IFRF Combustion Journal Article Number 200502, June 2005 ISSN 1562-479X

Theoretical Study of COG and COG/BOF Gas Injection in a Blast Furnace Dietmar Andahazy*, Gerhard Löffler*, Franz Winter*, Christoph Feilmayr+ and Thomas Bürgler+ *

Christian Doppler Laboratory for Chemical Engineering at High Temperatures Vienna University of Technology Vienna, Austria

+

Voestalpine Stahl GmbH Linz, Austria

Editors Note: This article is one of the papers presented as posters and published in the proceedings of the 25th IFRF Topic Oriented Technical Meetings (TOTeM25) held at Jerkontoret, Stockholm - Sweden

Corresponding Author(s): Franz Winter Christian Doppler Laboratory for Chemical Engineering at High Temperatures Institute of Chemical Engineering Getreidemarkt 9/166 A-1060 Vienna Austria Tel. No.: Fax No.: E-mail :

+43 1 58801 159 40 +43 1 58801 159 99 [email protected]

IFRF Combustion Journal Article No 200502

-2 -

Andahazy et. al. June 2005

ABSTRACT Because of the availability of gases from the coke production and the basic oxygen furnace (BOF) process in an integrated metallurgical plant, a project of utilization of these gases is proposed. The gases are injected into the blast furnace to substitute reducing agents like heavy oil. The aim of this paper is to study the combustion characteristics of coke oven gas (COG) and a mixture of COG and gas from the BOF. The modeling approaches used are the thermodynamic equilibrium, the plug flow reactor (PFR) model with detailed chemistry (with and without consideration of the mixing time) and Computational Fluid Dynamic (CFD) modelling. Each approach leads to additional aspects for an improved understanding of the combustion processes. The COG/BOF gas mixture ignites earlier than the COG gas because of its different composition. The combustion of COG leads to a higher temperature due to its higher net calorific value so that the thermal strain on the tuyere is higher. The combustion of the COG gas results in higher H2O, H2 and CO and lower O2 and CO2 concentrations. Keywords: blast furnace, gas injection, modeling, combustion

IFRF Combustion Journal Article No 200502

-3 -

Andahazy et. al. June 2005

INTRODUCTION Because of the availability of gases from coke production and the basic oxygen furnace (BOF) in an integrated metallurgical plant, a project of utilization of these gases was developed [1]. In the blast furnace process alternative reducing agents such as heavy oil which is purchased from outside sources can be substituted. The injection of reducing gases into the blast furnace results in better flexibility in gas utilization. The tuyeres of a blast furnace are exposed to high thermal stress. The combustion of injected gases including reducing species such as CO and H2 into these tuyeres can lead to overheating of the tuyere and burning of the lances. For this reason a complete understanding of the combustion characteristics is very important. Furthermore it is the goal of this work to show the differences in combustion characteristics between coke oven gas (COG) and a mixture of COG and gas from the basic oxygen furnace process (COG/BOF), not only in overheating of the tuyere but also, for example, in predicting the composition of these gases at the exit of the raceway. The calculation of the combustion process can be done using different models. These different modeling approaches can be distinguished by the limiting factor which can be the mixing and/or the reaction kinetics. In the case of H2 combustion the mixing is limiting factor, whereas the reaction kinetics is the limiting factor for the combustion of CO and CH4. For COG and COG/BOF gas it is to be expected that the time scales of mixing and reaction kinetics are in the same range. The modeling approaches were chosen in such a way as to take four options into consideration (Table 1). The four models chosen cover the different possibilities of combining limiting and instantaneous steps. All of these models are needed to describe the entire combustion process of the gases. Table 1: Modeling options used Mixing

Reaction kinetics

Equilibrium

instantaneous

instantaneous

PFR with full chemistry

instantaneous

limiting

PFR with full chemistry and mixing

limiting

limiting

CFD (local equilibrium)

limiting

instantaneous

IFRF Combustion Journal Article No 200502

-4 -

Andahazy et. al. June 2005

The simplest approach is thermodynamic equilibrium which assumes instantaneous mixing and reaction. Reasoning this way it is possible to estimate the temperature and the species concentrations in the combustion products. Using the model of the plug flow reactor (PFR) with full chemistry, the transport equations are simplified in a one-dimensional form which allows the utilization of a complex, elementary reaction mechanism. This mechanism (GRI-Mech v 3.0) by Bowman et al. [2] contains 325 reactions. Consequently, the ignition processes based on radical species chemistry can be described and the ignition delay defined. The effect of kinetically limited mixing on the ignition process can be introduced by a constant addition of gas into the air/gas-mixture or of air into the gas/air-mixture. Another way of introducing this effect can be accomplished by stochastic fluctuations about an averaged concentration. Thus a simple mixing model is combined with a complex description of the reaction mechanisms. With the consideration of the mixing time it is possible to study the influence of different mixing lengths on the resulting temperature and species concentrations at the end of the tuyere. Computational Fluid Dynamic modelling (CFD) allows a more detailed description of the predominant flow of gas injection. Because of the high resource demanding flow calculation, simplified models are used for the simulation of the reaction. The simplest assumption is the thermodynamic equilibrium, which is based on the local composition.

MODELLING The case in this study represents a tuyere with two lances for the gas inflow (Figure 1). The lances terminate 10 cm from the end of the tuyere. The different compositions of the two gases and their net calorific values as well as the predominant excess air ratios of the combustion process are listed in Table 2. The specified boundary conditions for one tuyere can be seen in Table 3. Figure 1: Tuyere with two lances for gas injection

IFRF Combustion Journal Article No 200502

-5 -

Andahazy et. al. June 2005

Table 2: Gas composition in mole-% and net calorific value of COG and COG/BOF Coke oven gas

COG/BOF

H2

62.90

45.21

H2O

0.60

0.91

CH4

22.70

16.38

C2H6

2.30

1.70

CO

6.30

20.03

CO2

1.30

5.06

N2

3.90

10.62

O2

-

0.08

42723

21679

Qn [J/mole]

45.1

37.7

λ [-]

1.69

2.09

Qn [kJ/kg]

Table 3: Boundary conditions of the calculated cases for one tuyere Fuel gas

Air (O2 enriched)

Temperature

30

1120

Pressure

3

2.5

Volume flow

1029

5765

Median diameter

0.022

0.141

Average velocity

~160

~198

The thermodynamic equilibrium was calculated using “Equil” as a part of the software package Chemkin v3.6 [3]. As input parameters, the important chemical species and their concentrations have to be defined. For the calculation of the equilibrium, only the enthalpy of these species is used. The PFR approach was used for three cases with different boundary conditions: isothermal, adiabatic and with a constant heat loss through the walls. The adiabatic and isothermal condition determines the upper and lower limit of the possible heat loss. The constant heat loss is set to approximately 130 kW per tuyere. Again, the software package Chemkin v3.6 is used for calculations with the program “Plug” as a part of this collection. For the PFR model with finite mixing, a continuous addition of air into the gas is done over a defined length (mixing length) in constant quantities (Figure 2). The addition of the air into the gas (and not the contrary) is considered because combustion occurs with an excess of fuel. These calculations are also done with the program “Plug” from the Chemkin v3.6 software package [3]. An adiabatic thermal boundary condition approach is used.

IFRF Combustion Journal Article No 200502

-6 -

m & air n

m & air m & gas

Andahazy et. al. June 2005

PFR1 PFR2

PFRi

PFRn-1 PFRn

m & total

mixing length L Figure 2: PFR model with finite mixing

The CFD calculations have been carried out using the commercial code FLUENT v6.0 [4] on the central computer (sc.zserv.tuwien.ac.at) of the ZID (Zentraler Informatikdienst) of the Vienna University of Technology. For these calculations equilibrium chemistry is chosen which uses the probability density function (PDF) models. First the chemical processes are calculated with the help of a pre-processor (prePDF) before the flow is simulated. The chemical equilibrium model expects that fuel and oxidizer are only separated through a negligible thin flame front. For the calculation of chemical equilibrium the enthalpy is used, therefore no defined reaction mechanism or detailed chemical kinetic rate data are required. It is sufficient to define the important chemical species. This model assumes that the reaction is mixing-limited and has a rigorous accounting of turbulence-chemistry interactions. For turbulence the ‘realizable k-ε model’ is chosen. The boundary conditions for the walls assume adiabatic and no-slip conditions. For the boundary of the raceway a porous layer is chosen to simulate the packed bed of coke of the blast furnace.

RESULTS AND DISCUSSION The thermodynamic equilibrium approach results in information on the temperatures reached and species concentrations (Table 4). After combustion - as expected - a high amount of H2O (mainly in COG) and CO2 (mainly in COG/BOF) is produced. Due to the excess of O2 in the enriched air a considerable quantity of O2 remains in the gases.

IFRF Combustion Journal Article No 200502

-7 -

Andahazy et. al. June 2005

Table 4: Composition [mole-%] and temperatures [°C] after combustion using the thermodynamical equilibrium Coke oven gas

COG/BOF

H2

0.32

0.10

H2O

17.13

12.75

CH4

0

0

C2H6

0

0

CO

0.58

0.31

CO2

4.93

6.80

N2

64.88

66.03

O2

8.77

11.55

O

0.40

0.21

H

0.13

0.03

OH

1.53

0.87

Temperature

2301

2145

The PFR approach gives information about the ignition process after mixing and the influence of different mixing lengths in terms of the temperatures and species concentration at the end of the tuyere. The resulting concentrations of the PFR without considering the mixing time calculations can be seen for both evaluated gases for an adiabatic process below in Figures 3 and 4 as a function of reactor length. 0.25

0.25 O2

O2

0.2

H2O

Volume Fraction [-]

Volume Fraction [-]

0.2

0.15 H2

0.1 CO2

CH4

0.05

H2O

0.15

0.1

H2

CO2

CH4

0.05

CO

CO

0

0 0

1

2

3

4

5

6

7

8

Length [cm] Figure 3: Species concentration profiles of COG in an adiabatic PFR

9

10

0

1

2

3

4

5

6

7

8

Length [cm] Figure 4: Species concentration profiles of COG/BOF in an adiabatic PFR

It can be seen that the COG/BOF gas mixture shows generally a similar behavior but ignites somewhat earlier than the pure COG comparing the points of inflection of the O2 curves.

9

10

IFRF Combustion Journal Article No 200502

-8 -

Andahazy et. al. June 2005

Under the same conditions the ignition propensity depends on gas composition; especially the three main species H2, CH4 and CO. H2 reacts fast because of its high diffusion coefficient and its chain-branching reaction chemistry. The other two main components CH4 and CO burn more slowly. Due to the higher H2 mole fraction in COG an earlier ignition of COG is expected. However, the COG/BOF gas mixture consists of a somewhat lower CH4 fraction and a relatively high fraction of CO compared to the COG and ignites somewhat earlier than the COG. Comparing the concentration curves with a PFR, assuming a constant heat loss, to those of the adiabatic process no difference can be obtained with respect to the ignition process. Only the temperature profiles (Figure 5) of the constant heat loss and adiabatic case show a difference, which however is less than 20°C. The approximate 160°C higher end temperature at the COG combustion in both cases is a result of the higher net calorific value of the COG gas. The flattening of the curve indicates that the complete combustion process requires a finite time to finish. 2600 2400

COG

Temperature [°C]

2200 2000 COG/BOF

1800 1600 1400 1200

adiabatic PFR PFR with constant heat loss

1000 800 0

1

2

3

4

5

6

7

8

9

10

Length [cm] Figure 5: Temperature profiles of an adiabatic and a constant heat loss PFR of COG and COG/BOF

The start temperature of the calculations for the PFR is averaged according to the inflow volumes of fuel gas and air and is equal to 943°C. The end temperatures do not really differ

IFRF Combustion Journal Article No 200502

-9 -

Andahazy et. al. June 2005

from those obtained using the thermodynamical equilibrium (Table 5). The model assuming thermodynamic equilibrium gives realistic temperature predictions. The product species composition is approximately in the same range which allows the assumption that the system is near thermodynamic equilibrium after 10 cm for the PFR.

Table 5: Temperatures of thermodynamical equilibrium and PFR model and ignition lengths COG

COG/BOF

T Thermodyn. Equilibrium

[°C]

2301

2145

T PFR adiabatic

[°C]

2270

2113

T PFR constant heat loss

[°C]

2256

2096

Ignition lengths

[cm]

2.4

2.1

The next model considered is a PFR with a finite mixing. A mixing length of 100 % amounts to 10 cm which is the distance between the end of the lances and the end of the tuyere (compare with Figure 2). All temperatures and species concentrations given are those at the end of the tuyere. The curve is not representing a profile over the length of the PFR. The longer the mixing length the less amount of oxygen is added at each step to the gas. Figures 6 and 7 show the influence of different mixing lengths on the temperature and the species concentrations reached at the end of the tuyere.

2000

0.2

1500 H2O

Temp. H2

0.1

1000 CO2 CH4

0.05

500

2500

2000 O2

0.15

1500 Temp.

0.1

1000

CO2

H2O H2

0.05

CH4

CO

500

CO

0 0

100

200

300

0 400

0 0

100

200

300

0 400

Mixing Length [%]

Mixing Length [%]

Figure 6: Influence of the mixing length on species concentrations and temperature at the end of the tuyere with COG

Figure 7: Influence of the mixing length on species concentrations and temperature at the end of the tuyere with COG/BOF

Temperature [°C]

Volume Fraction [-]

0.15

0.25

Volume Fraction [-]

O2

0.2

2500

Temperature [°C]

0.25

IFRF Combustion Journal Article No 200502

-10 -

Andahazy et. al. June 2005

For the case with a mixing length of 75 % all the oxygen is added within 7.5 cm. The ignition lengths of the COG and the COG/BOF gases, determined with the PFR model without mixing (see earlier section), are shorter than 2.5 cm. Consequently, below a mixing length of 75 %, the two gases had enough time to complete mixing and to ignite before the end of the tuyere. At a mixing length of 200% only half of the amount of oxygen is mixed with the gas at the end of the tuyere. It can be seen that at this mixing length the quantity of air is not high enough for ignition to proceed neither for the COG nor for the COG/BOF mixture. Changes to the values at higher mixing lengths at the end of the tuyere are the result of the mixing of gas and air. Since no thermal strain can be measured at the end of the tuyere in these cases the calculation interval is chosen to be large. Nevertheless it can be seen that the temperature curve decreases with increasing mixing lengths which means that the combustion process is not finished until the end of the tuyere. The COG shows a steeper decrease than the COG/BOF. The difference in the obtained temperatures between the mixing lengths of 75 % and 100 % amounts approximately 1000°C while for the COG/BOF combustion only a quarter of this value is obtained. This indicates that the COG/BOF gas mixture ignites somewhat earlier than the COG gas (compare with the earlier section). The greatest difference can be seen by comparing the results at a mixing length of 100 %. The COG gas ignites just a small distance before the end of the tuyere. This is indicated by the calculated temperature as well as the oxygen concentration. The COG/BOF gas mixture is nearly fully burned at this point (Figure 6 and 7). In the case with a mixing length of 100% the amount of oxygen the fuel gas needs to burn completely to end products is already available after 5 cm for the combustion of COG/BOF gas. For the combustion of COG gas this amount is not reached until 6 cm. This means that when changing the injected gases without adjusting other predominant parameters, the COG/BOF gas mixture will ignite earlier than the COG.

IFRF Combustion Journal Article No 200502

-11 -

Andahazy et. al. June 2005

A matter of particular interest is the local distribution of species concentrations and temperatures in the tuyere, in the raceway and at the exit of the raceway. Temperature is the critical factor for the thermal strain of the tuyere. The local species and temperature distribution can only be determined by using CFD. Figure 8 and 9 show the temperature distributions of the COG and COG/BOF combustion cases.

Figure 8: Temperature distribution in the raceway for the combustion of COG

Figure 9: Temperature distribution in the raceway for the combustion of COG/BOF

IFRF Combustion Journal Article No 200502

-12 -

Andahazy et. al. June 2005

The gas temperatures on the tuyere wall at the end of the tuyere have reached about 1900 °C in both cases whereas the temperature for the COG combustion is somewhat higher. For this combustion case a larger surface of the tuyere is affected by higher temperatures, which causes a higher thermal strain. It can be seen that for the COG combustion the highest temperature is reached at the end of the raceway while for the COG/BOF combustion a maximum is reached at 0.75 m after the end of the tuyere. All values resulting from the CFD simulation are calculated over the volume flow and the accompanying characteristic of each cell of the raceway. The averaged temperatures and gas velocities at the exit of the raceway (listed in Table 6) also have an effect on the subsequent blast furnace chemistry. The higher gas velocity at the COG combustion is solely a consequence of the higher mean temperature.

Table 6: Calculated averaged temperatures and gas velocities at the exit of the raceway COG

COG/BOF

Temperature [°C]

[°C]

2258

2097

Gas velocity [m/s]

[m/s]

2.76

2.61

Table 7 shows the composition of the gases leaving the raceway. It can be seen that no hydrocarbon species exist after the combustion of both gases. The combustion of the COG gas results in higher H2O, H2 and CO concentrations whereas the O2 and CO2 concentrations are higher after the combustion of the COG/BOF gas. These species are of high importance in the reduction process of iron ore and blast furnace chemistry. The tuyere wall temperature is calculated as well. The highest gas temperature at the tuyere wall is reached using the equilibrium model (see Table 5). Calculations with a temperature of about 2300 °C show that a maximum wall temperature of about 1050 °C is reached at the cladding of the tuyere. In the liquid cooled part of the tuyere the wall temperature has a maximum of about 400 °C. Therefore it can be expected that there are no problems due to the thermal strain under these operating conditions.

IFRF Combustion Journal Article No 200502

-13 -

Andahazy et. al. June 2005

Table 7: Resulting averaged species concentrations leaving the raceway COG

COG/BOF

H2

1.02

0.30

H2O

17.03

12.40

CH4

0

0

C2H6

0

0

1.44

0.87

CO CO2

4.18

6.22

N2

63.85

66.34

O2

9.98

12.43

O

0.43

0.27

H

0.37

0.11

OH

1.70

1.05

CONCLUSIONS The COG/BOF gas mixture ignites somewhat earlier than the COG gas because of its different composition and combustion conditions. The combustion of COG leads to a higher maximum and mean temperature due to its higher net calorific value so that the thermal strain on the tuyere is higher than for the COG/BOF gas utilization. Under operating conditions with appropriate cooling no problems are expected. The combustion of COG gas results in higher H2O, H2 and CO and lower O2 and CO2 concentrations. An effect on blast furnace chemistry is expected depending on which fuel is used.

ACKNOWLEDGEMEN T The R&D work of the CD-Laboratory for Chemical Engineering at High Temperatures at the Vienna University of Technology has been funded by the Christian Doppler Forschungsgesellschaft, Austria. The authors also acknowledge Dr. Michael Harasek, Institute of Chemical Engineering, Vienna University of Technology, for supporting the CFD calculations.

REFERENCES [1]

Bürgler, T.H., Brunnbauer, G., Ferstl, A. (2003): “Commissioning and First Operational Results of Blast Furnace Gas Injection at voestalpine Stahl GmbH”.

IFRF Combustion Journal Article No 200502

-14 -

Andahazy et. al. June 2005

Proceedings: METEC Congress 03, 3rd International Conference on Science and Technology of Ironmaking, 2003, Steel Institute VDEh, Düsseldorf, pp. 157-159. [2]

Bowman, C.T., Hanson, R.K., Davidson, D.F., Gardiner, W.C., Lissianski, V., Smith, G.P., Golden, D.M., Frenklach, M. and Goldenberg, M. (1996): http://www.me.berkeley.edu/gri_mech/

[3]

Kee,R.J., Rupley, F.M., Miller, J.A., Coltrin, M.E., Grcar, J.F., Meeks, E., Moffat, H.K., Lutz, A.E., Dixon-Lewis, G., Smooke, M.D., Warnatz, J., Evans, G.H., Larson, R.S., Mitchell, R.E., Petzold, L.R., Reynolds, W.C., Caracotsios, M., Stewart. W.E., Glarborg, P., Wang, C. and Adigun, O. (2000): “Chemkin Collection, Release 3.6”. Reaction Design, Inc., San Diego, CA

[4]

FLUENT Inc. (2001): “FLUENT 6.0 User’s Guide”.