Influence of Shape of Cohesive Zone on Gas Flow and Permeability in the Blast Furnace Analyzed by DEM-CFD Model

Abstract

Maintaining gas permeability is an important issue to realize low coke rate operation of blast furnace. In present study, the melting behavior of the iron ore and the layer structure in low coke rate operation were introduced in to the DEM-CFD model, and then behaviors of gas and moving bed in the blast furnace were simulated. Influence of shape of cohesive zone in low coke rate operation was investigated, and the effect of coke slit in the cohesive zone on gas flow was demonstrated in the calculation result. Surface area of cohesive zone and intersection angle between the burden layers and cohesive zone will determine the activity of coke slit.

1. Introduction

The significance of cohesive zone in which softening and melting of iron ore proceed was recognized in the dissection studies of blast furnace in 1970s.¹⁾ The cohesive zone usually locates in the middle part of the furnace, from lower part of the shaft down to the tuyeres. It was reported that the cohesive zone is formed in different shapes namely; asymmetric, inverted V-type or W-type,²⁾ depending on the operating condition of the blast furnace. The iron ore layer in the cohesive zone is composed with solid and liquid phases and the liquid phase is soak into the void space among the solid phase. The void fraction in the ore layer decreases and then the gas permeability is deteriorated. The gas permeability of cohesive zone is extremely lower than that of lumpy or dripping zone.

In conventional operation of blast furnace, layer structure of burden materials is formed to control the gas flow by charging the ore and coke alternately. The coke layer acts as the ventilation slits in the cohesive zone. Gas permeability of packed bed in the blast furnace determines the limits of the blast volume, and directly limits the productivity. Therefore, it is necessary to optimize the cohesive zone and the structure of the moving bed of burden in the furnace for decreasing gas flow resistance.

To mitigate CO₂ emissions, blast furnace operation with low reducing agent rate has been actively investigated.^3,4,5,6,7) To establish CO₂ reduction technology, the COURSE 50 project is considering using hydrogen in a blast furnace as a reducing agent.³⁾ Research on decreasing the reducing agent rate by enhancing the reaction rate using carbon iron ore composite^4,5,6) or ferrocoke³⁾ had been carried out. Substituting hydrogen gas as the reducing agent and improving the gas utilization ratio might decrease the reducing agent rate and thus allow operation with a low coke ratio of about 250 kg/thm compared to the 350 kg/thm used in the conventional operation. When the coke rate is reduced, volume ratio of coke slits in cohesive zone become relatively low; as a result, the permeability decreases. Control of gas permeability is important in low coke rate operation.⁸⁾

With the DEM-CFD model,^9,10) the gas flow within blast furnace can be calculated reflecting the distribution of the void fraction in the furnace. The mathematical model would be a preferable method for estimating gas flow in blast furnace. In the previous paper,¹⁰⁾ a method to express the softening and melting behavior of solid particles by changing the Young’s modulus of the spherical particles used in the DEM was reported. This DEM-CFD model would help to predict the in furnace phenomena under the low coke rate operation. In this paper, effects of cohesive zone shape on the gas permeability and gas flow behavior are evaluated by the DEM-CFD analysis.

2. DEM-CFD Model

Since the details of the mathematical model was described in the previous paper,¹⁰⁾ the outline of the model is explained. The motion of the solid and the gas phase were calculated by DEM and CFD, respectively. The interaction between solid and gas flow was evaluated as the momentum exchange.

2.1. Motion of Solid Particle¹⁰⁾

The motion of the packed particles in moving bed in blast furnace was calculated by the discrete element method (DEM) which tracks motions of all particles forming the bed. In DEM translational and rotational motions of each particle is described by the Newton’s second law of motion. When the particle contacts with the other particles and/or a wall, the contact forces from these objects acting on the particle at the contact points are taken into account in the equations of motion. The contact force between two contacting particles is represented by the Voigt model¹¹⁾ as shown in Fig. 1(a). The Voigt model approximates the contact force with using springs and dashpots. In order to reduce the calculation load, the DEM simulation uses enlarged particles as shown in Fig. 1(b). In order to reproduce the moving bed motion in actual blast furnace, the spring constant and the dashpot viscosity coefficient to calculate the contact forces F_c and F_cs were determined by the same method reported in the previous work.¹⁰⁾ Equations of motion of the particle in translational and rotational directions are shown as follows.   

$$m_p\frac{d{v}_p}{dt}=\sum _{j\ne i}^{N_c}{F}_c+F_f+F_g$$(1)

  

$$I_p\frac{d{\omega }_p}{dt}=r_p\sum _{j\ne i}^{N_c}(F_{cs}-F_r)$$(2)

where m_p, v_p, t, F_c, N_c, F_f, F_g, I_p, ω_p, r_p, F_cs and F_r denote mass of particle, particle velocity, time, inter particle contact force, number of contacting particles, force on particle from fluid, gravity on particle, moment of inertia, angular velocity, radius of particle, contact force in shear direction and rotational resistance force, respectively.

Fig. 1.

Discrete element model for the numerical simulation including gas flow.

2.2. Motion of Fluid

The fluid motion is estimated by the DEM-CFD model.⁹⁾ The fundamental equations describing the fluid flow are the continuity equation and the Navier-Stokes equations. The Ergun’s equation¹²⁾ or the Wen-Yu’s equation¹³⁾ was used to evaluate the flow resistance in the packed bed, corresponding to the void fraction. The flow resistance calculated by these equations is translated to the interaction force F_f in Eq. (2).   

$$\frac{\partial }{\partial t}=(\epsilon {\rho }_g)+\nabla \cdot (\epsilon {\rho }_g{u}_g)=0$$(3)

  

$$\frac{\partial }{\partial t}(\epsilon {\rho }_g)+\nabla \cdot (\epsilon {\rho }_g{u}_g{u}_g)=-\epsilon \nabla p_g+\epsilon {\mu }_g{\nabla }^2{u}_g+\epsilon {\rho }_{g}g+F_p$$(4)

where, ε, ρ_g, u_g, p_g, μ_g and F_p denote void fraction, density of fluid, velocity of fluid, pressure of fluid, viscosity and force on fluid from particle respectively. The evaluation of the interaction between gas and particle is depicted in shown in Fig. 1(b).

2.3. Interaction Force between Particle and Fluid

Flowchart of DEM-CFD calculation is shown in Fig. 2. Solid and gas flows are separately calculated by DEM and CFD, respectively. Due to the difference in the Courant number, the CFD calculation is made every several time steps of the DEM calculation. These sub-models exchange information through interaction force. Additionally the voidage distribution is transferred from the DEM to the CFD, and the gas velocity distribution is transferred from the CFD to DEM. The interaction force taken into account here is gas drag force, and is expressed as a linear function of velocity difference.   

$$F_f={\beta }_t(v_p-u_g)$$(5)

where, β_t is the momentum exchange coefficient which takes into account voidage, particle diameter, and so on. For bed voidage less than 0.8, β_t is derived by the Ergun’s equation,¹²⁾ otherwise by the Wen-Yu’s correlation.¹³⁾ In this study the particle diameters larger than the actual ones are used in the DEM calculation to reduce the calculation load. In the calculation of interaction force, it is assumed that the packed bed consists of actual size particles, and has the same void fraction to the DEM particle bed.

Fig. 2.

Flow chart of calculation including interaction between particles and fluids.

2.4. Physical Property of DEM Particle

Calculation conditions of DEM and CFD model are listed in Table 1. Although the Young’s modulus are specific physical properties of materials, they are adjusted to reproduce the overall flow characteristics of the burden materials in blast furnace. Particle size was enlarged for computational load reduction in DEM, however the interaction between the moving bed and the gas phase was derived depending on the actual particle size in CFD. Furthermore the softening behavior of the ore particles in the cohesive zone was represented by changing hardness of the particle. With these treatment, the Young’s modulus of the coke particle was set at 1.0 GPa¹⁰⁾ and one for iron ore particle was set at 1.0 GPa in stack region and was reduced to 0.02 GPa in the cohesive zone.¹⁰⁾ Particle number density is increased by the proximity of the particles and thus void fraction is changed. The rotation resistance¹⁴⁾ was employed to represent irregular shape of the burden particle. The friction coefficient and the rotational resistance coefficient were determined through the same manner reported in the previous report.¹⁵⁾ Although the size of the burden materials in actual blast furnace gradually changes with progress of reaction and degradation, the variation in the particle diameter was ignored in this study.

Table 1. Calculation conditions used in DEM-CFD.

Maximum Particle number in DEM		600000	[−]
Diameter of particle (coke, ore)	DEM	0.300, 0.150	[m]
	CFD	0.050, 0.020
Poisson’s of particle (coke, ore)		0.21, 0.24	[−]
Apparent density (coke, ore)		1.1, 4.0	[kg/m3]
Young’s modulus (coke)		1.00	[GPa]
Young’s modulus (ore)              above cohesive zone		1.00
in cohesive zone		0.02
Rotational resistance coefficient		0.1	[−]
Friction coefficient		0.3	[−]
Number of grids in CFD		790372 (121×46×142)	[−]
Inlet area of tuyere		0.0560	[m2]
Inlet velocity		87.2	[m/s]
Number of tuyere		20 (in semicircle)	[−]
Time step (DEM, CFD)		1.0×10−4, 7.2	[s]

2.5. Calculation Conditions

Shape of the blast furnace simulated in the present study, positions of raceway and cohesive zone are shown in Fig. 3. The computational domain is a half-cut model of the blast furnace with 5775 m³ of inner volume, and the cut plane was represented by a wall with no friction. The coke and the iron ore particles were charged from top of the furnace alternately to form layered packing structure. The numbers of charged coke and ore particles were adjusted to form 1-m-thick coke layer at the top of the burden and to set the coke rate at 240 kg/thm. The cohesive zone shapes tested in this study are shown in Fig. 4. In the figure, colored region of a), b) and c) denote inverted V-type, inverted V-type with low height, and W-type shaped cohesive zones, respectively. b) and c) represent softening behavior of the cohesive zone when the thermal reserve zone temperature is decreased and the gas flow ratio near the wall was increased, respectively. The ore particles were softened in these zones as described above, and the ore particles reached to the lower boundary of the cohesive zone are eliminated to represent the melting. The shape of the cohesive zone was examined for the same thickness of 4.0 m in vertical direction. Twenty tuyeres were set in the half-cut model used in the simulation. Spherical raceway with 1.6 m in diameter was put in front of each tuyere. The coke particles were successively eliminated at a constant interval as it reached to the raceway zone. The descending velocity of the moving burden bed was controlled by this elimination rate and the motion of the moving bed in the DEM simulation was artificially accelerated by a factor of 120 times in comparison with an actual furnace. It was confirmed that the influence of such acceleration was small.¹⁶⁾ For inflow boundary conditions of CFD calculation, the gas was blown horizontally from the tuyeres toward the center of furnace at constant flow rate, and the gas flow rates were the same for all tuyeres.

Fig. 3.

Schematic diagram of three-dimensional calculation region of blast furnace with inner volume of over 5000 m³.

Fig. 4.

Shape of cohesive zone in calculation condition.

The blast furnace was initially packed with the coke and the ore particles, then the gas blowing and the eliminations of the coke and the ore particles in the raceway and the cohesive zones were started. The burden materials were fed from the top of the furnace according to the burden descent as mentioned above. After the motions of the gas and the burden materials reached the stable state, the flow behaviors of the gas and the burden materials were evaluated.

3. Gas Flow Variation at the Cohesive Zone with Low Coke Rate Operation

The packed bed structures under the cases with coke rate of 240 kg/thm are shown in Fig. 5. Where, the cohesive zone shapes a), b) and c) are inverted V-type, half-height inverted V-type and W-type, respectively. In case b), lower cohesive zone case, the inclination angle of layer structure is slightly larger than that of case a). The ore layer shrank in cohesive zone, and then particle flows toward the center slightly, therefore angle of layer border of b) is slightly larger than that of a). In the W-type cohesive zone condition c), little change in the layer structure due to decrease in the coke rate in the lumpy zone was observed. The mixing of the coke and the ore particles, however, is observed in the central and the bottom parts of the cohesive zone.

Fig. 5.

Burden layer in blast furnace for each shape of cohesive zone. (a) Base, b) 1/2 height, c) W shape).

The distributions of the void fraction in the blast furnace with various shapes of cohesive zone are shown in Fig. 6. In the cohesive zone, void fraction is decreased due to softening of the particles. It is observed that green parts and fairly blue parts pile up alternately in the figure. This means that the coke slits are formed in the cohesive zone. In cases b) and c), the height and the shape of the cohesive zone affect on the void fraction in the cohesive zone, namely, the void fraction in the central part becomes higher and one in the peripheral part becomes lower. Moreover, in case c), the void fraction in the lower part of the cohesive zone is high, in which the mixing of the coke and the ore have been observed in Fig. 5. Also in the low coke operation case, it is considered that the mixed coke particle increases the void fraction of the ore layer.

Fig. 6.

Comparison of void fraction by shape of cohesive zone. (a) Base, b) 1/2 height, c) W shape).

The gas velocity vector distributions in the low coke rate operation are shown in Fig. 7. In case b), since the void fraction in central part is relatively high, the gas flow concentrates to this part. Furthermore, the gas flows upward since the coke slit becomes unclear due to the mixing of the burden materials. In the mid-radial part of the cohesive zone, the gas flows horizontally outward by the effect of the coke slits. The gas velocity and flow rate are low near the wall due to low void fraction.

Fig. 7.

Comparison of gas velocity vectors by shape of cohesive zone. (a) Base, b) 1/2 height, c) W shape).

The isobar planes for various shaped cohesive zones are shown in Fig. 8. The pressure distribution is formed depending on the gas flow and the void fraction. The pressure drop in case b) is larger than that of case a). In the low coke rate operation condition, the pressure drop increases with decrease in the coke slits. Especially, as shown in Fig. 8(b), the pressure drop increases significantly in the flat cohesive zone case. In case c), W-shaped isobar plane was seen. In the low coke rate operation condition, the pressure difference between the bottom of the cohesive zone and the furnace top increases to 240 kPa (Fig. 8(a)) from 150 kPa in the conventional condition.¹⁰⁾ Pressure differences of a) and c) in Fig. 8 are 240 kPa and 200 kPa, respectively, the difference from the coke rate is relatively small.

Fig. 8.

Comparison of isobar planes by shape of cohesive zone. (a) Base, b) 1/2 height, c) W shape).

In present study, influence of shape of the cohesive zone on the gas flow was demonstrated. In low coke rate operation, the relative thickness of the coke layer to the ore layer becomes thin, and the active volume of the coke slits in the cohesive zone decreases. And the function of coke slit for gas flow become small. In order to increase the permeability of the cohesive zone, the decrease in the cohesive zone thickness is desired,¹⁰⁾ and the mixing charge also is effective for decreasing pressure drop. On the other hand, as shown in present study, the control of the cohesive zone shape through the burden charging distribution would also be an effective way in enhancement of the gas flow in the blast furnace. When the thickness in the height direction of the softening part are the same, the intersection angle of the layer structure and the gas flow vector determines the length of the coke slit. Larger area of surface of cohesive zone and shorter coke slit increase gas permeability. It is considered that the permeability of the cohesive zone can be improved with increasing the slope angle of the cohesive zone that increases the flow area for gas across the cohesive zone.

4. Conclusions

Toward low reducing agent rate operation of the blast furnace, improvement of the operation has been required in the steel industry. It is necessary to understand the in furnace phenomena to achieve a low carbon operation, and then the development of a simulation model of high accuracy is desired. The melting behavior of the iron ore and the layer structure in low coke rate operation were introduced to the DEM-CFD model, and then behavior of gas and moving bed in the blast furnace was simulated. Influence of shape of cohesive zone in low coke rate operation was investigated and following conclusions are obtained.

(1) The length and number of active coke slit are changed with variation of the shape of the cohesive zone. The pressure drop at the cohesive zone and the gas velocity in the lower part of the furnace are increased in case of the height cohesive zone is lowered, and decreased with W shaped cohesive zone.

(2) In a region where coke and ore are mixed, the void fraction of the cohesive zone is kept relatively high. Mixed charging of coke into the iron ore layer would be effective to increase the gas permeability.

References

• 1)   S.  Kondo: 33th and 34th Nishiyama Memorial Seminar, ISIJ, Tokyo, (1975), 125.
• 2)   M.  Sasaki,  K.  Ono,  A.  Suzuki,  Y.  Okuno and  K.  Yoshizawa: Tetsu-to-Hagané, 62 (1976), 559.
• 3)   M.  Naito,  S.  Nomura and  K.  Kato: Tetsu-to-Hagané, 96 (2010), 201.
• 4)   S.  Nomura,  H.  Kitaguchi,  K.  Yamaguchi and  M.  Naito: ISIJ Int., 47 (2007), 245.
• 5)   A.  Kasai,  M.  Naito,  Y.  Matsui and  Y.  Yamagata: Tetsu-to-Hagané, 89 (2003),1212.
• 6)   S.  Kondo and  S.  Aoyama: Tetsu-to-Hagané, 59 (1973), S23.
• 7)   S.  Natsui,  S.  Ueda,  H.  Nogami,  J.  Kano,  R.  Inoue and  T.  Ariyama: ISIJ Int., 51 (2011), 51.
• 8)   S.  Ueda,  T.  Miki,  T.  Murakami,  H.  Nogami and  T.  Sato: Tetsu-to-Hagané, 99 (2013), 1.
• 9)   S.  Natsui,  H.  Nogami,  S.  Ueda,  J.  Kano,  R.  Inoue and  T.  Ariyama: ISIJ Int., 51 (2011), 41.
• 10)   H.  Kurosawa,  S.  Matsuhashi,  S.  Natsui,  T.  Kon,  S.  Ueda,  R.  Inoue and  T.  Ariyama: ISIJ Int., 52 (2012), 1010.
• 11)   P. A.  Cundall and  O. D. L.  Strack: Geotechnique, 29 (1979), 47.
• 12)   S.  Ergun: Chem. Eng. Prog., 48 (1952), 89.
• 13)   C. Y.  Wen and  Y. H.  Yu: Chem. Eng. Prog. Symp. Ser., 62 (1966), 100.
• 14)   Y.  Tsuji,  T.  Kawaguchi and  T.  Tanaka: Powder Technol., 77 (1993), 79.
• 15)   S.  Natsui,  S.  Ueda,  M.  Oikawa,  J.  Kano,  R.  Inoue and  T.  Ariyama: ISIJ Int., 49 (2009), 1308.
• 16)   S.  Natsui,  S.  Ueda,  Z.  Fan,  J.  Kano and  T.  Ariyama: Tetsu-to-Hagané, 96 (2010), 1.
• 17)   E. A.  Mousa,  A.  Babich and  D.  Senk: ISIJ Int., 51 (2011), 350.