Experimental and numerical study on flame

Powder Technology 266 (2014) 456–462

Contents lists available at ScienceDirect

Powder Technology journal homepage: www.elsevier.com/locate/powtec

Experimental and numerical study on flame propagation behaviors in coal dust explosions Weiguo Cao a, Wei Gao b, Yuhuai Peng a, Jiyuan Liang a, Feng Pan a,c, Sen Xu a,c,⁎ a b c

School of Chemical Engineering, Nanjing University of Science and Technology, Nanjing, Jiangsu 210094, PR China School of Chemical Machinery, Dalian University of Technology, Dalian, Liaoning 116024, PR China National Quality Supervision and Inspection Center for Industrial Explosive Materials, Nanjing, Jiangsu 210094, PR China

a r t i c l e

i n f o

Article history: Received 3 April 2014 Received in revised form 27 June 2014 Accepted 30 June 2014 Available online 9 July 2014 Keywords: Coal dust explosion Flame propagation Flame temperature Numerical simulation Flow velocity

a b s t r a c t To reveal the flame propagation behaviors during coal dust explosions, a kind of coal dust cloud was studied through experiment and numerical simulation in semi-enclosed vertical combustion tubes with different lengths. A high speed video camera and a thermal infrared imaging device were used to record the flame propagation process. The result indicated that the supreme flame propagation velocity and the highest flame temperature both rose gradually with the tube length increasing. Meanwhile, FLUENT was applied to numerical simulation of flame propagation behaviors during the coal dust explosions. The simulation result showed the flame combustion and the temperature varying process was consistent with the experimental result. It also revealed the distribution of flow velocity in the flow field during the combustion process, which indicated that flow velocity higher than flame propagation velocity was an important reason for dust re-entrainment and consistent explosion. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Attention on the dust explosion started from a report in 1795 which covered the explosion in a flour silo in Turin, Italy [1]. As for its great destructive power, the dust explosions have become an important research direction in the dust explosions in the recent years [2–4]. Although, many countries have made corresponding progress in the basic research of dust explosions [2,5], the tragedy caused by dust explosions has not been completely controlled yet. With the rapid development of powder technology, it has a great value in preventing and controlling major hazardous accidents in the process industries. Therefore, it is necessary to carry out further study on the characteristic parameter of dust explosions. Some recent researches have focused on the flame propagation process which is one of the characteristic parameters during dust explosions [6–11]. Based on the theoretic study of premixed flame and using starch, sulfur and aluminum powder as test samples, Gao [7,11] researched on the flame propagation process of the organic dust with different concentration, such as hexadecanol, octadecanol and eicosanol, using a high speed video camera. Han [9] studied the flame structure and the flame propagation process of lycopodium in vertical tube. Proust [10] studied the effects of thermal radiation for the flame

propagation process in terms of detonation, laminar and turbulent flame. However, flame propagation mechanisms of dust clouds are not well-understood and the current research is an experimental study due to the dust explosion that is a complex combustion process of two-phase flow, causing great difficulties in the simulation process of dust explosion. Fortunately, with the development of the mechanical model, the numerical simulation technology based on computational fluid dynamics (CFD) has become the powerful tool during the researching process [12–15]. It is hopeful that this technology may become a design tool for safeguard measures, replacing the common empirical equations and charts gradually. To promote understanding of flame propagation during dust explosions, a kind of coal dust cloud was applied to carry out systematic research on flame propagation through experiment and numerical simulation in semi-enclosed vertical combustion tubes of different lengths. Comparing with experiment, FLUENT was used for numerical simulation to provide a valuable simulation tool for dust explosion research and useful information for safety prevention design. 2. Experimental 2.1. Experimental apparatus

⁎ Corresponding author. Tel.: +86 25 84315898 8830; fax: +86 25 84431574. E-mail address: [email protected] (S. Xu).

http://dx.doi.org/10.1016/j.powtec.2014.06.063 0032-5910/© 2014 Elsevier B.V. All rights reserved.

The experimental apparatus is shown schematically in Fig. 1, which is composed of vertical combustion tubes of different lengths including 300 mm, 600 mm and 900 mm, a high pressure dispersion system, an

W. Cao et al. / Powder Technology 266 (2014) 456–462 Table 1 Proximate and ultimate analyses of the coal. Sample

Coal

Proximate analyses(%)

Ultimate analyses(%)

Mad

Aad

Vad

FCad

C

H

O

N

3.54

14.46

41.75

40.25

57.05

4.43

37.4

1.12

457

As shown in Fig. 2, according to laser particle size analysis, the particle size distributions of most coal particles are (10 ~ 100) μm and the median diameter is 34 μm. FESEM result shows that the coal particle size is nonuniform and the shape is irregular. The diameters of coal particles are all less than 100 μm,which are well consistent with the laser analysis result.

Mad: moisture content; Vad: volatile matters; Aad: ash; FCad: fixed carbon.

3. Numerical simulation

ignition system, a high-speed video camera, a thermal infrared imaging device, and a control system. The vertical combustion tubes have a 68 mm inner diameter with the top open. The coal particles were placed evenly in the tube base and dispersed by a high dispersion pressure powder spray machine, whose pressure was 0.7 MPa. The uniform coal dust clouds with the concentration of 500 g/m3 were formed in the combustion tube. The ignition system is positioned at 100 mm above the bottom of the combustion tube. The distance between the tips of the two electrodes is 6 mm. A high voltage transformer with an output of 8000 V was adopted to make an ignition spark with the ignition energy of 5 J. The frame-rate of the high-speed video camera and the thermal infrared imaging device are 1000 fps and 100 fps, respectively. The weighed coal particles were placed evenly at the bottom of the vertical combustion tube and dispersed into the tube under high pressure to form a uniform coal dust cloud. The suspended particles were ignited by an electric spark after reaching a height of 300 mm to guarantee a consistent concentration of coal dust cloud and reduce the influence of residual turbulence on the flame propagation. After ignition, the flame temperature and the flame propagation process were recorded with a thermal infrared imaging device and a high speed video camera, respectively.

3.1. Governing equations CFD is applied to numerical simulation of the coal dust explosions. It is supposed that the coal particles are spherical particles. Based on chemical reaction kinetics and fluid mechanics, the governing equations are established through mass conservation, energy conservation, momentum conservation, and chemical reaction balance [16]. The main equations are as follows: Mass conservation equation: ∂ρ ∂ρui ¼0 þ ∂t ∂xi

ð1Þ

Energy conservation equation: ∂ρh ∂ þ ∂t ∂xi

ρu j h−

μ e ∂h σ h ∂x j

¼

dP þ Sh dt

ð2Þ

Momentum conservation equation: ∂ρui ∂ þ ∂xi ∂t

ρui u j −μ e

∂ui ∂x j

2.2. Experimental materials The source of coal particles is from Huolinhe Coal Mine. The proximate and ultimate analyses of the coal particles are summarized in Table 1. The morphology of the coal particles was characterized by field emission scanning electronic microscopy (FESEM, JEOL JSM6700F) and the diameter distribution of coal particles was characterized by laser particle size analyzer (Mastersizer 2000).

!

!

! ∂u j ∂ρ ∂ þ μe ∂xi ∂x j ∂x j 2 ∂ ∂u δij ρk þ μ e k : − 3 ∂x j ∂xk

¼−

ð3Þ

Chemical reaction balance equation: ∂ ρY fu ∂t

þ

∂ ∂x j

ρu j Y fu −

μ e ∂Y fu σ fu ∂x j

! ¼ Rfu

ð4Þ

Fig. 1. Experimental apparatus. 1 electric spark generator; 2 programmable logic controller; 3 pneumatic piston; 4 combustion tube; 5 ignition electrodes; 6 nozzle; 7 powder injection valve; 8 gas tank; 9 air inlet valve; 10 high pressure air; 11 piston-actuated valve; 12 powder tank; 13 high speed camera; 14 infrared imager.

458

W. Cao et al. / Powder Technology 266 (2014) 456–462

Fig. 2. Scanning electron microscopy and diameter distribution of the coal particles.

where P is the pressure; t is the time; ρ is the density; Yfu is the mass component of the fuel; ui is the velocity; μ is the dynamic viscosity; and k is the turbulent kinetic energy; Sh is the source term; xi and xj represent the direction, when i or j = 1 corresponds to the x-direction, i or j = 2 corresponds the y-direction and i or j = 3 corresponds the zdirection; δij is the Kronecker delta; σ is the Prandtl number; and Rfu is the rate of chemical combustion reaction. 3.2. Turbulence model According to the flame propagation behaviors during coal dust explosions and practical experiences, k − ε model summarized by the previous researches is selected as turbulent calculation model, which is the two-equation turbulence model [17,18].

Dk ∂ ¼ ρ Dt ∂xi

ρ

Dε ∂ ¼ Dt ∂xi

μ ∂k þ Gk þ Gb −ρε−Y m μþ t σ k ∂xi μþ

μ t ∂ε ε ε2 þ C 1ε ðGk þ C 3ε Gb Þ−C 2ε ρ σ ε ∂xi k k

C þ O2 →CO2 :

In the model, it is assumed that the coal dust clouds are evenly distributed at the ignition time. After ignition, the distribution of the coal dust clouds is decided by the flow field variation of the reaction region. Assume that chemical reactions occur much faster than turbulent combustion that can mix reactants and heat into the reaction region. Considering that multi-fuels combust simultaneously, the reaction rate is limited by the turbulent combustion rate and the turbulent combustion rate is determined by the lowest combustion rate of the fuel composition. The calculation equation of turbulent combustion rate is shown as the following [19]:

Rfu;T ¼ − ð5Þ

ð6Þ

ð8Þ

1 = C R ρg fu 2 ε

k

ð9Þ

where Rfu,T is the rate of turbulent combustion; CR is the constant; gfu is the ripple mean square of the fuel mass fraction; ε is the dissipation rating of turbulent kinetic energy; ρ is the density; and k is the turbulent kinetic energy. 4. Results and discussion

where k is the turbulent kinetic energy; ε is the dissipation rating of turbulent kinetic energy; Gk and Gb are the turbulent energy caused by the average velocity gradient and buoyancy, respectively; YM is the effects on total dissipation rating by compressible turbulent flow pulsation; μ t is the turbulence viscosity coefficient; ρ is the density; t is the time; and C1ε,C2ε and C3ε are all constants. 3.3. Combustion model As the turbulent eddies at molecular level have decisive influence on the combustion process of the coal dust cloud, Eddy-Break-Up (EBU) turbulence combustion model that fully considers the effects of turbulence is selected as a computational model. Taking the burning of coal particles in the air into consideration: Cx Hy þ ðy=4 þ xÞO2 →y=2H2 Oþx CO2

ð7Þ

4.1. Behaviors of flame propagation The flame propagations of the coal dust clouds in and above the combustion tube are depicted in Fig. 3, as recorded by the high-speed video camera. However, there was almost no flame in the initial several milliseconds, the flame propagated slowly during the initial stage after ignition; at this point, the coal particles were heated and decomposed. When the spark energy reached to the particle surface, the particle surface temperature rose rapidly. After the temperature rose to a certain value, the granules pyrolyzed rapidly and produced flammable gases. These gases mixed with air to form an explosive gas mixture, meanwhile, gas phase reactions occurred and chemical reaction heat was released, resulting in the adjacent particles heating and evaporation and ignition. As the coal particles continued to volatilize and released combustible gases, the flame propagation gradually accelerated and reached


459

Fig. 3. High-speed photographs of coal dust cloud flame. Tube height: 300 mm; concentration: 500 g/m3; time from ignition.

the highest which was the flame acceleration phase [20]. With the reaction continuing, the flame propagation velocity became slower and the flame died out finally. The developing situation of flame in the computational domain in different stages during the numerical simulation process can be shown in Fig. 4. The simulation computational domain included the vertical combustion tubes and the flow area outside the tube. Initially, the flame propagated slowly from the ignition position and flame front was spherical. With the combustion time continued, the flame sprayed to ambient environment from the tubes. The comparisons between simulation and experimental results of the flame propagation behaviors are shown in Figs. 5 and 6. 4.2. Influence of tube length on the flame propagation progress The flame propagation velocities of coal dust clouds in combustion tubes of different lengths are shown in Fig. 5. When the heights were 300 mm, 600 mm and 900 mm, the flame propagation velocities peaked before decreasing overall. The velocities of coal dust clouds reached their maxima of 5.0 m/s, 12.6 m/s and 15.2 m/s at 95 ms, 115 ms and 115 ms after ignition, and the flame fronts rose to the heights of 398 mm, 743 mm and 1058 mm at 140 ms after ignition, respectively. The maximum velocity relative errors of the experimental and simulated conditions were 11.3%, 10.2% and 6.0% at 80 ms, 120 ms and 95 ms respectively. When the ignition side was closed, with the expanded role of combustion products, the flame propagation velocity accelerated gradually. The expansion of combustion products induced unburned mixture before the flame front to produce turbulent flow, in turn, the turbulent increased the burning velocity, and so repeated interaction, the flame propagation accelerated increasingly [21]. In addition, after reaching the maximum, the flame propagation velocity decreased gradually due to the open port and the concentration of particles declined caused by unburned particles discharged. The flame temperatures of coal dust clouds in combustion tubes of different lengths are shown in Fig. 6. It can be seen that the topmost

flame temperatures reached 1220 °C, 1270 °C and 1360 °C when the heights of combustion tubes were 300 mm, 600 mm and 900 mm. The maximum absolute errors of experiment and simulation temperatures were 140 °C, 148 °C and 124 °C, while the relative errors were 11.5%, 11.7% and 9.1%. The flame propagation velocities and temperatures in simulation were higher than those in experiment. The reasons for these errors are as follows: (1) Simplifying assumption of many models. (2) Difference of the distribution of coal dust clouds between the simulation and the experiment. (3) The selection of empirical constants which were suitable for specific situations. (4) Compared with the simulation, dust stratification as a result of the cooling effect of hot combustion gases by the tube walls was an important reason, not only made the flame propagation velocities decrease, but also caused the flame temperatures to drop in the experiment [22]. The flame propagation velocities and temperatures increased with the rising of combustion tubes. The possible reason is that the long constraint distance of the combustion tube increased the air turbulence and accelerated the coal particles combustion rate. Thus, heat accumulation in a short time made flame propagation velocities and temperatures increases. The other reason is that the expansions of the combustion products played an important role in promoting the flame propagation velocity due to the fact that the location of the ignition system was near to the bottom of combustion tube, and the acceleration result was more obvious with the tube length increasing. Similar phenomena were observed by Ding [23]. From the analysis above, the simulation results reflect the flame propagation process during the coal dust explosions, and the errors of flame propagation velocities and temperatures are acceptable. Also, the distribution of the surrounding air flows in the coal dust explosions is drawn, which Fig. 7 shows.

Fig. 4. The simulation photographs of coal dust cloud flame. Tube height: 300 mm; concentration: 500 g/m3; time from ignition.

460


Fig. 5. Flame propagation behaviors with different height of combustion tubes. Concentration: 500 g/m3.

4.3. Flow field analysis of the tube outlet The distribution conditions of the air flow velocity in the calculation field in different times are shown in Fig. 7. Flame began to propagate from the center to the edge after ignition 25 ms, and the velocity of air flow was relatively low and the highest value was 2 m/s in the flame front. The flame propagated faster in the pipe orifice in 60 ms, because of the restriction which produced by the sides and bottoms of the

Fig. 6. Relationships between the top flame temperatures and the height of combustion tubes. Concentration: 500 g/m3.

tube. Therefore, the higher air flow velocity was concentrated in the area between the position of ignition and pipe orifice. The direction of the air flow in vertical and the highest velocity were 10 m/s. 80 ms after ignition, flame accelerated, but the flame was limited by the closed bottom. And the highest air flow velocity, approximately 24 m/s, occurred when the flame front approached to the nozzle. Then, 110 ms after ignition, the combustion tube was filled with flame which escaped upwards. The highest velocity 40 m/s appeared in the nozzle. Because of the loss of limit of combustion tube, the air flow spread out, thus, the velocity slows down. 130 ms later, flame formed the typical mushroom


461

Fig. 7. Flow velocity distribution in the process of coal dust explosions. Tube height: 300 mm; concentration: 500 g/m3.

cloud shape outside tube (Figs. 3 and 4). As flame had all rushed out of the narrow tube space, the air flow velocity was slower than that in 80 ms, when the highest approached to 28 m/s. 140 ms after ignition, mushroom clouds continued to spread and coal dust was still combusting, and the highest air flow velocity decreased to about 12 m/s. After that, owing to the gradual decrease of the fuel concentration, the flame attenuated till it died out. The maximum velocities of air flow in the process of the coal dust explosions with different times after ignition are presented in Fig. 8. The velocity increased with the height of combustion tube. The maximum value was respectively 40 m/s, 72 m/s, and 110 m/s after ignition 110 ms, 120 ms, and 130 ms when the height of the combustion tubes were 300 mm, 600 mm and 900 mm. All the velocities were higher than that of the flame propagations at the same time (Fig. 5). Therefore, the flow of air is an important contributor to the dust explosions. The flowing air produced by dust explosions during its initial stage would raise the surrounding dust layer and form dust clouds of certain concentration in the new space. At the same time, the spreading flames and radiant heat served as ignition sources, which formed chain explosions, and at last the area where the dust placed were damaged by explosions. 5. Conclusions This aim of this study is to get the general understanding of the flame propagation behaviors during coal dust explosions process. The vertical

combusting tubes with different heights were used and numerical simulation was also included. The following results were obtained. The flame propagation velocity and temperature of coal dust clouds increased as the height of combustion tubes rose and the flame propagation velocity peaked before decreasing overall. When the heights of combustion tubes were 300 mm, 600 mm and 900 mm, the velocities of flame reached their maximums of 5.0 m/s, 12.6 m/s and 15.2 m/s and the highest flame temperatures reached 1220 °C, 1270 °C and 1360 °C, respectively. The simulation results reflect the varying behaviors of flame propagation velocity and temperature during coal dust explosions process. The errors between experiments and simulations vary from 6% to 12%, and simulation results conform well to the experimental finals. All these indicate that this numerical model is appropriate for the numerical study of dust explosions. The simulation finals also reveal flowing characteristics of surrounding air during the flame propagating process. At the same time level, air flowing velocity is obviously higher than that of flame propagation, which indicates that the air flow is one of the reasons that lift up dust layer and cause chain explosions.

Acknowledgments The authors appreciate the financial support from the Research and Innovation Project for College Graduates of Jiangsu Province (Grant No. CXZZ13_0216) and the Natural Science Foundation of China (Grant No. 11102091). Additionally, the authors appreciate the financial support by the “Fundamental Research Funds for the Central Universities” (DUT14RC(3)061).

References

Fig. 8. The relationship between the highest flow velocity and the height of combustion tubes.

[1] P.R. Amyotte, R.K. Eckhoff, Dust explosion causation, prevention and mitigation: an overview, J. Chem. Health Saf. 17 (2010) 15–28. [2] R.K. Eckhoff, Current status and expected future trends in dust explosion research, J. Loss Prev. Process Ind. 18 (2005) 225–237. [3] N.S. Kuai, J.M. Li, Z. Chen, Experiment-based investigations of magnesium dust explosion characteristics, J. Loss Prev. Process Ind. 24 (2011) 302–313. [4] W.G. Cao, L.Y. Huang, J.Y. Liang, Research on characteristic parameters of coal dust explosion in a spherical sealed container, J. Min. Sci. Technol. 43 (2014) 113–119. [5] W.L. Frank, Dust explosion prevention and the critical importance of housekeeping, Process Saf. Progr. 23 (2004) 175–184. [6] Y. Yin, J.H. Sun, Y.B. Ding, S. Guo, X.Ch. He, Experimental study on flames propagating through zirconium particle clouds, J. Hazard. Mater. 170 (2009) 340–344. [7] W. Gao, T. Mogi, J.H. Sun, J. Yu, R. Dobashi, Effects of particle size distributions on flame propagation mechanism during octadecanol dust explosions, Powder Technol. 249 (2013) 168–174. [8] R. Dobashi, K. Senda, Mechanism of flame propagation through suspended combustible particles, J. Phys. IV France 12 (2002) 459.

462


[9] O.S. Han, M. Yashima, T. Matsuda, H. Matsui, A. Miyake, T. Ogawa, Behavior of flames propagating through lycopodium dust clouds in a vertical duct, J. Loss Prev. Process Ind. 13 (2000) 449–457. [10] Ch. Proust, A few fundamental aspects about ignition and flame propagation in dust clouds, J. Loss Prev. Process Ind. 19 (2006) 104–120. [11] W. Gao, T. Mogi, J.H. Sun, R. Dobashi, Effects of particle thermal characteristics on flame structures during dust explosions of three long-chain monobasic alcohols in an open-space chamber, Fuel 113 (2012) 86–96. [12] Z.H. Chen, B.C. Fan, Flame propagation through aluminum particle cloud in a combustion tube, J. Loss Prev. Process Ind. 18 (2005) 13–19. [13] A.Di. Benedetto, P. Russo, Thermo-kinetic modelling of dust explosions, J. Loss Prev. Process Ind. 20 (2007) 303–309. [14] J. Wang, X. Li, S. Zhong, On the flame propagation along pipeline during dust explosion between interconnected vessels, J. Northeast. Univ. Nat. Sci. 31 (2010) 157–160 (in Chinese). [15] B.Y. Dong, X. Peng, Numeral simulation on venting explosion process of flammable gas in closed cylindrical vessel, China Saf. Sci. J. 21 (2011) 29–34 (in Chinese). [16] V. Giovangigli, M.D. Smooke, Extinction of strained premixed laminar flames with complex chemistry, Combust. Sci. Technol. 53 (1987) 23–49.

[17] T.H. Shah, W.W. Liou, A. Shabbir, A new k-ε Eddy viscosity model for high Reynolds number turbulent flows model development and validation, Comput. Fluids 24 (1995) 227–238. [18] B.E. Lauder, D.B. Spalding, The numerical computation of turbulent flows, Comput. Methods Appl. Mech. Eng. 3 (1974) 269–289. [19] C.Y. Zhang, X.F. Sun, Numerical simulation on vortex shedding and combustion oscillation in aero-engine afterburner, J. Aerosp. Power 25 (2010) 270–277 (in Chinese). [20] H. Hanai, H. Kobayashi, T. Niioka, A numerical study of pulsating flame propagation in mixtures of gas and Particles, Proc. Combust. Inst. 28 (2000) 815–822. [21] Q.M. Liu, C.H. Bai, X.D. Li, L. Jiang, W.X. Dai, Coal dust/air explosions in a large-scale tube, Fuel 89 (2010) 329–335. [22] W. Gao, R. Dobashi, T. Mogi, J.H. Sun, X.B. Shen, Effects of particle characteristics on flame propagation behavior during organic dust explosions in a half-closed chamber, J. Loss Prev. Process Ind. 25 (2012) 993–999. [23] Y.B. Ding, J.H. Sun, X.C. He, Flame propagation characteristic of zirconium particle cloud, Combust. Sci. Technol. 16 (2010) 353–357.