Fuel Processing Technology 196 (2019) 106165
Contents lists available at ScienceDirect
Fuel Processing Technology journal homepage: www.elsevier.com/locate/fuproc
Research article
Co-combustion of semicoke and coal in an industry ironmaking blast furnace: Lab experiments, model study and plant tests
T
Z.J. Hua,b,1, Y.R. Liuc,1, H. Xub, J.M. Zhub, S.L. Wua, , Y.S. Shenc, ⁎
⁎
a
School of Metallurgical and Ecological Engineering, University of Science and Technology Beijing, Beijing 100083, China Ironmaking Plant, Baoshan Iron & Steel Co Ltd, Shanghai 201900, China c School of Chemical Engineering, University of New South Wales, Sydney, NSW 2052, Australia b
ARTICLE INFO
ABSTRACT
Keywords: Semicoke CFD Plant test PCI Co-combustion Blast furnace
Low-rank coals can be upgraded to semicoke, and potentially used to partly replace expensive metallurgical coals in pulverised coal injection (PCI) in ironmaking blast furnaces (BFs). In this paper, an integrated research of “lab experiments – model study – plant test” is used to study the feasibility of injecting semicoke in BFs. First, lab experiments are conducted to characterise the metallurgical properties of semicoke. Secondly, a three-dimensional CFD model is developed to study in-furnace phenomena related to two PCI operations: single injection of semicoke and co-combustion of PCI coal and semi-coke under full-scale BF conditions. The simulation results indicate that the semicoke shows similar combustion profiles with the PCI coal, confirming the feasibility of using semicoke in BFs. Moreover, the co-combustion of semicoke and PCI coal with the blending ratios of 0–40% is studied, as a result, the optimal blending ratio of semicoke, 30%, is recommended. Then, plant tests of semicoke combustion of blending ratio 0–20% are conducted. It is indicated that ironmaking indexes remain stable largely, confirming the practical feasibility of semicoke co-injection operation. The study provides a combined view of co-injecting PCI coal and semicoke in terms of fundamental combustion profiles and plant performance in a commercial BF.
1. Introduction Pulverised coal injection (PCI) has been recognised as an effective technology in blast furnace (BF) ironmaking for many benefits including cost reduction, stable operation and coal supply flexibility [1]. In most cases, coal blends will be injected in PCI operations. In this process, pulverised coal particles are injected into the lower part of BFs via tuyere and combust in the void region, the so-called raceway. It presents a very intense and complicated multiphase and thermochemical process. At present, BF and PCI operations are facing the coal constraint problems, i.e. steel industry can use only a certain quotation of coal in ironmaking according to local government environmental policy [2–5]. Semicoke, also named Lantan, refers to the low-volatility solid carbonaceous products of the pyrolysis of high-volatility nonviscous or low viscous bituminous coals with a low temperature of 400–700 °C, where coal tar and retorting gas can also be obtained [6,7]. Notably, semicoke is not regarded as coal according to local government environment policy and is not included into the coal quotation in many countries including some regions in China [8]. In addition,
semicoke has several favourable characteristics, such as low-ash, lowsulphur, low-phosphorus, low-aluminium, high fixed-carbon, high chemical reactivity and high specific resistance. Therefore, semicoke has the potential to be used in PCI operations [9]. Moreover, the price of semicoke is comparatively lower than coke and anthracite coal, making semicoke injection in BFs very competitive. Thus, the co-combustion of semicoke and coal is recognised as a desired option for PCI operation for better coal supply flexibility [10,11], environmental policy relief [10], and moderate facility modifications [12,13]. On this basis, key questions related to this promising technology should be answered before massive implementation: i) Are the in-furnace phenomena of semicoke combustion comparable with the typical PCI operation? ii) If yes, what are the criteria of semicoke blending ratio for BF stability concern? To address these issues, different research methods, such as lab experiment, plant test, and model study, have been used. Firstly, several attempts were made based on lab-scale experiments and some basic properties related to PCI operation were studied. Geng et al. [6] studied the pyrolysis characteristics of Shenmu bituminous coal and its
Corresponding authors. E-mail addresses:
[email protected] (S.L. Wu),
[email protected] (Y.S. Shen). 1 Contribute equally as first authors. ⁎
https://doi.org/10.1016/j.fuproc.2019.106165 Received 1 April 2019; Received in revised form 21 June 2019; Accepted 30 July 2019 Available online 26 August 2019 0378-3820/ © 2019 Elsevier B.V. All rights reserved.
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
pyrolysates and inferred that the semicoke is a useful material for industrial production. Zou et al. [14] studied the quality of semicoke and confirmed it met the very basic requirement of the BF, while the grindability and combustion performance of semicoke fluctuate greatly. Li et al. [15] found that semicoke has a lower initial reaction temperature than coke, larger reaction rate than coke particles, and sufficient reactivity with CO2. Yang et al. [16] found that when the semicoke ratio reached 40%, the blended coal for PCI showed excellent metallurgical properties. The above properties and performance tested by lab experiments make semicoke a possible injectant in PCI operations. However, lab-scale experiments cannot provide detailed in-furnace phenomena related to semicoke combustion under the industry scale conditions. Secondly, some initial plant tests were conducted to confirm the feasibility of semicoke combustion in BFs. Some Chinese ironmaking industries, including Ansteel, Baotou steel and JISCO, carried out some preliminary plant tests using their commercial BFs and concluded some basic performance of semicoke injection in the commercial BFs, such as conveying performance, non-explosivity, and high heat value of semi-coke etc. For example, Ansteel's plant tests [17] confirmed that the transportation properties, combustion efficiency and theoretical replacement ratio of the optimised blended could meet the basic industrial standards. In the plant tests at Baotou Steel [18], semicoke obtained in the low-temperature pyrolysis can meet the basic PCI requirements such as metallurgical performance and cost reduction. A grindability of 58.22%–62.56%, a similar explosibility and higher combustibility than the current PCI coals, and a cost reduction of 50 RMB·t−1 were achieved when replacing the coal with semi-coke. JISCO [19] conducted a plant test on one BF to find the semicoke has combustibility between bituminous coal and anthracite and to confirm that pulverised semicoke used in PCI could guarantee the BF smooth operation. However, it is high risk and cost to test, and even higher if generalise mechanism and experience after plant test. Modelling study then is highly recommended to estimate metallurgists because of its cost-effective, and more detailed in-furnace is not provided [20–22]. Zhang et al. [23] extended a computational fluid dynamics (CFD) model to the virtual use of beneficiated Victorian brown coal and its semi-cokes in the 600 MWth plant. However, the PCI process largely differs from other coals. This complex process involves gas/powder/solid multiphase flow and heat/mass transfers related to specific chemical reactions. Liao et al. [24] developed a 3D industryscale CFD model to simulate the complicated in-furnace combustion phenomena relevant to the injection of Victorian brown coals, and used for investigating the feasibility of replacing conventional PCI coal with brown coal products in these validating plant tests. However, co-combustion of coal and semicoke, the most possible utilisation way of semicoke in BFs, has not been studied using a 3D industry-scale PCI model in the past. As a result, the detailed in-furnace phenomena relate to the co-combustion is not clear under industry-scale BF conditions. Moreover, the comprehensive view of using semicoke in BFs may need a multiscale study, by combining lab characterization, model studies and plant tests. Such multi-scale comprehensive study is rare in testing a new material for practice. In this study, an integrated research method of “Lab experiment Model study - Plant test (LMP)” is proposed and used to investigate the feasibility of semicoke combustion in PCI technology. Lab experiments are firstly used to identify its metallurgical properties. A 3D industrialscale CFD model is used to understand the in-furnace phenomena of flow and combustion in some co-firing semicoke and coal and confirm critical conditions of CFD studies for plant tests. A plant test is conducted to confirm the effect of co-combustion of semicoke and coal in real BF practice. The study aims to understand the applicability of metallurgical properties of semicoke by lab experiments, in-furnace phenomena related to combustion by the CFD model, and to examine
Fig. 1. The schematic of “Lab experiment - Model study - Plant test (LMP)” approach of this study.
the effectiveness of co-combustion by plant test in a commercial BF. 2. Methods The schematic of this integrated LMP research method, including i) lab experiments, ii) a 3D industrial-scale CFD model study; and iii) plant tests, is shown in Fig. 1. The three components at different scale represent different stages of this research. They are interconnected, where the lab experiments provide input parameters for the model study; the model study helps understand in-furnace phenomena under different conditions and more importantly, provides a safe and optimal design for the plant tests. 2.1. Lab experiments The chemical composition of semi-coke is obtained using coal proximate analysis procedure GBT 212-2008 and coal ultimate analysis procedure GBT 476–2001. For example, VM% is measured by heating a certain amount of air-dried coal sample at 900 ± 10 °C for 7 min (with air isolated) and calculated by Eq. (1). Ash% is measured by completely ashing the coal sample in a rapid ash tester from the preheating temperature of 815 ± 10 °C and calculated by Eq. (2). S% is measured by oxidizing various forms of sulphur in the coal into sulphur oxides and then trapped in a hydrogen peroxide solution, then formed sulphuric acid solution is titrated with a sodium hydroxide solution.
VM % =
Wloss × 100% Wsample
Ash% =
Wresidue × 100% Wsample
Mad
(1) (2)
where the Wloss is the weight loss of the coal sample, Mad is the moisture content in the air-dried coal sample, Wsample is the mass of the coal sample, Wresidue is the mass of the residue. The grindability is an indicator of coal hardness, reflecting the difficulty of raw coal processing and utilisation and affecting the energy consumption in pulverising coal in this study. CNK-60 mill is used for
2
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
testing Hardgrove grindability index (HGI) of the semi-coke. The semicoke samples are prepared to be in the particle size range of 0.63–1.25 mm and dried at 105 °C for 2 h. HGI is calculated based on the weight of original powders (W0), and weight of powders sieved by a 200-mesh sieve (W-200) after milled for (60 ± 0.25) r·min−1, as shown in Eq. (3).
HGI = 13 + 6.93 × (W0
W
investigated. Blowpipe-tuyere-raceway region is treated as a cavity, and the coke bed is treated as a porous medium. The following physicochemical processes are included in this model: (1) turbulent flow of gasparticle in the raceway and coke bed, (2) Pulverised fuel blends combustion process, including the devolatilization of coal, fuel gas combustion and char reactions, (3) gasification and combustion of coke, (4) heat transfer among gas-solid phase. The CFD model is developed on the platform of ANSYS-CFX 17.2.
(3)
200 )
The ignition temperature refers to the lowest temperature at which the volatile (fuel gas) released by coal will spontaneously ignite in a normal atmosphere without an external source of ignition. PCI requires the pulverised fuel to have a relatively high ignition temperature for the sake of the safety of coal storage, transport, milling and injection. The semicoke is milled to 3-5 mm and dried in 80 °C for 2 h, and then heated in the RBX-03 tester continuously. The temperature at which the gas volume in the heating tube suddenly expands or the rate of temperature rise suddenly increases is recorded as the ignition temperature of the sample. To ensure the safety of PCI, pulverised fuel is required to be nonexplosive or low explosive. At present, the explosiveness of pulverised fuel is mainly detected by the length of the return flame of the instantaneous explosion of pulverised fuel. Semicoke is prepared to be 0.074 mm or less and dried at 105 °C for 1 h. The length of the return flame is determined by using 1 g of pulverised semicoke sprayed to a fire source of 1050 °C (1323 K).
2.2.1. Model details A set of 3D, steady-state Reynolds averaged Navier-Stokes equations closed by the standard k-ε turbulence model equations is used to describe the gas phase flow. The Lagrangian method is used to track particles along the discrete particle trajectories. Changes in particle movements are calculated by Newton's second law, where drag force and turbulent dispersion are considered. The change of particle temperature is governed by three physical processes: convective heat transfer, latent heat associated with chemical reactions, and radiative heat transfer. For example, heat transfer associated with mass transfer QM driven by latent heat in this model:
QM =
dmp dt
Hreac
(4)
where the sum is taken over all components of the particle for which heat transfer is taking place. mp is the particle mass. Hreac is the latent heat loss/gain by reactions, including the heat loss due to devolatilization of raw coal and two gasification reactions of char, and heat gain due to char oxidation reaction. This is also used in Refs. [22, 25]. Chemical reactions related to the combustion of the blends of semicoke and coal are considered in Table 1, including preheating,
2.2. CFD model study The model has been described in details in Refs. [24, 25], and is outlined below for completeness. The model in this paper uses a single computational domain to cover the lower part of the commercial BF Table 1 Chemical reactions Reaction
Reaction rate expression
Reaction kinetics
Coal blends reactions
Coal Coal
kv1 kv2
1 Fuel
gas + (1
1 ) Char
2 Fuel
gas + (1
2 ) Char
A1 = 3.7 × 105s 1; E1 = 18000K A2 = 1.46 × 1013s 1; E2 = 30189K 1 = VM (daf ); 2= Q× 1 CS = 0.26
E RTp
k = A exp
mref d d = CS d 0 dt mref ,0
Fuel gas + O2 → CO2 + 2H2O
ri = CA min
ϕChar + O2 → 2(ϕ − 1)CO + (2 − ϕ)CO2
2( 2
Char + CO2 → 2CO
1)
Char + H2O → CO + H2
[ i] vt
MC e MO2
1
k2 = (1
e)
k3 = kc Tp
Coke reactions Coke + O2 ↔ CO2
dmcoke dt
kc Dp e
C
(k1 1 + (k2 + k3) 1) 1m c
coth
Dref fluid
Tp + Tg 2Tref
; kd =
Dref rcoke
Tcoke + Tg 2Tref
0.75
P ; PA
A c = 6069m s 1 K 1; Tc = 32406K AS = 0.0004; TS = 6240K
1
2
Tc Tp 0.5
2 = (kd 1 + kc _1coke ) 1 [i]4 rcoke
Coke + CO2 ↔ 2CO Note: Dp = effic × D; D =
A c = 202300m s 1 K 1; Tc = 39743K AS = 0.0004; TS = 6240K
kc rp
k c = A c Tp exp =R
A c = 14m s 1 K 1; Tc = 21580K AS = 2500; TS = 6240K
TS Tp
= AS exp 3
dmc = dt D k1 = 2 rp
CA = 4.0
( )
k c _coke = A c exp
Tc Tcoke
3
P PA
A c = 3.26 × 106kg m Tc = 10855K
2
s
1
A c = 4.71 × 109kg m Tc = 29018K
2
s
1
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
Table 2 Operating conditions. Operating conditions Working volume Productivity Tuyere number Reference pressure Boundary conditions O2 enrichment in blast Blast (23.7% O2, 15g·m−3 H2O) Cooling gas (air) Conveying gas PCI injection rate
4706 m3 2.23 tHM·m−3·day−1 (10,500 t·day−1)) 40 (total area of 0.5019 m2) 392 kPa (235 + 157 kPa) 15,000 Nm3·h−1 (2.7% O2 enrichment) 6700 Nm3·min−1 air 1523 K + 15,000 Nm3·h−1 O2 (1250 °C) 3 −1 3 −11 (100 m ·h for 298 K 4000 m ·h each tuyere) 3200 m3·h−1 air 353 K + 1500 m3·h−1 N2 190 kg·tHM−1 353 K (Blending ratio of semi-coke: 0%, 10%, 20%, 30%, 40%)
devolatilization, gaseous combustion, char gasification and oxidation, and coke gasification and combustion. The particle swelling due to the gas release during the devolatilization phase is considered in this model by assuming that the particle diameter change is in proportion to the volatiles released [26]. In this study, the coal and semicoke are treated as materials with low VM (also termed fuel gas in figures), according to the proximate analysis from the industry report (section 3.1). Coke is treated as non-VM material due to very low VM, < 2% [25,27], and thus the devolatilization is not considered for coke. Such treatment has been widely used in PCI or BF modelling works [25,28].
Fig. 3. Particle size distribution of the injectant mixture.
2.2.2. Model validation and simulation conditions The model is validated against measurements, in terms of burnout and gas composition in the previous studies for a range of coals, black coals [1,25] and brown coals [24]. Pulverised coal blends are injected into BF under the same operating conditions with different blending ratios of the semi-coke, as shown in Table 2. In this work, a stainless-steel co-axial coal lance is used for PCI operation. Two streams can be observed in this coal lance, including coal with conveying gas and cooling gas (Fig. 2 [29]). According to the industry report, in practice, different raw coals (including semicoke) are mixed firstly and then used in the grinding process. Thus, in this study, the blend of coal and semicoke is treated as one injectant with Fig. 4. Plan of plant test.
weighted average properties of coal and semicoke. Particle size distribution of the blends is described by Rosin-Rammler distribution (Fig. 3), where the mean particle size, de, is 115 μm, and a measure of the dispersion of particle sizes, the spread parameter, γ, is 3.5. 2.3. Plant tests A plant test of semicoke injection was conducted on a commercial BF. Based on the lab experiments and model study, detailed plan of plant test is made as showed in Fig. 4. That is, plant test was conducted for 49 days using one commercial BF. The blending ratio of semicoke is changed from 0%, 10%, 15%, 18% and 20%. During the plant test, other raw materials and operational conditions are maintained at a similar level.
Fig. 2. Lance configuration and dimensions [29].
4
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
Table 3 The proximate and ultimate analysis of semicoke and two PCI coals. Properties
Semicoke
Coal 1
Coal 2
Proximate analysis (ad.) Moisture, % Ash, % Volatile matter (VM), % Fixed carbon, % HGI Qnet_d
7.56 9.02 10.10 80.16 58 31,500
1.17 7.72 16.24 76.04 73 32,570
0.75 9.90 8.64 81.89 57 31,287
Ultimate analysis (daf.) C, % H, % N, % S, %
81.50 3.5 1.25 0.34
84.59 3.84 1.08 0.47
82.33 3.52 1.16 0.50
3. Results and discussion The LMP method is used to give a multi-scale view and full picture of the feasibility of co-combustion of semicoke and PCI coal. The lab experiments, model study and plant test are conducted and analysed from the aspects of metallurgical properties, in-furnace combustion profiles and BF-performance. 3.1. Lab experiments Metallurgical properties of semicoke are tested in the lab experiments, including the grindability, composition, ignition temperature and explosiveness. It is shown in Table 3 that the semicoke has the characteristics of lower ash content, lower sulphur content, higher fixed carbon content and higher calorific value, and is suitable for PCI operation. HGI of semicoke were tested as 58, 52, 57 and 53, which all meet the PCI requirement (> 50) [30]. Semicoke has a low VM content and sulphur content and a high ignition temperature of 420, 430, 440, 450 °C (693, 703, 713, 723 K). Compared with PCI coals, the relative high ignition temperature shows that semicoke is not easy to gasify and burn, thus helping to improve the safety for PCI operation. The returned fire length of semicoke is shorter related to the low VM content, namely, 180, 220, 200, and 120 mm. In this study, the returned fire length of semicoke is even shorter than the low VM content, namely the explosiveness of semicoke is weak and semicoke is safe in transportation in view of explosiveness. The weak explosiveness of semicoke helps to improve the safety of PCI. On this basis, the semicoke not only meets the PCI requirement in terms of the grindability and composition but also helps to improve the safety of PCI operation with its high ignition temperature and weak explosiveness. Qnet_d is used in this study to represent net calorific value by combusting a specified quantity (i.e. net heat of combustion). The selected semicoke is subjected to the model study with these input parameters.
Fig. 5. Particle trajectories coloured by particle temperature: (a) semicoke and (b) Coal 2. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
3.2.1. Single injections of semicoke and coal The in-furnace phenomena of a single injection of semicoke and Coal 2 are compared and analysed to investigate the possibility of replacing PCI coal (anthracite) with the semi-coke, in terms of the flow field within the raceway region and combustion profiles of the pulverised fuels. As a measure of the combustion efficiency, the burnout is calculated according to the ash balance, as shown in Eq. (5).
3.2. Model study
Burnout = 1
Based on the measurements from lab experiments, the CFD model is used for studying the semicoke combustion under full-scale BF conditions. First, the single injection of semicoke and Coal 2 in Table 3 are studied, and their in-furnace behaviours are compared for exploring the feasibility of replacing Coal 2 with the semicoke. Then, more practically, the co-combustion of semicoke and PCI coal (Coal 1) is studied for identifying the suitable blending ratio of semicoke in the blends in PCI technology.
ma,0 /(1 ma
ma,0)
(5)
where ma,0 is the ash content of the original coal blends and ma is the ash content of the burnt residual collected. As defined, the burnout represents the total weight loss of the semicoke or coal particles. In this study, the burnout can be calculated along the particle plume and over the entire tuyere-raceway region, respectively. Fig. 5 compares the flow pattern and particle temperature between the two single injection cases. It is indicated that the flow patterns are
5
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
Fig. 6. Comparison of semicoke and Coal 2 in terms of the evolution of burnout along the particle plume within the raceway region.
very similar. Two parts can be observed in both cases: particle plume of lower temperature and large-scale recirculation of higher temperature. That is, the inclined particle plume is formed after powder exiting lance into the tuyere, followed by a recirculation within the raceway centre. Along the particle plume, the temperature increases from the room temperature and remains at a relatively low level. It is less than the hot blast temperature of 1523 K in the upstream, indicating the particles are heated by hot blast firstly before reaching the peak particle temperature near the endpoint of raceway as a result of the exothermal reactions related to combustion. In the recirculation region, the particle temperature starts from the highest temperature, near 2800 K at the boundary of the raceway to a moderate temperature around the raceway centre. Therefore, it is indicated that using semicoke in PCI technology has an insignificant impact on the in-furnace flow and temperature fields. Fig. 6 compares the evolutions of burnout along the particle plume between the two single injection cases. It shows that the combustion profiles of the two cases are similar. Both injectants start to burn from around 0.8 m away from the lance tip, where semicoke gives a slightly earlier start because of its higher VM content and the resulted stronger gas combustion. The two increase rapidly and continuously until the endpoint of the raceway. Similarly, semicoke has a slightly higher burnout than Coal 2 along the particle plume. However, anthracite is a typical coal component in the coal blends in PCI technology. It means that due to similar combustion profiles, the semicoke has the potential to replace anthracite as one of the components in the coal blends.
Fig. 7. Particle trajectories of co-combustion coloured by burnout under the blending ratio of semicoke of: (a) 0%, (b) 40%. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 7 compares the combustion profiles of two blends: 0% semicoke (namely 100% Coal 1) vs. 40% semicoke. It is indicated that the flow patterns also include two parts: i) an inclined particle plume of lower burnout level. ii) a large scale of recirculation starts from the boundary of the raceway, with burnout up to 100%. The qualitative similar flow pattern of two cases indicates it is possible to use semicoke in the blend in PCI operations. To identifying the safe and suitable semicoke blending ratio, the effect of the blending ratio of semicoke on the in-furnace phenomena of co-combustion is further studied quantitatively. Fig. 8 compares the fuel gas molar fractions qualitatively and then quantitatively. It is indicated that in the cases of 0% and 40% semicoke blending ratio, the fuel gas only appears near the end of particle plume, indicating a fast devolatilization and gases combustion process. However, using a higher blending ratio of semi-coke, the devolatilization is delayed, and less fuel gas is produced. Moreover, four semicoke blending ratios are compared along the centreline. The difference of peak values between the cases of 0–20% and cases of > 30% are relatively large. This indicates a relatively larger amount of unburnt coal in the raceway and thus a much worse permeability in the surrounding coke bed, when > 30% semicoke is used, because the volatiles yield plays a key role in determining combustion efficiency of the coal blend [25]. Thus a blending ratio of semicoke below 30% is recommended in
3.2.2. Co-combustion of semicoke and PCI coal The single injection simulations indicate that the semicoke has the potential to replace anthracite as one of the components in the coal blends. In this part, the semicoke will be used to replace Coal 2, anthracite, in the blend and investigate the effect of blending ratio on infurnace phenomena for identifying the safe and suitable range of blending ratio which will be adopted in the plant tests. The in-furnace phenomena of co-combustion of semicoke and Coal 1 are investigated with a range of blending ratios of 0–40% semicoke, in terms of gas and particle phases, such as temperature field, flow field, combustion efficiency.
6
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
7
(caption on next page)
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
Fig. 8. Effect of blending ratio of semicoke on the co-combustion in terms of devolatilization and fuel gas combustion: the contour of fuel gas molar fraction under blending ratio of semicoke of 0% (a) and 40% (b), the evolution of fuel gas molar fraction along the particle plume (c).
the co-injection of semicoke and PCI coal in BF practice. Fig. 9 compares the gas temperature profiles under different semicoke blending ratios qualitatively and then quantitatively, especially in the downstream of particle plume. Overall, gas temperature increases significantly along the particle plume from upstream to downstream (around 2300 K), followed by a level off near the raceway end. Quantitative profile of temperature is compared along the particle plume in Fig. 9(c). The gas temperature profiles are similar before 1.2 m, corresponding to the process of particles heated by the hot blast. After that, the gas temperature increases faster and to a higher value when the less semicoke is used in the blends. This is mainly because that the ignition and combustion of higher VM of coal blends release more heat and will ignite and oxidise the residual char, generating more heat and leading to the higher temperature in the raceway [25]. Moreover, the endothermic char gasification in coal and semicoke combustions in the raceway and coke gasification in the coke bed also contribute the lower gas temperature in this region, especially when more semicoke (of lowVM and high char) is used in the blend. Fig. 10 compares the burnout profiles under different semicoke blending ratios, where two burnout calculations are used, respectively. The final burnout represents the unburnt particles that flow into coke bed through the endpoint of the raceway region. Average burnout is a more comprehensive indicator to represent the combustion efficiency of coal blends over the entire tuyere-raceway region [31]. Both two burnout calculations decrease with the higher blending ratio of semicoke, where final burnout decreases from 39.26% to 30.16% by 9.1% and the average burnout decreases from 68.33% to 62.83% by 5.5%. Moreover, it is found that both burnouts are much lower when over 30% of semicoke blending ratios are used. This infers that under the conditions of the semicoke blending ratios of up to 30%, the burnout does not change significantly, indicating up to 30% semicoke blending ratio should be used in the plant test for safety reasons.
the conveying gas. On this basis, it is inferred that using semicoke in the PCI blends has no significant effect on PCI supply system, thus ensuring the precondition of the stable BF operation. 3.3.2. Effect of semicoke injection on BF performance Fig. 13 compares two key BF performance indicators: gas utilisation rate and K value. The gas utilisation rate is an important indicator to measure the level of energy consumption in BFs, and the K value (namely permeability index) is an important parameter of BF stability. During the plant test, the productivity and PCI rate are remained at 10500 t·d−1 and 190 kg·tHM−1, respectively. In Fig. 13(a), the gas utilisation rate remains at a relatively stable level, ~52% on average, indicating a stable BF operation during the plant test when the semicoke blending ratio increases from 0% to 20%. Moreover, although some fluctuations could be found, there is no observed correlation between the blending ratio of semicoke and K value. This indicates the up to 20% semicoke injection in the blend does not affect the BF efficiency and stability significantly. Then the effect of semicoke injection on BF product is studied. [Si], the mass fraction of Si in hot metal (i.e. liquid iron), is an indicator of the temperature of BF that provides a guideline to adjust ironmaking operation. During the plant test, Fig. 14(a), [Si] largely stays at a level of 0.40% in the early stage but increases to a bit higher value in the last few days. This shows the semicoke can be used in PCI, while the higher [Si] is related to the operations dealing with the problem of elevated sidewall temperature. It can be seen from Fig. 14(b) that as the blending ratio of semicoke is increased, the sulphur load in the BF is reduced from 2.95 kg·t−1 to 2.82 kg·t−1, and the [S] of the hot metal is simultaneously decreased. The Sulphur content of the coal blends is lower due to the low-S content of semicoke, so that it is beneficial to reduce Sulphur load in BF, thereby reducing the [S] in the hot metal. Fig. 15 compares the effect of semicoke blending ratio on the fuel efficiency during the plant in terms of fuel ratio and corrected coke ratio, as in Table 4. Overall, fuel consumption increases during the plant test, which is also affected by changes in furnace conditions caused by elevated sidewall temperatures, changed fuel structure, newly mixed lump ore and external coke test at the same time. Therefore, it is difficult to give the quantitative relationship of the fuel ratio during the plant test. Under this circumstance, Table 4 compares the product data during the period of blending ratio of 10%–15% corresponding to the relatively stable operating conditions and raw materials conditions. When increasing the blending ratio of semicoke from 10% to 15% in this period, the fuel ratio is increased by 2.5 kg·t−1. However, the fuel ratio is affected by multiple factors, such as blast temperature, moisture, heat load, furnace temperature and [Si]. After all these comprehensive considerations, the corrected coke ratio is calculated and found to be reduced by 0.1 kg·t−1.
3.3. Plant test Based on the lab experiments and model study, detailed plan of plant test is made as showed in Fig. 11(a). That is, plant test was conducted for 49 days using one commercial BF. The blending ratio of semicoke is changed from 0%, 10%, 15%, 18% and 20%. The effects of blending ratio of semicoke are investigated in aspects of the PCI supply system and BF performance, including BF efficiency, stability and products. 3.3.1. Effect of semicoke injection on PCI supply system Fig. 11 shows the blend properties during the plant test, including volatile matter (VM%), ash% and S% (STD, the sulphur content in dry basis) under different semicoke blending ratios conditions. As increasing the blending ratio of semi-coke, the fluctuations can be found in the contents of VM, ash and sulphur where VM and S show slight decrease due to the lower VM and sulphur contents in the semicoke compared to Coal 1. That is, the addition of semicoke into the blends does not affect the blend properties significantly. The effect of semicoke blending on PCI supply system is studied first by the plant test, in terms of pressure of injecting tank and consumption of conveying gas, as shown in Fig. 12. It is indicated that the pressure of injecting tank varies with the PCI rate regardless of the blending ratio of the semicoke. The consumption of the conveying gas remains at a stable level, ~3200 m3·h−1, when increasing the blending ratio of the semicoke. This can also be reflected by the relatively stable N2 pressure in
4. Conclusions An integrated research method of “Lab experiment – Model study – Plant test (LMP)” is used to study the feasibility of injecting semicoke in a commercial BF. The metallurgical properties, in-furnace phenomena of flow and combustion, and effect of semicoke injection on BF performance are studied. The main conclusions are summarized below. (1) The lab experiments indicate that the semicoke can meet PCI requirement in terms of the grindability and composition and also can improve the safety of PCI operation considering its high ignition
8
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
9
(caption on next page)
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
Fig. 9. Effect of blending ratio of semicoke on the co-combustion in terms of gas temperature: the contour of gas temperature under blending ratio of semicoke of 0% (a) and 40% (b), the evolution of gas temperature along the particle plume (c).
Fig. 11. Effect of blending ratio of semicoke on coal blends properties: (a) VM% and ash%, (b) STD. Fig. 10. Effect of blending ratio of semicoke on the co-combustion in terms of two calculation of burnouts: (a) final burnout at the endpoint of the raceway, (b) average burnout over the entire tuyere-raceway region.
efficiency. Both final burnout and average burnout are much lower when the blending ratio is over 30%. This suggests it is better to use a blending ratio of semicoke of < 30% in the plant test. (3) The plant test confirms that the semicoke injection has a slight influence on the PCI system and BF performance when the blending ratio of semicoke is < 20%.
temperature and low explosiveness. (2) The model study indicates that the semicoke could fully replace the anthracite in the PCI process, their flow patterns are similar, including an inclined particle plume followed by a large-scale of recirculation around the raceway centre. The semicoke gives a slightly better combustion performance than the anthracite. Moreover, co-combustion of semicoke and PCI coal with the blending ratio ranging 0%–40% are investigated for identifying the safe and optimal ratio. The higher blending ratio of semicoke gives worse devolatilization performance and lower combustion
Overall, it is feasible to replace PCI coal with semicoke partially in terms of safe metallurgical properties, similar in-furnace phenomena of flow and combustion, and stable BF operation and performance. The LMP research method is an effective research method to test new raw materials and develop new technology for large scale industry including ironmaking.
10
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
Fig. 14. Effect of blending ratio of semicoke on the BF products: (a) temperature of the hot metal and Si content, (b) basicity and [S].
Fig. 12. Effect of blending ratio of semicoke on the PCI supply system in terms of (a) pressure of injecting tank, and (b) consumption of conveying gas.
Fig. 15. Effect of blending ratio of semicoke on the fuel ratio.
Fig. 13. Effect of blending ratio of semicoke on the BF performance: gas utilisation rate and K value.
11
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
Wloss
Table 4 Effect of blending ratio of semicoke on the fuel ratio and coke ratio. Blending ratio of semicoke 10% 15%
Productivity, t·d−1 10,544 10,504
PCI rate, kg·t−1
Fuel ratio, kg·t−1
189.7 192.8
487.2 489.7
Wsample Wresidue W-200 W0
Corrected coke ratio, kg·t−1 456.6 456.5
weight loss of the coal sample during heating with air isolated mass of the coal sample mass of the ashing residue weight of powders sieved by a 200-mesh sieve weight of original powders
Greek letters α α1, α2 ϕ γ ρ ρc ρfluid ρ∞
Nomenclature A1, A2 Ac As a CA Cs D Dref Dp daf. d0 de e E1, E2 Hreac [i] k kv1, kv2 k1 k2 k3 kc kd ma ma,0 mc mcoke ṁref mref,0 Mad Mc MO2 p P PA Q QM Qnet_d R rcoke rp ri T Tc Tcoke Tg Tp Tref Ts VM vi
pre-exponential factors of devolatilization reactions, s−1 pre-exponential factors in Gibb model, m s−1 K−1 constant in Gibb model, 0.0004 exponent in Gibb model, 0.75 Eddy dissipation model constant, 4 Swelling coefficient external diffusion coefficient of oxygen in Gibb model, m2 s−1 reference dynamic diffusivity in Gibb model, 1.8e−5 kg m−1 s−1 Pore diffusivity dry and ash free particle diameter at start of devolatilization the volume-equivalent-sphere diameter, μm void fraction of char particles activation energy of devolatilization reactions, K reaction heat, J kg−1 molar concentration of component i turbulent kinetic energy, m2 s−2 devolatilization rate constant, s−1 rate of external diffusion in Gibb model, s−1 rate of surface reaction rate in Gibb model, s−1 rate of internal diffusion and surface reaction in Gibb model, s−1 carbon oxidation rate in Gibb model, m s−1 diffusion rate of coke reactions in Field model, kg m−2 s−1 ash mass fraction original ash mass fraction mass of char, kg mass of coke, kg rate of change of mass of the reference material mass of the reference material at the start of devolatilization Moisture content in air-dried coal sample molecular weight of carbon molecular weight of oxygen molecule pressure, Pa local pressure, Pa atmospheric pressure, Pa Q-factor for devolatilization model heat transfer associated with mass transfer net calorific value of coal sample resistance to flow in porous media (R = 0 for cavity) coke particle radius, m particle radius, m reaction rate of gas species i, mol m−3 s−1 temperature, K activation energy in Gibb model, K coke particle temperature, K gas temperature, K particle temperature, K reference temperature in Gibb model, 293 K constant in Gibb model, 6240 K volatile matter stoichiometric coefficient of species i.
volume/internal surface area ratio in Gibb model volatile yield mechanism factor in Gibb model spread parameter in Rosin-Rammler distribution density, kg m−3 Char density Fluid density Far field concentration of reacting gas for time averaged value obtained from gas phase calculation
Subscripts c coke g p
char coke gas particle
Acknowledgements The authors gratefully acknowledge the financial support from Baosteel Group Corporation and Australian Research Council (LP150100112). References [1] Y. Shen, B. Guo, A. Yu, D. Maldonado, P. Austin, P. Zulli, Three-dimensional modelling of coal combustion in blast furnace, ISIJ Int. 48 (2008) 777–786, https:// doi.org/10.2355/isijinternational.48.777. [2] D. Wu, P. Zhou, H. Yan, P. Shi, C.Q. Zhou, Numerical investigation of the effects of size segregation on pulverized coal combustion in a blast furnace, Powder Technol. 342 (2019) 41–53, https://doi.org/10.1016/j.powtec.2018.09.067. [3] K. Luo, J. Xing, Y. Bai, J. Fan, Universal devolatilization process model for numerical simulations of coal combustion, Energy Fuel 31 (2017) 6525–6540, https:// doi.org/10.1021/acs.energyfuels.7b00970. [4] K. Dong, A. Yu, S. Liu, D. Pinson, J. Tsalapatis, Z. Zhou, Numerical investigation of burden distribution in a blast furnace, Steel Res. Int. 86 (2015) 651–661, https:// doi.org/10.1002/srin.201400360. [5] C. Wen, N. Karvounis, J.H. Walther, Y. Yan, Y. Feng, Y. Yang, An efficient approach to separate CO2 using supersonic flows for carbon capture and storage, Appl. Energy 238 (2019) 311–319, https://doi.org/10.1016/j.apenergy.2019.01.062. [6] C. Geng, S. Li, C. Yue, Y. Ma, Pyrolysis characteristics of bituminous coal, J. Energy Inst. 89 (2016) 725–730, https://doi.org/10.1016/j.joei.2015.04.004. [7] Y. Zhuo, T. Wang, C. Li, Y. Shen, Numerical study of the pyrolysis of ellipsoidal lowrank coal briquettes, Energy Fuel 32 (2018) 4189–4201, https://doi.org/10.1021/ acs.energyfuels.7b03224. [8] J. Tian, H. Ni, Y. Han, Z. Shen, Q. Wang, X. Long, Y. Zhang, J. Cao, Primary PM2.5 and trace gas emissions from residential coal combustion: assessing semi-coke briquette for emission reduction in the Beijing-Tianjin-Hebei region, China, Atmos. Environ. 191 (2018) 378–386, https://doi.org/10.1016/j.atmosenv.2018.07.031. [9] Y. Yu, M. Xu, H. Yao, D. Yu, Y. Qiao, J. Sui, X. Liu, Q. Cao, Char characteristics and particulate matter formation during Chinese bituminous coal combustion, Proc. Combust. Inst. 31 II, 2007, pp. 1947–1954, , https://doi.org/10.1016/j.proci.2006. 07.116. [10] J.M. Lee, D.W. Kim, J.S. Kim, Characteristics of co-combustion of anthracite with bituminous coal in a 200-MWe circulating fluidized bed boiler, Energy 36 (2011) 5703–5709, https://doi.org/10.1016/j.energy.2011.06.051. [11] Y. Shen, A. Yu, P. Zulli, CFD modelling and analysis of pulverized coal injection in blast furnace: an overview, Steel Res. Int. 82 (2011) 532–542, https://doi.org/10. 1002/srin.201100045. [12] P. Wang, G. Wang, J. Zhang, J.-Y. Lee, Y. Li, C. Wang, Co-combustion characteristics and kinetic study of anthracite coal and palm kernel shell char, Appl. Therm. Eng. 143 (2018) 736–745, https://doi.org/10.1016/j.applthermaleng.2018.08.009. [13] J. Yu, A. Tahmasebi, Y. Han, F. Yin, X. Li, A review on water in low rank coals: the existence, interaction with coal structure and effects on coal utilization, Fuel Process. Technol. 106 (2013) 9–20, https://doi.org/10.1016/j.fuproc.2012.09.051. [14] C. Zou, C. Ma, J. Zhao, L. Wen, C. Bai, Research status and suggestion of mid-low
12
Fuel Processing Technology 196 (2019) 106165
Z.J. Hu, et al.
[15] [16] [17] [18] [19] [20] [21]
[22] [23]
temperature pyrolysis semi-coke as the PCI fuel in blast furnace, Clean Coal Technol. 23 (2017) 57–64. P. Li, J. Zhang, R. Xu, T. Song, Representation of characteristics for modified coal simi-coke and coke used in blast furnace injection, Energy Metall. Ind. 34 (2015) 41–45. S. Yang, W. Cai, H. Zheng, J. Liang, S. Zhang, Q. Xue, Performance analysis of semicoke for blast furnace injection, Chin. J. Process. Eng. 14 (2014) 896–900. L. Zhang, W. Ren, D. Liu, W. Zhang, Z. Wang, W. Deng, Study on semi-coke used as pulverized coal for injection into blast furnace, Angang Technol. (2015) 13–17. Y. Wang, Y. Zhang, J. Zhang, Lab study of low-temperature semicokes for injection, Sci. Technol. Baotou Steel. 38 (2012) 6–8. Y. Jiao, B. Hu, Y. Gui, Feasibility analysis of semi-coke as JISCO blast furnace with coal injection, Energy Metall. Ind. 30 (2011) 20–22. D. Rangarajan, T. Shiozawa, Y. Shen, J.S. Curtis, A. Yu, Influence of operating parameters on raceway properties in a model blast furnace using a two-fluid model, Ind. Eng. Chem. Res. 53 (2014) 4983–4990, https://doi.org/10.1021/ie301936r. Y. Shen, B. Guo, S. Chew, P. Austin, A. Yu, Three-dimensional modeling of flow and thermochemical behavior in a blast furnace, Metall. Mater. Trans. B Process Metall. Mater. Process. Sci. 46 (2015) 432–448, https://doi.org/10.1007/s11663-0140204-y. T.J. Taha, A.F. Stam, K. Stam, G. Brem, CFD modeling of ash deposition for cocombustion of MBM with coal in a tangentially fired utility boiler, Fuel Process. Technol. 114 (2013) 126–134, https://doi.org/10.1016/j.fuproc.2013.03.042. J. Zhang, Q. Wang, Y. Wei, L. Zhang, Numerical modeling and experimental investigation on the use of brown coal and its beneficiated semicoke for coal blending combustion in a 600 MWe utility furnace, Energy Fuel 29 (2015) 1196–1209,
https://doi.org/10.1021/ef502287c. [24] J. Liao, A.B. Yu, Y. Shen, Modelling the injection of upgraded brown coals in an ironmaking blast furnace, Powder Technol. 314 (2017) 550–556, https://doi.org/ 10.1016/j.powtec.2016.11.005. [25] Y. Shen, B. Guo, A. Yu, P.R. Austin, P. Zulli, Three-dimensional modelling of infurnace coal/coke combustion in a blast furnace, Fuel 90 (2011) 728–738, https:// doi.org/10.1016/j.fuel.2010.08.030. [26] J. Yu, J.A. Lucas, T.F. Wall, Formation of the structure of chars during devolatilization of pulverized coal and its thermoproperties: a review, Prog. Energy Combust. Sci. 33 (2007) 135–170, https://doi.org/10.1016/j.pecs.2006.07.003. [27] M. Zaharia, V. Sahajwalla, B.-C. Kim, R. Khanna, N. Saha-Chaudhury, P. O’kane, J. Dicker, C. Skidmore, D. Knights, Recycling of rubber tires in electric arc furnace steelmaking: simultaneous combustion of metallurgical coke and rubber tyres blends, Energy Fuel 23 (2009) 2467–2474, https://doi.org/10.1021/ef8010788. [28] H. Nogami, M. Chu, J. Yagi, Multi-dimensional transient mathematical simulator of blast furnace process based on multi-fluid and kinetic theories, Comput. Chem. Eng. 29 (2005) 2438–2448, https://doi.org/10.1016/j.compchemeng.2005.05.024. [29] Y. Shen, A. Yu, Characterization of coal burnout in the raceway of an ironmaking blast furnace, Steel Res. Int. 86 (2015) 604–611, https://doi.org/10.1002/srin. 201400333. [30] M. de L.I. Gomes, E. Osório, A.C.F. Vilela, Thermal analysis evaluation of the reactivity of coal mixtures for injection in the blast furnace, Mater. Res. 9 (2006) 91–95 http://www.scielo.br/pdf/mr/v9n1/28577.pdf (accessed March 8, 2019). [31] Y. Liu, Y. Shen, Computational fluid dynamics study of biomass combustion in a simulated ironmaking blast furnace: effect of the particle shape, Energy Fuel 32 (2018) 4372–4381, https://doi.org/10.1021/acs.energyfuels.7b03150.
13