Heat Recovery Process from Packed Bed of Hot Slag Plates

Abstract

Reduction of CO₂ emissions from the steelmaking process is strongly required for prevention of global warming. One promising heat resource that is estimated to have great potential for energy saving is the waste heat of steelmaking slag, which has a temperature of above 1673 K in the molten state. This molten slag can be solidified in a plate-like shape by feeding it on the surface of water-cooled rolls, and the heat of the plate-like slag can be recovered easily in spite of its low heat conductivity. When these hot slag plates are packed in a slag chamber, the heat of the slag can be recovered by heat exchange with a counter current gas flow. Because the efficiency of gas-slag heat transfer changes depending on the shape of the packed slag, it is difficult to estimate the efficiency of slag heat recovery without evaluating the accuracy of the heat transfer coefficient in the bed. In this work, the effect of the slag shape on the accuracy of the heat transfer equation was evaluated by conducting both laboratory-scale and pilot-scale slag heat recovery experiments and performing a fitting analysis by using a slag packed bed heat transfer simulation model. A comparison of the experimental results and calculation results confirmed that the heat transfer coefficient can be estimated by using Johnson-Rubesin’s equation modified by a correction factor in case the packed materials are plate-like. The effect of the correction factor on the efficiency of slag heat recovery at the industrial scale was also estimated.

1. Introduction

CO₂ emissions from the steel industry account for 15% of total CO₂ emissions in Japan and have a large impact on the global warming problem. Under these circumstances, Japanese steel companies are participating in a national project called COURSE50 with the aim of developing technologies to reduce CO₂ emissions from steel works.^1,2) One of the principal development subjects in the COURSE50 project is waste heat recovery to provide thermal energy for CO₂ separation. One promising heat resource that can be recovered is the waste heat of steelmaking slag, which has both a higher temperature and a higher calorific value than other types of waste heat and is unused in the current process. Although many slag heat recovery processes, including air atomizing^3,4,5,6) and mechanical atomizing,^7,8,9,10) were tried previously,^11,12) most were not successful at the industrial scale due to the instability of the physical properties of slag, such as viscosity, which changes drastically depending on both the temperature and the chemical composition of the slag. Furthermore, the heat conductivity of slag is also relatively low compared with that of other materials like coke, from which heat is recovered by coke dry quenching (CDQ), and this reduces the efficiency of heat recovery from a packed bed of slag particles. As recent progress in slag heat recovery, a twin-roll type continuous slag solidification process was introduced.¹³⁾ This process can improve heat recovery efficiency, as the thickness of the slag which is solidified on the roll is small enough to reduce the slag surface temperature drop caused by the low heat conductivity of the slag. However, the heat transfer behavior in a packed bed of plate-like slag was not clear due to the shape difference compared with traditional packed bed materials like coke in CDQ and sintered ore in a blast furnace. Packed bed heat recovery is widely used in variety of industrial processes, and considerable amount of study has been carried out by using experimental methods and simulations. As the results, it was clarified that the gas-solid heat transfer in the packed bed changes considerably due to the effects of the shape of packed materials.^{14,15,16,17,18)} In this work, we evaluated the accuracy of the slag heat transfer equations in the slag packed bed and estimated the effect of the shape of the packed materials on the heat transfer coefficient by comparing gas temperature histories in both experiments and simulations. Subsequently, we conducted a pilot-scale slag heat recovery test by using slag from an actual steel mill to verify the feasibility and efficiency of this process.

2. Slag Heat Recovery Experiments

2.1. Laboratory-scale Slag Heat Recovery Experiment

As an initial study, we conducted laboratory-scale slag heat recovery experiments.¹⁹⁾ Figure 1 shows the experimental method used in the preparation and reheating of the sample slag plates. Figure 2 shows the experimental method for slag heat recovery. Sample slag particles in a graphite crucible were melted by a high-frequency induction furnace and solidified on a steel plate. The thickness of the slag was controlled to 7 mm by a steel roller. The solidified slag plate was crushed and sorted by plane size, and 20 kg slag plates of the size within 30 to 50 mm were prepared. These slag plates were reheated with a steel container in an electric furnace to T₀ [K], and were then charged into a slag chamber having an inner diameter of Φ300 mm. After charging the slag plates, air was blown from holes in a perforated plate at the bottom of the heat recovery chamber at the flow rate V [L/min]. To prevent temperature drop in the slag, the reheated slag plates were charged within 1 min after removal from the furnace, which was located close to the chamber. The experimental conditions are listed in Table 1. These experiments were conducted under the conditions of different slag reheating temperature T₀ and air flow rate V.

Fig. 1.

Laboratory-scale slag heat recovery experiment (slag casting, sizing, and reheating processes).

Fig. 2.

Laboratory-scale slag heat recovery experiment (slag heat recovery chamber).

Table 1. Experimental conditions.

	Slag weight M [kg]	Slag reheating temperature T0 [K]	Gas flow rate V [L/min]
Case 1	20	823	60
Case 2	20	823	200
Case 3	20	1273	200

2.2. Pilot-scale Slag Heat Recovery Experiment

As a next step to verify the feasibility of this process, pilot-scale slag heat recovery experiments were carried out with a heat recovery chamber having an inner diameter 6.5 times larger than the previous laboratory-scale device.^20,21) Figure 3 is a schematic diagram of this pilot plant. The left side of Fig. 3 shows a twin-roll type slag continuous solidification pilot plant (hereafter, “Twin-roll plant”), which was used to provide hot slag plates for heat recovery. The right side of Fig. 3 is a slag heat recovery pilot plant (hereafter, “Slag heat recovery plant”), into which the slag plates were charged. Figure 4 shows the exterior of the Twin-roll plant. The plant specifications are listed in Table 2. This plant consists of two water-cooled rolls, which are arranged close together and rotate to the outward direction, a ladle tilting machine, which is used to control the flow rate of molten slag onto the surface of the rolls and has the maximum capacity of 2 t/min, and an apron conveyer to transfer the solidified slag plates. The slag used in this experiment was obtained from a chromium ore smelting reduction furnace. The ladle tilting machine can accommodate a ladle that is transferred from the actual steel mill, and the increment of its tilting angle can be controlled precisely by hydraulic cylinders.

Fig. 3.

Process drawing of Slag heat recovery plant.

Fig. 4.

Twin-roll plant.

Table 2. Specifications of pilot plants.

Equipment		Specifications
Twin-roll plant	Cooling roll	Dimensions	Φ1.6 m×W1.5 m
		Number of rolls	2
		Material	Cu
		Rotation speed	Max. 20 rpm
		Cooling water flow rate	125–130 m3/h/roll
	Ladle tilting machine	Tilting speed	Max. 6.5°/min
		Load	Max. 140 t
	Conveyer	Dimensions	W1.3 m×L14.5 m
		Lifting height	5.5 m
		Speed	25 m/min
		Material	SUS304
Slag heat recovery plant	Crusher	Capacity	1.0 t/min
	Bucket elevator	Transport capacity (Slag charging rate)	1.0 t/min
	Heat recovery chamber	Chamber size	L1.5 m×W2.0 m×H2.5 m
		Capacity	Max. 6 t
	Blower	Gas flow rate	Max. 6000 Nm3/h
		Motor	75 kW
		Number of blowers	2
	Cyclone	Size	Φ2.2 m×H7.5 m

Figure 5 shows the exterior of the Slag heat recovery plant. The plant specifications are listed in Table 2. This plant collects hot slag plates at a temperature above 1273 K at the end of the apron conveyer by means of a sliding chute and crushes the plates with a hot crusher for direct transportation. To prevent temperature drop during transportation, the crushed slag plates are transferred directly to the heat recovery chamber by a bucket elevator. After the slag plates are charged into the chamber, air for heat recovery is blown from the chamber bottom by two blowers. Since this experimental plant has no heat utilization equipment, after measurement of the gas temperature profile, the hot gas is cooled by water, dedusted by a cyclone, and released into the atmosphere.

Fig. 5.

Slag heat recovery plant.

The pilot test conditions are listed in Table 3. Two kinds of tests, Case A and Case B, were conducted at different chances. The ladle tilting speed of the Twin-roll plant was controlled to keep the slag feeding rate of 1.0 t/min, and the rotation speed of the cooling rolls was set to 10 rpm to stabilize the speed at which the slag was drawn up on the roll surface. The amount of slag charged into the slag heat recovery chamber was controlled to 1.7 t or 4.8 t by the sliding chute. The maximum gas flow rate was 7200 Nm³/h.

Table 3. Operating conditions of slag heat recovery pilot test.

		Case A	Case B
Twin-roll plant	Ladle tilt machine Tilt speed [mm/s]	0.8	1.0
	Cooling roll Rotation speed [rpm]	10	10
Slag heat recovery plant	Amount of slag charged [t]	1.7	4.8
	Gas flow rate [Nm3/h]	6000	7200

3. Results of Experiments

3.1. Results of Laboratory-scale Slag Heat Recovery Experiments

Figure 6 shows the temperature profiles of the gas obtained as a result of the experiments under the conditions in Table 1. The peak gas temperature of Case 2 is higher than that of Case 1, and the gradient of the gas temperature decrease after the peak is also larger in Case 2. The peak temperature was observed a few minutes later than the initial temperature rise because the gas temperature in the early stage was decreased by the effect of inner wall of the chamber. In spite of the shorter gas residence time in the packed bed under the condition of a higher gas flow rate, the gas temperature of Case 2 was higher than that of Case 1. This is explained by the increase of the heat transfer coefficient, which increases together with the gas flow rate. The highest gas temperature, which was observed in Case 3, is the result of the larger heat transfer due to the temperature difference between the gas and slag.

Fig. 6.

Temperature profile of hot air obtained in slag heat recovery experiments.

3.2. Results of Pilot-scale Heat Recovery Experiments

Figure 7 shows a photograph of slag plates that were crushed by the hot crusher immediately after transportation by the apron conveyer. The crushed slag plates are transferred directly to the bucket elevator located at the left side in Fig. 7. Figure 8 is a photograph taken by a monitoring camera which was installed to observe the slag inside the chamber, and shows that the slag plates keep a high temperature during charging. The slag temperature at the slag charging hole measured by infrared thermography was about 1373 K. Figure 9 shows the level of the charged slag in this pilot test. Figure 10 shows the gas temperature profiles obtained by the heat recovery pilot tests. The gas temperature increased with the amount of slag, and the maximum gas temperature was 989 K in the Case B test. Under the assumption that recovered heat with a temperature of over 413 K is available for actual use, the heat recovery ratio as a relative value against the total heat of the molten slag was 43%. Although increasing the gas flow rate sometimes causes fluidization of the bed, especially when the packed materials are plate-like, fluidization was not observed in these experiments. Figure 11 is a photograph of the discharged slag plates, showing that they retained their shape during heat recovery. The average thickness of the representative slag samples was 7 mm. Figure 12 shows the particle size distribution of the 60 kg slag plates in Fig. 11 measured by using standard sieves. From the results in Fig. 12, most of the slag plates were fractured to a size over the average thickness of 7 mm, and the percentage of fine particles with sizes of less than 7 mm was small. Because of the flat shape of the slag, it is estimated that cracking does not occur in a direction that would split the thickness of the slag plate; therefore, the value of the horizontal axis in Fig. 12 is regarded as the plane size L of the slag after crushing. Eq. (1) shows the relationship between the particle size d_p and the incipient fluidization velocity U_mf. U_mf changes depending on the shape factor ϕ of the particle. When the shape of the slag is approximated as a rectangular plate with dimensions of T×L×L [mm] and Φ is assumed to be a volume shape factor, the value of U_mf depends only on the plate thickness T in Eq. (1), where the equivalent circle diameter of the rectangular plate was substituted for d_p. From this equation, the U_mf of the 7 mm thickness slag plate is estimated to be 10 m/s. As the maximum gas superficial velocity in this pilot test was 2.4 m/s, which is lower than U_mf, fluidization of the slag packed bed was suppressed during heat recovery.   

$$\begin{array}{l}{\mathrm{U}}_{\mathrm{m}\mathrm{f}}=\frac{{\varphi }^2{\mathrm{d}}_{\mathrm{p}}{}^2\left({\rho }_{\mathrm{s}}-{\rho }_{\mathrm{g}}\right)\mathrm{g}}{1\mathrm{\hspace{0.17em} }650{\mu }_{\mathrm{g}}}=\frac{{\pi }^2{\mathrm{T}}^2\left({\rho }_{\mathrm{s}}-{\rho }_{\mathrm{g}}\right)\mathrm{g}}{26\mathrm{\hspace{0.17em} }400{\mu }_{\mathrm{g}}}\\ \left[\because \varphi =\frac{\mathrm{V}}{{\mathrm{d}}_{\mathrm{p}}{}^3}\hspace{1em}\mathrm{V}=\mathrm{T}{\mathrm{L}}^2\hspace{1em}{\mathrm{d}}_{\mathrm{p}}=2\mathrm{L}{\pi }^{-1/2}\right]\end{array}$$(1)

Φ: Particle shape factor

d_p: Particle size [m]

ρ_s: Particle density [kg/m³]

ρ_g: Gas density [kg/m³]

g: Gravitational acceleration [m²/s]

μ_g: Gas viscosity [Pa·s]

Fig. 7.

Appearance of hot slag plates crushed before charging into chamber.

Fig. 8.

Hot slag plates being charged into chamber.

Fig. 9.

Level of charged slag in chamber.

Fig. 10.

Temperature profile of hot air.

Fig. 11.

Appearance of slag plates discharged after heat recovery.

Fig. 12.

Distribution of slag particle size measured after heat recovery.

4. Discussion

4.1. Slag Packed Bed Heat Transfer Simulation Model

In order to calculate the gas temperature profile and its distribution inside the slag packed bed during heat recovery, we created slag packed bed heat transfer simulation model. Figure 13 shows schematic drawings of this model. In this model, the chamber is divided in the height direction one dimensionally, and the amount of heat exchanged between the gas and slag is calculated for each layer, which has the thickness of Δz. The total heat transfer surface in one layer is defined by the relevant input data, which include the slag shape, porosity of the bed, and bulk specific gravity of the packed slag. The gas-slag heat transfer coefficient on the surface and the temperature distribution inside a slag particle are both calculated by using a slag single particle heat transfer model which is defined separately in this simulator. For calculation of heat conduction in a slag single particle, the following Eq. (2) or Eq. (3) is used selectively, depending on the shape of the slag. Equation (4) is the heat transfer equation between the gas and slag surface. The heat transfer coefficient h_s in Eq. (4) is calculated by the equations detailed later. The heat radiation from the surface of the slag is simplified by converting the real heat flux to the average heat flux to the extended side wall, as shown by the dotted arrows in Fig. 13. This simulator has two calculation modes, stationary calculation for a stable operation, in which operating conditions such as the slag charging rate and discharging rate, gas flow rate, and temperature distribution in the bed are fixed, and nonstationary calculation, in which these operating conditions are not stable. Since the laboratory-scale and pilot-scale heat recovery experiments were batch type operations, the gas temperature profiles were calculated by nonstationary calculations under the condition that the slag charging rate and discharging rate are both zero. On the other hand, the gas temperature profile of an actual-scale plant in continuous operation is calculated by stationary calculation.   

$$\frac{\partial \left({\rho }_{\mathrm{s}}{\mathrm{H}}_{\mathrm{s}}\right)}{\partial \mathrm{t}}=\frac{\partial }{{\mathrm{r}}^2\partial \mathrm{r}}\left({\mathrm{r}}^2{\mathrm{\Gamma }}_{\mathrm{s}}\frac{\partial {\mathrm{H}}_{\mathrm{s}}}{\partial \mathrm{r}}\right)$$(2)

  

$$\frac{\partial \left({\rho }_{\mathrm{s}}{\mathrm{H}}_{\mathrm{s}}\right)}{\partial \mathrm{t}}=\frac{\partial }{\partial \mathrm{r}}\left({\mathrm{\Gamma }}_{\mathrm{s}}\frac{\partial {\mathrm{H}}_{\mathrm{s}}}{\partial \mathrm{r}}\right)$$(3)

  

$${\mathrm{Q}}_{\mathrm{s}-\mathrm{g}}={\mathrm{h}}_{\mathrm{s}}\left({\mathrm{T}}_{\mathrm{s}}-{\mathrm{T}}_{\mathrm{g}}\right)$$(4)

H_s: Slag specific enthalpy [J/kg]

Γ_s: k_s/C_s

C_s: Slag specific heat [J/kgK]

ρ_s: Slag density [kg/m³]

k_s: Slag thermal conductivity [W/mK]

T_s: Slag surface temperature [K]

T_g: Gas temperature [K]

t: Time [s]

r: Distance in depth direction [m]

Q_s−g: Heat flux from slag to gas [W/m²]

h_s: Heat transfer coefficient on surface of slag [W/m²K]

Fig. 13.

Schematic diagram of model for calculation of heat recovery in slag packed bed.

4.2. Results of Slag Heat Recovery Experiments and Evaluation of Accuracy of Gas-slag Heat Transfer Coefficient

To evaluate the effect of the shape of packed slag on the gas-slag heat transfer coefficient h_s at the slag surface, the following Ranz-Marshall’s equation²²⁾ (R-M equation, Eq. (5)) and Johnson-Rubesin’s equation²³⁾ (J-R equation, Eq. (6)) are used selectively in the above-mentioned model. The R-M equation is commonly used for calculation of the heat transfer in a packed bed of materials which can be approximated as spherical, such as coke in CDQ and sintered ore in a blast furnace. The J-R equation is used for calculation of forced convection heat transfer on the surface of plate-like materials. The correction factors α and β in the respective equations are related to the effect of the packing condition and particle size distribution. For calculation of the heat transfer in a blast furnace, the value of α in the R-M equation is estimated to be about 0.2,²⁴⁾ whereas that for a single particle is 1.0 in the ideal state. In this report, the J-R equation is multiplied by the correction factor β in the same manner as the R-M equation.   

$$\begin{array}{l}{\mathrm{h}}_{\mathrm{s}}=\alpha \frac{{\mathrm{k}}_{\mathrm{g}}}{{\mathrm{D}}_{\mathrm{p}}}\left(2.0+0.6{Pr}^{1/3}{Re}^{1/2}\right)\\ Pr={\mathrm{C}}_{\mathrm{g}}\frac{\mu }{{\mathrm{k}}_{\mathrm{g}}}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }Re={\mathrm{D}}_{\mathrm{p}}\frac{{\mathrm{V}}_{\mathrm{g}}}{\mathrm{\nu }}\end{array}$$(5)

  

$$\begin{array}{l}{\mathrm{h}}_{\mathrm{s}}=\beta \frac{{\mathrm{k}}_{\mathrm{g}}}{{\mathrm{L}}_{\mathrm{m}}}\left(0.037{Pr}^{1/3}{Re}^{4/5}\right)\\ Pr={\mathrm{C}}_{\mathrm{g}}\frac{\mu }{{\mathrm{k}}_{\mathrm{g}}}\hspace{1em}Re={\mathrm{L}}_{\mathrm{m}}\frac{{\mathrm{V}}_{\mathrm{g}}}{\mathrm{\nu }}\end{array}$$(6)

C_g: Gas specific heat [J/kgK]

k_g: Gas thermal conductivity [W/mK]

V_g: Gas velocity [m/s]

μ: Gas viscosity [Pa-s]

ν: Gas kinematic viscosity [m²/s]

D_p: Average particle diameter (R-M equation)

L_m: Average particle side length (J-R equation)

α, β: Correction factors (single particle: α=1.0, infinite plate: β=1.0)

These two equations have different exponents of the Reynolds number (Re). The exponent of Re in the J-R equation is closer to 1.0 than that in the R-M equation, meaning that the h_s of the J-R equation increases more linearly with the increase of Re. Re is proportional to the gas flow rate V_g. When the calculations are fitted to the results of experiments conducted with different V_g, the stability of α and β means the accuracy of the heat transfer equations. The inlet gas flow rate directly relates to V_g, and the slag temperature T also affects the value of V_g because of the thermal expansion of the gas during heat recovery. Therefore, the Table 1 experiments were conducted under conditions in which the values of V_g and T were changed, and the gas temperature profiles were also calculated by the slag packed bed heat recovery model to compare the stability of α and β after fitting them to the experimental results. Considering the batch operation illustrated in Fig. 14, nonstationary calculations were used for the calculations of gas temperature in these experiments. The size of the packed materials for the calculation with the J-R equation was assumed to be 7×30×30 [mm], and that with the R-M equation was defined by the diameter of an approximated spherical particle that has a surface area equivalent to the same slag plate. Figures 15, 16, 17 are the results of calculations under the conditions in Table 1. Figures 15 and 16 are the results when the calculations are fitted to the measured data by adjustingα in the R-M equation and β in the J-R equation. Figure 17 is the rate of change expressed as a relative value against the fitted values of α and β in Case 1 (α1, β1). From the results of Fig. 17, β is more stable than α, meaning that the J-R equation is more reliable when the packed slag is plate-like. The value of β in Case 3, which was conducted at the highest temperature, was 0.25. Similarly to the above-mentioned α in the R-M equation, this value is lower than the value for an ideal infinite plate, 1.0.

Fig. 14.

Operating conditions of slag heat recovery experiment for calculation.

Fig. 15.

Results of α fitting analysis (R-M equation).

Fig. 16.

Results of β fitting analysis (J-R equation).

Fig. 17.

Comparison of stability of values of α and β.

4.3. Evaluation of Accuracy of Heat Transfer Coefficient in Pilot-scale Test

Figure 18 shows the comparison of the gas temperature profiles obtained by the pilot test in Case B and the calculation by the slag packed bed heat recovery model. The calculation conditions are listed in Table 4. The value of β was changed in the range of 0.25 to 0.42. In this pilot plant, the gas temperature changes considerably due to the effect of the inner wall of the chamber compared with the laboratory scale chamber because the free area before the gas temperature measuring points is longer. Heat storage in the inner wall can be estimated from the wall heat capacity and wall temperature measured by thermocouples on the outside of the wall. Therefore, for the comparison with the test results in Fig. 18, the gas temperature decrease caused by heat storage in the inner wall was subtracted from the calculated gas temperature. From the results in Fig. 18, the shapes of the calculated gas temperature profiles are similar to the measured profile. In this pilot test, the peak value of the measured gas temperature agreed with the calculation result when the value of β was 0.42. The difference between β in the laboratory-scale experiment and in this simulation is estimated to be due to the different shape of the slag or different packing conditions. As in the evaluation in the laboratory-scale experiment, the value of β in the pilot-scale test was also lower than 1.0. Thus, the results confirmed that the value of β in the packed bed of plate-like slag becomes lower than the ideal value 1.0 of an infinite plate.

Fig. 18.

Comparison of measured and calculated gas temperature.

Table 4. Conditions of slag heat recovery calculations.

Calculation method	Transient
Calculation time [min]	80
Chamber cross section [m2]	3
Slag weight [t]	4.8
Slag temperature [K]	1373
Slag bed porosity	0.55
Slag bed bulk density [t/m3]	1.0
Gas flow rate [Nm3/h]	7200
J-R eq. Correction value β	0.25, 0.33, 0.42

4.4. Relationship between Correction Factor β and Efficiency of Slag Heat Recovery at Industrial Scale

The effect of the aforementioned correction factor β on the efficiency of slag heat recovery at the industrial scale was estimated by performing a calculation with the slag packed bed heat transfer model. The calculation was performed with the stationary calculation model, in which the slag charging rate and discharging rate are both stable and the temperature distribution of the slag packed bed does not change.

Figure 19 shows the assumed operating conditions in this calculation. The assumptions for the industrial scale operation were defined as slag charging rate, 60 t/h, slag charging temperature, 1373 K, packed bed size, Φ2.5 [m]×H [m] (height: 3 m or 6 m), slag bed porosity, 0.55, and gas flow rate, 40000 [Nm³/h] (gas superficial velocity: 8.0 m/s at 973 K). The heat loss from the chamber wall was fixed at 10 [kW/m²]. Figures 20 and 21 show the calculation results. The slag heat recovery ratio η decreases depending on the decrease of β. In the results in Fig. 20, when β decreases from 1.0 to 0.2, η decreases from 48% to 33%. A decrease of the gas temperature and increase of the slag discharge temperature are also observed with the decrease of β, meaning that the efficiency of heat exchange deteriorated and the heat of 60 t/h of slag was not recovered sufficiently during the residence time in the chamber. To decrease the slag discharge temperature and increase η, it is effective to increase the height of the slag packed bed so as to secure a longer residence time, on the condition that the slag plates retain their shape under the increased packing pressure. The results in Fig. 21, which was calculated for the case where the slag bed height was increased from 3 m to 6 m, show that the slag discharge temperature decreases, and as a result, the slag heat recovery ratio ηincreases. The rate of increase is larger, especially at the left side of the graph, which means that the plant performance remains stable against changes in β.

Fig. 19.

Operating conditions of actual slag heat recovery plant for calculation.

Fig. 20.

β sensitivity analysis for slag heat recovery system (height of packed bed: 3.0 m).

Fig. 21.

β sensitivity analysis for slag heat recovery system (height of packed bed: 6.0 m).

5. Conclusion

In this report, we evaluated the heat transfer behavior in a slag packed bed in laboratory- and pilot-scale experiments and by simulation with a newly-proposed slag packed bed heat transfer model. As a result, it was confirmed that the gas-slag heat transfer coefficient in a packed bed of plate-like slag can be estimated by a Johnson-Rubesin’s equation modified by a correction factor β. The value of β tends to become lower than 1.0 in the packed bed. The same effect was observed in the results of subsequent pilot-scale slag heat recovery tests with slag from an actual steel mill. The value of β affects the efficiency of the slag heat recovery process at the industrial scale. In case the value of β decreases, the slag heat recovery ratio also decreases because of the slag discharge temperature increases. A higher slag packed bed height reduces the effect of β and improves the stability of plant operation.

Acknowledgement

This research was supported as a part of the NEDO project in Japan, “Environmentally Harmonized Steelmaking Process Technology Development (COURSE50)”.

References

• 1)   T.  Miwa and  H.  Okuda: J. Jpn. Inst. Energy, 89 (2010), 28.
• 2)   S.  Tonomura: Proc. 11th Int. Conf. on Greenhouse Gas Control Technologies (GHGT-11), Elsevier, Amsterdam, (2012), 1G-1.
• 3)   J.  Ando,  T.  Nakahara,  H.  Onous,  S.  Ichimura and  M.  Kondo: Mitsubishi Heavy Industries Tech. Rev., (1985), 136.
• 4)   M.  Yoshinaga,  K.  Fujii,  T.  Shigematsu and  T.  Nakata: ISIJ Int., 22 (1982), 823.
• 5)   S. J.  Pickering,  N.  Hay,  T. F.  Roylance and  G. H.  Thomas: Ironmaking Steelmaking, 12 (1985), No. 1, 14.
• 6)   W. B.  Featherstone: Iron Steel Eng., (1998), 42.
• 7)   T.  Mizuochi,  T.  Akiyama,  T.  Shimada,  E.  Kasai and  J.  Yagi: ISIJ Int., 41 (2001), No. 12, 1423.
• 8)   N.  Maruoka,  T.  Mizuochi,  H.  Purwanto and  T.  Akiyama: ISIJ Int., 44 (2004), No.2, 257.
• 9)   H.  Purwanto,  T.  Mizuochi,  H.  Tobo,  M.  Takagi and  T.  Akiyama: Mater. Trans., 45 (2004), No.1,23286.
• 10)   Y.  Qin,  X.  Lv,  C.  Bai,  G.  Qiu and  P.  Chen: JOM, 64 (2012), No.8, 997.
• 11)   H.  Zhang,  H.  Wang,  X.  Zhu,  Y.  Qiu,  K.  Li,  R.  Chen and  Q.  Liao: Appl. Energ., 112 (2013), 956.
• 12)   M.  Rycroft: Energize, April (2014), 55.
• 13)   H.  Tobo,  Y.  Ta,  M.  Kuwayama,  Y.  Hagio,  K.  Yabuta,  H.  Tozawa,  T.  Tanaka,  K.  Morita,  H.  Matsuura and  F.  Tsukihashi: Tetsu-to-Hagané, 99 (2013), No.12, 683.
• 14)   S.  Messai,  M. E.  Ganaoui,  J.  Sghaier and  A.  Belghith: Therm. Sci., 18 (2014), No.2, 443.
• 15)   J.  Yang,  J.  Wang,  S.  Bu,  M.  Zeng,  Q.  Wang and  A.  Nakayama: Chem. Eng. Sci., 71 (2012), 126.
• 16)   J.  Yang,  Q.  Wang,  M.  Zeng and  A.  Nakayama: Chem. Eng. Sci., 65 (2010), 726.
• 17)   E.  Schröder,  A.  Class and  L.  Krebs: Exp. Therm. Fluid Sci., 30 (2006), 545.
• 18)   P. X.  Jiang,  R. N.  Xu and  W.  Gong: Chem. Eng. Sci., 61 (2006), 7213.
• 19)   N.  Shigaki and  S.  Ozawa: Proc. PRICM8, The Minerals, Metals & Materials Society (TMS), Warrendale, PA, (2013), 769.
• 20)   S.  Ozawa,  N.  Shigaki and  H.  Tobo: Proc. PRICM8, The Minerals, Metals & Materials Society (TMS), Warrendale, PA, (2013), 761.
• 21)   H.  Tobo,  N.  Shigaki and  Y.  Hagio: JFE Tech. Rep., (2014), No.19, 126.
• 22)   W. E.  Ranz and  W. R.  Marshall: Chem. Eng. Prog., 48 (1952), 173.
• 23)   H. A.  Johnson and  M. W.  Rubesin: Trans. ASME, 71 (1949), No.5, 447.
• 24)   M.  Hatano and  K.  Kurita: Tetsu-to-Hagané, 66 (1980), 1898.