Numerical Simulation of the Acenaphthylene Compound in an Atmospheric Plasma Reactor to Treat Cooking Fumes

Acenaphthylene (chemical formula of C12H8, Acpy), also known as cyclopenta[de]naphthalene, is a polycyclic aromatic hydrocarbon (PAH) with 3 aromatic rings. The Acpy compound is a PAH on the Environmental Protection Agency’s (EPA’s) priority pollutant list. This study presents a simulation of an atmospheric plasma reactor (APR) using a method based on computational fluid dynamics (CFD). A commercial CFD tool was used to solve mass, momentum, and energy equations. The commercial FLUENT code was then used to simulate the Acpy compound using a 3D APR to treat the cooking fume exhaust emitted from a restaurant kitchen. The simulations in this study adopted the APR size and operating parameters from a self-designed atmospheric plasma reactor in a previous study (NSC95-2221-E020-021). An in-house reduced chemical mechanism was coupled with the CFD code for improved computational runtime. The reactivity of the system was considered with the RNG k-ε turbulence model and the classical Eddy Dissipation Concept combustion model. The simulation results were compared with the experimental temperature measurement and the removal efficiency of Acpy. The simulated average removal efficiency of Acpy was 61.3%.


INTRODUCTION
Polycyclic aromatic hydrocarbons (PAHs) are organic pollutants widely distributed in the environment (Yang et al., 1998;Li et al., 2003;Lai et al., 2010;Wu et al., 2010;Bari et al., 2011;Masih et al., 2012).As a class, PAHs are widely produced by natural sources, such as forest fires and volcanoes, and from humanmade sources, including the burning of wood in homes, cooking fume, automobile and truck emissions, tobacco smoke, and the production of coal tar and coal tar products (ATSDR, 1990;Li et al., 2003;Chang et al., 2011;Tsai et al., 2011;Liu et al., 2012).Acenaphthylene, also known as cyclopenta [de]naphthalene, is a polycyclic aromatic hydrocarbon (PAH) with 3 aromatic rings.The United States Environmental Protection Agency (U.S. EPA) has ranked acenaphthylene the 16 priority air pollutant among polycyclic aromatic hydrocarbons (U.S. EPA, 1990).Therefore, the emission of acenaphthylene from cooking behaviour/activity on both air qualities and health issues is of critical importance.
A number of PAHs cause tumors in laboratory animals that are exposed to PAHs through their food, from breathing contaminated air, and when a PAH is applied to the skin.A study on the effects of Acpy on kidney, liver, blood, the reproductive system, and the lungs was conducted by oral subchronic toxicity tests on animals.After 40 d of continuous oral Acpy administration at a dosage of 0.6 g/d, the kidney function of small rats was affected by changes in distal blood (Knobloch et al., 1969).The lung function of large rats was affected after taking 176 mg/kg/d for 2 months (Rotenberg and Mashbits, 1965).Oral toxicity tests conducted for 90 d showed a correlation between the mortality increase of female rats and the dose of Acpy, after orally administering male and female rats with 100, 200, or 400 mg/kg of Acpy each day.Other effects included the volume ratio decrease of hemes and corpuscles, the decrease in soterocytes (male) and leukocytes (female), increase in liver quantity, hypertrophy of hepatocytes, kidney disease, and other relevant pathological changes.The administration of Acpy also reduced ovary quality, decreased the activity of ovaries and wombs, and resulted in atretic corpora lutea (U.S. EPA, 1989).The erythrocyte quantity of male rats and the liver quality of female rats both increased.In addition, female rats exhibited hypertrophy of hepatocytes and other relevant pathological changes when the dose was 200 mg/kg/d.The results of Acpy oral acute toxicity testing on animals showed that the LD 50 (median lethal dose) of large rats is 3 g/kg, and that of small rats is 1.76-2.2g/kg (Rotenberg and Mashbits, 1965;Knobloch et al., 1969).However, toxicity data indicate absorption, and data from structurally related PAHs (primarily benzo[a]pyrene) suggest that acenaphthylene is absorbed readily from the gastrointestinal tract, lungs, and skin (U.S. EPA, 1991).
The application of plasma technology in handling environmental pollutants primarily involves pollutants in the solid, liquid, and gaseous phase (Hsieh et al., 1998;Moustakasa et al., 2005;Magureanu et al., 2010).With its high-energy gas molecules, plasma can decompose tenacious, toxic molecules and harmful organic pollutants, and react with chemical compounds such as MTBE, BTEX, and other PAHs in automobile exhaust.The advantages of plasmadecomposing toxic substances at high temperature can decrease the emissions concentration of organic compounds.However, plasma technology requires high electricity consumption during operation, thus increasing its cost.Few studies have used an atmospheric pressure plasma reactor to manage cooking fumes because of the difficulty in sampling and limits of obtaining test data (Chang et al., 2011).Therefore, there is a lack of understanding regarding the actual burning conditions inside a plasma reactor when using the plasma technology to treat PAHs produced during the cooking process.
In recent years, computational fluid dynamics (CFD) is widely applied to the design of various types of engine combustors, refrigeration units, and air conditioning systems.This technique is also useful for predicting the diffusion of air and water pollutants.It is widely applied in academia and is a useful tool for industrial solutions to solve engineering problems.Many reports show that the joint application of process simulation and CFD is a helpful tool in the design and optimization of complex and innovative concepts in chemical engineering.Miltner et al. (2007) showed that process simulation and CFD using validated models with a sound physical basis can significantly reduce the development costs and the time-to-market of innovative chemical engineering concepts.
The numerical investigation of the corona plasma region in negative wire-to-duct corona discharge and a novel computation method calculating the plasma region thickness was presented with the plasma region model (Kim et al., 2010).The proposed method consists of a state-of-the-art tool employing the CFD technique to improve indoor pollution and ventilation methods.The proposed method focuses on the optimum sizing and sitting of ventilation schemes and the provision of practical suggestions as a part of an indoor pollution management system (Panagopoulos et al., 2011).
This study investigates the combustion in a plasma reactor used to treat Acpy emission from the cooking process by applying the CFD method.With the appropriate calculation models, the CFD results can be used to optimize both the plasma reactor geometry and the combustion parameters regarding the improvement of efficiency and the decomposition of PAH emissions in existing systems.Environmental engineers are typically confronted with 2 difficulties.The first is the simulation of a plasma combustion system using CFD modeling, which remains a challenging domain.The second is the relatively little progress, until recently, in the application of plasma technology to treat PAH emissions.The conceptual framework and working methods of treating the Acpy compound in an atmospheric plasma reactor remain unclear.Therefore, this study uses the numerical value simulation methods to discuss reactions by applying CFD numerical simulation, and determines the temperature of gas, the flow field and rate, and the Acpy removal efficiency of the plasma system.The solution proposed in this paper is to adopt simplified plasma torch calculations instead of detailed, complex calculations because plasma torch calculations can be generated rapidly and provide information on the removal efficiency of air toxins.This rapid assessment allows immediate modification of the APCD operation system in situ.In the future, these results can assist engineers in improving plasma systems or achieve energy savings during experimental plasma reactor operation.

METHODS
Fig. 1 shows the simulation process.The simulations in this study adopted the atmospheric plasma reactor size and operating parameters from the self-designed atmospheric plasma reactor (APR) in our previous study (Hsieh, 2007;Chang et al., 2011).Table 1 shows the system parameters for simulation.Fig. 2 shows the full-scale reactor.The parameter setting value was based on practical experimental data.We discuss the distributed concentration in the flow field of the reactor and the removal efficiency of Acpy compounds.The results can serve as a reference for decisionmaking in engineering applications.

Mesh Independent Study
A mesh was generated using the commercial gridgeneration tool GAMBIT.The 3D geometry was meshed using the mesh generator GAMBIT 2.4.6 (GAMBIT, 2007).Fig. 3 shows the 3D meshed model.The mesh was a quadrilateral mesh.To seek an appropriate grid system, this study tests and compares 3 different grid systems of 493,204, 980,000, and 3,460,000 elements.Table 2 shows the 3 grid systems and the distributions of temperature.The outlet temperature discrepancies are small.Therefore, this study adopts the grid system of 493,204 elements.Fig. 4 shows the grid point distribution (XX plane) for treating Acpy using an atmospheric plasma reactor.

Atmospheric Plasma Reactor Material Boundary Conditions
The simulations in this study adopted the following boundary settings: (1) Inlet: Initial calculation was conducted to simulate the process of the Acpy compounds of fugitive cooking fumes entering the torch reaction area under atmospheric pressure.The boundary conditions of the velocity inlet were obtained here.The inlet velocity was 6.23 m/s.(2) Outlet: At the outlet, the process of cooking fumes entering the airspace after passing through the torch reaction area under atmospheric pressure plasma was simulated, and the pressure outlet was applied.The outlet pressure was 1 atm.
(3) The duct vent was composed of stainless steel.Therefore, "stainless steel" was chosen as the fluent parameter, and the pipe thickness was set to 3 mm.

Numerical Method
To predict the outlet temperature and mass fraction of Acpy in the previous models, the commercial software ANSYS FLUENT 12.1 was used to solve the governing equations and the associated boundary conditions.In the software, the momentum, the energy, the species transport with the Eddy Dissipation Concept (EDC) for turbulent combustion was used.First-order upwind discretization was used for both momentum and energy equations with the semiimplicit method for pressure-linked solutions (SIMPLE) algorithm for pressure-velocity coupling.Although a second-order scheme would yield better accuracy, it is not explored because of computational time constraints.The resulting solver is based on the finite-volume method and employs a pressure-based solver with double precision.When the relative residues of all physical and chemical scales were smaller than 10 -6 , convergence was presumed to have reached and the calculations were terminated.

RESULTS AND DISCUSSION
This study simulates the inlet of the plasma reactor as the initial operation setting level based on the concentration of experimental data (Li et al., 2003;Chang et al., 2011).Thus, it is similar to the emission concentration of PAHs in cooking fumes.The simulated Acpy concentration is used for calculation with 2.18 × 10 -9 (mass fraction).Table 3 shows the molecular structures of the species used in this study.Because the practical combustion reaction is a complex chain reaction, this study adopts the simplified chemical reactions shown in Table 4.
To define the different relative distances used in CFD simulations, Fig. 4 uses a simple diagram to represent the different target position in the plasma reactor.Assuming that a box (ABCD volume) is the main plasma reaction of activity, the shading in the rectangle (i.e., the ABCD section) designates the plasma reaction zone in the reactor.The reactant species enter the AB-section, pass through the plasma torch, and continue through the BC-section to the exit vent.We compare the simulated results with the experimental data from a real plasma reactor.The measured volume flows, temperatures, and acenaphthylene compounds in flue gas were averaged over a quasi-stationary period.

Temperature Profiles
Figs. 5(a) and 5(b) show a comparison of 2 curves.The averaged levels of temperature varied in the vertical direction (Y-axis).The curve in Fig. 5(a) represents the averaged levels of temperature variation at different positions on the Y-axis (in the XY-plane), 7.35 cm from the torch position.The curve in Fig. 5(b) represents the average temperature variation at a position 37.35 cm from the torch (in the XY plane).These simulation results show that the variation trend of the curve at 7.35 cm is considerably greater than that of the curve at 37.35 cm.
When inputting plasma power at 24 A (Fig. 6), the highest temperature area in the flow field is in the central position of the plasma torch, where the simulated temperature reaches 2243 K. Fig. 6(a) shows the temperature distribution in the x direction (horizontal plane) from x = -2.0cm (left of the torch) to x = +37.5 cm (right of the torch) after 0.1 s simulation.The temperature distribution varies in the plasma torch zone, which can be explained by the efficient turbulent heat transfer of the torch.Similarly, there is no significant temperature gradient in the bulk zone away from the position of x = +18 cm.Moreover, low-temperature zones (less than 400 K) exist in the upper region of the duct, where temperatures are the lowest and no gradients exist.This can be explained by the effect of the less turbulent flow in this part of the duct and by the cooling wall temperature on the steel duct boundary.The temperature difference between the center zone of the torch and near the wall of the boundary zones is approximately 2000 K. On a practical scale, the temperature gradient between the bulk and the boundary zone might be expected to be accepted in this study.Figs. 6(b) and 6(c) show the temperature profiles in the vertical plane at 7.35 cm and 37.35 cm after 0.1 s numerical simulation, respectively.The upper bulk region has a uniform temperature profile with no temperature gradients.Because of diffusion, the temperature of the hottest mixture fractions decreases when the X-distance from the center of the plasma torch increases.After 0.1 s, mixtures nearing stoichiometry reach the steady stage (after finishing the main decomposition of Acpy) at X = 7.35 cm, whereas these mixtures are in the low torch regime on the left side of the profile (at positions from Y = -7 to Y = -1 cm) at X = 3.75 cm (Fig. 6(b)).The most reactive mixture fraction is not the center of the torch area (i.e., the chemical reactions involved are supposed to be the most active).
However, the temperature gradient within the plasma torch zone is larger than that of the post-torch zone.Temperature differences of approximately 600 K and 350 K appear between the torch area and temperature near the upper areas at 7.35 cm and 37.35 cm, respectively.This is the result of air stream shifting from the torch nozzle.

Concentration Profiles
The accuracy of CFD simulations depends on many factors.For example, the effects of appropriate geometry boundaries, the inlet turbulence parameters, and the selected turbulence models (such as the low-Reynolds number k-ε model, renormalization group k-ε model, or Shear Stress Transport k-w model) are important (Rong et al., 2011).Based on the characteristics of a plasma torch, particular attention should be paid to the boundaries chosen for the plasma-reaction domain.This is because both the position of the domain boundaries and the type of condition imposed (e.g., pressure node, velocity node, and non-reflecting surface) have significant effects on the results of modal analysis.These limits are likely to be the same limits of the computational domain in the CFD simulation, which is performed in the successive steps of the decomposition of acenaphthylene compound in the APR.The contour plots in Fig. 7 show the predicted levels of mass fraction of Acpy, with the progress variable on the horizontal axis (xdirection) or the vertical symmetry plane near the plasma torch.

• Acpy (C 12 H 8 )
In the plasma-torch zone (i.e., the ABCD-box in Fig. 4), temperature distribution gradients affect the concentration of Acpy profiles within a cross-section of the duct.High thermal energy is the dominant factor in the decomposition process, affecting both the gas kinetics and the final product performance.A comparison of temperature distribution shows that higher temperatures yield greater removal efficiency.Fig. 7 shows the concentration profile of Acpy (mass fraction %) within the reactor.After calculation, the residence time of reaction is approximately 0.12 s.According to this profile, Acpy was decomposed and destructed through the reaction zone of the plasma torch.In general, when the distance in the X direction increases (i.e., moving away from the plasma torch), the concentration of the reactant mixture decreases.This phenomenon is due to the transport of temperature and reactive species from hottest mixtures to coldest mixtures, which increases with the dissipation of high energy.
We sampled and measured the concentration of Acpy in the inlet of the duct at 2.18 × 10 -9 (mass fraction %).Based on this feeding level, the simulated concentration of Acpy in the duct ranged from 8.49 × 10 -10 to 1.5 × 10 -9 (mass fraction %), and the simulated removal efficiencies of Acpy ranged from 31.1% to 61.05%, respectively.Comparing the average simulated removal efficiency (approximately 61.05%) of Acpy to the actual experimental data (averaging approximately 65.2%) measured in the reactor shows that the CFD simulations slightly underestimated removal efficiency.Furthermore, faster mixing and a higher local temperature characterize the decomposed configuration the correlation of both temperature and velocity related to the lower section at X = 7.35 cm (as shown in Figs.6(b) and 7(b)) showed that the removal decomposition efficiency might differ only slightly from Y = -7.35 to + 7.35 cm (as Fig. 7(b) shows) because the plasma torch had a small injection angle (less than 5°).
For flame temperatures < 1550 K, the key reaction for one ring molecule formation is H 2 CCCH + C 2 H 2 = C 5 H 5 .At this temperature, the propargyl radical reacts with C 2 H 2 to produce C 5 H 5 (Slavinskaya and Frank, 2009).Cyclopentadienyl further forms A1 by reacting with methyl radicals in the main reaction zone and in the post-flame zone.Zhong and Bozzelli (1997) proposed a submechanism of important cyclopentadienyl (C 5 H 6 ) radical reactions and tested it in an elementary model for benzene combustion.

• C 6 Mechanism
The common pathways considered in most mechanisms involve propargyl (C 3 H 3 ) recombination and the addition of C 2 Hx on C 4 Hy molecules (Zhang et al., 2009).Benzene is primarily produced by propargyl (C 3 H 3 ) recombination (R#30) (Wang and Frenklach, 1997).Sharma and Green (2009) studied a new pathway (R#31) for the formation of benzene through the addition of CH 3 on cyclopentadienyl (C 5 H 5 ) radicals (Sharma and Green, 2009).The most important steps involved in this PAH formation process require phenyl (C 6 H 5 ) and naphthyl (C 10 H 7 ) to be oxidized by O 2 (Marinov et al., 1998) .Ethynylbenzene (C 8 H 6 ) is a key component in the subsequent growth of PAH.The hydrogen abstraction/acetylene addition (HACA) mechanism, through the phenyl radical (C 6 H 5 ) addition to acetylene, is the main route to form ethynylbenzene (C 8 H 6 ).This reaction pathway explains more than 75% of benzene depletion.Ethynylbenzene is also formed (approximately 20%-25%) by the addition of the C 4 H 5 radical to C 4 H 2 (Granata et al., 2002).
The radicals H and CH 3 formed from the reactions R#76 and R#77 can attack toluene, accelerating the consumption of toluene.During the benzene decomposition period (R#84, R#85), a pathway beginning with the formation of biphenyl is followed by the sequential addition of acetylene, first to Phenanthrene (C 14 H 10 ), and then to Pyrene (Richter and Howard, 2000).

• C 10 Mechanism
One of the important pathways (R#40) for the formation of naphthalene is through the self-combination of cyclopentadienyl radicals (C 5 H 5 ) (Dean, 1990;Kislov and Mebel, 2007).The key step in the naphthalene production process is phenyl oxidation by O 2 .This step produces a phenoxy radical that decomposes to cC 5 H 5 + CO.The cyclopentadienyl radicals self-combine and then undergo H-atom shifts and 2 H-atom ejections, leading to naphthalene production (Marinov et al., 1998).The reaction (R#94) of 1-naphthyl with acetylene produces acenaphthylene (A2R5, C 12 H 8 ) (Richter et al., 1999).
A reactive model based on CFD code must be just detailed enough to predict the effects of Acpy oxidation on the flow field.Thus, more computational resources can be dedicated to simulating the dynamics of flow, the concentration of the main species, and the heat and radioactive exchange.
The experiments and CFD comparisons in this study show satisfactory agreement on Acpy removal efficiency, especially in the horizontal axis (x-direction).However, adding more reaction mechanisms to the simulation may be necessary to accommodate the combustion reactions of complex PAHs in a plasma reactor.

CONCLUDING REMARKS
This study presents numerical simulations of decomposing acenaphthylene in an atmospheric plasma reactor to treat cooking fumes.We also compared experimental results with the simulation data.The online measured temperature levels in the exhaust of the reactor ranged from 350 to 600 K, which is similar to the simulation levels.The simulation results slightly underestimated Acpy removal efficiency compared to the experimental data.Actual measurements of acenaphthylene decomposition indicate that the maximum relative error of calculations is less than 6.3%.These results show that the CFD method is a useful tool in the rational design of an atmospheric plasma reactor to treat cooking fumes.

Fig. 4 .
Fig. 4. A description of grid relating to the atmospheric plasma reactor.(XY profile).

Fig. 5 .
Fig. 5.The simulated gas temperatures at different Y distances far from the center of plasma torch paralleled the BCsection at X= 7.35 cm (a) and X= 37.35 cm (b).

Fig. 6 .
Fig. 6.Temperature profiles (K): (a) at different X distances far from the center of plasma torch; (b-c) the distribution profile paralleled the BC-section at X= 7.35 (b) and X= 37.35 cm (c); (d) comparison of temperature levels between at X= 7.35 and X= 37.35 cm.

Fig. 7 .
Fig. 7. Concentration of Acpy profiles (%): (a) at different X distances far from the center of plasma torch; (b-c) the distribution profile paralleled the BC-section at X= 7.35 (b) and X= 37.35 cm (c); (d) comparison of C 12 H 8 levels between at X= 7.35 and X= 37.35 cm.

Table 1 .
System Parameters for simulation.

Table 2 .
Summary of grid independence test.

Table 3 .
Molecular structures of species used in the present work.

Table 4 .
Reaction, reaction rate parameters, and references for Acenaphthylene.