Scholarly article on topic 'Development of 2nd Generation Oxyfuel CFB Technology – Small Scale Combustion Experiments and Model Development Under High Oxygen Concentrations'

Development of 2nd Generation Oxyfuel CFB Technology – Small Scale Combustion Experiments and Model Development Under High Oxygen Concentrations Academic research paper on "Chemical engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Energy Procedia
OECD Field of science
Keywords
{"carbon capture" / oxyfuel / "circulating fluidised bed" / "bubbling fluidised bed"}

Abstract of research paper on Chemical engineering, author of scientific article — Toni Pikkarainen, Jaakko Saastamoinen, Heidi Saastamoinen, Timo Leino, Antti Tourunen

Abstract The 1st generation oxyfuel CFB (Circulating Fluidised Bed) technology has been demonstrated up to 30MWth scale and the commercial concept for 300 MWe air/oxy flexible CFB power plant is available. Currently, the development of 2nd generation oxyfuel CFB technology is ongoing aiming to reduce significantly – around 50% – the overall efficiency penalty of CO2 capture in power plants compared to 1st generation concepts. The 2nd generation oxyfuel CFB plants are designed only to oxyfuel operation with CCS. The experimental results of the test campaigns with 0.1 MWth pilot scale CFB unit and laboratory scale BFB (Bubbling Fluidised Bed) unit under high oxygen concentrations of feed gas are presented. The pilot scale tests were carried out with Spanish anthracite and petroleum coke mixture and with Polish bituminous coal. Two Spanish limestone types were used for in- furnace sulphur capture. Test matrix of pilot scale CFB unit contained 11 test balances with varying feed gas O2 concentrations (between 21…42 vol-%) and O2 staging to primary and secondary gas feeds (primary gas O2 share 50…80 vol-%) at different bed temperature levels (820…920°C). Combustion performance and emission formation was studied at air combustion and varying oxyfuel combustion conditions. The fuel impulse tests with laboratory scale BFB included 16 tests with Spanish anthracite and Polish bituminous coal in varying feed gas O2 concentrations (5…50 vol-%) with two fuel size fractions (0.5-2.0 mm and 4.0-8.0 mm) at constant bed temperature level (850°C). The main objective was study how high O2 concentration effects on char reactivity and formation of nitrogen oxide emissions. A one dimensional model for laboratory scale BFB was used to further develop existing sub-models’ descriptions and parameters in the one dimensional model (1D-model) for pilot scale CFB in order to improve modelling capabilities in oxygen combustion conditions. Firstly the sub-models’ equations for different phenomenon – e.g. pyrolysis, char combustion and reactions of nitrogen species – were implemented and validated with bench scale experimental data. Secondly, the sub-model parameters found in bench scale modelling for char reactivity and NOx formation were implemented to the 1D-model of pilot scale CFB combustor. According to the results of the validation modelling against pilot scale experimental results, the combustion was successfully scaled up as the modelled temperature profiles and flue gas oxygen contents were well in line with measurements. Anyhow, further validation of the combustion model is needed by modelling of dynamic responses of pilot scale experiments. The pilot scale 1D-model was not able to predict NOx formation with the sub-model adapted from the bench scale model. Further model development and experimental work are needed with different particle size fractions at different operating conditions. The improved modelling capabilities under high oxygen concentrations can be utilized in the development and designing of 2nd generation oxyfuel CFB power plant concepts. Generally, the CFB technology appears to be ideally suited to oxyfuel combustion. The flexibility of the fluid-bed process offers an outstanding benefit for CFB in retaining uniform furnace temperature profiles, metal temperatures and local heat production rates in air-firing as well as in oxygen-firing and under varying load conditions. These benefits make also possible to apply higher oxygen concentrations which is a key element in the 2nd generation high efficiency oxyfuel CFB concepts.

Academic research paper on topic "Development of 2nd Generation Oxyfuel CFB Technology – Small Scale Combustion Experiments and Model Development Under High Oxygen Concentrations"

<®

CrossMark

Available online at www.sciencedirect.com

ScienceDirect

Energy Procedía 63 (2014) 372 - 385

GHGT-12

Development of 2nd generation oxyfuel CFB technology - small scale combustion experiments and model development under high

oxygen concentrations

Toni Pikkarainena*, Jaakko Saastamoinena, Heidi Saastamoinena, Timo Leinoa and

Antti Tourunenb

aVTT Technical Research Centre of Finland, P.O. Box 1000, FI-02044 VTT, Finland bReteres Oy, Kytomaantie 32, FI-41390 Aijala, Finland

Abstract

The 1st generation oxyfuel CFB (Circulating Fluidised Bed) technology has been demonstrated up to 30MWth scale and the commercial concept for 300 MWe air/oxy flexible CFB power plant is available. Currently, the development of 2nd generation oxyfuel CFB technology is ongoing aiming to reduce significantly - around 50% - the overall efficiency penalty of CO2 capture in power plants compared to 1st generation concepts. The 2nd generation oxyfuel CFB plants are designed only to oxyfuel operation with CCS.

The experimental results of the test campaigns with 0.1 MWth pilot scale CFB unit and laboratory scale BFB (Bubbling Fluidised Bed) unit under high oxygen concentrations of feed gas are presented. The pilot scale tests were carried out with Spanish anthracite and petroleum coke mixture and with Polish bituminous coal. Two Spanish limestone types were used for in-furnace sulphur capture. Test matrix of pilot scale CFB unit contained 11 test balances with varying feed gas O2 concentrations (between 21...42 vol-%) and O2 staging to primary and secondary gas feeds (primary gas O2 share 50...80 vol-%) at different bed temperature levels (820...920°C). Combustion performance and emission formation was studied at air combustion and varying oxyfuel combustion conditions.

The fuel impulse tests with laboratory scale BFB included 16 tests with Spanish anthracite and Polish bituminous coal in varying feed gas O2 concentrations (5... 50 vol-%) with two fuel size fractions (0.5-2.0 mm and 4.0-8.0 mm) at constant bed temperature level (850°C). The main objective was study how high O2 concentration effects on char reactivity and formation of nitrogen oxide emissions.

* Corresponding author. Tel.: +358-400-872153 E-mail address: toni.pikkarainen@vtt.fi

1876-6102 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.Org/licenses/by-nc-nd/3.0/).

Peer-review under responsibility of the Organizing Committee of GHGT-12

doi: 10.1016/j.egypro.2014.11.040

A one dimensional model for laboratory scale BFB was used to further develop existing sub-models' descriptions and parameters in the one dimensional model (ID-model) for pilot scale CFB in order to improve modelling capabilities in oxygen combustion conditions.

Firstly the sub-models' equations for different phenomenon - e.g. pyrolysis, char combustion and reactions of nitrogen species - were implemented and validated with bench scale experimental data. Secondly, the sub-model parameters found in bench scale modelling for char reactivity and NOx formation were implemented to the ID-model of pilot scale CFB combustor. According to the results of the validation modelling against pilot scale experimental results, the combustion was successfully scaled up as the modelled temperature profiles and flue gas oxygen contents were well in line with measurements. Anyhow, further validation of the combustion model is needed by modelling of dynamic responses of pilot scale experiments. The pilot scale ID-model was not able to predict NOx formation with the sub-model adapted from the bench scale model. Further model development and experimental work are needed with different particle size fractions at different operating conditions. The improved modelling capabilities under high oxygen concentrations can be utilized in the development and designing of 2nd generation oxyfuel CFB power plant concepts.

Generally, the CFB technology appears to be ideally suited to oxyfuel combustion. The flexibility of the fluid-bed process offers an outstanding benefit for CFB in retaining uniform furnace temperature profiles, metal temperatures and local heat production rates in air-firing as well as in oxygen-firing and under varying load conditions. These benefits make also possible to apply higher oxygen concentrations which is a key element in the 2nd generation high efficiency oxyfuel CFB concepts. ©2014 TheAuthors. PublishedbyElsevierLtd.Thisis an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

Peer-review under responsibility of the Organizing Committee of GHGT-12

Keywords: carbon capture; oxyfuel ; circulating fluidised bed; bubbling fluidised bed

1. Introduction

Due to growing interest to decrease greenhouse gas emissions in order to control the climate change, CO2 capture and oxygen combustion technology has been under intensive study during the recent years [1-4]. Fluidized bed combustion is efficient way to produce energy from wide variety of fuels, including low quality fossil fuels. However, carbon capture by oxygen combustion technology causes significant efficiency penalty in power production. Fluidized bed combustion in high oxygen concentrations is promising option to decrease this penalty.

In this study experimental work under high oxygen content fluidised bed oxyfuel combustion and reference conditions was carried out at two scales, bench and small pilot scale. Bench scale bubbling fluidized bed (BFB) combustion tests in oxygen combustion conditions with 5-50 vol-% of oxygen in feed gas and pilot scale circulating fluidised bed (CFB) combustion test campaign up to 42 vol-% O2 in feed gas with two fuel mixtures were carried out. The produced experimental data serves the understanding of the phenomenon as well as the development and validation of bench scale sub model for oxygen combustion conditions at high oxygen combustion conditions which further serves the pilot and full scale modelling.

2. Experimental set-up

2.1. Bench scale BFB

A photograph and schematic diagram of the bench scale bubbling fluidized bed (BFB) combustor is shown in Fig. 1. The height of the riser tube is 670 mm and the inner diameter of the combustor is 36 mm in the lower part and 53 mm in the upper part. The main flue gas compounds were measured by FTIR spectrometer and on-line analyzers after the filter. Instead of CFB combustor, BFB combustor was selected to have the most representative conditions from the fuel particle point of view in batch combustion procedure. The target of the tests was to solve how high O2 concentrations affect the char reactivity of coal and emission formation in fluidized bed combustion. In CFB, part of the particles is in circulation loop causing e.g. time delay in combustion, cooling of the particles and secondary fragmentation by mechanical stress.

Fig. 1. Photograph and schematic diagram of the bench scale bubbling fluidized bed (BFB) combustor.

Batches of Spanish anthracite and Polish bituminous coal were fed in to bubbling fluidized sand bed. Fluidization gas O2 concentrations of 5, 15, 30, and 50% of were examined, and the rest of the feed gas was CO2 without flue gas recirculation. Pre-heated combustion gas was fed with 50/50 vol-% shares to primary and secondary air. Two fuel size fractions were tested, 0.5-2.0 mm and 2.0-4.0 mm. The temperature in the combustor was set to 850°C with heating appliance in all tests and temperature change due to coal combustion was measured from six points along the riser tube. Totally test matrix contains 16 tests with two fuel quality and two particle size at varying feed gas O2 concentrations and at constant bed temperature. Composition of the fuels is in Table 1.

Table 1. Fuel analysis of bench scale BFB tests.

Spanish anthracite Polish bituminous coal

Analysis Unit

0.5-2.0 mm 2.0-4.0 mm 0.5-2.0 mm 2.0-4.0 mm

Moisture w-%, wet 3.0 1.6 8.9 7.7

Ash (815°C) w-%, dry 35.3 28.2 22.4 21.5

Volatiles w-%, dry 8.3 7.8 29.3 30.6

Carbon w-%, dry 57.5 65.1 59.5 60.1

Hydrogen w-%, dry 2.4 2.4 3.6 3.8

Nitrogen w-%, dry 0.92 1.03 0.84 0.89

Sulphur w-%, dry 1.23 1.02 1.86 1.37

2.2. Pilot scale CFB

A schematic diagram of the pilot scale circulating fluidized bed (CFB) combustor is shown in Fig. 2. The test rig can be operated in air and oxygen combustion modes with fuel thermal input ranging between 20-140 kW. The height of the riser is 8 m and the inner diameter is 167 mm. The combustor is equipped with several separately controlled electrically heated and water/air-cooled zones in order to control the process conditions (for example oxygen level, temperature and load) almost independently. During the oxygen combustion tests under high oxygen content additional water-cooled bed heat exchanger was installed to control the temperature level. The combustor is controlled with an automation system on which all measurement data is stored.

Fuel can be fed into the combustor through two separate fuel feeding lines and third feeding line is for additives such as limestone. Fuel and additive containers are mounted on the top of scales which enables the determination of mass flow rates for solid materials as a weight loss against time. Bed material (bottom ash) can be sampled above the grid via sampling tube and circulation material sample can be taken below the primary cyclone and from the loop seal. Fly ash samples can be taken from the secondary cyclone, gas cooler, extra gas cooler and bag house filter. In addition, several ports for gas and solid material sampling are located in the combustor freeboard area.

The feed gas is divided into primary, secondary and tertiary gases. Primary gas is fed through the grid, and the secondary and tertiary gases can be fed to three different levels on the combustor. The composition of primary and secondary gases can be adjusted independently. In this work air and different mixtures of oxygen and recirculated flue gas were used as feed gases. The flue gas composition was analyzed by a FTIR-spectrometer and by conventional on-line gas analyzers for main flue gas compounds.

Fig. 2. Schematic diagram of the pilot scale circulating fluidized bed (CFB) combustor.

The test campaign contained two set of tests:

1. Spanish anthracite and petcoke blend as fuel and Spanish limestone 1 for in-furnace sulphur capture

2. Polish bituminous coal as fuel and Spanish limestone 2 for in-furnace sulphur capture

The test matrix is presented in Table 2 and the main variable of each test is highlighted by yellow. The fluidization velocity and flue gas oxygen content were kept constant in all tests by a adjusting the fuel power. The first set with anthracite/petcoke blend contains reference tests in air combustion and oxyfuel combustion test with low (28 vol-%) feed gas O2 (test 1 and 2). Tests 3-5 were high oxygen content (42 vol-%) tests at different temperature levels controlled by furnace cooling power. In test 6-7 oxygen staging to primary and secondary gas feeds was varied while total feed gas oxygen content and total primary and secondary gas shares and flows remained constants. The second set with bituminous coal contains similar high feed gas oxygen content tests than in the first set with anthracite/petcoke blend, but no reference tests in air and low feed gas O2 oxyfuel conditions. Fuel and limestone compositions are presented in Table 3 and Table 4, respectively.

Table 2. Test matrix of pilot scale CFB tests.

Test Mode Fuel Limes. Fuel power Ca/S total Bed temp. Feed gas O2 Prim. gas O2 Sec. gas O2 Prim. gas Grid fluid. veloc.

# air/oxy type type kW mol/mol °C % wet % wet % wet share-% m/s

1 air 55 3.1 889 21 21 21 2.3

2 e o 79 2.8 873 28 28 28 2.2

3 c u la 121 2.6 918 42 42 42 2.3

4 % j 118 2.6 856 42 42 42 70 2.2

5 oxy ci3 Is £ 121 2.7 817 42 42 42 2.1

6 I 121 2.7 863 42 48 29 2.2

7 118 2.5 853 42 37 55 2.2

10 al 132 3.4 920 42 42 42 2.3

11 o c al c er 131 3.4 863 42 42 42 2.2

12 oxy to s ^ c 132 3.6 841 42 42 42 60 2.2

13 * I ut rB 132 3.9 869 42 48 32 2.2

14 bi 132 3.8 866 42 36 51 2.2

Table 3. Fuel compositions of pilot scale CFB tests.

Analysis Unit Spa nish anthracite Spa nish petcoke Polish bit. coal

Moisture w-%, wet 3.6 2.4 20.4

Ash (815°C) w-%, dry 30.9 0.45 13.0

Volatiles w-%, dry 8.0 10.1 35.9

Fixed carbon w-%, dry 61.1 89.5 51.1

Higher heating value MJ/kg, dry 24.09 35.11 26.64

Lower heating value MJ/kg, dry 23.60 34.37 25.70

Carbon w-%, dry 62.8 87.9 65.6

Hydrogen w-%, dry 2.3 3.5 4.4

Nitrogen w-%, dry 1.0 2.1 1.1

Sulphur w-%, dry 1.2 5.3 2.0

Calcium w-%, dry 0.71 0.04 0.23

Table 4. Limestone compositions of pilot scale CFB tests.

Analysis Unit Limestone 1 Limestone 2

Moisture w-%, wet 0.1 0.3

CaCO3 w-%, dry 98.1 93.6

MgCO3 w-%, dry 0.8 2.1

Inert w-%, dry 1.1 4.2

3. Experimental results

3.1. Bench scale BFB

According to bench scale BFB combustion tests the increase in oxygen concentration in feed gas increased the combustion rate. In high oxygen concentrations volatile release and combustion as well as char gasification and combustion occurred almost simultaneously, as in low oxygen concentration these phases separate. This was especially clear with bigger fuel size fraction 2.0-4.0 mm. Combustion of bituminous coal occurred faster compared to anthracite due to higher volatile content of fuel. Flue gas oxygen responses are illustrated in Fig. 3.

Fig. 3. Flue gas oxygen responses.

Feed gas 5 15 30 50

O2-% Anthracite 0.5-2.0 mm

5 15 30 50

Anthracite 2.0-4.0 mm

5 15 30 50

Bit. coal 0.5-2.0 mm

N2O NO

5 15 30 50

Bit. coal 2.0-4.0 mm

Fig. 4. Fuel nitrogen conversion to flue gas NO and N2O.

The peak temperature increased along with the increase in oxygen concentration. This affected the rate of NO formation, which increased along the increase in oxygen feed. Release of nitrogen strictly followed the volatile release and char combustion. Although the rate of NO formation increased, there was not significant change in total nitrogen release to NO and N2O (Fig. 4). In tests with 50% feed gas oxygen concentration, nitrogen conversion to NO decreased slightly and conversion to N2O increased.

3.2. Pilot scale CFB

In pilot scale CFB combustion tests temperature profiles were more uniform with low volatile anthracite/petcoke fuel blend than with higher with higher volatile bituminous coal (Fig. 5). The effect of oxygen share variations on temperature profiles was not significant for either of the fuel types. Solids densities along the furnace height and solids circulation rates were high which enhanced the heat transfer and smoothed the temperature profiles. At comparable conditions the solids density profiles were rather similar for both fuel types.

In oxyfuel conditions with higher CO2 and SO2 concentrations and longer gas residence time in the flue gas recirculation loop the sulphur capture was better than that at air-firing conditions. The best sulphur capture performance by limestone was measured in in oxygen combustion conditions where high CO2 partial pressure hinders the calcination but not prevents it, practically in temperatures 850...880oC. The improvement in sulphur capture in this indirect sulphur capture at high CO2 concentration is explained by enhanced diffusion of SO2 when delayed flow of CO2 from calcination through the sulphation product layer hinders the formation of dense CaSO4 layer. The dense CaSO4 is the main reason for relatively low calcium utilisation for sulphur capture in fluidised bed combustion applications. It has been observed also in earlier experimental work [5-6] that the sulphur capture is worse in direct sulphation region but capture is enhanced by higher CO2 partial pressure in indirect sulphation region. Only conditions in test 5 were clearly on CaCO3 side of CaCO3-CaO equilibrium curve, when sulphur is directly captured by CaCO3 without calcination. The sulphur capture was clearly poorer in test 5 compared to other oxygen combustion tests. Fig. 6 illustrates SO2 emissions, calcium utilisations, sulphur retentions (RET) and total Ca/S molar ratios (Ca/S). Calcium utilization is defined as RET/(Ca/S) and it tells the share of calcium in limestone and fuel utilized in SO2 capture.

Fig. 5. Temperature profiles in the riser in medium temperature pilot scale CFB tests.

! 120 S 100 i 80

I 20 ! 0

SO2 emission [mg/MJ] ^Calcium utilisation [%]

Ca/S total [mol/mol] ♦ SO2 retention [%]

94.6 97.6 98.2 99.0 1. ♦ 98.7 99.1

♦ A 2.8

▲ 2.6 A 2.6 A 2.7 ▲ 2.7 ▲ 2.5

35.1 37.6 38.1 4. 36.9 39.3

■ ■ ■

94 40 31 17 40 21 14

Test 1 Test 2 Test 3 Test 4 Test 5 Test 6 Test 7

Air -medium T Oxy 28% -medium T Oxy 42% -high T Oxy 42% -medium T Oxy 42% -low T Oxy 42% -high prim. O2 Oxy 42% -low prim. O2

SO2 emission [mg/MJ] ^Calcium utilisation [%]

86.9 ♦

3.4 ▲

25.2 207

Ca/S total [mol/mol] ♦ SO2 retention [%]

3.6 ▲

98.0 ♦

® 100

z m 60

Test 12

Test 14

Oxy 42% -high T

Oxy 42% -

Oxy 42% - I Oxy 42% -low T I high prim. O2

Oxy 42% -w prim. O2

Fig. 6. SO2 emissions, calcium utilisations, sulphur retentions and total Ca/S molar ratios of pilot scale CFB tests.

Both NO and N2O emissions were much lower in pilot scale oxyfuel combustion tests due to reduction of nitrogen oxides to elementary nitrogen when flue gas is recirculated through bed material and reducing gas atmosphere. At comparable conditions in oxyfuel combustion the feed gas oxygen content had no remarkable effect on emissions of nitrogen oxides. In anthracite/petcoke blend tests significant part of nitrogen oxides' emissions were in form of N2O which is typical for high fixed carbon (low volatile content) fuels. In high oxygen content oxyfuel tests NO emission increased with increasing temperature whereas N2O decreased with increasing temperature. The tested oxygen staging had no significant effect on nitrogen oxide emissions.

80 70 60 i 50 i 40

1 30 1

m 20 10 0

□ NO emission [mg/MJ] DN2O emission [mg/MJ] »Bed temperature

Test 2 Test 3 Test 4 Test 5 Test 6 Test 7

Oxy 28% - Oxy 42% - Oxy 42% - Oxy 42% - Oxy 42% - Oxy 42% -

medium T high T medium T low T high prim. low prim. O2

860 T¡

„ 25 1

I 15 1 m 10

lest 1

Air -medium T

Fig. 7. NO and N2O emissions and bed temperatures of pilot scale CFB tests.

4. Modelling

4.1. Bench scale BFB

4.1.1. Model description

The experimental and modelling work around the bench scale fluidized bubbling fluidized bed combustor combustor is utilized in developing the combustion and emission formation sub-models for pilot scale CFB. The model for the combustor is based on VTT's one dimensional model for pilot scale CFB presented in [7]. Model calculates the mass balance of gas species (O2, CO2, CO, H2O, and NO) along the furnace riser. In order to calculate the balance, the combustion chamber is divided vertically to control volumes. Model includes equations for drying, volatile release and combustion, char combustion and CO combustion.

Since in this study, the bench scale combustor is operated in bubbling fluidized bed mode, fuel particles are assumed to stay in the bottom bed part of the combustor. Therefore drying, volatile release and char combustion are

assumed to occur in the lowest calculation cell of the combustor. The shrinking particle model in [9] is applied for char combustion and the model parameters are determined from the experimental fuel batch responses. The effect of char primary fragmentation is included in the apparent model parameters. Both CO and CO2 are formed in char combustion. Volatile combustion is assumed to be rapid compared to char combustion, and it occurs immediately after volatile release. CO combustion is modelled using a reaction rate defined in [10] in all the control volumes.

In the conditions applied in this study, all of the measured nitrogen is from the fuel combustion, since feed gas does not contain N2. In the modelling approach, fuel nitrogen is assumed to spread to volatiles and char according to the volatile content in the fuel fraction. Since significant amounts of precursors HCN and NH3 were not measured in the experiments, released nitrogen was assumed to form NO. A similar approach is used widely in the literature [11-13].According to the literature, NO can be reduced in the gas phase reaction with NH3 on the char surface, and in the CO catalysed reaction with solids in the reactor [14-17]. In this study, formed NO is assumed to reduce to N2 on the char surface. The bench scale model applied is described more detailed in [8].

4.1.2. Modelling results

The combustion model developed for pilot scale CFB combustor operated in air combustion conditions works relatively well in modelling the bench scale BFB combustor results in oxygen combustion conditions at 850°C, as can be seen in Fig. 8 illustrating the combustion of anthracite and bituminous coal fraction 0.5-2.0 mm in 5, 15, 30 and 50% vol-O2 feed gas. The modelled flue gas O2 responses from fuel batches are close to measured values.

Feed gas 02 = 5 vol-%

° 4 -

modelled, anthracite —measured, anthracite -- modelled, bituminous —measured, bituminous

200 250 300 350 400 450

Feed gas 02 = 30 vol-%

—modelled, anthracite —measured, anthracite --modelled, bituminous --measured, bituminous

350 400 450

Feed gas 0)2 = 15 vol-%

50 1,49

'■§ 48

—modelled, anthracite —measured, anthracite

— modelled, bituminous

— measured, bituminous

300 t[s]

Feed gas 02 = 50 vol-%

------------ iM

modelled, anthracite measured, anthracite modelled, bituminous measured, bituminous

50 200 250

300 t[s]

350 400 450

Fig. 8. Measured and modelled flue gas oxygen responses in varying feed gas oxygen content at 850°C temperature level for anthracite and bituminous coal batch feed tests of size fraction 0.5-2.0 mm in bench scale BFB combustor.

The model for NO formation on char particle surface [12] with reduction rate model for NO to N2 on char surface [16] works well in bench scale fuel impulse experiments for anthracite. However, modelling results for bituminous coal were not as favorable due to significantly higher volatile content of fuel. Two approaches to volatile nitrogen modelling were taken. In the first one, volatile nitrogen forms NO which reacts on char particle surface by rate defined reaction for char-NO reduction. In the second approach, all formed volatile NO reduces to N2. The measured

and modelled flue gas NO responses with two volatile nitrogen modelling approaches for anthracite and bituminous coal batch feed tests are illustrated in Fig. 9 and in Fig. 10, respectively. The latter approach works better in predicting the NO formation in bituminous coal combustion.

180r ,160-.140120-'10080604020

NO concentration in 15% O2 feed

—volatile N forms NO —volatile N forms N2 —measured

200 250

300 t[s]

350 400 450

180 ,160 .140 120 '100 80 60 40 20

NO concentration in 50% O2 feed

—volatile N forms NO —volatile N forms N2 —measured

200 250

300 t[s]

350 400 450

Fig. 9. Measured and modelled flue gas NO responses in 15 and 50 vol-% feed gas oxygen content at 850°C temperature level for anthracite batch feed tests of size fraction 0.5-2.0 mm in bench scale BFB combustor; two modelling approaches for volatile nitrogen [8].

Fig. 10. Measured and modelled flue gas NO responses in 15 and 50 vol-% feed gas oxygen content at 850°C temperature level for bituminous coal batch feed tests of size fraction 0.5-2.0 mm in bench scale BFB combustor; two modelling approaches for volatile nitrogen [8].

4.2. Pilot scale CFB

4.2.1. Model description

Models in zero and one dimension can usually adequately describe the overall behavior of the processes for many practical purposes at same time keeping the computational time short. One dimensional time dependent model (ID-model) with mass balance equations for chemical species and combustion reactions has been found to describe combustion process accurately enough in modelling VTT's pilot scale CFB combustor. In the ID-model (Fig. 11) the furnace is divided into 80 control volumes. The mass balances for main gas components (CO, CO2, O2, H2O, N2), volatile compounds as well as emissions (SO2, NOx) are calculated for each control volume against time. Char is divided into particle size fractions and zero dimensional mass balance for each char fraction is calculated. The 1D-model is described more detailed in [7].

The sub-model parameters found in bench scale modelling (see Chapter 4.1) for char reactivity and NOx formation were implemented in the 1D- model of pilot scale CFB combustor. The target was to find out if the bench scale experimental results can be directly utilized in the pilot scale modelling. The validation of the pilot scale 1D-model was done with experimental data presented in Chapter 3.2.

Flue gas

to stack Fluegas

Fig. 11. Schematic layout of the dynamic ID-model of CFB-pilot

4.2.2. Modelling results

The results for oxygen profiles in the furnace showed that the combustion and heat release profiles can be modelled fairly well in both air and oxygen firing cases (Fig. 12-14). Some further study must be done for the heat release in high oxygen combustion conditions (Fig. 14). As it was expected the greatly simplified first approach for the NO emission formation produced 4 times greater emissions in air combustion and 10-15 greater in oxygen combustion due to utilisation of recirculated flue gas. The NO formation tests in bench scale must be done in several temperatures and atmospheres to cover the range of conditions particle experiences in the CFB combustor. In addition the reducing effect of char and CO must be added to the NO emission sub-model.

Fig. 12. The measured and modelled O2, temperature and NO profiles in air combustion test with Polish bituminous coal.

Fig. 13. The measured and modelled O2, temperature and NO-profiles in oxygen combustion test with 28 vol-% wet O2 content in feed gas with Polish bituminous coal.

n 15 O

§ 850,

Distance from the grid (m)

Distance from grid (m)

9000 8000 7000 ) 6000 ^ 5000

a 4000

z 3000 2000 1000 0

Measured ~ Simulated

Distance from the grid (m)

Alter pnm

Alter prim

Fig. 14. The measured and modelled O2, temperature and NO-profiles in oxygen combustion test with 42 vol-% wet O2 content in feed gas with Polish bituminous coal.

5. Conclusions

According to bench scale BFB combustion tests the increase in oxygen concentration in feed gas increased the combustion rate. In high oxygen (> 30 vol-%) concentrations volatiles release and combustion as well as char gasification and combustion occurred almost simultaneously, as in low oxygen concentrations char reactions last much longer than volatiles combustion.

In pilot scale CFB combustion experiments the difference in combustion performance between air and oxygen combustion modes were relatively small, as the high solids densities along the furnace height and high solids circulation rates enhanced the heat transfer and smoothed the temperature profiles. On the contrary, in emission performance clear differences were measured. In oxyfuel conditions with higher CO2 and SO2 concentrations and longer gas residence time in the flue gas recirculation loop the sulphur capture was better than that at air-firing conditions. In oxygen combustion, both NO and N2O emissions were much lower due to reduction of nitrogen oxides to elementary nitrogen when flue gas is recirculated through bed material and reducing gas atmosphere. The best sulphur capture performance by limestone was measured in in oxygen combustion conditions where high CO2 partial pressure hinders the calcination but not prevents it. The worst sulphur capture in oxygen combustion was measured in direct sulphur capture conditions, where high CO2 partial pressure in low temperature prevents the calcination of CaCO3. Therefore the design of the furnace heat transfer and temperature levels has significant effect on in-furnace sulphur capture in CFB combustor, especially in oxygen combustion conditions. At comparable conditions in oxygen combustion the feed gas oxygen content or oxygen staging (to primary and secondary feed gases) had no remarkable effect on SO2 capture or emissions of nitrogen oxides.

The one dimensional combustion model (ID-model) developed for pilot scale CFB combustor operated in air combustion conditions works relatively well in modelling the bench scale BFB combustion results in oxygen combustion conditions at 850°C temperature level. Modelled flue gas O2, CO2, CO, H2O and bed temperature responses were well in line with measured values. For modelling the NO emission formation, two approaches to volatile nitrogen modelling were considered: in the first one, volatile nitrogen forms NO which reacts on char particle surface by rate defined reaction for char-NO reduction, and in the second one, all formed volatile NO reduces to N2. Latter approach gave better results in case of bituminous coal with higher volatiles content and with anthracite with low volatiles content there was not big difference between the approaches.

The sub-model parameters found in bench scale modelling for char reactivity and NOx formation were implemented in the 1D- model of pilot scale CFB combustor. The results of validation model simulations showed that the combustion and heat release profiles can be modelled fairly well in both air and oxygen firing cases. Anyhow, further validation of the combustion model is needed by modelling of dynamic responses of pilot scale experiments. As expected, the greatly simplified first approach for the NO emission formation must be further developed by experimental work covering temperatures and atmospheres particle experiences in the CFB combustor. In addition the reducing effect of char and CO must be added to the NO emission sub-model.

Generally, the CFB technology appears to be ideally suited to oxyfuel combustion. The flexibility of the fluid-bed process offers an outstanding benefit for CFB in retaining uniform furnace temperature profiles, metal temperatures and local heat production rates in air-firing as well as in oxygen-firing and under varying load conditions. These benefits make also possible to apply higher oxygen concentrations which is a key element in the 2nd generation high efficiency oxyfuel CFB concepts.

Acknowledgements

The research leading to these results has received funding from the European Union Seventh Framework Programme FP7/2012 under grant agreement no 295533 (O2GEN, "Optimization of Oxygen-based CFBC Technology with CO2 capture") and grant agreement n° 239188 (FLEXI BURN CFB, "Development of High Efficiency CFB Technology to Provide Flexible Air/Oxy Operation for Power Plant with CCS").

References

[1] Anthony, E., J.: Greenhouse Gas Sci Technol. 3 (2013), pp. 106-123.

[2] Buhre, B., Elliot, L., Sheng, C., Gupta, R., Wall, T.: Progress in Energy and Combustion Science (2005), vol. 31, pp. 283-307.

[3] Scheffknecht, G. Al-Makhadmed, L., Schnell, U., Maier, J. (2011). International Journal of Greenhouse Gas Control 55, pp. S16-S35.

[4] Toftegaard, M., Brix, J., Jensen, P., Glarborg, P., Jensen, A.: Progress in Energy and Combustion Science 36 (2010), pp. 581-625.

[5] Takkinen, S., Hyppanen, T., Saastamoinen, J., Pikkarainen, T. Experimental and Modeling Study of Sulfur Capture by Limestone in Selected Conditions of Air-Fired and Oxy-fuel Circulating Fluidized-Bed Boilers. Energy & Fuels, 2011, 25, 2968-2979. 06/2011; D0I:10.1021/ef200416e

[6] Rahiala, S., Hyppanen, T., Pikkarainen, T. A bench scale and modeling study of sulphur capture by limestone in typical C02 concentrations and temperatures of fluidised bed air and oxy-fuel combustion. Energy & Fuels, 2013, 27, 7664-7672. dx.doi.org/10.1021/ef401971m

[7] Tourunen, A. (2010). A Study of Combustion Phenomena in Circulating Fluidized Beds by Developing and Applying Experimental and Modelling Methods for Laboratory-Scale Reactors. Thesis for the degree of Doctor of Science. Lappeenranta University of Technology, pp. 34-38.

[8] Saastamoinen, H., Leino, T., Saastamoinen, J., Tourunen, A., Pikkarainen, T. Analysis and modelling of bench-scale fluidized bed combustion experiments in high oxygen concentrations. Submitted to Fuel (2014).

[9] Saastamoinen, J., Tourunen, A. Energy&Fuels (2012), pp. 407-416.

[10] Saastamoinen, J., Kilpinen, P., Norstrom, T. (2000). Energy&Fuels (14), pp. 1156-1160.

[11] Jones, J., Patterson, P., Pourkashanian, M., William, A.: Carbon (1999), vol. 37, pp. 1545-1552.

[12] Saastamoinen, J., Huttunen, M., Kilpinen, P., Kjaldman, L., Oravainen, H., Bostrom, S. Progress in Computational Fluid Dynamics (2006), vol. 6, pp. 209-216.

[13] Goel, S., Zhang, B., Sarofim, A. Combustion and Flame 104 (1996), pp. 213-217.

[14] Aarna, I., Suuberg, E.: Energy&Fuels (1999), vol. 13, pp. 1145-1153.

[15] Desroches-Ducarne, E., Dolignier, J., Marty, E., Martin, G., Delfosse, L.: Fuel (1998), vol. 77, pp. 1399-1410.

[16] Zevenhoven, R., Hupa, M.: Fuel (1998), vol. 77, pp. 1169-1176.

[17] Liu, H., Gibbs, B.: Fuel (2002), vol. 81, pp. 271-280.