1). The concept aims to reduce the necessary external ue gas recirculation rate  from the common value of 6080% [1, p. 589] down to 50%. This relates to a reduction of the recirculated ue gas mass ow of 2060%. The ue gas recirculation rate is dened as

_ recirculation m _ furnace end m

Corresponding author.
E-mail address: becher@tum.de (V. Becher).

_ recirculation is the mass ow of cold recirculated ue gas and where m _ furnace end is the total mass ow at the end of the furnace including m the recirculated mass ow.

0010-2180/$ - see front matter 2011 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.combustame.2010.12.029

V. Becher et al. / Combustion and Flame 158 (2011) 15421552

1543

3000

Adiabatic flame temperature [C]

Feasible temperature range

33% Recirculation

2500
50% Recirculation

2000

1500
Burner Operation for CSNB

composition, type of recirculation (wet or dry) and design ue gas exit temperature. If these two base cases are compared with the CSNB case with around 50% recirculation, 49% more heat has to be transferred in the furnace to the water-cooled wall. The tendency in the shift in heat transfer is independent of fuel type and only a function of the reduced mass ow through the steam generator. The main questions for the experimental validation of the concept presented in this article are: 1. Can the ame temperature and the heat ux be controlled by the stoichiometry? 2. What is the difference in wall heat uxes and radiation for the Controlled Staging concept in comparison to combustion with air and to oxyfuel combustion with high recirculation? 3. Is the mixing of the gases in the combustion chamber sufcient to ensure complete combustion at the furnace exit? 4. What other implications have to be considered for a reduction of the recirculation rate? 2. Methods The concept was studied experimentally in a 210 kW natural gas test rig. Natural gas was chosen as fuel as the additional costs and complexities of operation of a coal test rig could be avoided. 2.1. Air cooled combustion chamber The cylindrical combustion chamber had a height of 4 m and an inner diameter of 700 mm (Fig. 4). It was divided into four parts which were each split up in two sections. Each section had its own independently controlled wall air cooling and four access ports, which were each 90 apart. The inner wall of the combustion chamber was a 1.5 mm thick oxidized austenitic high temperature steel (1.4876/X10NiCrAlTi32-21/heat resistant 61100 C). The air cooling for the walls consisted of two cold air supply lines and two hot air exhaust lines for each section. The air owed along a 90 section of the wall. Mass ow controllers in the cold air supply line controlled and measured the amount of cooling air. The temperatures of the cold supply air and of the hot exiting air were recorded and allowed over an energy balance the calculation of the

1000
66% Recirculation 75% Recirculation

500 0.125

0.25

0.5

Stoichiometry
Fig. 1. Theoretical adiabatic ame temperature vs. stoichiometry; calculated from factsage database [8]; Trecirculation = 300 C, wet recycle, Toxygen = 300 C, fuel = hard coal, overall combustion stoichiometry = 1.05 [9].

An arrangement of over- and substoichiometric burners has to be used in the furnace to ensure complete combustion with a minimal oxygen excess at the end of the furnace (Fig. 2). Non-stoichiometric burner operation in this sense means that the fuel and oxidant streams to the each burner result in an equivalence ratio for the combustion reaction of substantially less or more than one. The oxidant stream is a mixture of oxygen and recirculated ue gas. The heat of reaction is transferred to the furnace wall before more reaction energy is released in the next burner. The temperature in the furnace never exceeds critical values. A shift in the heat transfer distribution in the whole steam boiler is a direct effect of the lowered mass ow in the furnace. The furnace exit temperature has to be kept constant due to the fuel ash melting behavior. Therefore more heat has to be transferred to the steam cycle in the furnace through radiation if the recirculation rate is lowered (Fig. 3). A recirculation rate  of 67% in oxyfuel combustion results in a similar heat transfer distribution between the radiative and convective part as in air combustion. This theoretical value calculated based on the sensible heat of the ue gas at the exit of the furnace is in the range of recirculation rates experimentally found [1, p. 593] and is very dependent on coal

Fig. 2. Arrangement of burners in furnace for concept of Controlled Staging with Non-stoichiometric Burners (CSNB) to reach complete combustion (kcombustion % 1) at the exit of the furnace. kburner is the stoichiometry of the fuel and oxidant streams to the burner, kcombustion is the combustion stoichiometry in the furnace and tgas is the gas temperature inside the furnace.

1544

V. Becher et al. / Combustion and Flame 158 (2011) 15421552

3000

0% r
2500

ecirc

Flue gas temperature [C]

30%
50%

ulatio

recir
reci

culat

ion
n

2000

67%

rcul

atio

78%
1500 1200 1000

reci Ai r

recir

rcul

culat

atio

n
Furnace end temperature

ion

firin g

500

0 0% Flue gas recirculation 78% 67% / Air 50% 30% 0%

Fig. 3. Change of heat transfer in an oxycoal boiler from the convective section to the furnace section with a change of the recirculation rate [9].

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Transferred Heat

Furnace wall

Convective heat transfer section

Flue gas

Water Air cooled wall Recirculation fan

Natural gas

Oxygen Air

Fig. 4. 3 70 kW air cooled oxyfuel natural gas combustion test rig.

total heat ux to the wall. Two 180 sections were measured independently: One section opposite of the burners with ame and hot ue gas impingement and one section around the burner quarls with mostly radiation heat transfer. Only the radiative side is presented in this article as there is no ame impingement in large scale combustion furnaces for power production. Three natural gas burners were installed horizontally at different heights on one side of the test rig. UV-probes were installed in the ports on the opposite side for ame detection. The two remaining access ports in the burner levels were used for the insertion of measurement probes.

The combustion chamber could be operated with air, oxygen enriched air or in dry ue gas recirculation mode for combustion in a O2/H2O/CO2 atmosphere (Fig. 4). All cold ows from outside (air, oxygen and natural gas) were measured with thermal ow sensors. The natural gas had a methane concentration of 97.6 vol.%. The ue gases and the oxidant mixtures of all three burners were measured by orices. The ue gas was sucked out on the top end of the combustion chamber, its temperature measured with a type K thermocouple (main parameter for stationary operation), cooled down to around 90 C by a ue gas cooler and vented to the atmosphere through

V. Becher et al. / Combustion and Flame 158 (2011) 15421552

1545

the chimney. The condensate stream from the ue gas cooler was measured by a scale. If the test rig was in recirculation mode, an additional fan recirculated part of the ue gas after the cooler back into the mixing line, where it was enriched with oxygen. The oxygen was supplied from a liquid oxygen tank and had a purity of 99.5 vol.%. A slight under pressure in the combustion chamber was controlled by the ue gas fan after the cooler. 2.2. Burner The natural gas burners for the test rig had to fulll certain requirements: 1. Flame stability in a stoichiometric band from at least 0.3 to 4. 2. Thermal material stability for oxyfuel combustion with oxygen concentrations over 50 vol.% in the oxidant. 3. And a stable ame in the range from 30 to 100 kW fuel input. As there were no commercial burners available which met all these requirements a natural gas ignition burner for full scale pulverized coal burners was modied (Fig. 5). The Hegwein burner (Model ZA0) had originally a stoichiometric operation range from 0.3 to 0.5 in air combustion. This range was extended to the required stoichiometries with the addition of another duct as a secondary oxidant outlet. An exchangeable swirl generator allowed the optimization of the ame shape. When the burner was operating in sub stoichiometric conditions only the primary oxidant outlet was used. The natural gas nozzle in the center had to be modied. Originally it had small radial holes in addition to the central axial nozzle to create a stoichiometric zone in the ame root for stabilizing of the air ames. These stoichiometric zones were too hot for the burner materials for oxygen enriched combustion. A new natural gas nozzle was designed without these holes. The stability of the oxyfuel ames was not an issue during the tests. The experiments for the air combustion reference case were done with the original nozzles with radial holes. The burners were arranged with a vertical distance of 1 m between (Fig. 4). The bottom burner was placed 0.75 m above the chamber bottom and the top burner 1.25 m below the furnace roof and ue gas exit point. 2.3. Measurement equipment The dry ue gas composition (CO, CO2, NO, NO2, SO2 and O2) was measured at the furnace exit with an extractive on-line ABB Gas-Analyzer type AO2000 (URAS 26, LIMAS 11, MAGNOS 206). The steam content of the combustion ue gases at the end of the combustion chamber and at the burner inlet in recirculation mode was calculated via a theoretical combustion calculation. In recirculation mode a part of the steam in the ue gas was condensed and

separated before recirculation (Fig. 4). The lost H2O mass ow was accounted for in the calculations. A video surveillance system was installed for visual observation of the ames and ame shapes. It consists of a surveillance camera in a water and air cooled measurement port. The camera had a Sony 1/300 color EXVIEW HAD CCD chip (ICX254AK) with a high sensitivity in the visual and near infrared spectral range (0.381 lm). Temperature proles were measured with an IFRF type suction pyrometer (Fig. 6) [10]. Radiation spectra emitted along a line-of-sight were measured with Fourier-Transform Infrared Emission Spectroscopy in a very narrow angle (Fig. 7). An Oriel MIR8025 Modular IR Fourier Interferometer was used with an InSb detector. The measurements were done through an open path optical setup (Fig. 7). For windows and lenses CaF2 (focusing lens in measurement port/f = 500 mm, windows and beam splitter of MIR8025) was used. Sapphire (window combustion chamber and lens detector) was used due to its high transmittance in the measured spectral range and its heat resistance [11]. A water-cooled black target was mounted in the access port on the opposite side of the FTIR. The black target was aligned to the chamber wall. Reections from other furnace parts into the optical path were assumed to be negligible due to the low temperature (around 100 C) and the black coating of the front of the black target. A black body furnace type Cyclops 878 from Isothermal Technology Ltd. was used for calibration of the setup. The calibrations were done at a constant temperature of 900 C. The spectra were recorded with a resolution of 16 cm1, an oversampling factor of four and a mirror speed of 25.3 mm/s. 100 scans were averaged to increase the signal-tonoise ratio. The resulting acquisition time was 7.9 s. The recorded interferogram was shortened to a resolution of 25 cm1 and weighted with the NortonBeer strong apodization function [12, p. 33]. The nal spectrum resulted after a Fast Fourier transformation and a phase correction with the Mertz method [12, p. 88ff]. All spectral calculations were done in a self written MATLAB program. 2.4. Measurement errors Measurement errors of the test rig sensors were calculated according to DIN 1319 [13,14]. The resulting uncertainty was compared with the maximal deviation of the value during stationary

Water cooled suction pyrometer 320mm Area of gas suction Burner

Ceramic head with 3 radiation shields and Typ B thermocouple

Swirl generator (45) Secondary oxidant Primary oxidant Natural gas 96

Distance to centerline

700 mm
73 68 42 16

Fig. 5. Natural gas burner for oxyfuel operation with high oxygen contents and extreme stoichiometries. All dimensions are given in mm.

Fig. 6. Cross section of temperature measurements with the IFRF type suction pyrometer.

1546

V. Becher et al. / Combustion and Flame 158 (2011) 15421552

Measurement port / burner Combustion chamber insulation Water cooled black background Line-of-sight CornerCube FTIR CaF2 lens Saphir combustion chamber window Combustion chamber LnSb Detector

Grey shaded areas were purged with CO2/H2O free air

Fig. 7. Setup of FTIR measurements of emitted radiation.

times of the test rig. The given uncertainty is the maximum value of both. The uncertainty of the FTIR emission measurements given is the root-mean-square of the spectral noise calculated from calibration measurements [12, p. 181ff]. Additional sources of uncertainty as alignment of FTIR mirrors, beam splitter and lenses, beam divergence and phase effects were not calculated explicitly. 2.5. Test cases Three different combustion cases were characterized for an estimation of the potential of the Controlled Staging with Non-stoichiometric Burner concept: Air combustion: Slightly fuel lean operation (k = 1.1) of all three burners. The oxidant was air. Oxyfuel combustion: Slightly fuel lean operation (k = 1.1) of all three burners. The oxidant was a mixture of recirculated ue gas and oxygen. The recirculation rate of 70% was the maximum possible recirculation rate while maintaining stable ames. CSNB combustion: Extreme fuel rich and fuel lean burner operation (Table 1). The oxidant was a mixture of recirculated ue gas and oxygen with a xed oxygen content. The fuel stream to all burners was kept constant and the amount of oxidant supplied to the burners was varied. The overall recirculation rate  of the test rig was xed to 50%. The total stoichiometries for the rst two cases were set to an optimum between low remaining oxygen and low CO emissions in the ue gas. The primary burner stoichiometry was set to 0.64 for all fuel lean operating burners to get a ame length well into the measurement plane (Fig. 9). All burners had the same thermal power of 70 kW. The stoichiometries for the CSNB case were set to get similar to the air and oxyfuel case an optimized combustion, but also to get lowest maximum ame temperatures with stable combustion (Table 1). Additional variations of the middle burner stoichiometry were done in the CSNB case. The emitted ame radiation was measured
Table 1 Burner stoichiometries kburner for CSNB combustion case. Burner Top Middle Bottom Primary stoichiometry 0.49 (0.02) 0.30 (0.02) 0.63 (0.03) Total stoichiometry 0.49 (0.02) 0.30 (0.02) 2.38 (0.11)

with the FTIR spectrometer. Only the primary oxidant outlet of the middle burner was used for these tests to ensure a ame well into the line-of-sight from the spectrometer. The total stoichiometry of the bottom burner was varied to maintain an overall furnace stoichiometry of 1.05 at the exit of the furnace. The wall cooling was kept constant for all cases with a constant cooling air mass ow. 3. Results 3.1. Test rig parameters The oxidant for both cases with recirculated ue gas consisted mostly of O2, CO2 and H2O (Table 2). The remaining nitrogen in the system was very low with only around 1.6 vol.%. The amount of oxygen was a function of the recirculation rate. More than 50 vol.% oxygen were in the oxidant in the case of 50% recirculation (CSNB). For 70% recirculation (Oxyfuel case) only 32 vol.% oxygen were in the oxidant. The H2O/CO2 ratio in the oxidant was not the normal value from methane combustion of two but 0.22 for the oxyfuel case and 0.24 for the CSNB case due to the H2O removal in the condenser before the recirculation line. The burnout from the combustion was good. The value of oxygen in the ue gases was around 2.5 vol.% (wet) for all three cases. The CO emissions were very uctuating with peak values of 400 ppm for the oxyfuel case (Table 3). The NO emissions were for all cases within the measurement uncertainty and therefore negligible. The remaining nitrogen in the ue gases at the exit of the combustion chamber was very low for both recirculation cases with a value of around 2 vol.%. The temperature value from the thermocouple directly in the ue gas duct after the combustion chamber was decreasing with decreasing ue gas mass ow in the three cases from 625 C (air) over 595 C (oxyfuel) to 570 C (CSNB). This fact is consistent with an increase of the total wall heat ux from the air case over the oxyfuel case to the CSNB case. The measured total wall heat ux was dominated by heat transfer from ame impingement to the wall on the side opposite of the burners (see Sections 2.1 and 3.3) and therefore very specic for the test rig. The oxidant temperatures were all around 60 C and had no signicant inuence on the furnace exit temperature. 3.2. Oxygen consumption The overall combustion stoichiometry calculated including the recirculated oxygen was chosen during operation of the test rig.

V. Becher et al. / Combustion and Flame 158 (2011) 15421552 Table 2 Calculated test rig values. Value Oxidant composition (wet) CO2 O2 H2O N2 Flue gas composition at furnace exit (wet) CO2 O2 H2O N2 Overall combustion stoichiometry Total wall heat ux Unit vol.% vol.% vol.% vol.% vol.% vol.% vol.% vol.% kW Air combustion 20.9 (0.1) 79.1 (0.1) 8.7 (1.2) 2.3 (0.4) 17.4 (0.8) 71.6 (1.3) 1.09 (0.04) 139 (11) Oxyfuel combustion 54.6 (5.3) 31.7 (1.1) 12.1 (4.8) 1.6 (1.9) 60.1 (5.2) 2.6 (0.4) 35.6 (3.7) 1.8 (2.1) 1.10 (0.03) 155 (12) CSNB combustion 38.0 (2.5) 51.1 (1.6) 9.3 (2.6) 1.6 (1.0) 48.9 (2.7) 2.8 (0.4) 46.3 (1.9) 2.0 (1.3) 1.05 (0.03) 168 (13)

1547

Table 3 Measured test rig values. Value Flue gas composition at furnace exit (dry) CO2 O2 CO NO Flue gas exit temperature Flue gas ow at end of furnace Recirculated ue gas ow Additional oxygen Unit vol.% vol.% ppm g/m3 C kg/h kg/h kg/h Air combustion 10.6 (1.5) 2.8 (0.5) 152 (158) 27 (75) 626 (14) 299 (11) Oxyfuel combustion 93.3 (2.3) 4.0 (0.7) 412 (206) 10 (75) 595 (7) 252 (9) 176 (6) 61.4 (2.6) CSNB combustion 91.1 (1.6) 5.1 (0.7) 300 (222) 23 (75) 570 (11) 154 (6) 76 (3) 62.2 (2.6)

The aim was to get the lowest possible O2 content in the ue gases and at the same time the lowest possible CO emissions. A low O2 content in the ue gas is necessary in an oxyfuel process to avoid unnecessary power losses for the production of oxygen and for the compression of the ue gases. The wet O2 concentrations in the ue gases for the air and the oxyfuel case with 70% recirculation were similar with an overall combustion stoichiometry of 1.1 (Table 2). The overall combustion stoichiometry for the CSNB case with 50% recirculation was lower with 1.05, but the remaining oxygen in the ue gas was even 0.2 vol.% higher than in the oxyfuel case. As the total mass ow to the stack is nearly independent from the recirculation rate (around 40 kg/h) we got a higher O2 loss for the 50% case. This is also reected in a 1.3% higher O2 mass ow from the oxygen supply tank. The reason for this is the fact that we recycled in the high recirculation oxyfuel case more oxygen from the ue gases back into the combustion and needed therefore less excess oxygen to reach a given combustion stoichiometry (Fig. 8 and Eq. (2)).

10

1.1

Oxygen excess [%]

combustion

= 1.1

1.08

1.06

combustion

= 1.05
1.04

1.02

20

40

60

1 80

Recirculation rate [%]

Fig. 8. Dependency of necessary oxygen excess and process stoichiometry (excl. recirculated oxygen) on recirculation rate and combustion stoichiometry (incl. recirculated oxygen).

O2;ex kprocess 1 kcombustion 1 1 

2
air and oxyfuel case were supposed to be very similar to the middle and bottom ame as the burner operation parameter were the same. A clear indication of the ame or hot ue gas impingement on the chamber wall opposite of the burners was the bright glow of the refractory in the blind measurement ports (right side of pictures). The effect of the swirl generator in the secondary oxidant outlet of the burner (Section 2.2) could be seen clearly in the over stoichiometric bottom burner of the CSNB case. The ame reached the measurement plane only barely and there was no impingement on the opposite wall. Another observation for the CSNB case was the ame luminosity difference between the middle burner and the top burner. At the middle burner ue gases from the bottom burner with a high content of oxygen mixed and reacted with the natural gas from the middle burner. In the top burner this luminosity was less

where O2,ex is the oxygen excess production necessary as ratio of the necessary oxygen for stoichiometric combustion, kprocess is the stoichiometry of the process without recirculated oxygen, kcombustion is the overall stoichiometry in the furnace with recirculated oxygen and  is the recirculation rate of the process (Eq. (1)). This fact has direct implications on the efciency of the whole oxyfuel power plant. The production of excess oxygen results in additional power losses and the higher oxygen concentration in the ue gas results in higher compression losses. 3.3. Flame imaging All ames reached well into the measurement plane of the FTIR measurements, which is the line between the camera and the middle of the measurement port at the opposite side of the combustion chamber (Fig. 9). The not recorded ames of the top burner for the

process

[-]

1548

V. Becher et al. / Combustion and Flame 158 (2011) 15421552

Air combustion

Oxyfuel combustion

CSNB combustion

tur pic No

ed! ord c e r No

tur pic

rd e eco er

d!

Top Middle Bottom

Fig. 9. Flame pictures of experimental cases.

3.4. Spectral radiation measurements

12

Intensity [kW/(m2 str m)]

10 8 6 4 2 0 2

CSNB combustion Oxyfuel combustion Air combustion

2.5

3.5

4.5

5.5

Wavelength [m]
Fig. 10. Spectral radiation at middle burner; height = 1.75 m; resolution = 25 cm1.

Intensity [kW/(m2 str m)]

CSNB combustion Oxyfuel combustion Air combustion

The lower limit of the measured spectral range of 2 lm is governed by the sensitivity of the InSb detector and the upper limit of 5.5 lm is governed by the transmissivity of the sapphire combustion chamber window. The three measured bands result from H2O radiation (2.42.63 lm), from CO2 radiation (4.04.7 lm) and from overlapping H2O and CO2 radiation (2.633.6 lm). In the region of 4.7 lm a weak CO band is located and between 4.5 lm and 5.5 lm a weak H2O band. The background radiation from the cooled target can be estimated in the clear window between the bands (3.64.0 lm) (Table 4). The 4.3 lm CO2-band has the interesting property of widening with an increase of gas temperature, the so called hot bands (Figs. 10 and 11). A result was the observable strong absorption from the cold gas layer close to the wall in the middle of the band and two peaks from hot gas regions in the band wings. The absorption was higher in both recirculation cases as the emissivity in this band is nearly 1 for CO2 concentrations higher as 10 vol.% and for path length larger than 0.5 m [1517]. An interesting property was the nonexistence of continuous soot radiation in the low wavelength regions (around 2 lm). All measured spectral radiation could be traced back to gas molecules. The increase of radiation intensity from air over oxyfuel to CSBN combustion was evident in nearly all measurement ports (Fig. 12). The radiation from the ames was 35 times higher as the radiation from the ue gases between and after the burners. The ue gas radiation values were comparable for all three cases. 3.5. Gas temperatures One main objective of the experiments was to ensure that ame temperatures with the CSNB concept could be kept in realistic limits. The measured gas temperature proles were in the same range for all three cases (Figs. 13 and 14). The lowest temperatures were measured in the air case, slightly higher temperatures in the oxyfuel case and highest temperatures in the upper two burners of the CSNB case.

0 2

2.5

3.5

4.5

5.5

Wavelength [m]
Fig. 11. Spectral radiation between middle and top burner; height = 2.25 m; resolution = 25 cm1.

pronounced due to the higher concentration of inert combustion products from the two burners below. This behavior could still be seen besides the lower stoichiometry of the middle burner of 0.3 compared to a stoichiometry of 0.5 of the top burner.

V. Becher et al. / Combustion and Flame 158 (2011) 15421552 Table 4 Characteristic spectral ranges in observed part of spectrum; Molecule/ property H2O H2O; CO2 Background CO2 CO H2O Band center position (lm) 2.5 2.7 3.8 4.3 4.7 6.3 Intensity Strong Strong Strong Weak Weak Notes

1549

 is the spectral emissivity, xCO

is the molar fraction of CO2, s is the path length and p the total pressure.

Strong overlapping

 = 1 for xCO2 > 0:05, s > 0.5 m and p = 1 bar [17,16]; strong absorption of cold gas layers close to the wall; widening of band with increase of temperature Strong overlapping with CO2 and H2O bands

12.5

CSNB combustion Oxyfuel combustion Air combustion

Band intensity [kW/(msr)]

10

7.5

2.5

Bottom burner
0 0.5 1 1.5

Middle burner
2 2.5

Top burner
3 3.5

Combustion chamber height [m]

Fig. 12. Integrated band intensity in the range of 25.5 lm.

Gas temperature [C]

Gas temperature [C]

1500 1250 1000 750 500 250 0 0 50

1500 1250 1000 750 500 250 0 0 50

Gas temperature [C]

Air combustion

Oxyfuel combustion

CSNBcombustion
1500 1250 1000 750 500 250 0 0 50 After top burner / h = 3.25 m Top burner / h = 2.75 m 100 150 200 250 300 350

After top burner / h = 3.25 m Top burner / h = 2.75 m 100 150 200 250 300 350

After top burner / h = 3.25 m Top burner / h = 2.75 m 100 150 200 250 300 350

Radial position [mm] Gas temperature [C] Gas temperature [C]

1500 1250 1000 750 500 250 0 0 50 100 150 200 250 300 350 After middle burner / h = 2.25 m Middle burner / h = 1.75 m 1500 1250 1000 750 500 250 0 0 50 100

Radial position [mm] Gas temperature [C]

1500 1250 1000 750 500 250 0 150 200 250 300 350 0 50 100

Radial position [mm]

After middle burner / h = 2.25 m Middle burner / h = 1.75 m

After middle burner / h = 2.25 m Middle burner / h = 1.75 m 150 200 250 300 350

Radial position [mm] Gas temperature [C]

1500 1250 1000 750 500 250 0 0 50 100 150 200 250 300 350 After bottom burner / h = 1.25 m Bottom burner / h = 0.75 m

Radial position [mm] Gas temperature [C] Gas temperature [C]

1500 1250 1000 750 500 250 0 0 50 100 150 200 250 300 350 After bottom burner / h = 1.25 m Bottom burner / h = 0.75 m 1500 1250 1000 750 500 250 0 0 50 100

Radial position [mm]

After bottom burner / h = 1.25 m Bottom burner / h = 0.75 m 150 200 250 300 350

Radial position [mm]

Radial position [mm]

Radial position [mm]

Fig. 13. Measured temperature proles.

3.6. Wall heat uxes The second important result was the wall heat ux (Fig. 15). The heat ux should not raise considerably with the new concept.

The air case had the lowest heat ux and the CSNB case the highest heat ux. The heat ux increased downstream from the bottom burner for the air and the oxyfuel reference cases. The already in the temperature measurements observed hottest ame in the middle burner from the CSNB case was also observed in a

1550

V. Becher et al. / Combustion and Flame 158 (2011) 15421552

1600

1400

CSNB combustion Oxyfuel combustion Air combustion

Max. gas temperature [C]

1200

1000

800

600

Bottom burner
400 0.5 1 1.5

Middle burner
2 2.5

Top burner
3 3.5

Combustion chamber height [m]

Fig. 14. Maximum measured gas temperatures.

22.5

20

CSNB combustion Oxyfuel combustion Air combustion

Wall heat flux [kW/m]

17.5

15

12.5

10

Bottom burner
7.5 0 0.5 1

Middle burner
1.5 2

Top burner
2.5 3 3.5 4

Combustion chamber height [m]

Fig. 15. Specic wall heat uxes in combustion chamber; side with no ame impingement.

peak of the resulting wall heat ux. The total wall heat ux of the combustion chamber (Table 2) equated to 66%, 73.6% and 80.1% of the input heat. The decrease of ue gas mass ow in the furnace and therefore thermal heat stored in the ue gases exiting the combustion chamber was reected in these values. The values were higher as common values for large scale steam generators (Fig. 3) because the test rig had a considerable higher surface to volume ratio with similar wall temperatures (around 500 C) as a large scale steam generator. The heat transfer was additionally increased due to ame impingement. This resulted in a much lower exit gas temperature (around 600 C) out of the furnace section compared to large scale steam generators (around 1200 C depending on fuel type). 3.7. Flame stoichiometry variations The integrated band intensity from the middle burner had the maximum value at a stoichiometry of 0.7 (Fig. 16). The integrated band radiation decreased from this value to both sides. 4. Discussion The emission of CO and NOx in the CSNB case were in the same range or even lower as in the oxyfuel case. Therefore, the burnout

of the CSNB concept for natural gas combustion was comparable to normal unstaged combustion. The low O2 content in all cases is another indication for good mixing and burnout. The high CO emissions for all cases were due to the fact, that the burner was not optimized for low CO and NOx emissions. Two opposing observations were the high luminosity from the ame imaging and the nonexistence of soot radiation in the IR spectra. Soot forms only if the C/O-ratio is higher as 0.45 [18, p. 232]. The C/O-ratio in the test rig was always below 0.25 as the combustion stoichiometry in the test rig was always above one (Fig. 2) and the used natural gas consisted mostly of CH4. Methane as a fuel has the lowest soot formation propensity of all hydrocarbon fuels for turbulent ames [18, p. 240]. The addition of H2O and CO2 also lead in former experiments to a reduction of soot formation [18, p. 236]. These facts from literature support the experimental observation from the spectral measurements of nonexistence of soot in the ame. The high ame luminosity in the CCD camera pictures were due to the high sensitivity of the surveillance camera in the near infrared. The fact of higher oxygen consumption for a reduced recirculation rate is a drawback for the concept. The experiments showed that this higher oxygen consumption can be compensated by a lower combustion stoichiometry due to the higher concentration of oxygen compared to an oxyfuel process with high recirculation.

V. Becher et al. / Combustion and Flame 158 (2011) 15421552

1551

14

Band intensity [kW/(m sr)]

13 12 11 10 9 8 0.3

0.5

0.7

0.9

1.1

1.3

Burner total stoichiometry [-]

Fig. 16. Integrated band intensity (25.5 lm) of middle burner for stoichiometry variations.

150%
Wall heat flux Radiation intensity 110% 100% 100% 129% 121% 133%

100%

50%

0% Air Oxyfuel (70%reci.) CSNB (50%reci.)

Fig. 17. Comparison of total wall heat ux from non-impingement side and integrated band radiation intensity (25.5 lm).

The ame temperatures for the CSNB concept were in the same range as for air combustion and oxyfuel combustion (Fig. 13). The middle burner of the CSNB case showed the highest temperatures, but they were still in acceptable levels. The oxygen concentration in this zone from the upstream burner ue gas was higher as at the top burner. In the experiments this burner was operated with the lowest stoichiometry to account for this effect. The limits of the stoichiometric burner operation range were only a question of burner design. The variations of the middle burner stoichiometry showed the controlling effect of the burner stoichiometry on the ame radiation. The integrated band radiation decreased if the burner stoichiometry was decreased or increased from 0.7 (Fig. 16). The shifted maximum peak from 1 to 0.7 was a result of operating a fuel rich burner in a oxygen rich gas atmosphere. The basic relation of higher heat transfer in the radiative section with lowered recirculation rate (Fig. 3) could be shown experimentally (Fig. 17). From oxyfuel to CSNB combustion the heat ux increased by 4 11% with a decrease of the recirculation rate from 70% to 50%. The decrease in furnace exit temperatures from air over oxyfuel to CSBN combustion was another indication of this relation. The main reason for this increased heat ux was the increased radiation from the hot ames (Fig. 12). The radiation from the hot ue gases between and after the burner showed no large deviations for all cases. Therefore, the heat ux to the wall can be controlled by the ame temperatures which-as shown with the ame stoichiometry variations-can be controlled by the burner stoichiometry.

These results prove that a reduction of the ue gas recirculation rate in oxyfuel natural gas combustion from 70% down to 50% is possible while avoiding inadmissible high ame temperatures with the concept of Controlled Staging with Non-stoichiometric Burners. Other effects of a reduced recirculation rate are an increase of the heat transfer in the radiative section and higher oxygen consumption. Latter effect can be compensated by a lowered combustion stoichiometry. These effects have to be accounted for in the selection of the optimum recirculation rate during the design of the steam generator and the whole plant. Acknowledgments This work was carried out under the research project FriendlyCoal (RFC-CR-06007) with a nancial grant from the Research Fund for Coal and Steel of the European Community and in the framework of the TUM Graduate School. References
[1] M. Toftegaard, J. Brix, P. Jensen, P. Glarborg, A. Jensen, Prog. Energy Combust. Sci. 36 (5) (2010) 581625, doi:10.1016/j.pecs.2010.02.001. [2] R.M. Davidson, S.O. Santos, Oxyfuel Combustion of Pulverised Coal, Tech. Rep. CCC/168, IEA Clean Coal Center, 2010. [3] T. Wall, Proc. Combust. Inst. 31 (1) (2007) 3147, doi:10.1016/ j.proci.2006.08.123. [4] T. Wall, Y. Liu, C. Spero, L. Elliott, S. Khare, R. Rathnam, F. Zeenathal, B. Moghtaderi, B. Buhre, C. Sheng, R. Gupta, T. Yamada, K. Makino, J. Yu, Chem. Eng. Res. Des. 87 (8) (2009) 10031016, doi:10.1016/j.cherd.2009.02.005.

1552

V. Becher et al. / Combustion and Flame 158 (2011) 15421552 [12] P.R. Grifths, J.A. de Haseth, Fourier Transform Infrared Spectrometry, second ed., John Wiley & Sons, Inc., Hoboken, NJ, 2007. [13] DIN Deutsches Institut fr Normung e.V., DIN 1319-3: Grundlagen der Metechnik - Teil 3: Auswertung von Messungen einer einzelnen Megre, Meunsicherheit, 1996. [14] DIN Deutsches Institut fr Normung e.V., DIN 1319-4: Grundlagen der Metechnik - Teil 4: Auswertung von Messungen; Meunsicherheit, 1999. [15] J. Bak, S. Clausen, J. Quant. Spectrosc. Radiat. Transfer 61 (5) (1999) 687694, doi:10.1016/S0022-4073(98)00056-9. [16] S. Bharadwaj, M. Modest, J. Quant. Spectrosc. Radiat. Transfer 103 (1) (2007) 146155, doi:10.1016/j.jqsrt.2006.05.011. [17] S. Clausen, J. Bak, J. Quant. Spectrosc. Radiat. Transfer 61 (2) (1999) 131141, doi:10.1016/S0022-4073(97)00205-7. [18] B. Haynes, H. Wagner, Prog. Energy Combust. Sci. 7 (4) (1981) 229273, doi:10.1016/0360-1285(81)90001-0.

[5] B. Buhre, L. Elliott, C. Sheng, R. Gupta, T. Wall, Prog. Energy Combust. Sci. 31 (4) (2005) 283307, doi:10.1016/j.pecs.2005.07.001. [6] R. Tan, G. Coraggio, S. Santos, Technology Review Oxy-coal Combustion with Flue Gas Recycle for the Power Generation Industry, Tech. Rep. G 23/y/1, International Flame Research Foundation, 2005. [7] A. Kather, C. Hermsdorf, M. Klostermann, VGB PowerTech. 4 (2007) 8491. [8] C.W. Bale, P. Chartrand, S.A. Degterov, G. Eriksson, K. Hack, R. Ben Mahfoud, J. Melanon, A.D. Pelton, S. Petersen, Calphad 26 (2) (2002) 189228, doi:10.1016/S0364-5916(02)00035-4. [9] H. Spliethoff, Power Generation from Solid Fuels, Springer, Berlin Heidelberg, 2010, doi:10.1007/978-3-642-02856-4. [10] N. Fricker, How Do I Measure Local Flame Temperatures with the IFRF Suction Pyrometer? IFRF Online Combustion Handbook. <http://www.handbook. ifrf.net/handbook/>. [11] D. Harris, Infrared Phys. Technol. 39 (4) (1998) 185201, doi:10.1016/S13504495(98)00006-1.