Possibility of Refuse Derived Fuel Fire Ineption by Spontaneous Ignition Lijing Gao 1, Takashi Tsuruda 1, Takeshi Suzuki 1, Yoshio Ogawa 1, Chihong Liao 1 and Yuko Saso 1 1 National Researh Institute of Fire and Disaster, Mitaka, Tokyo 181-8633, Japan Abstrat The possibility of fire ineption aused by spontaneous ignition of piled RDF was eluidated. Various properties of RDF were measured. Pyrolysis behavior of RDF was examined by TG-DTA and apparent ativation energy of RDF was alulated. In addition, the ritial ignition temperature (T ) of RDF was predited by the theory of thermal explosion by Frank-Kamenetskii, and the relation between the T and the height of RDF pile was alulated. The results show that the relation between the T and the height of RDF pile is almost idential for RDFs of different states, although the thermal ondutivity and bulk density of RDF sample depend on its state. This implies that the powdering of RDF does not lower the T largely. 1 1. Introdution On August 23, serious fire and explosion ourred at a refuse derived fuel (RDF) power station in Mie Prefeture, Japan. In the fire and explosion, two fire fighters were killed and 5 workers wounded. The fire ourred in a RDF storage, where no ignition soure was reported. In the initial stage of investigation, the ause of the fire was onsidered to be spontaneous ignition of piled RDF [1-6]. The proess of spontaneous ignition of this fire needs to be onfirmed by experiments and/or Corresponding Author- Tel.: +81-422-44-8331; Fax: +81-422-44-8393 E-mail address: gao@fri.go.jp theoretial analyses. The present study has been onduted to investigate this proess [6]. It is true that effetive energy reovery of muniipal solid waste is important for protetion of global environment. One of these methods is the use of RDF, whih is produed by shredding, sreening, drying and pelletising. For effetive energy reovery, a large-sale power station is needed. In suh a power station, a large apaity storage is onstruted for ontinuous operation. The objetive of this study is to eluidate the possibility of fire ineption aused by spontaneous ignition of piled RDF. Various properties of RDF were measured. Pyrolysis behavior of RDF was Copyright International Assoiation for Fire Safety Siene
examined by TG-DTA and apparent ativation energy of RDF was alulated. In addition, the ritial ignition temperature (T ) of RDF was predited by the theory of thermal explosion by Frank-Kamenetskii, and the relation between the T and the height of RDF pile was alulated. 2. Experimental Three kinds RDF were used to predit the possibility of fire ineption by spontaneous ignition. RDF-A was the sample from the rash site in Mie Prefeture. RDF-B and RDF-C were made from muniipal solid waste by different manufaturing fatories of RDF. Figure 1 shows the photographs of RDF. Table 1 is the results of analyses of three kinds RDFs. The result of analysis shows that the various properties of RDFs are almost similar exept for the moisture. To examine the pyrolysis and ombustion harateristi of RDF by TG-DTA analyses, RDF was rushed. The experimental onditions of TG-DTA are as Digital amera X-ray CT image Fig.1 Photographs of RDF Table 1 The results of analysis of tested RDF analysis item unit RDF-A RDF-B RDF-C moisture 26.8 2.9 7.3 ash 12.9 11.4 11.7 (wt%) volatile matter 72.6 76.3 74.7 fixed arbon 14.5 12.3 13.6 higher alorifi value 22.84 22.94 21.44 (MJ/kg) lower alorifi value 21.22 21.34 19.96 C 51.8 47.6 46.2 H 7.11 7.4 6.6 N 1.11 1.12 1.3 (wt%) S.54.59.48 Cl.8.9.53 Ca 2.2 2.1 2.7 follows: the reation atmospheres are nitrogen and air for pyrolysis and ombustion, respetively; The gas flow rate is 5 ml / min; About 1 mg of rushed RDF sample is heated at onstant heating rate of 1 C / min; The final temperature is raised to 1 C. The thermal ondutivity of RDF was measured by the laser flash method (state of RDF is pellet) and the needle probe method (state of RDF is powder). The heat of reation of RDF is alulated and the value of ΔH is assumed to be 5.5 % higher alorifi value than the measured one beause the mass loss rate is 5.5 % at the ignition temperature of RDF by TG urve. The speifi heat of RDF is measured using the DSC (Differential Sanning Calorimeter). The various properties of RDF are shown in Table 2. Table 2 Condutivity and bulk density of RDF sample properties of RDF unit powder pellet measurement temperature, K (K) 298 298 gas onstant, R (J mol -1 K -1 ) 8.31 8.31 bulk density, ρ (kg m -3 ) 4.21 1 2 9.91x1 2 speifi heat, (J kg -1 K -1 ) 1.42 1 3 1.42x1 3 thermal diffusivity, α (m 2 s -1 ) 9.8 1-8 1.21x1-7 thermal ondutivity, λ (J s -1 m -1 K -1 ) 5.87 1-2 1.62x1-1 apparent ativation energy, E (J mol -1 ) 89.82 1 3 - apparent frequeny fator, A (s -1 ) 3.6 1 5 - heat of reation, Δ H (J kg -1 ) 1.26 1 6 - δ (ube) - 2.52 - δ (infinite ylinder) - 2. - δ (infinite plane layer) -.88 -
3. Results and Disussions 3.1 Pryolysis behavior of RDF Pyrolysis behavior of RDF was examined by TG-DTA. Figure 2 shows a TG urve of RDF in the air atmosphere. Pyrolysis of RDF takes plae with about three stages. It seems that RDF dehydrated in a range of temperatures from room temperature to 15 C, and pyrolysis of organi element and ombustion took plae in a range of temperatures from 15 C to 55 C. Carbide omponent and inorgani element in RDF burned in a range of temperatures more than 55 C. Pyrolysis proesses of tested RDFs from other origins are similar to that presented in Fig.2. Weight, / % 12 1 8 6 4 2 RDF- A RDF- B 3.2 Calulation of apparent ativation energy The ombustion reation of the ombustibles in RDF an be expressed as: A + O 2 B + C(g) RDF- C 2 4 6 8 1 Temperature, / Fig.2 TG urve of RDF in the air atmosphere Sine the onentration of O 2 is exessive, the rate for the disappearane of reatant of A from the mixture an be expressed as follows: dα n = k( 1 α), (1) dt where α is the onversion of ombustion reation, k is the reation rate onstant, n is the reation order, and t is reation time. Aording to the Arrhenius expression, the reation rate onstant an be expressed as k = Ae E / RT, (2) while heating rate an be given by φ = dt dt (3) Substitutions of Eqs.2 and 3 into Eq.1 lead to the following equation: dα A e E/ RT n = ( 1 α), (4) dt φ where E is the apparent ativation energy, A is the apparent frequeny fator and φ is the heating rate, respetively. In determination of the kineti parameters by means of the above equations, various analysis methods have been suggested [7-11]. In the present study, the nonisothermal analysis method proposed by Ito et al. was used [11]. Introduing logarithm into Eq.4, the following equation is obtained: L M NM O P Q d φ α ln dt E M ln A ( α ) n 1 P = RT (5) The values of dα /dt, α, and T an be obtained from results of TG analysis. α an be alulated from the weight-losing urve of TG using the following equation: ( w w ) α = ( w w ), (6)
where w, w, and w are the sample weight before the ombustion reation, in the ombustion reation, and after the ombustion reation, respetively. To determine the apparent kineti parameters in Eq.5, ombustion is generally analyzed as a first-order reation. Assuming that n is equal to 1, the alulated results of the left-hand term in Eq.5 are plotted against the reiproal of absolute temperature. It shows a straight line in ln((1-α) -n (dα /dt)φ) versus 1/T, and then the apparent ativation energy E and the apparent frequeny fator A an be obtained from slope and interept of the straight line aording to Eq.5. It is seen that the apparent ativation energies and the apparent frequeny fators are different in different reation stages. To predit the possibility of RDF fire ineption by spontaneous ignition, it is important that onsidering the effet of the apparent ativation energy under ignition temperature on spontaneous ignition of RDF. Sine the ignition temperature of RDF is 24 C, while RDF was dehydrated from room temperature to 15 C, so the apparent ativation energy of RDF between 15 C-24 C was alulated. The value of the apparent ativation energy E and the apparent frequeny fator A are 89.82 kj / mol and 3.6 1 5 1 / s, respetively. 3.3 Predition of ritial ignition temperature (T ) on the basis of the thermal explosion theory of Frank-Kamenetskii The ignition phenomenon has been explained or predited on the basis of thermal explosion theories, suh as the Semenov, Frank-Kamenetskii and Thomas theories [12-18]. Considering that RDF is pratially stored with a large pile (the size of storage of RDF in Mie Prefeture is 15 m in inside diameter and 25 m in height), in whih temperature is not uniform due to low thermal ondutivity. Therefore, in the present analysis using the Frank-Kamenetskii theory, an assumption of non-uniform temperature is adopted. The theory of Frank-Kamenetskii is shown by the following expressions [13]: 2 n ΔH E r A C E δ = exp, (7) 2 λ RT RT L NM O Q P where ΔH, E, A, R and C are the heat of reation, the apparent ativation energy, the apparent frequeny fator, the gas onstant and the initial onentration of sample, respetively. The RDF is solid, and the initial sample density is onsidered to be equal to the bulk density of RDF. n is a reation order, and in the present study, the pyrolysis reation of RDF an be assumed to be of a first-order, i.e., n is equal to 1. In Eq.7, δ, λ, and r are the ritial parameter for explosion, thermal ondutivity of sample, and radius of the RDF pile, respetively. T is the ritial ignition temperature. Napierian logarithm is taken and arranged in both sides of the Eq.7. Then it beomes as follows: E 1 ΔH A E C lnt + = lnr + ln 2RT 2 δ λ R L NM O QP (8) Eah parameter about RDF shown in Table 2 is substituted to Eq.8. Consequently, the relation between the T and the height of pile of RDF an be predited. Figure 3 shows the result. It is seen that T dereases with the height of RDF pile. The T is lowest in the ase of infinite plane layer of T in the ases of other onfiguration of similar thiknesses of RDF piles. For the RDF stored in a storage, the predition for an infinite ylinder was adopted to investigate the relation between the T and the height of the RDF pile. For example, when RDF is assumed to store up to 5 m in height, the T
is about 4 C. While in the ase of Mie Prefeture, RDF was stored in a storage that T and the height of RDF pile are the heat of reation ΔH, the apparent ativation energy 15 15 T, / 12 9 6 Cube (δ =2.52) infinite ylinder (δ =2.) infinite plane layer (δ =.88) T, / 12 9 6 Powder pellet 3 3 5 1 15 2 25 height of RDF pile, / m Fig.3 The relation between the T and the height of RDF pile the inside diameter is 15 m, the T will be about 2 C. It means that the proess to store RDF in a large pile, there is a possibility that a fire happens even though the surrounding environmental temperature is the room temperature. In fat, serious fire and explosion have ourred at the RDF power station in Mie Prefeture, Japan on August 23. 3.4 Effet of the RDF state on the possibility of fire ineption It was onsidered that rushed RDF auses the fire by spontaneously igniting in a RDF pile at the initial stage of investigation. To onfirm effet of the RDF state on the possibility of fire ineption, two kinds of RDF with different states (powder and pellet) are used. It is seen that the thermal ondutivity and bulk density of RDF sample depend on its state as shown in Table 2. The thermal ondutivity of rushed RDF (powder) is lower than that of the pellet, and the bulk density of rushed RDF is smaller than pellet of RDF. When the ritial parameter δ for explosion is onstant, that is, in ase of the state of the RDF pile is onstant, the fators influening the relation between the 5 1 15 2 height of RDF pile, / m Fig.4 Effet of the RDF state on the relation between the T and the height of RDF pile E, the apparent frequeny fator A, the thermal ondutivity λ, and the initial onentration C of sample (or bulk density ρ of sample), respetively. The effet of E is the largest beause other influene fators are in the logarithm term. In addition, in the ase when the bulk density and the thermal ondutivity of sample derease at the same time, it an be presumed that the value of ln (ρ / λ) sarely hanges. Figure 4 shows the alulation result of the relation between the T and height of RDF pile for various bulk densities and thermal ondutivities. It is seen that the relation between the T and the height of RDF pile is almost idential for powder and pellet of RDF. This indiates that the powdering of RDF does not appreiably lower the T. 4. Conlusions 1. Pyrolysis proesses of RDFs from different soures are similar and take plae in three stages. The main proess is pyrolysis of organi element and ombustion that takes plae in a temperature range of 15 C-55 C. 2. The apparent ativation energy E at the
ignition temperature of RDF was alulated, whih is 89.82 kj / mol. 3. The ritial ignition temperature (T ) of RDF was predited on the basis of the Frank-Kamenetskii theory, and the relation between T and the height of RDF pile was alulated. It is shown that 4 C is enough to ignite RDF if the pile is 5 m in height. 4. Although the thermal ondutivity and bulk density of RDF sample depend on its state, the relation between T and the height of RDF pile is almost independent of the state. This result implies that the powdering of RDF sarely lower T. Aknowledgement The authors express there thanks to Toray tehno Ltd. for their help in the analyses of RDF. Referenes 1. RDF power station aident investigation tehnial ommittee, Mie Prefeture, RDF power station aident investigation interim report, 23, 1-2 2. Audit ommittee, Mie Prefeture, Regular audit debrief report (Mie Prefeture RDF power station audit debrief report) in 23 fisal year, 23,1-5 3. RDF propriety management study ommittee, The Ministry of the Environment, Proper management strategy of RDF, 23,1-15 4. Eletri power safety subommittee RDF power station aident investigation working group, The Ministry of Eonomy, Trade and Industry, Eletri power safety subommittee RDF power station aident investigation working group report, 23,1-42 5. RDF power station aident investigation tehnial ommittee, Mie Prefeture, RDF power station aident investigation finality report, 23,1-46 6. The Ministry of Publi Management, Home Affairs, Posts and Teleommuniations Fire and Disaster Management Ageny, Examination report of safety measures investigation of failities related to RDF, 23,1-234 7. E. S. Freeman, B. Carroll, Fuel 1958, 62, 394-41 8. H. J. Borhardt, F. J. Daniels, J. Am. Chem. So. 1957, 79, 41-46 9. A. W. Coats, J. P. Redfern, Nature 1964, 21, 68-72 1. C. D. Doyle, J. Appl. Polum. Si. 1961, 5, 285-291 11. N. Ito, K. Obata, T. Hakuta, H. Yoshitome, Kagaku Kougaku Ronbunshun, 1983, 9, 434-439 12. N. N. Semenov, Chemial Kinetis and Chain Reations, Oxford Univ. Press, 1935 13. D. A Frank-Kamenetskii, Diffusion and Heat Transfer in Chemial kinetis, Plenum Press, 1969 14. P. Gray, P. et al, Trans. Farad. So., 1959, 55, 581 15. P. Gray, P. et al, Comb. and Flame, 1959, 13, 461 16. C. L. Chambre, J. Chem. Phys., 1952, 2, 1795 17. J. W. Enig, Comb. and Flame, 1966, 1, 197 18. P. H. Thomas, Trans. Farad. So. 1958, 54, 6