Thermophysics and Aeromechanics, 2011, Vol. 18, No. 3

Numerical investigation of heat and mass transfer in the reactor of a plasma-chemical plant purposed for polychlorinated biphenyls treatment* 1

1

1

2

I.А. Bassina , Yu.P. Malkov , G.A. Troshchinenko , and I.М. Zasypkin 1

FGUP Russian Scientific Center “Applied Chemistry”, St. Petersburg, Russia

2

Khristianovich Institute of Theoretical and Applied Mechanics SB RAS, Novosibirsk, Russia

E-mail: [email protected], [email protected] (Received July 1, 2010) Gas-dynamic and thermal characteristics of the gas flow in the flow part of a small-scale plasmachemical reactor for trichlorbiphenyl decomposition were calculated numerically. The investigations were performed with no regard to the chemical interaction of the components: in the calculations, the treated substance was replaced by a simulator (water steam), water steam was also used as an oxidant. Mathematical model of the flow is based on the complete system of Navier — Stokes equations in the context of axisymmetric task statement, with due regard to the gas flow swirling. The calculation results enabled us to choose the optimum geometrical parameters of the reactor design. Key words: plasma-chemical reactor, pyrolysis, decontamination, trichlorbiphenyl, gas dynamics and heat exchange, reactor design, optimal geometrical parameters.

In recent years, we have been facing the environmental problem of global level. It is related to the necessity of developing an industrial technology of safe decomposition of various halogen-organic compounds, which were fabricated before in large quantities and were not supposed to be dangerous. Today, the fabrication of these compounds is prohibited, they must be destroyed. Among them, above all, there are polychlorinated biphenyls (PCB), ozone-depleting freons, and also industrial waste containing these freons and other fluoro- and chloro-organics. To solve this task, different technologies were proposed; they involve both thermal and chemical methods. Concerning these methods, special attention should be given to the plasmachemical method, which possesses a number of advantages over the other ones. The major ones are: *

The work was supported by the Federal Agency of Science and Innovations for the Program Activity 1.5 in the framework of Federal Target Program “Investigations and Developments for the Top-Priority Goals of the Development of Scientific and Technological Complex of Russia”. © I.А. Bassina, Yu.P. Malkov, G.A. Troshchinenko, and I.М. Zasypkin, 2011

487

− the possibility to reach high mass-average temperatures (above 5 500 K) in the plasma jet; − the possibility to use almost any technological gas as a plasma medium; − the possibility to adjust the temperature mode within very wide limits. Realization of high-temperature technologies causes big problems which occur when the reactor is designed. The problems result from the presence of corrosion-active components in decomposition products. Water cooling of reactor walls is the most effective way to protect them from the corrosion, but in this case, there is a threat that a relative cold near-wall layer will appear near the walls; in this area, the process of treated substance destruction will be ineffective, and secondary high-toxic compounds will form there. It is possible to reduce this threat by means of a heat-resistant thermal-protective material used to cover the internal surface of the reactor, which allows to increase significantly the wall temperature and, hence, the near-wall layer temperature. Heat-resistant ceramics is utilized intensively in furnace aggregates which are used for treatment of various industrial and domestic waste. During the operation, the temperature in the zone of treated substances decomposition does not exceed 1 500 K, but even in this case it is needed to repair periodically (normally at least once per year) and renew the refractory material which undergoes strong thermal, chemical, and mechanical action of the gas. When plasma-chemical technologies are used, the process of treated substances decomposition is realized at higher temperatures, thus, the action of the corrosion medium on the reactor walls from a heat-resistant material will be much more intensive, which will cause more frequent repairing and renewal of defected reactors. This serious disadvantage of the heat-resistant thermal isolation stopped its application in the plasma-chemical plant reactor which was developed in RSC “Applied Chemistry”. In order to reduce the heat losses in the reactor, we had to use the method which divided the decomposition and oxidation processes in two stages [1, 2]. The first stage was realized outside the reactor and consisted of the preliminary heating of the halogen-organic substances mixed with the oxidant (water steam) up to the temperatures approaching but not exceeding the temperature of treated substances decomposition. The mixture was hence prepared for the second (main) stage; the mixture was injected into the high-temperature area in the reactor, where, due to the preliminary preparation of the halogen-organic compounds, their decomposition could be performed within the minimal residence time of the decomposition products in the reactor working channel. Thus, the process of destruction of the treated substances could be realized in a small-scale reactor with watercooled walls at relatively low heat losses. Development of this plasma-chemical plant required preliminary numerical investigations of the influence of main reactor parameters (including its dimensions) on the gas-dynamic and thermal characteristics of the gas flow, which is formed in the working channel when the PCB/ oxidant mixture is supplied into it. The purpose of this research was to choose the optimum dimensions of the reactor and modes of its operation. The numerical investigations of the gas flow characteristics in the reactor which dictate the conditions of PCB decomposition and oxidation ignored the chemical interaction of the components. This assumption was necessary because today there are no reasonable data concerning the physical and chemical properties of the PCB at the temperatures above their melting point, as well as concerning the rate constants of corresponding chemical reactions which are needed to evaluate the influence of the chemical action on 488

the mixture flow in the reactor. At the same time, it can be assumed that this influence is insignificant, since the high temperatures of the gas in the reactor result mainly from the gas heating by the plasma jet, not from the thermal effect of the chemical reactions. The PCBs contain much carbon, and a significant quantity of the oxidant is needed to oxidize it into СО2, which results in great heat consumptions for these substances decomposition and oxidation. Among various kinds of PCBs which were and still are widely used in the industry, trichlorbiphenyl (С12Н7Cl3) is the richest in carbon (55.96 %). Trichlorbiphenyl is used as an insulating liquid in industrial capacitors. Trichlorbiphenyl was chosen for our research as a treated substance because, as compared to the other PCBs, its mixture with the oxidant is heated by the most amount of heat. Due to a lack of information concerning the physical and chemical properties of trichlorbiphenyl (TCB), we had to substitute it with a stimulant – water steam, which possesses high heat capacity comparable to the heat capacity of this substance. As was mentioned before, water steam was also utilized as an oxidant. To evaluate the conditions of the gas flow generation in the reactor and to determine temperature, speed, and concentration fields, we chose the model of the flow part of the reactor working channel which is shown in Fig. 1. During the calculations of the gas flow characteristics in the reactor, it was presumed that a plasma jet is injected into it coaxially from the plasmatorch EDP-109/200М, its output channel diameter was 0.016 m. Nitrogen was used as a plasma gas. Its mass flow rate (0.01 kg/s) and mass-average temperature of the plasma jet (5 273 K) were taken to be corresponding to the rated mode in which this plasmatorch usually operated in industrial conditions. Thermal and gas-dynamic characteristics of the gas flow in the reactor were defined under the assumption that the reactor walls are water-cooled. In this connection, the temperature of the inner surface of the working channel was taken to be of 300 K. The calculations were done for three values of the PCB simulator flow rate: 0.003 kg/s, 0.005 kg/s, and 0.007 kg/s. Amount of the oxidizer needed to oxidize carbon into СО2 was founded with due regard to the excess oxidant ratio of 1.2. For the simulator flow rate of 0.003 kg/s, the water steam flow rate was 0.00604 kg/s, for the second mode (simulator flow rate 0.005 kg/s) it was 0.01007 kg/s, and for the third one (simulator flow rate 0.007 g/s) ⎯ 0.0141 kg/s. Presumably, the simulator and oxidant were pre-heated up to 773 K before the injection into the reactor. The investigations were carried out for four values of the inner channel diameter: 0.05 m, 0.07 m, 0.1 m, and 0.15 m. It was thought that the reactor length

Fig. 1. Schematic of the reactor working channel. 1 — nitrogen plasma jet, 2 — jet of treated substance mixed with oxidant, 3 — decomposition products. Diameter of nitrogen plasma jet channel d' = 0.016 m.

489

remained constant and equal to 1.2 m. It was also thought that the mixture of simulator and oxidant was swirled for the injection, and the longitudinal component of the mixture speed was equal to the tangential one. The gas pressure in the reactor was taken to be constant and equal to 105 Pa. The thermal and gas-dynamic processes in the reactor were described within the limits of the axisymmetric task statement, with due regard to the gas flow swirling based on the complete system of Reynolds equations involving the k−ε model of turbulence [3]. The system includes the continuity equation, three momentum equations, the energy equation, and the mass balance equations for individual chemical components of the mixture, namely nitrogen, simulator, and oxidant, as well as two equations which correspond to the chosen k−ε turbulence model. The systems of boundary conditions correlated with the chosen modes of the reactor operation. Uniform temperature profiles (5 273 K) and longitudinal velocity component (779.13 m/s, which correlates with the nitrogen jet flow rate of 0.01 kg/s) were assigned as the boundary conditions in the output cross section of the plasmatorch channel (cross section 1 in Fig. 1); the rest velocity components were presumably equal to zero. Under the character of the boundary conditions in the simulator/oxidant input cross section of the reactor (cross section 2 in Fig. 1), the uniform profiles of the following values were taken: the longitudinal velocity component which is defined by the value of the assigned mass flow rate of the mixture; swirling speed which is equal to the longitudinal velocity component (the radial component was presumable equal to zero); simulator and oxidant temperature (773 K) and concentrations. In the output cross section of the reactor working channel (cross section 3 in Fig. 1), the assigned boundary conditions were soft, i.e., the derivatives of the defined functions with respect to the longitudinal coordinate were presumably equal to zero. The velocity components on the solid channel walls was thought to be equal to zero, the gas mixture temperature was taken to be equal to the wall temperature (300 K). The absence of the mass flows of the gas mixture components on the walls was preconditioned, too. Numerical integration of the given system of equations was realized by the standard method on a nonuniform mesh refined near the solid walls in order to obtain the most precise resolution of the near-wall layer [4]. The results of the performed calculations are shown in the following Figures. The curves shown in Fig. 2 demonstrate the variation of the mass-average temperature of the gas flow Т over the reactor working channel length at different values of the channel diameter d, simulator (water steam) flow rate Gsim and oxidant flow rate (water steam, too) GH 2 O . Fig. 2. Variation of the mass-average temperature along the reactor channel length. Simulator flow rate Gsim = 0.003 (1), 0.005 (2), 0.007 (3) kg/s. Oxidant flow rate G H 0 = 2

= 0.00604 (1), 0.01007 (2), 0.0141 (3) kg/s. Diameter of reactor working channel d = 0.05 (dotted lines), 0.07 (solid lines), 0.10 (dash-dot), 0.15 m (double dash-dot).

490

The calculation results show that, as was expected, the value of the simulator/oxidant mixture flow rate takes the strongest effect on the gas temperature in the reactor in the chosen operation mode of the plasmatorch. In every case, we observe the significant temperature reduction over the reactor length, which is caused by the heat losses into the reactor water-cooled wall, the temperature of which was taken as 300 K. It is also seen from Fig. 2 that the mass-average temperature of the gas in the reactor slightly depends on the channel diameter. The reason is that as the diameter, and, hence the internal area of the channel surface, increases, specific heat fluxes in the channel wall decrease simultaneously, which results from the gas flow speed reduction. The area extension is almost fully balanced by the specific heat fluxes decrease. According to the calculation results presented in Fig. 2, in the chosen reactor length of 1.2 m, the only mode corresponding to Gsim = 0.003 kg/s and GH 2 O = 0.00604 kg/s, satisfies the condition, at which, when the system is utilized for TCB destruction, the generation of secondary acute toxic substances is unlikely. In the other two modes, the probability of appearance of the secondary toxic compounds is higher because in these modes, the gas flow temperature in the output section of the reactor is below 1 500 K. Since the extension of the working channel diameter from 0.05 m to 0.15 m slightly influences the mass-average temperatures of the gas, it could be presumed that the reactor with maximum possible diameter should be utilized, because in this case the residence time of the decomposition products in the reactor increases significantly, which should stimulate the more effective process of treated substances decomposition and oxidation. However, according to the calculations, this hypothesis fails, since the working channel diameter extension changes the temperature fields in such a way that the high-temperature area decreases. To illustrate this fact, Fig. 3 shows the temperature contours in the working channels with the diameters of 0.07 m and 0.15 m, which are formed in the mode with Gsim = 0.003 kg/s and GH 2O = 0.00604 kg/s. When comparing the fields in the Figure, we see that almost all the gas passing through the reactor with d = 0.07 m, in its leading section, is heated up to at least 1 900 K, whereas in the reactor with d = 0.15 m, only a minor part of the gas flow attains this temperature, the temperature of the rest gas does not exceed 1 700 K. Taking this into account, the optimum dimensions of the reactor should be chosen with due regard to the fact that the effectiveness of the treated substances decomposition depends significantly not only on the gas mixture residence time in the working channel, but also on the temperature level in the region wherein this process predominantly realizes. Considered results of the study of the processes in the reactor were obtained under the assumption of the uniform distribution of the temperature profile of the nitrogen plasma jet which comes into the reactor from the plasmatorch. In reality, however, this profile differs significantly from the taken one. This difference is especially evident when the plasmatorches with vortex stabilization of the arc discharge are utilized; the plasmatorch EDP-109/200M installed on the plasma-chemical plant belongs to this category. In the plasmatorch of this design, the supplied gas is heated mainly due to flow around the arc discharge, which is located along the axis of the plasmatorch working channel. In this context, the central area of the plasma jet, which is under the direct action of the arc discharge, has the temperature much higher than in periphery layers which did not pass near or through the arc discharge. Because of the high gas 491

Fig. 3. Gas temperature contours in the symmetry plane of the reactor channel. Simulator flow rate Gsim = 0.003 kg/s, oxidant flow rate GH 0 = 0.00604 kg/s, channel diameter d = 0.07 (а), 2

0.15 (b) m.

density in the jet periphery, the essential part of the plasma gas flow rate has in this case the lower temperature than the gas passing through the central area. Typical distribution of the gas temperature on the output of the plasmatorch with the vortex stabilization of the arc discharge is given in [5, 6]. This distribution is shown in Fig. 4. The Figure shows the dependence of the ratio Т/Тmax, where Т is the local temperature value, and Тmax is its maximum value on the relative coordinate R/R0, where R is the radial coordinate, and R0 is the radius of the plasmatorch output channel. To determine this dependence, special experiments were carried out; during them, the temperature was measured by the gas-dynamic method, with the aid of a rake designed for this purpose. As is seen from Fig. 4, this temperature profile at the plasmatorch channel input differs significantly from the uniform pro-

Fig. 4. Temperature profile in the output cross section of the plasmatorch with the vortex discharge stabilization used for the calculations (nodes show experimental data). 492

Fig. 5. Gas temperature contours in the symmetry plane of the reactor channel. Channel diameter d = 0.07 m, simulator flow rate Gsim = 0.003 kg/s, oxidant (water steam) flow rate GH 0 = 0.00604 kg/s. Speed profile of the plasma jet: а ⎯ uniform, b — experimental. 2

file of the calculations. For this reason, we had to evaluate the effect of this difference on the thermal and gas-dynamic characteristics of the gas flow in the reactor. To do this, we chose the mode considered before which corresponded to the mass flow rate of the PCB simulator Gsim = 0.003 kg/s and oxidant (water steam) flow rate GH2O = 0.00604 kg/s. The calculations were carried out for the reactor with d = 0.07 m. Figure 5 presents the gas temperature contours in the symmetry plane of the reactor channel for the uniform profile of the plasma jet speed and for the profile shown 493

in Fig. 4. Comparison of these contours proves that the effect of essentially different temperature fields of the plasma jet, as the mass-average temperature value is equal (5 273 K), on the temperature distribution in the reactor working channel is not large. Analysis of the above numerical results enables us to conclude that the reactor with the working channel diameter of 0.07 m is the best to satisfy the optimum conditions of the heat and mass transfer process. In the inlet section of this channel, a hightemperature area forms; this area propagates over the whole width of this section (see Fig. 3, а), and, as the channel diameter extends, this area contracts dramatically (see Fig. 3, b). The contraction of the working channel diameter to 0.05 m, according to the calculations, takes almost no effect on the high-temperature area extension over the channel length and radius, but in this case, the residence time of the decomposition products in the reactor decreases significantly. When the optimum diameter of the working channel is chosen, the reactor input part was designed and applied in the experimental plasma-chemical plant with the plasmatorch EDP-109/200М. This structure is presented in Fig. 6. As was demonstrated by the experiments performed on the experimental plant, the input part of the reactor must be electrically insulated from the earthed units of the plant, among which is the anode unit of the plasmatorch connected to the input part. Otherwise there is a big threat of the electric discharge breakdown onto the wall of the reactor input part, which causes local overheating, melting, and rupture. The proposed design of the reactor input part satisfies this condition. However, to cancel this threat completely, not only electrical insulation is needed, but also the proper width of the channel through which the treated substance/oxidant mixture is supplied into the reactor input part because the threat of breakdown remains actual at the small width of the channel. On the other hand, as the channel width increases, the speed of mixture injection into the reactor decreases, and this must affect the thermal and gas-dynamic characteristics of gas flow in the reactor. To evaluate the influence of this size on the gas temperature distribution in the reactor, the flow characteristics were calculated for two values of the channel width: 0.0034 m

Fig. 6. Input part of the reactor. α — width of the channel for treated substance/oxidant mixture supply. Working channel diameter d = 0.07 m, length L = 0.26 m. 1 — nitrogen plasma jet supply, 2 — PCB/oxidant supply, 3 — electric insulator, 4 — plasmatorch anode.

494

Fig. 7. Gas temperature contours in the symmetry plane of the reactor channel. Channel diameter d = 0.07 m, simulator flow rate Gsim = 0.003 kg/s, oxidant (water steam) flow rate GH 0 = 0.00604 kg/s, α = 0.0034 (а), 0.015 (b) m. 2

and 0.015 m. In the calculations, the mode corresponded to the mass flow rate of the simulator Gsim = 0.003 kg/s and oxidant (water steam) flow rate GH 2 O = 0.00604 kg/s. Figure 7 presents the results of these calculations. It can be noted that such an essential change of the input channel width does not take any significant influence on the thermal characteristics of the gas flow. That is why the channel width of 0.01 m 495

was chosen; with this width, the electric discharge breakdown is impossible. The proposed design can be considered as completely protected from the destructive effect of the electric discharge. It is impossible to determine precisely the optimal length of the reactor fit for the decomposition and oxidation of any type of halogen-organic compounds on the only base of the analysis of the results of the considered numerical study of the chemically frozen flow; the reason is that the length providing the needed residence time of the decomposition products in the reactor may depend on the treated substance kind and character of the chemical reactions between components. However, according to further experimental investigations of the PCB treatment performed in RSC “Applied Chemistry” on the plant with the reactor which working channel diameter was 0.07 m and length 1.2 m, these dimensions resulting from the numerical study turned out to be optimal. Finally note that during the development of the plasma-chemical plant for the treatment of various halogen-organic compounds, the optimum reactor design is one of the most complicated challenges. Experimental researches which are needed for this task solution require big material consumptions and special safety measures. In this connection, an alternative, cheaper and faster way was used to solve the problem; it consisted of the numerical investigations. Analysis of the obtained results enables us to make the following general conclusions. 1. The value of the gas mass-average temperature in the reactor and its variation along the reactor length in the chosen modes of the plasmatorch operation, mass flow rate, and content of the treated substance/oxidant mixture is feasibly independent of the diameter of the reactor working channel within the studied range of its sizes (from 0.05 to 0.15 m). 2. In order to extend the residence time of the gas mixture in the reactor, it is desirable to utilize the working channels of big diameter, but the high-temperature area in the reactor input section (Т ≥ 1 900 K) contracts as the diameter is increased. Thus, the optimum channel diameter which is recommended for the reactor, is 0.07 m. Almost all gas passing through the reactor is heated up to 1 900 K in the leading section of the reactor. 3. The width of the channel for the treated substance/oxidant mixture supply into the reactor takes almost no effect on the gas-dynamic and thermal characteristics of the reactor within the studied range of values (from 0.0034 to 0.015 m). REFERENCES 1. Patent of the Russian Federation 2105928 MPK6 F23G7/00. Plasma-chemical method of treatment of gaseous and liquid halogen-organic waste / Davidian А.А. et al.; applicant and patent holder: Russian Scientific Center “Applied Chemistry”, No. 96100669/03, application from 10.01.1996; published on 27.02.1998. 2. Patent of the Russian Federation MPK6 F23G7/00. Plasma-chemical method of treatment of gaseous and liquid halogen-organic substances and waste containing them / Malkov Yu.P. et al.; applicant and patent holder: FGUP Russian Scientific Center “Applied Chemistry”, No. 2002115550/032002115550, application from 10.06.2002; published on 10.02.2004. 3. Yu.V. Lapin and M.Kh. Strelets, Internal Flows of Gas Mixtures, Nauka, Moscow, 1989. 4. S.B. Koleshko, Yu.V. Lapin, D.A. Nikulin, M.Kh. Strelets, and Yu.S. Chumakov, Physical and mathematical bases of the numerical simulation of hydro- and aerodynamics and thermal exchange. A complex of special courses for the users of application program package COOLIT. OKDAYL, St. Petersburg, 2005. 5. А.S. Koroteev, V.M. Mironov, and Yu.S. Svirchuk, Plasmatorches: Designs, Characteristics, Calculations, Mashinostroenie, Moscow, 1993. 6. M.F. Zhukov, I.M. Zasypkin, A.N. Timoshevsky et al., Electric-arc Generators of Thermal Plasma, Nauka, Sib. Predpr. RAN, Novosibirsk, 1999.

496