Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

International Journal of Minerals, Metallurgy and Materials

Volume 25, Number 1, January 2018, Page 37


https://doi.org/10.1007/s12613-018-1544-4

Gas–liquid mass transfer and flow phenomena in the Peirce–Smith


converter: a water model study

Xing Zhao1,2), Hong-liang Zhao2,3), Li-feng Zhang2,3), and Li-qiang Yang2,4)


1) Institute of Process Engineering, Chinese Academy of Science, Beijing 100190, China
2) School of Metallurgical and Ecological Engineering, University of Science and Technology Beijing, Beijing 100083, China
3) Beijing Key Laboratory of Green Recycling and Extraction of Metal, Beijing 100083, China
4) Chambishi Copper Smelter Ltd., Beijing Representative Office, Beijing 100029, China
(Received: 18 May 2017; revised: 12 July 2017; accepted: 9 August 2017)

Abstract: A water model with a geometric similarity ratio of 1:5 was developed to investigate the gasliquid mass transfer and flow charac-
teristics in a Peirce–Smith converter. A gas mixture of CO2 and Ar was injected into a NaOH solution bath. The flow field, volumetric mass
transfer coefficient per unit volume (Ak/V; where A is the contact area between phases, V is the volume, and k is the mass transfer coeffi-
cient), and gas utilization ratio (η) were then measured at different gas flow rates and blow angles. The results showed that the flow field
could be divided into five regions, i.e., injection, strong loop, weak loop, splashing, and dead zone. Whereas the Ak/V of the bath increased
and then decreased with increasing gas flow rate, and η steadily increased. When the converter was rotated clockwise, both Ak/V and η in-
creased. However, the flow condition deteriorated when the gas flow rate and blow angle were drastically increased. Therefore, these para-
meters must be controlled to optimal conditions. In the proposed model, the optimal gas flow rate and blow angle were 7.5 m3·h−1 and 10°,
respectively.

Keywords: Peirce–Smith converter; water model; mass transfer; flow phenomena; volumetric mass transfer coefficient

1. Introduction gle, tuyere number, and tuyere diameter [7–10]. The PSC
provides three forms of energy, namely, kinetic, buoyancy,
A Peirce–Smith converter (PSC) is a cylindrical, hori- and expansion. Whereas earlier calculations of the flow
zontal reactor used to remove Fe and S from the matte field, phase, and field distribution were characterized by
phase. It is widely used in Cu, Ni, and Pt production [1]. numerical simulation, flow fields and bubble behaviors are
Mixing and mass transfer are the key parameters of the now directly characterized using high-speed photography
converter. Several research studies in the field of and particle image velocimetry [11–13].
gas–liquid multiphase systems using physical and numeri- In previous studies, the mixing time was usually selected
cal methods have been conducted. Although the PSC as the standard by which mixing intensity and efficiency
technology has been employed and studied for decades, were evaluated. Tmix is the total mixing time required to
some problems continue to persist, resulting in diminished achieve a value within ±5% of the tracer concentration at
outputs [2]. As reported in Refs. [3–6], physical and every nodal location in the system after tracer introduction
chemical processes, such as mixing, mass transfer, heat in a well-mixed bath. Stapurewicz and Themelis [14] intro-
transfer, oxygen effects, sloping, splashing, conversion, duced black dye at the top of the bath and positioned a pho-
dispersion, and accretion growth, in a PSC are affected by tocell at the bottom of the vessel to detect changes in trans-
the air injected into it. These processes are also affected by mittance in the converter. The volumetric mass transfer
several operational factors, such as bath height, blow an- coefficient per unit volume (Ak/V; where A is the contact

Corresponding author: Hong-liang Zhao E-mail: zhaohl@ustb.edu.cn


© University of Science and Technology Beijing and Springer-Verlag GmbH Germany, part of Springer Nature 2018
38 Int. J. Miner. Metall. Mater., Vol. 25, No. 1, Jan. 2018

area between phases, V is the volume, and k is the mass Q 2 g


transfer coefficient ), and utilization ratio of CO2 (η), during Fr   (1)
g l  πd 4  D 2 H
2
gas absorption can also be employed to simulate mixing and
mass transfer in the converter. However, very few studies in Thus,
this direction have been conducted.
(2)
In this work, measurements were conducted on a 1:5 scale
model of a PSC, and a high-speed camera was used to record
where Fr′ is the modified Froude number; Qm and Qp
the flow field and directly capture its characteristics. Ak/V
represent the gas flow rate in the model and prototype, re-
and η were then calculated and analyzed to determine optimal
spectively (m3·h−3); g and 1 represent the density of gas
mixing and mass transfer conditions in the PSC model.
and liquid phase, respectively (kg·m−3); g,p is the density of
2. Materials and methods the oxygen-enriched air used in practice (kg·m−3); 1,p is the
density of the molten matter (kg·m−3); g,m is the density of
2.1. Apparatus injected air (kg·m−3); 1,m is the density of water (kg·m−3); d
A partial model with three tuyeres was built, and air was is the tuyere diameter (m); H is the characteristic length; D
injected into this model through nozzles placed on its side- is the hydraulic diameter (m); and  is the similitude ratio
wall. The model could rotate along the axial direction. which equals 0.2.
Geometric, kinematic, and dynamic similarities must fulfill In this work, the experimental apparatus was a part of the
the similarity principle by considering the related dimen- prototype and consisted of three tuyeres; the prototype itself
sionless numbers. For a gas injection system, the modified was equipped with 54 tuyeres. Thus, the Fr′ of the prototype
Froude number of the industrial prototype should be equal and model was not identical (Table 1). If the Fr′ of this part
to that of the experimental model [15–16]. The modified in the prototype was calculated, the result would match that
Froude number can be calculated as of the water model.

Table 1. Physical properties of a generic industrial PSC and the proposed PSC model

Similarity Parameter Prototype Model


Diameter of converter / m 3.17 0.634
Length of converter / m 11.7 0.0912
Height of liquid / m 1.8 0.36
Geometric similarity Number of tuyere 54 3
Diameter of tuyere / m 0.05 0.01
Height of tuyere / m 0.625 0.125
Distance between tuyeres / m 0.152 0.0304
14.46 (air)
Gas flow rate / (m3·h−1) 30000
12.19 (CO2)
17.06 (air)
Tuyere gas velocity / (m·s−1) 78.64
14.38 (CO2)
Dynamic and kinematic similarity Modified Froude number 2.66 3.84
−3
Density of liquid / (kg·m ) 4500 998
1.29 (air)
Density of gas / (kg·m−3) 1.37
1.82 (CO2)

to determine the influence of the gas flow rate and blow an-
2.2. CO2 absorption in the NaOH solution and
gle on the absorption rate coefficient between the injected
high-speed video technology
gas and liquid. The blow angle is defined as the angle be-
The experimental apparatus is schematically illustrated tween the blow direction and the horizontal line. Normally,
in Fig. 1. The CO2–NaOH system was selected as the a chemical reaction rate depends on the slowest step in the
model reaction. The main objective of this investigation is reaction; in this work, mass transfer in the bath restricts CO2
X. Zhao et al., Gas–liquid mass transfer and flow phenomena in the Peirce–Smith converter: a water model study 39

absorption [17]. The reaction rate in the absorption process


can be calculated as [18]
ln[(Ce  Ct ) / (Ce  C0 )]  ( Ak / V )t (3)
where A is the area of the interface (m2); V is the volume of
the NaOH solution (m3); k represents the mass transfer coef-
ficient (m·s–1); and Ce, C0, and Ct represent the concentra-
tions of CO2 in the balanced state, in the initial state, and
during the process, respectively (mol·m−3). As reported in
Ref. [19], Ce = 0.049 mol·L−1 when the pH value varies
from 12 to 8. The relationship between the pH value and
CO2 concentration can be determined as [19]
Ct 
 K H 2 O  K1 K 2  K1c(H+ )  [c(H+ )]2 Fig. 2. Variations of pH value with time.
+ +
 c(H )  c (Na )  
 c(H+ )  2 K1 K 2  K1c (H+ ) Parameter η could be described as [20]
(4)   [V (Ct  C0 ) / t ] / [  Q / M ] (8)
+
pH   lg c(H ) (5) where ρ is the density of CO2 and equals 1.977 × 10 kg·m−3,
−9

where c represents the concentration (mol·m−3); K1 and K2 Q is the gas flow rate (m3·h−1), and M is the molecular mass
are the respective ionization constants of the following two of CO2.
reactions:
H 2 CO3  H+ +HCO- 3 (6) 3. Results and discussion
HCO3 H 
+CO32 (7) 3.1. Characterization of the flow field
−6.325 −10.329 −14
In this work, K1 = 10 , K2 = 10 , and K H 2O = 10 .
A high-speed camera (HiSpec5) that could take 500
frames per second was used in this work, as shown in Fig. 3.
After injection into the bath, the mixed gas rose against the
sidewall, leading to wave formation and splashing. Bub-
bles drove the nearby liquid to flow upward toward the
bath surface, and the liquid flowed toward the tuyeres. In
the bath, the upward flow declined, causing circular flow
in the region close to the tuyeres. This phenomenon is
known as the strong loop. Beside the strong loop was a
weak loop driven by a strong circular flow; no clear boun-
dary could be observed between these two regions. The
bath was completely covered by these two homogenous
Fig. 1. Schematic of the experimental apparatus.
circular flows. Over the bath surface, a spilling and
splashing region may be found; this region should be con-
In the CO2 absorption experiment, a gas mixture trolled because it would extend the mixing process other-
(20vol% CO2 + 80vol% Ar) was injected into the PSC wise. Flow in the region above the weak loop was fairly
model through the tuyeres installed in the model sidewall; sluggish and could hardly be observed. This region is de-
air was also injected into the model during recording of fined as the dead zone in this PSC.
the flow field. Ar was mixed with CO2 to simulate the During the decline of rising waves, a large amount of tiny
concentration of O2 in air, which is used in practical bubbles formed and moved with the circular flow, thereby
equipment. Changes in pH value were measured by three extending the interface between the gas and liquid phases
probes located in the middle of the model; these probes and accelerating the reaction. These bubbles did not
were installed 10, 100, and 200 mm (P1, P2, P3) from the aggregate or form larger bubbles that could rise to the
bottom of the model. Variations in pH value with time surface and escape from the system. The experimental
(Fig. 2) were recorded by the probes connected to the results showed that a certain amount of injected air drove
monitor of the system. the bath to flow upward and escaped through the liquid
40 Int. J. Miner. Metall. Mater., Vol. 25, No. 1, Jan. 2018

surface. Some of the air formed tiny bubbles that moved phase and bubbles, thereby promoting mass transfer and
with the flow and became involved in the reaction. These the reaction in the PSC. However, spilling and splashing
bubbles were concentrated in the strong and weak loops. were intensified by increasing the gas flow rate and
Increasing gas flow rate increased the initial velocity allowed more air to escape from the system, resulting in a
and immersion depth, as shown in Fig. 4; the mixing decline in η. Under high-temperature conditions, the
intensity and mass transfer in the bath also increased. system consumes a larger cold charge; strong circular
Increasing the gas flow rate also extended the volumes of flows in the bath may also erode the reactor and reduce its
the injection region and strong loop. This adjustment life. Therefore, the gas flow rate should be controlled to
helped homogenize the distributions of the gas–liquid optimal levels.

Fig. 3. Flow field (a) and regionalization (b) in the PSC.

Fig. 4. Flow fields obtained at various gas flow rates: (a) Q = 9 m3·h−1; (b) Q = 15 m3·h−1.

Fig. 5 shows that the PSC model clockwise increased the air stirring the bath more vigorously. Rotating the tuyeres
distance between the tuyeres and the liquid surface, thereby anticlockwise may cause the convergence of tiny bubbles, as
increasing the distance over which the air could rise. As a shown in Fig. 5.
result, the bubbles formed generated more energy to drive
3.2. Absorption rate of CO2
the bath. The formation of larger amounts of bubbles in the
bath extended the reaction and mixing processes. However, During the investigation, the gas flow rate was varied from
waves and splashing also increased as a result of the injected 6 to 12 m3·h−1 when the mixing gas was injected horizontally,
X. Zhao et al., Gas–liquid mass transfer and flow phenomena in the Peirce–Smith converter: a water model study 41

and the blow angle was varied from −15° to 15° when the gas than in other regions. After the reaction in this region reached
flow rate was kept at 12 m3·h−1. Fig. 6 shows that the volume- equilibrium, Ak/V and η could be expected to become con-
tric mass transfer and η dropped rapidly at first and then even- stant. However, as a stream of higher NaOH concentration
tually stabilized. Probes at different immersion depths indi- in the weak loop flowed into the strong loop driven by the
cated variations in Ak/V and η. When CO2 was injected into concentration gradient, mass transfer and the reaction rate
the NaOH solution, the concentration of CO2 increased. While accelerated in the strong loop; thus, Ak/V and η increased
the dispersed CO2 accelerated the reaction and mass transfer in once more. This result explains why the P3 curve shows an
the bath, the reaction tended to equilibrate after some time, initial downward trend followed by an upward trend.
causing the reaction rate to slow down and mass transfer to As shown in Fig. 7, Ak/V increased with the gas flow rate,
decrease. The characteristics of the flow field in different re- and η initially increased and then rapidly decreased. The
gions of the PSC varied significantly, so the Ak/V measured by concentration of CO2 in the NaOH solution and the mixing
the probes varied at different immersion depths. intensity improved with increasing gas flow rate. Both of
Compared with those of P1 and P2, the Ak/V and η curves these factors could accelerate the reaction rate and mass
of P3 show an upward trend during the mixing process (e.g., transfer. As more CO2 was injected into the bath, the excess
Q = 6 m3·h−1 from 140 to 200 s). This result reveals that P3 gas formed a gas core and escaped from the bath through
was located in the strong loop, a region in which mass this core instead of driving the flow field and reacting with
transfer and the desired reaction occurred more extensively the NaOH solution. As a result, η decreased.

Fig. 5. Flow fields obtained at various blow angles: (a) α = 9; (b) α = 9; (c) α = 18; (d) α = 18. Positive angle for anticlockwise
rotation and negative for anticlockwise rotation.
42 Int. J. Miner. Metall. Mater., Vol. 25, No. 1, Jan. 2018

Fig. 6. Variations of Ak/V and η (utilization ratio of CO2) with time at different gas flow rates: (a,b) Q = 6 m3·h−1; (c,d) Q = 9 m3·h−1;
(e,f) Q = 12 m3·h−1.

The blow angle was another factor that influenced mix-


ing and mass transfer in the system, as shown in Figs. 8 and
9. Ak/V and η increased with increasing blow angle from
negative to positive. A negative angle here implies clock-
wise rotation of the PSC, and a positive angle implies anti-
clockwise rotation. As the PSC was rotated clockwise, the
immersion depth of the tuyere increased and the mixing gas
could be injected deeper into the system, thereby improving
the reaction time and volume between the two phases. The
driving force acted over a longer distance and improved the
mixing process; waves and splashing in the bath were also
Fig. 7. Variations of the average Ak/V and η (utilization ratio decreased. Although Ak/V increased with increasing blow
of CO2) with gas flow rate. angle, the latter should still be controlled. The results of
X. Zhao et al., Gas–liquid mass transfer and flow phenomena in the Peirce–Smith converter: a water model study 43

Fig. 8. Variations of Ak/V and η(utilization ratio of CO2) with time at different blow angles: (a,b) α = 15; (c,d) α = 0; (e,f) α = 15.
Positive angle is for anticlockwise rotation and negative for clockwise rotation.

high-speed photography indicate that the tuyeres should be


rotated clockwise by less than 10°.

4. Conclusions

(1) The Ak/V and η were calculated by the CO2 absorp-


tion method. These indices could be employed to simulate
the industrial process of a PSC and estimate operation con-
ditions.
(2) The mass transfer coefficient increased with increas-
Fig. 9. Variations of the average Ak/V and η (utilization ratio ing gas flow rate and blow angle. Parameter η first increased
of CO2) with blow angle. Positive angle is for anticlockwise ro- and then decreased with increasing gas flow rate, and a cer-
tation and negative for clockwise rotation. tain amount of the injected gas escaped from the system in-
44 Int. J. Miner. Metall. Mater., Vol. 25, No. 1, Jan. 2018

stead of reacting with the NaOH solution in the bath. Al- [8] L. Shui, Z.X. Cui, X.D. Ma, M.A. Rhamdhani, A.V. Nguyen,
though η may be increased by increasing the blow angle, and B.J. Zhao, Understanding of bath surface wave in bottom
bubbles could form and gather in the middle of the bath, blown copper smelting furnace, Metall. Mater. Trans. B,
slowing down the reaction. 47(2016), No. 1. p. 135.
(3) Five regions could be divided in the bath: injection, [9] A. Valencia, M. Rosales-Vera, and C. Orellana, Fluid dy-
namics in a Teniente type copper converter model with one
strong loop, weak loop, splashing, and dead zone. Mixing
and two tuyeres, Adv. Mech. Eng., 5(2013), No. 1, p. 323.
and mass transfer varied in different regions, and the optim-
[10] A.K. Hasanzadeh-Lileh, M. Halali, M. Askari, and M.T.
al flow condition with the most extensive mass transfer was
Manzari, Investigation and modeling of splashing in the
observed in the strong loop. Peirce Smith converter, Chem. Prod. Process Model.,
3(2008), No. 1, p. 2.
Acknowledgements [11] T. Yamaguchi, M. Iguchi, and T. Uemura, Behavior of a
small single bubble rising in a rotating flow field, Exp. Mech.,
This work was financially supported by the National 44(2004), No. 5, p. 533.
Natural Science Foundation of China (No. 51504018), the [12] H. Bai and B.G. Thomas, Bubble formation during horizontal
China Postdoctoral Science Foundation (2015M580986), gas injection into downward flowing liquid, Metall. Mater.
and the Fundamental Research Funds for the Central Uni- Trans. B, 32(2001), No. 6, p. 1143.
versities (FRF-TP-17-038A2). [13] L.F. Zhang, S.B. Yang, K.K. Cai, J.Y. Li, X.G. Wan, and
B.G. Thomas, Investigation of fluid flow and steel cleanliness
References in the continuous casting strand, Metall. Mater. Trans. B,
38(2007), No. 1, p. 63.
[1] D.K. Chibwe, G. Akdogan, C. Aldrich, and P. Taskinen, [14] T. Stapurewicz and N.J. Themelis, Mixing and mass transfer
Characterisation of phase distribution in a Peirce-Smith con- phenomena in bottom injected gas-liquid reactors, Can. Me-
verter using water model experiments and numerical simula- tall. Q., 26(1987), No. 2, p. 123.
tion, Miner. Process. Extr. Metall., 120(2011), No. 3, p. 162. [15] L.Q. Zhang, X.H. Feng, J.J. Wang, and L. Zhou, Research on
[2] J. Vaarno, J. Pitkälä, T. Ahokainen, and A. Jokilaakso, Mod- the optimization of tundish by water model, Baosteel Tech.
elling gas injection of a Peirce–Smith converter, Appl. Math. Res., 3(2009), No. 2, p. 31.
Modell., 22(1998), No. 11, p. 907. [16] G.A. Sheikhzadeh, R.D. Yazdeli, and M.S. Kashani, Fluid
[3] D.K. Chibwe, G. Akdogan, and P. Taskinen, Numerical in- dynamics in a copper converter: an investigation on mixing
vestigation of combined top and lateral blowing in a phenomena in an experimental model, IJE Trans. A,
Peirce-Smith converter, Chem. Prod. Process Model., 29(2016), No. 1, p. 118.
8(2013), No. 2, p. 119. [17] J. Zhang, Y. Liu, T.A. Zhang, and H.L. Zhao, Study of
[4] O. Haida and J.K. Brimacombe, Physical model study of the gas-liquid absorption for water model experiment under ec-
effect of gas kinetic energy in injection refining processes, centric stirring, [in] The 16th Academic Conference on Me-
Trans. Iron Steel Inst. Jpn., 25(1985), No. 1, p. 14. tallurgy Reaction Engineering, Tangshan, 2012, p. 371.
[5] A. Valencia, R. Paredes, M. Rosales, E. Godoy, and J. Orte- [18] Y. Liu, T.A. Zhang, S. Masamichi, Q. Wang, and J.C. He,
ga, Fluid dynamics of submerged gas injection into liquid in a Study on absorption rate by eccentric mechanical stirring in
model of copper converter, Int. Commun. Heat Mass Trans- gas injection refining for iron and steel making, J. Iron Steel
fer, 31(2004), No. 1, p. 21. Res. Int., 18(2011), No. S2, p.166.
[6] H.T. Ling, F. Li, L.F. Zhang, and A.N. Conejo, Investigation [19] Y. Liu, T.A. Zhang, H.L. Zhao, S.C. Wang, Z.H. Dou, X.L.
on the effect of nozzle number on the recirculation rate and Jiang, and J.C. He, Study on absorption of CO2 bubble disin-
mixing time in the RH process using VOF+DPM model, tegration in NaOH solution, Chin. J. Process Eng., 9(2009),
Metall. Mater. Trans. B, 47(2016), No. 3, p. 1950. No. 6, p. 185.
[7] D.K. Chibwe, G. Akdogan, C. Aldrich, and R.H. Eric, CFD [20] Y. Liu, S. Masamichi, T.A. Zhang, and J.C. He, Mechanical
modelling of global mixing parameters in a Peirce-Smith stirring for gas injection refining in iron and steel making: 1.
converter with comparison to physical modelling, Chem. Intensification of bubble disintegration, [in] The 154th ISIJ
Prod. Process Model., 6(2011), No. 1, p. 22. Meeting, Gifu, 2007, p. 4.

You might also like