Anda di halaman 1dari 12

DE GRUYTER International Journal of Chemical Reactor Engineering.

2017; 20150068

Hongliang Zhao1,2 / Lifeng Zhang1,2 / Pan Yin1 / Sen Wang1,3

Bubble Motion and Gas-Liquid Mixing in


Metallurgical Reactor with a Top Submerged
Lance
1
School of Metallurgical and Ecological Engineering, University of Science and Technology Beijing, Beijing 100083, P.R.China,
E-mail: zhanglifeng@ustb.edu.cn
2
Beijing Key Laboratory of Rare & Precious Metals Green Recycling and Extraction of Metal, Beijing 100083,P.R.China, E-mail:
zhanglifeng@ustb.edu.cn
3
Chambishi Copper Smelter LTD, Beijing Representative Oce, Beijing 100029, P.R China

Abstract:
The multiphase uid mixing and bubble motion in a scale-down Isa smelting furnace with a top submerged
lance were investigated through cold water model experiments. The mixing time in the liquid bath was deter-
mined by a novel criterion, and the bubble formation frequency and gas-liquid contact area were also measured
by high-speed photographing. The empirical equations of mixing time (T m ) relating to mixing energy () were
m,u = 9130.5 for solid tracer and m,u = 6240.48 for liquid tracer, respectively. The mixing time trended
down with the increasing of bubble formation frequency, aected slightly by the bubbles with small size and
decreased obviously when increasing the gas-liquid contact area. The optimal conditions of gas ow rate, bath
height, lance submerged height and lance diameter to guarantee high-eiciency mixing and low-level surface
uctuation/splashing were obtained.
Keywords: gas-liquid ow, mixing, metallurgical reactor, top submerged lance
DOI: 10.1515/ijcre-2016-0139

1 Introduction
The top submerged lance (TSL) technology was invented by Commonwealth Scientic and Industrial Research
Organization (CSIRO) in Australia in early 1970s and further developed and applied by Ausmelt and Mount Isa
Mines (Floyd 1996). Besides low capital and operating costs and good environmental control, TSL outperforms
the traditional blowing process in heat and mass transfer in the molten bath as well as high gas-liquid reaction
rates(Mounsey and Robilliard 1994). Thus, TSL is widely used in nonferrous pyrometallurgy (Bakker, Nikolic,
and Mackey 2011; Fountain et al. 1993; Mounsey and Kocherginsky 1997; Mounsey and Piret 2000), such as
copper, nickel, lead and zinc, and is extended to the industries of waste processing (Hagelken 2006; Hughes
2000) and iron production (Dash and Das 2009).
With the TSL technology, gas (oxygen-enriched air) can be injected into a molten bath though a submerged
vertical lance within a cylindrical furnace. The ow behavior and mixing characteristic involved in gas-liquid
ows have signicant eects on chemical reaction, impurity separation and metal enrichment during the smelt-
ing process. The uid ow and mixing time in the vessel stirred by various gas injection types have been exten-
sively studied. Kochi et al. investigated the mixing time and swirl motion in a cylindrical water bath agitated by
horizontal air injection through an L-shaped top lance (Kochi et al. 2011). They pointed out the practical impor-
tance of swirl motion and proposed an empirical equation related to bath depth/diameter, lance exit position
and gas ow rate for calculation of mixing time. Iguchi et al. reported the eects of surface ow control on uid
ow and mixing time in a bottom blown bath and measured the mixing time under three types of surface ow,
which strongly changed the mixing (Iguchi et al. 1999). Su et al. studied the mixing behaviors in the bottom gas
blowing process with multiple tuyeres and analyzed the eects of tuyere size, number of tuyeres and gas ow
rate on mixing time in order to optimize the operating parameters (Su et al. 2010). By measuring the mixing
time in a side-blown converter using a cold water model, Ternstedt et al. found the mixing time was shortened
with an increased gas ow rate as well as a decrease of vessel diameter, and stated the eect of the bath depth
on the mixing time could be negligible (Ternstedt et al. 2010). Ooyabu et al. (2009) and Hiratsuka et al.(2008)
investigated the Bubble Formation in a rectangular vessel using a multi-hole TSL.

Lifeng Zhang is the corresponding author.


2017 by De Gruyter.

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM
Zhao et al. DE GRUYTER

In summary, the bubble motion and mixing are important parameters for gas injection. As an intensively-
used blowing technology, the use of TSL should also concern the bubble motion and mixing in the vessel. In
this study, a cold water model was established and used to simulate the multiphase ow and mixing process
(heat transfer was not considered (Ma et al. 2016; Sun et al. 2016)) in an Isa copper smelter and to optimize the
operating conditions for the top-submerged injection process.

2 Experimental
2.1 Apparatus

The experimental reactor was established geometrically 10% scale-down from the industrial Isa smelting fur-
nace of Chambishi Copper Smelter. The furnace body was made of plexiglass and the lance model was made
of aluminium alloy. And a siwiler was installed inside of the lance, and it was made of photopolymer by using
3D printing technology. The experimental reactor is shown in Figure 1.

Figure 1: Experimental reactor of Isa smelting furnace (a) Whole model (b) Lance (c) Swirler.

Air and water were used to simulate the gas-melt multiphase ow in the Isa furnace. The gas ow rate in the
experiment was calculated by considering the modied Froude Number (Fr) so as to guarantee the dynamic
similarity between the experimental reactor and the industrial furnace (Krishnapisharody and Irons 2013). Fr
is dened as follows:
u 2
= (1)
(u u )

where g and l are the gas- and liquid-phase densities (kg/m3 ), respectively; u is the gas ow velocity at
the lance outlet; H is the bath height. The gas ow velocity can be calculated as follows:

4
= (2)
2
where Q is the gas ow rate (Nm3 /s) and d is the diameter of the lance outlet (m).
Under the same Fr between experimental and practical operating conditions (Frm = Frp ), the gas ow rate
in the experiment can be calculated as follows:

u,u 0.5 u,u u,u 0.5


0.5

2
u = u


u u (3)
u,u u,u u,u u u
Brought to you by | University of South Carolina Libraries
Authenticated
Download Date | 2/24/17 2:44 PM
DE GRUYTER Zhao et al.

The parameters of dimensions, material properties and operating conditions between the Isa furnace and
the experimental reactor are listed in Table 1.

Table 1: Comparison of simulation and plant data.


Parameter Isa smelting furnace Experimental reactor
Dimension
Furnace height (m) 7.5 (Body) + 5.6 (Flue) 0.75
Furnace inner diameter, D (m) 3.84 0.384
Lance inner diameter, d (m) 0.3 0.4 0.035
Material properties
Liquid density, uu (kg/m3 ) 40004700 (matte) 998 (water)
30003700 (slag)
Gas density, uu (kg/m3 ) 1.36 (oxygen enriched air) 1.225 (air)
Operating conditions
Bath height, H (m) 1.852 0.1650.225
Gas ow rate, Q (Nm3 /h) 2500035000 3555
Lance immersion depth, h (m) 0.20.4 0.010.07

The mixing time was measured by a conductivity meter and the bubble motion was photographed by a
high-speed camera. The vertical distance of measuring point away from the bottom and the radial distance
away from the side wall are both 0.01m. Figure 2 shows a schematic of the experimental system of the cold
water model.

Figure 2: Schematic diagram of the experimental system of the cold water model.

2.2 Measuring method

The parameter of mixing time is widely adopted to evaluate the mixing eiciency in chemical or metallurgical
reactors. In most of the existing research, the mixing time was determined by measuring the conductivity,
and saline solution such as KCl or NaCl was usually used as the tracer. In an Isa smelting process, the copper
concentrate powder was dropped from a top inlet into the slag phase. The powder rst suspended on the
surfaces and then was stirred into slag phase to react with the dissolved O. Since a saline solution is usually
as the tracer, the solid motion along with the liquid ow as well as the mass transfer between solid and liquid
phases can be ignored. However, the melting and diusion of concentrate powder are important for smelting
reactions. In addition, most of the salt grains have larger specic gravity than water. Thus, it is diicult to
simultaneously simulate the solid suspension and dissolution in the cold model experiment. In this study, the
special salt powder with a hollow structure provided by SODA-LO was used as a solid tracer, and its lower
density and smaller size keeps momently oating at liquid surface and follows uid motion well.

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM
Zhao et al. DE GRUYTER

A HiSpec5 high-speed camera adopted in this work could take more than 1000 frames in a second. The
bubble characteristics including bubble formation frequency and gas-liquid contact area were investigated from
the static pictures and videos by high-speed photographing and image processing technologies.

3 Results and discussion


3.1 Comparison on mixing time criteria

In the existing research, the mixing time is generally determined by the 95 pct mixing time criterion. Here we
propose a novel criterion which determines the mixing time from the maximum and minimum of conductivity
uctuation. Figure 3 shows the schematic of the two criteria, which are dened as follows:

Figure 3: Two criteria for determination mixing time (a) within 5% of concentration change (b) within maximum and
minimum of uctuation.

Criterion 1: The uctuation of conductivity is within 5% of the dierence between initial and stable values.
Criterion 2: The conductivity is less than or equal to the maximum and minimum of uctuation when the
conductivity is stable.
Figure 4 shows the mixing time determined by two criteria against tracer addition. The mixing time deter-
mined by criterion 2 has a smaller measuring error compared with criterion 1. At low trace addition, the values
of mixing time between the two criteria are very close. With the increase of tracer addition, the mixing time
of criterion 1 is reduced while that of criterion 2 remains stable. The decrease of mixing time from criterion 1
is attributed to the enlarged conductivity variation range of uniform mixing following the increase of tracer
addition.

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM
DE GRUYTER Zhao et al.

Figure 4: Mixing time against tracer addition (H = 0.185m, Q = 45 Nm3 /h, h = 0.04m, d = 0.035m).

3.2 Mixing time and gas-liquid distribution

Figure 5(a) shows the mixing time determined by criterion 2 against gas ow rate (Q) with 1g powder tracer
and 50 ml liquid tracer (20g/L NaCl solution). In the Isa smelting furnace, the mixing power is mainly pro-
vided by the impinging of gas ow and the motion of the rising bubbles. With the increase of Q, stronger gas
impinging and bubble motion may improve the solid solution-diusivity and liquid mixing, but also aggravate
the splashing and uctuation on the liquid surfaces, which may disturb the diusion of the tracer. The mixing
time with powder tracer decreases initially due to the increase of Q and varies slightly when Q increases to
a certain value (45 Nm3 /h). When using liquid tracer, the saline solution is without dissolution and diuses
more rapidly than solids. With the increasing of Q, the liquid tracer diuses rapidly to the splashing zone due
to the strong turbulence, rather than pushing away from the splashing zone by bubble motion like solids. So
the mixing time increased largely when Q is above 45 Nm3 /h. Figure 5(b) shows the mixing time against the
bath height (H). The mixing time is prolonged with an increase of bath liquid. This is because higher H results
in larger liquid volume, where the tracer has to diuse to a larger range for uniform mixing. The liquid tracer
diuses more quickly than powder tracer at low bath height. With the increasing of H, the agitation at bottom is
weakened as well as the turbulent diusion. So the dierence between two tracers becomes less. Increasing the
lance immersion depth (h) may also enhance the gas agitation and gas-liquid mixing, leading to the decrease of
mixing time (Figure 5(c)). The mixing time with liquid tracer decreases slightly when increasing h from 0.05m
to 0.06m, which is because of the splashing to protract the time for uniform mixing and diusion. To avoid
excessive uctuation and splashing on liquid surface, h should be controlled within a reasonable range, which
is 0.040.05m in this study. Figure 5(d) shows the mixing time against lance diameter with solid and liquid trac-
ers, respectively. Increasing lance diameter, the lower velocity of gas at lance outlet leads less agitation power
for liquid mixing. So the mixing time is increased when increasing d. The mixing time for solid and liquid tracer
varies similarily with the increasing of d.

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM
Zhao et al. DE GRUYTER

Figure 5: Mixing time with dierent conditions (a) H = 0.185m, h = 0.04m, d = 0.035m (b) Q = 45 Nm3 /h, h = 0.04m, d =
0.035m (c) Q = 45 Nm3 /h, H = 0.185m, d = 0.035m (d) Q = 45 Nm3 /h, H = 0.185m, h = 0.04m

In a gas injection reactor, the mixing time is related to the stirring intensity. Several researches veried that
the mixing time, T m (s), and agitation energy, (W/m3 ), are correlated by the following equation: m = ab
where a and b are constants which depend on the mixing characteristics, the vessel dimensions and measur-
ing procedure. In most studies, the exponent b has been found to have a value between 0.25 and 0.5. (Akdogan
and Eric 1999; Eric 2008; Sinha and McNallan 1985)
Equation for the top gas injection process to evaluate the agitation energy is given in eq. 4). (Iso et al. 1988;
Koria 1992)

u 3
0.623 106 cos ( 60 )
= u 2 2 3 (4)

4

where is the angle of lance ( = 0 for vertical lance), M is the molecular weight (M = 29 for air) and n is
the nozzle number on a lance tip (n = 1 for Isa lance).
Figure 6 shows the mixing time against mixing energy with powder and liquid tracers, respectively. Under
the conditions with limited inuence of splashing, T m is decreased with the increasing of . Compared to the
solid tracer, the liquid tracer diusion rate decreases more slowly with larger due to the surface splashing and
uctuation. The mixing time for powder and liquid tracer can be calculated as follow:

m,u = 9130.5 (5)

m,u = 6240.48 (6)

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM
DE GRUYTER Zhao et al.

Figure 6: Mixing time against

3.3 Bubble characteristics

3.3.1 Bubble formation frequency

The above analysis indicates that the bubble motion is directly correlated with the mixing, uctuation and
splashing. Thus, we analyzed the bubble characteristics in detail from the high-speed photographing results.
In the top submerged injection process, bubbles are formed when the gas is blown to the liquid bath, and then
rapidly rise away from the lance, which shows periodicity. In this work, the bubble formation frequency (f B ) is
dened as the inverse of period between adjacent bubbles. A smaller f B indicates the longer residence time of
bubbles in the bath. Figure 7(a) shows the average of f B against gas ow rate. Increasing Q could generate more
injection power, and also strengthens the liquid-phase impinging, leading to the decrease of f B . The downtrend
of f B slows down with further increase of Q. Increasing bath height, f B decreases at rst, and then changes
slightly, as shown in Figure 7(b). It seems that the bath liquid high and the gas ow rate aect the bubble
formation in a very minor way. The bubble residence time is estimated within 0.075 to 0.1 s with the h of 0.04m.
By contrast, f B was inuenced largely by the lance immersion depth. With a small value of h, the stirring of gas
ow is relatively weak. Even many bubbles rise out of the bath without stirring, which forms a short circuit
air column. Therefore, the residence time of gas ow in the liquid bath is extremely short (about 0.04 s at h
= 0.01m). When increasing the lance immersion depth, the gas stirring eect is improved, the f B is reached a
stable level of about 10 to 12 Hz and the gas residence time is about 0.08 to 0.1 s (Figure 7(c)). As shown in
Figure 7(d), f B is increased with the increasing of lance diameter. The velocity at lance tip is lower with larger
d, which causes the bubble grows up weakly in vertical direction and forms rapidly.

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM
Zhao et al. DE GRUYTER

Figure 7: Bubble formation frequency with dierent conditions (a) H = 0.185m, h = 0.04m, d = 0.035m (b) Q = 45 Nm3 /h,
h = 0.04m, d = 0.035m (c) Q = 45 Nm3 /h, H = 0.185m, d = 0.035m (d) Q = 45 Nm3 /h, H = 0.185m, h = 0.04m

As show in Figure 8, the mixing time was positively correlated with f B . The higher is the f B , the shorter is
the time for gas impinging the liquid bath. The mixing energy transfers from gas phase to liquid phase when
bubble agitating the bath. So the time for uniform mixing will be prolonged with large f B .

Figure 8: Mixing time against fB

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM
DE GRUYTER Zhao et al.

3.3.2 Gas-liquid contact area

The maximum bubble supercial area was calculated by image processing (Figure 9). We chose the partial zone
of bubbles at the lance outlet. First, the bubble boundary was identied and coordinated by (xn , yn ). Then the
bubble supercial area could be determined by integrating the coordinate value when the bubble was enlarged
to the maximum size:
u
max = n n n1 (7)

Figure 9: Computing method of gas-liquid contact area

Figure 10(a) shows the maximum gas-liquid contact area against the gas ow rate. The area increases slightly
with the increase of gas ow rate. Thus, increasing the gas ow rate is not the best way to improve the smelting
reaction. A larger gas-liquid contact area is obtained at higher bath height (H) (Figure 10(b)). Although increas-
ing H could prolong the bubble residence time and enlarge the gas-liquid contact area, the mixing time in the
bath is longer. Therefore, the maximum operating height of the molten bath should be limited, and the optimal
value of H in the cold model experiment is 0.185m. Increasing the lance submerged depth (h), which is the most
important factor for smelting process, could greatly enlarge the gas-liquid contact area (Figure 10(c)). Combin-
ing the results of ow eld, bubble characteristics and mixing time, we think the lance should be controlled
within an appropriate range so as to guarantee high- eiciency smelting/separation (slag-matte) and low-level
uctuation/splashing. The best value of h in the cold model experiment is 0.040.05m. With an increases of d,
the bubble size will decrease in vertical direction while increase in radial direction. So the gas-liquid contact
area varies slightly when changing the lance diameter, as shown in Figure 10(d).

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM
Zhao et al. DE GRUYTER

Figure 10: Gas-liquid contact area with dierent conditions (a) H = 0.185m, h = 0.04m, d = 0.035m (b) Q = 45 Nm3 /h, h =
0.04m, d = 0.035m (c) Q = 45 Nm3 /h, H = 0.185m, d = 0.035m (d) Q = 45 Nm3 /h, H = 0.185m, h = 0.04m

Figure 11 shows the mixing time varying with the gas-liquid contact area, Amax . Under the conditions with
low gas-liquid contact area, the mixing time is relatively long, but no obvious regularity between them is ob-
tained. When increasing the gas-liquid contact area, the mixing time decreases rapidly and is stable to a certain
value. Thus it can be seen that the operating conditions should be controlled to maintain an appropriate size
of blowing bubble.

Figure 11: Mixing time against Amax

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM
DE GRUYTER Zhao et al.

4 Conclusions
Based on the experimental results of mixing time and bubble characteristics, we get the following conclusions:
(1) The mixing time determined by the criterion of maximum and minimum of conductivity uctuation is
more insensitive to tracer addition and bath volume compared to the traditional criterion.
(2) The mixing time using solid tracer is higher than it using liquid tracer under a certain condition and
liquid tracer is easier to be aected by surface splashing and uctuation.
(3) The mixing time is decreased with the increasing of mixing energy and increased with the increasing
of bubble formation frequency. The mixing time is slightly aected with small size bubbles, and increasing the
gas-liquid contact area will shorten the time for uniform mixing obviously.
(4) The optimal conditions to guarantee high-eiciency mixing and low-level surface uctuation/splashing
in this study are: Q = 45 Nm3 /h, H = 0.185m, d = 0.0350.04m and h = 0.040.05m.

Funding

The authors would like to acknowledge the nancial support for this research work from the National Natu-
ral Science Foundation of China, (Grant/Award Number: 51504018), the China Postdoctoral Science Founda-
tion, (Grant/Award Number: 2015M580986)and the Fundamental Research Funds for the Central Universities,
(Grant/Award Number: FRF-TP-15-069A1).

References
Akdogan, G, and R.H Eric. 1999. Model study on mixing and mass transfer in ferroalloy rening processes. Metallurgical and Materials Trans-
actions B 30: 231239.
Bakker, M.L, S Nikolic, and P.J Mackey. 2011. ISASMELT TSLapplications for nickel. Minerals Engineering 24: 610619.
Dash, R.N, and C Das. 2009. Recent developments in iron and steel making industry. Journal of Engineering Innovation and Research 1: 2033.
Eric, R.H. 2008. Physical modeling of ferroalloy/stainless steel rening reactors. Materials and Manufacturing Processes 23: 769776.
Floyd, J.M. 1996. The third decade of top submerged lance technology. In The Howard Worner International Symposium on Injection in Pyromet-
allurgy. Minerals, Metals & Minerals, edited by M Nilmani, and T Lehner, 417429. Warrendale: M Society.
Fountain, C.R, J.M.I Tuppurainen, N.R Whitworth, and J.K Wright. 1993. New developments for the copper ISASMELT process. Extractive
Metallurgy of Copper, Nickel and Cobalt 2: 14611473.
Hagelken, C. 2006. Recycling of electronic scrap at umicore precious metals rening. Acta Metallurgica Slovaca 12: 111120.
Hiratsuka, A, S Ooishi, R Tsujino, M Hashimoto, K Ohmi, and M Iguchi. 2008. Bubble formation from a two-hole nozzle attached to a top
lance. Materials Transactions 49: 26182624.
Hughes, S. 2000. Applying ausmelt technology to recover Cu, Ni, and Co from slags. Jom 52: 3033.
Iguchi, M, R Tsujino, K.I Nakamura, and M Sano. 1999. Efects of surface fllow control on flluid fllow phenomena and mixing time in a bottom
blown bath. Metallurgical and Materials Transactions B 30: 631637.
Iso, H.I, Y Ueda, T Yoshida, S Osada, S Eto, and K Arima. 1988. Rening control of top-and bottom-blowing converter by manipulating
bottom-blown gas fllow rate. Transactions of the Iron and Steel Institute of Japan 28: 372381.
Kochi, N, K Mori, Y Sasaki, and M Iguchi. 2011. Mixing time in a bath in the presence of swirl motion induced by horizontal gas injection with
an L-shaped lance. ISIJ International 51: 344349.
Koria, S.C. 1992. Studies of the bath mixing intensity in converter steelmaking processes. Canadian Metallurgical Quarterly 31: 105112.
Krishnapisharody, K, and G.A.A Irons. 2013. Critical review of the modied froude number in ladle metallurgy. Metallurgical and Materials
Transactions B 44: 14861498.
Ma, J, Y.S Sun, B.W Li, and H Chen. 2016. Spectral collocation method for radiativeconductive porous n with temperature dependent
properties. Energy Conversion and Management 111: 279288.
Mounsey, E.N, and D.M Kocherginsky. 1997. Application of ausmelt technology in the Nickel-Cobalt industry. In Proceedings of the Nickel-
Cobalt 97 international symposium, edited by C.A Levac, and R.A Beryman. Canada: Canadian Institute of Mining, Metallurgy and
Petroleum.
Mounsey, E.N, and N.L Piret. 2000. A review of ausmelt technology for lead smelting. Lead-Zinc 2000: 149170.
Mounsey, E.N, and K.R Robilliard. 1994. Sulde smelting using ausmelt technology. Jom 46: 5860.
Ooyabu, H, A Hiratsuka, R Tsujino, and M Iguchi. 2009. Frequency of bubble formation from a multi-hole nozzle attached to a top lance.
Materials Transactions 50: 18121819.
Sinha, U.P, and M.J McNallan. 1985. Mixing in ladles by vertical injection of gas and gas-particle jets a water model study. Metallurgical
and Materials Transactions B 16: 850853.
Su, C.J, J.M Chou, S.H Liu, and C.H Chiang. 2010. Efect of gas bottom blowing conditions on mixing of molten iron inside an ironmaking
smelter. Materials Transactions 51: 15941601.
Sun, Y.S, J Ma, B Li, and Z Guo. 2016. Predication of nonlinear heat transfer in a convective-radiative n with temperature-dependent prop-
erties by the collocation spectral method. Numerical Heat Transfer, Part B: Fundamentals 69: 6883.

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM
Zhao et al. DE GRUYTER

Ternstedt, P, A Tilliander, P.G Jnsson, and M Iguchi. 2010. Mixing time in a side-blown converter. ISIJ International 50: 663667.

Brought to you by | University of South Carolina Libraries


Authenticated
Download Date | 2/24/17 2:44 PM

Anda mungkin juga menyukai