Effect of flame propagation regime on pressure evolution of nano and micron PMMA dust explosions

Effect of flame propagation regime on pressure evolution of nano and micron PMMA dust explosions

Journal of Loss Prevention in the Process Industries 63 (2020) 104037 Contents lists available at ScienceDirect Journal of Loss Prevention in the Pr...

3MB Sizes 0 Downloads 31 Views

Journal of Loss Prevention in the Process Industries 63 (2020) 104037

Contents lists available at ScienceDirect

Journal of Loss Prevention in the Process Industries journal homepage: http://www.elsevier.com/locate/jlp

Effect of flame propagation regime on pressure evolution of nano and micron PMMA dust explosions Xinyan Zhang a, b, *, Wei Gao c, Jianliang Yu c, Yansong Zhang a, Jie Zhang a, Xingwang Huang a, Jinshe Chen a a b c

School of Mining and Safety Engineering, Shandong University of Science and Technology, Qingdao, 266590, China Mine Disaster Prevention and Control-Ministry of State Key Laboratory Breeding Base, Shandong University of Science and Technology, Qingdao, 266590, China College of Chemical Engineering, Dalian University of Technology, Dalian, 116024, China

A R T I C L E I N F O

A B S T R A C T

Keywords: Nano and micron dust explosions Flame propagation regime Explosion pressure evolution Combustion reaction intensity Flame temperature distribution

Experiments using an open space dust explosion apparatus and a standard 20 L explosion apparatus on nano and micron polymethyl methacrylate dust explosions were conducted to reveal the differences in flame and pressure evolutions. Then the effect of combustion and flame propagation regimes on the explosion overpressure char­ acteristics was discussed. The results showed that the flame propagation behavior, flame temperature distribu­ tion and ion current distribution all demonstrated the different flame structures for nano and micron dust explosions. The combustion and flame propagation of 100 nm and 30 μm PMMA dust clouds were mainly controlled by the heat transfer efficiency between the particles and external heat sources. Compared with the cluster diffusion dominant combustion of 30 μm dust flame, the premixed-gas dominant combustion of 100 nm dust flame determined a quicker pyrolysis and combustion reaction rate, a faster flame propagation velocity, a stronger combustion reaction intensity, a quicker heat release rate and a higher amount of released reaction heat, which resulted in an earlier pressure rise, a larger maximum overpressure and a higher explosion hazard class. The complex combustion and propagation regime of agglomerated particles strongly influenced the nano flame propagation and explosion pressure evolution characteristics, and limited the maximum overpressure.

1. Introduction With rapid development of the modern industry, many industries such as forestry, grain, mechanical, chemical and pharmaceutical have used solid powders as raw materials for processing and production. Besides, some industrial intermediates or products are also powders. The types of powders involved are numerous and the using dosage is greatly increased, which can greatly increase the possibility of dust explosion accidents. Especially with the gradual refinement of powders and the application of nano materials, the potential explosion danger of ultrafine powders has attracted a lot of attention. For ultrafine dust particles, the considerable small size with large effective specific surface area can rapidly pyrolyze and combust, dominating the flame propagation (Zhang et al., 2016). When the particle approaches to nano scale, the van der Waals force increases. Because of the larger particle forces, the dispersion and explosion characteristics of combustible nano dusts are very different from those of micron dusts. The sensitivity of nano dust flame and its rapid development are believed to greatly increase the

likelihood of nano scale dust explosions (Krietsch et al., 2015; Azha­ gurajan et al., 2012; Boilard et al., 2013; Mittal, 2014). However, nano dust particles are usually confronted with severe agglomeration effects (Tang et al., 2017; Balbudhe et al., 2015). When agglomeration of nano particles changes them from nano to micron scale, certain nano scale properties change. When the agglomeration of dust particles is quite severe, the suspended agglomerates can settle down quickly. The par­ ticle mass density actually participating in combustion reaction is greatly reduced. In addition, the large-sized agglomerates are difficult to be ignited. So the weak flame is difficult to maintain self-sustaining propagation and tends to quickly quench in a short time (Zhang et al., 2017). In this case, the nano scale dust is more difficult to explode than the micron dust. Moreover, the pyrolysis and combustion of agglomer­ ates in nano dust cloud are very complex. As a result, it has a significant influence on the flame microstructure and propagation characteristic (Zhang et al., 2017), which in turn can affect the dust explosion char­ acteristics to a large extent. The uncertainty of the explosion charac­ teristics makes it difficult to predict the destructive strength of nano

* Corresponding author. School of Mining and Safety Engineering, Shandong University of Science and Technology, Qingdao, Shandong, 266590, China. E-mail address: [email protected] (X. Zhang). https://doi.org/10.1016/j.jlp.2019.104037 Received 26 October 2019; Received in revised form 19 December 2019; Accepted 22 December 2019 Available online 28 December 2019 0950-4230/© 2019 Elsevier Ltd. All rights reserved.

X. Zhang et al.

Journal of Loss Prevention in the Process Industries 63 (2020) 104037

scale dust explosions, and there’s no uniform conclusion on explosion hazards till now (Mittal, 2014; Jiang et al., 2011; Dobashi, 2009; Vignes et al., 2012; Eckhoff, 2012; Bouillard et al., 2010). The consequence of nano dust explosion is difficult to predict, which is extremely detri­ mental to safe production. It is of great significance to carry out in-depth research on utilizing the basic theory of micron dust explosion for reference to predict the danger of nano scale dust explosion based on its essential mechanism. In this study, the differences in the flame evolution behaviors in the initial stage of nano and micron PMMA dust explosions were experi­ mentally studied. The combustion and flame propagation regimes were comparatively revealed. Then the effect of flame propagation regime on the pressure evolution of nano and micron PMMA dust explosions was discussed for predicting the explosion consequences accurately and developing the effective protections essentially.

pressure acquisition. 2.2. Experimental material 100 nm (Series: MP-300) and 30 μm (MZ-30H) polymethyl methac­ rylate (PMMA) dust particles (Soken Chemical & Engineering Co., Ltd. of Japan) dried in an air dry oven for 24 h were used. The corresponding thermo-physical parameters were listed in our former study (Zhang et al., 2017). 2.3. Particle size distribution and morphology characteristic The particle size distributions of 100 nm and 30 μm PMMA dust particles were measured by the Laser Diffraction Particle Size Analyzer (Malvern Mastersizer 2000) and Phase Doppler Particle Analyzer (TSI INCORPORATED). The particle morphologies were observed by the Scanning Electron Microscopy (FEI Nova NanoSEM 450). The results are shown in Fig. 3 and Table 1. For 100 nm PMMA dust particles, the volume mean diameter D [4,3] and surface mean diameter D [3,2] were 24.877 μm and 10.486 μm, respectively. The cumulative distribution diameter D10 was 6.151 μm. The single spherical particle was almost 100 nm in diameter, whereas the natural particles were existed as the irregular agglomerates in size scale from 25.179 μm to 28.251 μm. When dispersed the particles to form dust cloud, the D [4,3], D [3,2] and D10 of dust particles suspended near the ignition electrodes were 24.896 μm, 22.893 μm and 15.060 μm, respectively, indicating a serious agglomeration. For 30 μm particles, most particles were in the size scale from 22.440 μm to 25.179 μm before dispersion. The D [4,3], D [3,2] and D10 were 28.974 μm, 24.080 μm and16.509 μm, respectively. The spherical particles were discretely distributed with non-agglomeration. After dispersion the D [4,3], D [3,2] and D10 of dust particles suspended near the ignition electrodes were 25.637 μm, 24.094 μm and17.446 μm, respectively. The charac­ teristic diameters were almost the same as those before dispersion, indicating a pretty well dispersity of 30 μm dust cloud. Although the size of the individual single particle was in nano and micron scale and the 100 nm dust contained much more small particles, the particle size distributions of 100 nm and 30 μm dust clouds were in the same size scale with similar diameter characteristics.

2. Experimental apparatus and material 2.1. Experimental apparatus The open space dust explosion experimental apparatus illustrated in Fig. 1 was used to study the flame propagation characteristics of nano and micron scale dust explosions. The detailed configuration was described in our previous study (Zhang et al., 2017). The high speed photography was used to track the dynamic flame evolution features. The macrophotography and the “Z" shaped parallel reflective schlieren optical system were used to observe the flame microstructures. A fine thermocouple comprised of 25 μm diameter Pt–Pt/Rh13% wires was used to detect the flame temperature. An ion current probe comprised of 100 μm diameter Pt wires was used to detect the combustion reaction intensity. The 20 L spherical explosion apparatus illustrated in Fig. 2 was used to test the dynamic pressure evolutions of nano and micron scale dust explosions according to the internationally accepted ASTM E1226 2010 standards. The detailed configuration was described in our previous study (Gao et al., 2015). The pressure in the explosion chamber was initially vacuumed to 0.056 MPa. After 50 ms of packing pressure, the dust particles were dispersed by 1.5 MPa air flow. The pressure in the explosion chamber gradually increased to zero and the particles were fully dispersed during the ignition delay period of 60 ms. Then the dust cloud was ignited, the flame formed and the pressure started to increase. A high frequency pressure sensor (EH PMC131) was used for explosion

Fig. 1. Open space dust explosion experimental apparatus. 2

X. Zhang et al.

Journal of Loss Prevention in the Process Industries 63 (2020) 104037

Fig. 2. 20 L spherical explosion apparatus.

Fig. 3. Particle morphologies and size distributions of 100 nm (Zhang et al., 2016b) and 30 μm PMMA dusts. Table 1 Characteristic diameters of 100 nm and 30 μm PMMA dust particles. Product name

Particle morphology

Particle size distribution Characteristic diameter

Before dispersion

During dispersion

MP-300 (100 nm)

D [4,3] (μm) D [3,2] (μm) D10 (μm)

24.877 10.486 6.151

24.896 22.893 15.060

MZ-30H (30 μm)

D [4,3] (μm) D [3,2] (μm) D10 (μm)

28.974 24.080 16.509

25.673 24.094 17.446

2.4. Thermal pyrolysis and oxidation characteristic

under air atmosphere mainly experienced two stages. The rapid weight loss stage was mainly dominated by the thermal pyrolysis reaction, and the slow weight loss stage mainly occurred the thermal oxidation reac­ tion of pyrolysis products. Compared with the thermal pyrolysis char­ acteristic of 30 μm particles, the 100 nm particles had a slower rapid pyrolysis process, a smaller maximum weight loss rate and a higher

The weight loss and thermal characteristics of 100 nm and 30 μm PMMA dusts were analyzed by using a Synchronous Thermal Analyzer (SDT- Q600). As shown in Fig. 4, the results indicate that the weight loss of 100 nm and 30 μm PMMA particles with the heating rate of 10 � C/min 3

X. Zhang et al.

Journal of Loss Prevention in the Process Industries 63 (2020) 104037

Fig. 4. Thermal pyrolysis and oxidation characteristics of 100 nm and 30 μm PMMA dust particles.

temperature corresponding to the maximum weight loss rate. This was possibly caused by the complicated pyrolysis of agglomerates. The py­ rolysis first occurred on the particle surface. As the temperature increased, the inner particles were gradually heated and pyrolyzed, which required more time and energy to maintain the progress of py­ rolysis. Because of the particular existence of nano agglomerates, the instant phase change and pyrolysis of large agglomerates occurred accompanied with a small endothermic peak after the foremost exothermic oxidation reaction, leading to a smaller weight loss rate and a lower oxidation release heat of the remained quite large agglomerates. However, considering the advantage of the single 100 nm particles and the separated agglomerates, the slow weight loss proceeded quicker and the oxidation occurred earlier. In addition, the strong oxidation reaction of 100 nm and 30 μm particles occurred at the end of the rapid weight loss stage and slow weight loss stage, respectively, which revealed the effect of particle size and agglomeration on the oxidation process.

3. Results and discussion 3.1. Propagating flame structure and propagation regime The flame evolutions of 100 nm and 30 μm PMMA dust clouds with a concentration of 364 g/m3 are displayed in Fig. 5. As discussed in (Zhang et al., 2016), 100 nm PMMA dust flame continuously propagated with slight disturbing deformation in an approximately spherical shape. Whereas 30 μm dust flame was characterized by the cluster yellow flames with the irregular flame shape. The micro morphologies of 100 nm and 30 μm dust flames are showed in Fig. 6. The flame surface of 100 nm PMMA dust cloud was smoothly upward propagated. The flame preheat zone thickness, defined as the distance from the edge of luminous flame to the edge of schlieren flame, was about 3 mm–4 mm. Though propagated as cluster luminous flames in the direct micro photograph, the micro schlieren photograph revealed a continuous propagating flame surface of 30 μm dust cloud. The flame preheat zone thickness was approximately 5 mm–6 mm. The cellular flame surface wrinkled originating from the

Fig. 5. Propagating flames of 100 nm and 30 μm PMMA dust clouds. 4

X. Zhang et al.

Journal of Loss Prevention in the Process Industries 63 (2020) 104037

Fig. 6. Micro morphologies of 100 nm and 30 μm PMMA dust flames.

non-uniform distribution of reaction components and the thermal heat diffusion. The regionally localized development feather corresponding to the cluster-like flame of the micron dust cloud was remarkable, revealing the diffusion controlled mode of the propagating micron dust flame. Based on the features of the propagating flame structure, the flame propagation regimes of 100 nm and 30 μm dust clouds are depicted in Fig. 7. The 100 nm dust flame mainly contained four zones, which were preheat zone, thin premixed reaction zone I, thick agglomerates reaction zone II, and post flame zone. Under high ignition energy activation, the particles instantaneously vaporized and were ignited. At the same time, the adjacent suspended smaller particles in preheat zone were instan­ taneously heated to pyrolyze. The pyrolysis volatiles were fully mixed with air and ignited at the ignition temperature, thus the preheat zone at t1 developed to the thin premixed reaction zone I at t2, where the ho­ mogenous premixed-gas combustion occurred. The larger particles in preheat zone at t1 required more heat to completely pyrolyze so that the particles could pyrolyze in the reaction zone and reacted in premixed

gas-phase in the presence of homogenous luminous flame in the thick agglomerates reaction zone II. The particular large agglomerates could ignite while pyrolyze in the thick agglomerates reaction zone II, forming luminous spot flames in the regime of diffusion combustion coupled with local premixed-gas combustion. Then the combustion reaction completed and the combustion products formed in the post flame zone. During flame evolution, the turbulence caused by heat and mass transfer disturbed the flame surface slightly. The smaller particles in preheat zone played an important role in flame propagation and dominated the flame upward propagation velocity. The larger agglomerates in com­ bustion reaction zone determined the burning time of nano dust flame. As for the 30 μm dust flame, the propagation regime was quite different. The cluster flame mainly contained the preheat zone, com­ bustion reaction zone and post flame zone. Under high ignition energy activation, a similar spherical flame formed in the initial stage. In the flame self-sustaining propagation stage, the smaller particles in preheat zone firstly pyrolyzed. The mass transfer of pyrolysis volatiles and ox­ ygen occurred intensively. The strong turbulence in the combustion flow

Fig. 7. Flame propagation regimes of 100 nm and 30 μm PMMA dust clouds. 5

X. Zhang et al.

Journal of Loss Prevention in the Process Industries 63 (2020) 104037

field of 30 μm PMMA dust flame led to the non-uniform concentration distribution of local flammable volatiles in the flame front. Besides, compared with the 100 nm dust particles, the pyrolysis time of 30 μm particles in larger sizes was much longer, and there was no enough time for the sufficient mixing of the non-uniform volatiles. So the flame propagated in the diffusion combustion regime to the unburnt regions with local high concentration volatiles in preheat zone, forming the local cluster flames. As the cluster flames developed, larger particles were heated and pyrolyzed instantaneously in the combustion reaction zone, where sufficient mixing with oxygen occurred, leading to the local premixed-gas combustion. The flame zone of the cluster flames con­ nected and propagated. The flame surface continuously propagated and wrinkled as the cluster flames developed. As the combustion reaction completed, the combustion products formed in the post flame zone. During flame evolution, the non-uniform concentration distribution of flammable volatiles in the flame front determined the diffusion com­ bustion of the cluster flames and dominated the flame upward propa­ gation velocity. The flame structure and propagation regime of the nano and micron dust clouds were quite different. The smaller the particle size was and the larger the number density of smaller particles was, the easier the pyrolysis and oxidation combustion occurred, resulting in the smoother flame surface and the more continuous flame luminous zone. Hence the dominant flame propagation regime could be the premixed-gas com­ bustion. However, when the particle size was in micron scale, the dominant flame propagation regime was the diffusion combustion. In addition, the complex pyrolysis and combustion of agglomerates in nano scale dust clouds affected the flame structure and propagation regime to a great extent.

turbulence caused by the thermal expansion and flame self-instability all affected the flame pulsation propagation characteristic. Compared with 100 nm particle, the 30 μm particle thermal capacity was larger and its reaction heat was weaker, leading to a longer time for particle pyrolysis. In addition, the thermal-diffusion instability of the 30 μm dust flame was relatively larger, resulting in a more intense pulsation propagation. 3.2.2. Flame temperature distribution The flame temperature distributions of 100 nm and 30 μm PMMA dust clouds are shown in Fig. 9. In the 100 nm dust flame preheat zone, the flame temperature rapidly increased from the room temperature to particle pyrolysis temperature. The particles were heated and pyrolyzed. The pyrolysis volatiles were fully mixed with air to form the premixed combustible gases. As the temperature increased to ignition tempera­ ture, the premixed-gas combustion reaction occurred in the thin pre­ mixed reaction zone I. As the combustion proceeded, the temperature increased from the ignition temperature to an initial maximum tem­ perature of 800 � C at the end of this zone. Then the flame temperature increased in a pulsatile way in the thick agglomerates reaction zone II due to the complex combustion of agglomerates. At the end of this zone, the temperature reached the maximum value of 1551 � C. In the post flame zone, there was almost no combustion reaction and no more heat was released. The temperature of the combustion products gradually decreased due to heat losses. As for the 30 μm dust flame, because the temperature measuring point of the installed thermocouple was 2 cm above the ignition electrodes, the thermocouple was capable of measuring the temperature distribution of one cluster flame in the di­ rection of flame propagation. In the preheat zone of 30 μm dust flame, the temperature rapidly increased from the room temperature to particle ignition temperature. In the flame combustion reaction zone, the flame temperature continuously increased to the maximum temperature of 1108 � C. In the post flame zone, the temperature of the combustion products gradually decreased due to heat losses. Compared with the 30 μm dust flame, the large amounts of small dust particles in the 100 nm dust flame preheat zone could instantaneously pyrolyze and vaporize as the flame temperature increased rapidly from the room temperature to ignition temperature, resulting in a thinner preheat zone. During flame propagation, the larger the pyrolysis reaction rate and the more com­ plete the combustion reaction, the higher the flame temperature.

3.2. Flame propagation characteristic 3.2.1. Flame propagation velocity The flame front position and propagation velocity were discussed. As shown in Fig. 8, the flame front of 100 nm PMMA dust cloud propagated upward almost linearly with the propagation time. Whereas the flame front of 30 μm dust cloud stepped propagated (Zhang et al., 2016), reflecting the development of cluster flames. The cluster diffusion feature was more obvious during the flame acceleration stage. The average flame propagation velocities of 100 nm and 30 μm PMMA dust clouds with a concentration of 364 g/m3 were 0.68 m/s and 0.33 m/s, respectively, indicating that the propagation velocity of nano dust flame was about twice as the micron dust flame. Additionally, the feedback regime between the particle energy bal­ ance and flame propagation velocity in preheat zone during flame propagation, the initial turbulence generated by dust injection, and the

3.2.3. Flame ion current distribution The typical ion current distributions of 100 nm and 30 μm PMMA dust flames are shown in Fig. 10. For 100 nm dust flame, in the flame preheat zone the ion current was almost zero, indicating that there was no reactive ion and only pyrolysis occurred. In the thin premixed reac­ tion zone I, the ion current started to increase and an initial ion current

Fig. 8. Flame front positions and propagation velocities of 100 nm and 30 μm PMMA dust clouds with a concentration of 364 g/m.3. 6

X. Zhang et al.

Journal of Loss Prevention in the Process Industries 63 (2020) 104037

Fig. 9. Flame temperature distributions of 100 nm and 30 μm PMMA dust clouds with a concentration of 364 g/m.3.

Fig. 10. Flame ion current distributions of 100 nm and 30 μm PMMA dust clouds.

peak was obtained, revealing an intense combustion reaction of the premixed pyrolysis volatiles. In the thick agglomerates reaction zone II, the ion current was with multiple peaks. This characteristic was caused by the gradual combustion of large agglomerates. As the agglomerates pyrolyzed while combusted, the intermediate ion products of Hþ, OH , CHx and so on were incessantly produced and consumed, resulting in a pulsatile increase of ion current corresponding to the pulsatile increase of flame temperature in the thick agglomerates reaction zone II. In the post flame zone, the combustion products were with no ion current. As for the 30 μm dust flame, the ion current distribution of one cluster flame in the direction of flame propagation was detected. In the flame preheat zone the ion current was zero. In the combustion reaction zone, the

oxidation reaction proceeded as the intermediate ion products were actively produced. The ion current reached a peak value and then decreased to zero in the post flame zone. The combustion reaction in­ tensity of 100 nm dust flame was obviously stronger than that of 30 μm dust flame. The thickness of the combustion reaction zone for 100 nm dust flame was thicker than that for 30 μm dust flame due to the com­ bustion of agglomerates in reaction zone. The differences in ion current distributions of 100 nm and 30 μm PMMA dust flames were significant, indicating the different flame structures and flame propagation regimes during nano and micron dust explosions.

7

X. Zhang et al.

Journal of Loss Prevention in the Process Industries 63 (2020) 104037

3.3. Time-scale analysis for pyrolysis and combustion

ratios of the particle surface area diameter and volume diameter to the Sauter diameter, respectively. The pyrolysis products combustion characteristic time (tcomb) during flame propagation is calculated as � λg T 2i Tf Ti tcomb ¼ (5) Cpg ρu v2f T 2u ðTi Tu Þ

The dominant regimes of the pyrolysis and combustion for 100 nm and 30 μm PMMA dust flames were evaluated by a time-scale analysis. Three dimensionless numbers of the Biot number (Bi), Thiel number €hler number (Da) were discussed (Bidabadi et al., 2010; (Th), and Damko Russo et al., 2013; Di Benedetto et al., 2010; Di Blasi, 1999; Ballal, 1983; Gao et al., 2013). The Bi was calculated to evaluate the controlling regime of the particle heating process. The Th was calculated to evaluate the controlling regime of the particle pyrolysis process. The Da was calculated to evaluate the controlling regime of the dust combustion process. The relationship of each dimensionless number is shown in Fig. 11. The particle heat transfer characteristic times during flame propa­ gation are listed as followed.

where Tu, Ti and Tf are the atmospheric temperature, ignition temper­ ature and flame temperature, respectively. vf is the flame propagation velocity. ρu is the air density. The calculated values of the Bi, Th and Da for 100 nm and 30 μm PMMA dust flames with various dust concentrations are showed in Table 2. The results indicated that the Bi of 100 nm and 30 μm PMMA dust flames were all smaller than 1, revealing that the particle internal heat transfer rate was much quicker than the external heat transfer rate. The particle heating process during pyrolysis was controlled by the particle external heat transfer rate. The Bi of 100 nm dust flame was larger than that of the 30 μm dust flame due to the quicker external heat transfer rate, indicating that the effect of the internal heat transfer was enhanced. This was attributed to the complex heating and pyrolysis of the inner agglomerated particles in 100 nm dust flame. The Th of 100 nm and 30 μm PMMA dust flames with various dust concentrations were all larger than 1, revealing that the particle pyrol­ ysis rate was much quicker than the thermal external heat transfer rate. The external thermal resistance of dust particles was so large that the pyrolysis process was controlled by the heat transfer efficiency between the particles and external heat sources. In addition, although the thermal heat transfer rate and pyrolysis rate were both larger, the Th of 100 nm dust cloud during flame propagation with various dust concentrations were always larger than those of 30 μm dust flame accordingly, demonstrating a much quicker pyrolysis reaction and a more important role external thermal heat transfer played. The calculated Da of 100 nm and 30 μm dust flames with various dust concentrations were all larger than 1. It is inferred that during the particle combustion and flame propagation, the combustion reaction of pyrolysis volatiles was so fast that the particle thermal heating and py­ rolysis steps were the controlling regime. The particle pyrolysis rate controlled the overall combustion process. The pyrolysis and combus­ tion mode of the large amounts of agglomerated particles existed in nano dust clouds was extremely complex (Zhang et al., 2017). Consequently the pyrolysis characteristic time of 100 nm particles was prolonged, resulting in a large Da value, and even larger than the Da of 30 μm dust flame. This result indicated that the flame propagation process of 100 nm dust cloud was more sensitive to the pyrolysis rate due to the agglomeration effect. In addition, the Da of 100 nm dust flame decreased with increasing dust concentration, revealing the dominant combustion regime had a tendency of transferring from the agglomerates diffusion

I. External convection heat transfer characteristic time (tconv) is calculated as tconv ¼

ρs Cps Lc

(1)

hc

where ρs and Cps are the density and specific heat of solid-phase particle at the atmospheric pressure, respectively. Lc is the characteristic length. hc is the heat transfer coefficient. II. External radiation heat transfer characteristic time (trad) is calculated as trad ¼

ΔT ρs Cps Lc Qr

(2)

where ΔT is the temperature gradient between the particle-self and surrounding gas. Qr is the radiation heat exchange. III. Internal heat transfer characteristic time (tcond) is calculated as tcond ¼

ρs Cps L2c

(3)

λs

where λs is the thermal conductivity of solid-phase particle. The particle pyrolysis characteristic time (tpyro) is calculated as tpyro ¼

C3 ρ D2 � . 3 s �32 8C1 λg Cp g lnð1 þ BÞ

(4)

where λg and Cpg are the thermal conductivity and specific heat of gasphase PMMA at the atmospheric pressure, respectively. B is the mass transfer number. D32 is the particle Sauter diameter. C1 and C3 are the

Fig. 11. Time-scale analysis for pyrolysis and combustion. 8

X. Zhang et al.

Journal of Loss Prevention in the Process Industries 63 (2020) 104037

Table 2 Calculated tint, text, tpyro, tcomb, Bi, Th and Da of 100 nm and 30 μm PMMA dust flames with various dust concentrations. Characteristic number

MP- 300 (100 nm) 193 g/m

tint (ms) text (ms) tpyro (ms) tcomb (ms) Bi Th Da

4.92 11.03 3.03 0.27 0.446 3.64 11.22

3

Characteristic number 285 g/m 4.92 10.95 2.93 0.27 0.449 3.73 10.94

3

3

377 g/m 4.92 10.60 2.69 0.27 0.464 3.94 9.96

tint (ms) text (ms) tpyro (ms) tcomb (ms) Bi Th Da

combustion to the premixed gas-phase dynamic combustion. This is quite different from the 30 μm dust flame as the Da increased with increasing dust concentration, signifying the dominant diffusion com­ bustion mode gradually significant.

MZ- 30H (30 μm) 223 g/m3

314 g/m3

406 g/m3

5.45 12.33 3.93 0.71 0.442 3.14 5.56

5.44 12.23 3.71 0.47 0.445 3.29 7.85

5.44 12.12 3.56 0.30 0.449 3.40 11.83

in the front of the flame surface and developed. The explosion over­ pressure rose rapidly in the explosion chamber. So the pressure rise rate of 100 nm dust explosion was quicker, and the maximum pressure rise rate was much larger than those of 30 μm dust explosion with the cluster diffusion flame propagation regime. In addition, the premixed-gas dominant combustion of 100 nm dust flame determined a smaller particle pyrolysis and combustion charac­ teristic time scale, a quicker combustion reaction rate, a faster flame propagation velocity, a stronger combustion reaction intensity, a quicker heat release rate, and a higher amount of released chemical reaction heat. The energy required to activate the pyrolysis and com­ bustion reaction of 100 nm particles was lower. During explosion, the energy gathered in large quantities, the temperature rose quickly, and the explosion overpressure enhanced strongly. With sustainable support of the combustion heat, the explosion combustion reaction of 100 nm dust cloud with large effective reaction surface could be more complete, and the overpressure gathered should be much greater. However, as has been proved, the100 nm dust existed serious agglomeration. Although the 100 nm individual particles were precisely in nano size, the particle size distributions of 100 nm and 30 μm dust clouds were almost in the same size scale with similar diameter characteristics. Therefore, although the 100 nm dust explosion possessed a larger maximum overpressure than 30 μm dust explosion, the maximum overpressure difference between the two was as small as 0.06 MPa. Those results revealed that the explosion overpressure of nano scale dust was rapidly gathered and increased in a short period of time, which resulted in a faster thermal heat and shock wave damage. However, due to the serious affection of nanoparticles agglomeration, the complex combustion and propagation regime of nano agglomerates with the diffusion combustion coupled with local premixed combustion during explosion strongly affected the flame temperature and explosion pressure evolution char­ acteristics, and limited the maximum explosion overpressure. So the nano scale dust explosion characteristic was dependent on the particle size distribution, particle pyrolysis oxidation characteristic and flame propagation regime. The nano dust explosion damage spread fast, but the damage was not necessarily much greater than that of micron scale dust explosion. However, a faster explosion pressure rise rate led to a higher explosion hazard class for nano scale dust, which was obviously much more dangerous than that of other scales dusts. So it acquired more attentions on prevention and mitigation of the nano scale dust explosion hazards.

3.4. Effect of flame propagation regime on explosion pressure evolution The dynamic pressure evolutions of 100 nm and 30 μm PMMA dust explosions with a concentration of 500 g/m3 is shown in Fig. 12. As for the 100 nm dust explosion, the pressure started increasing rapidly at about 30 ms after ignition, and reached the maximum overpressure of 0.76 MPa at about 60 ms after ignition. The maximum explosion pres­ sure rise rate of 84.65 MPa/s was obtained at about 50 ms after ignition during the rapid pressure rise stage. As for 30 μm dust explosion, the pressure started increasing rapidly at about 50 ms after ignition, and reached the maximum overpressure of 0.70 MPa at about 90 ms after ignition. The maximum explosion pressure rise rate of 28.25 MPa/s was obtained at about 120 ms after ignition during the rapid pressure rise stage. In the pressure reduce stage, both 100 nm and 30 μm dust ex­ plosion pressures decreased quickly. The calculated explosion index (Kst) of 100 nm dust was 22.98 MPa m/s, and the explosion hazard class of dust was St2- Strong explosion according to NFPA 68 2018. The calculated Kst of 30 μm dust was 7.67 MPa m/s, and the corresponding explosion hazard class of dust was St1- Weak explosion. Considering the combustion and flame propagation regimes of 100 nm and 30 μm dust explosions, the explosion pressure evolution char­ acteristics were closely related. Compared with the 30 μm dust explo­ sion, the 100 nm dust particles with larger specific surface area in rather smaller size instantaneously pyrolyzed and reacted with oxygen, resulting in an earlier pressure rise. The premixed-gas combustion flame propagated rapidly and released amounts of chemical reaction heat. The cold gas around flame surface was heated and expanded. As the thermal expansive gas expanded and moved forward, the pressure wave formed

4. Conclusions The propagating flame structures, flame evolution behaviors and flame propagation regimes of nano and micron PMMA dust explosions were comparatively studied. The effect of flame propagation regime on the explosion pressure evolution was discussed. The conclusions ob­ tained are as follows: (1) Due to the complex configuration of 100 nm particles under affection of serious agglomeration, not only the pyrolysis reaction was passivized and more energy was required for pyrolysis, but

Fig. 12. Pressure evolutions of 100 nm and 30 μm PMMA dust explosions. 9

X. Zhang et al.

(2)

(3) (4)

(5)

Journal of Loss Prevention in the Process Industries 63 (2020) 104037

also the oxidation reaction was more violent to release more chemical reaction heat than those of 30 μm particles. The flame propagation behavior, flame temperature distribution and ion current distribution all demonstrated the premixed-gas combustion dominant flame structure including the preheat zone, thin premixed reaction zone I, thick agglomerates reaction zone II, and post flame zone for 100 nm PMMA dust explosion, and the cluster diffusion gas combustion dominant flame struc­ ture including the preheat zone, combustion reaction zone and post flame zone for 30 μm PMMA dust explosion. The combustion and flame propagation of 100 nm and 30 μm PMMA dust clouds were mainly controlled by the heat transfer efficiency between the particles and external heat sources. Compared with the cluster diffusion dominant combustion of 30 μm dust flame, the premixed-gas dominant combustion of 100 nm dust flame determined a quicker pyrolysis and combustion reac­ tion rate, a faster flame propagation velocity, a stronger com­ bustion reaction intensity, a quicker heat release rate, and a higher amount of released chemical reaction heat, resulting in an earlier pressure rise, a larger maximum overpressure and a higher explosion hazard class. The complex combustion and propagation regime of agglomer­ ated particles strongly influenced the flame propagation regime and explosion pressure evolution characteristic, and limited the maximum overpressure during nano scale dust explosions.

References ASTM E1226, 2010. Standard Test Method for Explosibility of Dust Clouds. The American Society of Mechanic Engineers. Azhagurajan, A., Selvakumar, N., Mohammed Yasin, M., 2012. Minimum ignition energy for micro and nano flash powders. Process Saf. Prog. 31, 19–23. Ballal, D.R., 1983. Flame propagation through dust clouds of carbon, coal, aluminium and magnesium in an environment of zero gravity. P ROY SOC A-MATH PHY. 385, 21–51. Balbudhe, K., Roy, A., Chakravarthy, S.R., 2015. Computer modelling of nano-aluminium agglomeration during the combustion of composite solid propellants. P COMBUST INST 35, 2471–2478. Bidabadi, M., Haghiri, A., Rahbari, A., 2010. The effect of Lewis and Damk€ ohler numbers on the flame propagation through micro-organic dust particles. Int. J. Therm. Sci. 49, 534–542. Boilard, S.P., Amyotte, P.R., Khan, F.I., Dastidar, A.G., Eckhoff, R.K., 2013. Explosibility of micron- and nano-size titanium powders. J LOSS PREVENT PROC 26, 1646–1654. Bouillard, J., Vignes, A., Dufaud, O., Perrin, L., Thomas, D., 2010. Ignition and explosion risks of nanopowders. J. Hazard Mater. 181, 873–880. Di Benedetto, A., Russo, P., Amyotte, P., Marchand, N., 2010. Modelling the effect of particle size on dust explosions. Chem. Eng. Sci. 65, 772–779. Di Blasi, C., 1999. Transition between regimes in the degradation of thermoplastic polymers. POLYM DEGRAD STABIL 64, 359–367. Dobashi, R., 2009. Risk of dust explosions of combustible nanomaterials. J. Phys. Conf. Ser. 170, 1–6. Eckhoff, R.K., 2012. Does the dust explosion risk increase when moving from μm-particle powders to powders of nm-particles? J LOSS PREVENT PROC 25, 448–459. Gao, W., Mogi, T., Sun, J.H., Yu, J.L., Dobashi, R., 2013. Effects of particle size distributions on flame propagation mechanism during octadecanol dust explosions. Powder Technol. 249, 168–174. Gao, W., Yu, J.L., Zhang, X.Y., Li, J., Wang, B., 2015. Characteristics of vented nanopolymethyl methacrylate dust explosions. Powder Technol. 283, 406–414. Jiang, B.Y., Lin, B.Q., Shi, S.L., Zhu, C.J., Li, W.X., 2011. Explosive characteristics of nanometer and micrometer aluminum-powder. MINING SCI TECHNOL (China) 21, 661–666. Krietsch, A., Scheid, M., Schmidt, M., Krause, U., 2015. Explosion behaviour of metallic nano powders. J LOSS PREVENT PROC 36, 237–243. Mittal, M., 2014. Explosion characteristics of micron- and nano-size magnesium powders. J LOSS PREVENT PROC 27, 55–64. NFPA 68, 2018. Standard on Explosion Protection by Deflagration Venting. National Fire Protection Association. Russo, P., Amyotte, P.R., Khan, F.I., Di Benedetto, A., 2013. Modelling of the effect of size on flocculent dust explosions. J LOSS PREVENT PROC 26, 1634–1638. Tang, Y., Kong, C.D., Zong, Y.C., Li, S.Q., Zhuo, J.K., Yao, Q., 2017. Combustion of aluminum nanoparticle agglomerates: from mild oxidation to micro explosion. P COMBUST INST 36, 2325–2332. Vignes, A., Mu~ noz, F., Bouillard, J., Dufaud, O., Perrin, L., Laurent, A., Thomas, D., 2012. Risk assessment of the ignitability and explosivity of aluminum nanopowders. PROCESS SAF ENVIRON 90, 304–310. Zhang, X.Y., Yu, J.L., Gao, W., Zhang, D.W., Sun, J.H., Guo, S., Dobashi, R., 2017. Effects of particle size distributions on PMMA dust flame propagation behaviors. Powder Technol. 317, 197–208. Zhang, X.Y., Yu, J.L., Sun, J.H., Gao, W., 2016. Effects of turbulent intensity on nanoPMMA flame propagation behaviors. J LOSS PREVENT PROC 44, 119–124. Zhang, X.Y., Yu, J.L., Yan, X.Q., Xie, Q.F., Gao, W., 2016. Flame propagation behaviors of nano- and micro-scale PMMA dust explosions. J LOSS PREVENT PROC 40, 101–111.

CRediT authorship contribution statement Xinyan Zhang: Methodology, Investigation, Formal analysis, Writing - original draft, Project administration, Funding acquisition. Wei Gao: Supervision, Conceptualization, Writing - review & editing. Jianliang Yu: Supervision, Resources, Data curation. Yansong Zhang: Writing - review & editing. Jie Zhang: Writing - original draft. Xing­ wang Huang: Validation, Visualization. Jinshe Chen: Writing - review & editing. Acknowledgements The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China (NO. 51904170), Natural Science Foundation of Shandong Province (NO. ZR2018BEE006), Sci­ entific Research Foundation of Shandong University of Science and Technology for Recruited Talents (NO. 2017RCJJ007). The authors also thank Soken Chemical & Engineering Co., Ltd. of Japan for providing the experimental PMMA particles.

10