1、NA-04-1 -4 Assessing Combustion Hazard of Flammable Gases: Revision of RF-Number Shigeo Kondo, Ph.D. Kenji Takizawa, Ph.D. A kif um i Ta ka hash i Kazuaki Tokuhashi ABSTRACT Revision of the RF-number has been made to give RF2 for assessing more precisely the combustion hazard in treating flammable g
2、ases and their mixtures. RF2 represents the expectancy of the combustion hazard, which takes into account the heat of combustion, flammability limits, and burning veloc- ity in a systematic way. INTRODUCTION Information on flammability limits is indispensable for safe treatment of combustible gases
3、and vapors. Normaliza- tion of the flammability limits enables one to develop system- atic discussion of flammability characteristics. The normalized flammable range is given by F-number (Kondo et al. 2001). F= i -(L/ r40.5 (1) Here, Uand L are the upper and lower flammability limits. F- number take
4、s values ranging from zero to unity depending on the degree of flammability. In addition, F-number is conve- nient because it can be expressed by an empirical equation in terms of parameters related to molecular structure and is used to predict the flammability limits of unknown compounds. However,
5、F-number alone is not sufficient to express the total combustion hazard of flammable gases. There is another index, called R-index, which is defined by the following equa- tion (Kataoka 2001), R = CJL, (2) where C, is the stoichiometric concentration. On the whole, the stronger the fuel, the larger
6、the value of R-index as well. In this sense, it can be taken as another way of expressing the Akira Sekiya, Ph.D. flammability range. However, a concern about this index is that the value is distorted by the effect of selective diffusion, especially for compounds with heavy molecules. ASHRAE is cons
7、idering a criterion for classifying refrig- erants according to their flammability characteristics (ASHRAE 2000). The criterion is to set threshold limit values for the lower flammability limit and heat of combustion. The value for the lower flammability limit is 0.10 kg/m3, and that for heat of com
8、bustion is 19000 kJ/kg. If the lower flamma- bility limit is larger than O. 1 Okg/m3 and the heat of combustion is smaller than 19000kJ/kg, class 2 is assigned to the refriger- ant. On the other hand, if the lower limit is smaller than 0.10 kg/m3 or the heat of combustion is larger than 19000, class
9、 3 is assigned to the refrigerant. The International Organization for Standardization (ISO) is also considering a similar method for classification of refrigerants (Walter 2000), except that IS0 assumes the criterion for the lower flammability limit to be 3.5 vol % instead of O. 10 kg/m3. The first
10、item of these crite- ria can be considered a factor of probability and the second a factor of intensity. In a previous paper (Kondo et al. 2002), we introduced a combustion hazard index called RF-number, systematically organizing the ASHRAE criterion. At first, the ignition prob- ability part of the
11、 ASHRAE criterion was reexamined and improved. Then, the probability factor and the intensity factor were combined together to give an expectancy of total combustion hazard per unit mass of combustible materials. RF-number is given by the following equation, ( uL)0.5 - L Q L M RF = (3) where, Uand L
12、 are the upper and lower flammability limits, Q S. Kondo, K. Takizawa, and A. Takahashi are researchers, K. Tokuhashi is a team leader, and A. Sekiya is vice-director, Flourine Chemistry Center, National Institute of Advanced Industrial Science and Technology, Tsukuba, Ibaraki, Japan. 534 02004 ASHR
13、AE. 350 I v - H2 . . .g 200 I - - - , - O - $ 150 . ._. O 1 O0 200 300 400 500 RF-number (kJ/g) Figure I Plot of maximum burning velocity vs. RF-number for a number of compounds. is molar heat of combustion, and Mis molecular weight. Also, there has been discussion that the assessment of the combust
14、ion hazard of refrigerants should take into account the burning velocity as well (Clodic 2001). In fact, when combus- tible materials are burnt in a laboratory or anywhere, the threatening power of combustion felt by the observers may come from the production rate ofheat, sound, and light. These qua
15、ntities are direct functions of burning velocity and may be related to the destructive power of combustion as well. There- fore, it seems desirable to reconstruct the combustion hazard index using the burning velocity as one of the major parts so that the index may express more precisely the power o
16、f flames supported by the individual materials. In the present paper, we have revised RF-number to give RF2 from this point of view. RESULTS AND DISCUSSION To begin with, we have investigated to find correlation between RF-number and the maximum burning velocity. Since both of these quantities are c
17、losely related to the combustion power of substances, it is very probable that a close correlation exists between them. If that is the case, there may be no need to explicitly add information about the burning velocity to the assessment of combustion hazardmade in terms of RF-number. Figure 1 shows
18、the result. It has been found that the maximum burning velocity changes almost proportionally to RF-number from compound to compound, though there are some exceptions. Figure 2 is similar to Figure 1 but corre- sponds to smaller values ofparameters. In this figure, there are multiple points for R-32
19、, NH, R-152a, CH, and R-290. The values of flammability limits corresponding to the black circles of CH, R-l52a, and R-290 were obtained with the ASHRAE method (ASHRAE 2000). It is noted that the point for carbon monoxide is at a distance from the general trend followed by other compounds. In the co
20、mbustion reaction of methane, the oxidation reaction may proceed through CH, CH20, CHO, and CO to reach CO,. Carbon monoxide appears 50 h 40 E v 30 .- o - 20 .- c m 10 f O C2H6 0 o “t O R600 0.3 o CH4 O 10 20 30 40 50 60 RF-number (kJ/g) Figure 2 Plot of maximum burning velocity vs. RF-number in a r
21、egion corresponding to smaller values of parameters. in the final stage ofthis series, and its oxidation ends with only one step, while that of methane proceeds in a series of five consecutive reactions. Whatever the counterpart of reaction, oxidation of carbon monoxide may proceed in one step or so
22、 in the combustion reaction of CO itself as well. This may be the reason why the burning velocity of carbon monoxide is fairly large in spite of its relatively small heat of combustion. The burning velocities of oxygen-containing compounds are in general a little larger than the ones of correspondin
23、g pure hydrocarbon compounds. At any rate, considering the nature of burning velocity and the definition of RF-number, it is not necessary that the two quantities be in complete correlation with each other. Now, in order to assess combustion characteristics of gases and vapors, various experimental
24、quantities are avail- able. The fundamental ones are heat of combustion, flamma- bility limits, burning velocity, minimum ignition energy, quenching distance, maximum flame temperature, maximum pressure, and rate of pressure rise in a closed vessel. Each of these quantities reflects a certain aspect
25、 of combustion phenomenon. However, it is not necessary to take into account all of them together for the assessment of combustion hazard because at times there are correlations among these quantities. In the following, we have examined correlations between vari- ous pairs from such quantities as he
26、at of combustion, flamma- bility limits, burning velocity, minimum ignition energy, quenching distance, and RF-number. The correlation between the minimum ignition energy and quenching distance is well known. Lewis and von Elbe (1 96 1) at first reported that the minimum ignition energy is proportio
27、nal to the square of quenching distance. Later they revised this and reported that the minimum ignition energy is proportional to the cubic power of quenching distance for extremely strong flames and proportional to the quadratic power for not so strong flames (Lewis and von Elbe 1987). In this case
28、, even the flames supported by stoichiometric hydro- carbon-air mixtures are classified as not so strong flames. The ASHRAE Transactions: Symposia 535 350 300 h . 250 o v O 50 01 I O IO 20 30 40 50 60 Reciprocal minimum ignition energy (mJ-) -e0 Figure 3 Plot of maximum burning velocity vs. reciproc
29、al minimum ignition energy for a number of compounds. quadratic correlation stands for quite a wide range of compounds. On the other hand, the minimum ignition energy is given by the heat loss from the surface of minimal flame within a time interval of Ma, where F is the width of flame front and Sa,
30、 the burning velocity averaged from unbumt gas temperature Tu to burnt gas temperature T,. From this, one obtains the following equation (Kondo et al. 2003): (4) Here, d is the diameter of minimal flame and ha, the heat conductivity averaged for the temperature range from Tu to T,. Figure 3 shows a
31、plot of reciprocal minimum ignition energy against burning velocity for a number of typical fuel compounds for which the data of both burning velocity and minimum ignition energy are available. A fairly good straight line is obtained. This indicates that the quantity excluding Sa, from the right-han
32、d side of Equation 4 does not change much from compound to compound. Similarly, the plot of reciprocal square of quenching distance against burning velocity makes a fairly good straight line as well. This is expected because the minimum ignition energy has a good linear relationship with the square
33、of quenching distance and the reciprocal minimum ignition energy has a fairly good correlation with burning velocity as well. On the other hand, consideration of heat of combustion is crucial for the assessment of combustion hazard. There is a loose correlation between the heat of combustion and rec
34、ip- rocal lower flammability limit of hydrocarbon compounds, which is called the Burgess-Wheeler plot (Burgess and Wheeler 191 1). We investigated further to find out additional correlations of heat of combustion with other quantities, but 536 the attempt was not successful. As for F-number, a simil
35、ar investigation was made but no close correlation was found with any quantities. Now, for revision ofRF-number, it is desirable to hold fast to the original philosophy that the expectancy of combustion hazard in treating flammable gases is given by a product of ignition probability factor and inten
36、sity factor. The minimum ignition energy is a typical quantity related to ignition probability of combustible gas mixtures. In order to cause ignition, an energy larger than a certain minimum value must be given to a fuel-air mixture. Before discussing the ener- getic condition, however, the materia
37、l condition must be satis- fied, i.e., the concentration of mixture must be within the flammable range. When a leakage occurs, a large concentra- tion gradient is produced at the leak spot, and gas mixtures of various concentrations are produced. As a first approximation, the probability of ignition
38、 can be considered proportional to the width of flammable range. In addition, the smaller the lower flammability limit, the easier the flammable mixtures are produced around the leak spot. In the end, we assumed that the total ignition probability is proportional to the lower half of the flammable r
39、ange divided by the lower flammability limit. ( UL)0.5 - L/L = ( U/L)0.5 - 11 = Fi( 1 - F) (5) where, U and L are the upper and lower flammability limits and F is given by Equation 1. At this stage, there would be a discussion of whether the stoichiometric concentration C, could be a substitute for
40、the geometric mean of upper and lower flammability limits. If so, we may not need the data of upper flammability limit. Unfortunately, however, this is not acceptable. As stated above, the effect of selective diffusion makes the value of (CS,-L) definitely different from the lower half of the flamma
41、ble range. It is not rare that the discrepancy between the geometric mean and the stoichiometric concen- tration reaches 30% or more. In fact, it was shown in our previ- ous paper that the relative deviation between the geometric mean and C, can be expressed by 0.0047(M-32) for a compound with molec
42、ular weight of A4 (Kondo et al. 2001). The value of this term may become as much as 32% if M equals 100 and 55% ifMequals 150. As for the intensity factor, it may be preferable that the burning velocity be taken into consideration. In fact, the burn- ing velocity is one of the most important indices
43、 that are related to combustion intensity. In addition, for the laboratory experiments recently conducted to investigate the actual flames supported by various refrigerant materials, the combustion power of flame itself seemed to be the main concern of the people (Clodic 2001). The combustion power
44、of flame may essentially be determined by the heat production rate. In some cases, the rate of pressure rise due to explosion may matter as well. The rate of pressure rise is also a function of the molar ratio of materials before and after the combustion reaction (Richard 2003). This ratio may be re
45、latively large for fluorinated compounds. For example, if the combustion reac- ASHRAE Transactions: Symposia Table 1. RF2 Value for a Number of Selected Compounds Substance Ammonia Difluoromethane 1,1 -Difluoroethane Methane n-Butane i-Butane Ethane Propane i-Pentane n-Pentane Cyclohexane Carbon mon
46、oxide n-Hexane Propene Ethene Ethylene oxide Hydrogen Acetylene HOC Stoichio-metric LFL UFL Burning velocity RF2 Mol Wt (kJ/mol) (volo) (voio) (vol%) (m/s) (kJ/mol)(m/s) 17.0 317 21.8 15.0 28 0.06 1.5 52.0 489 17.4 13.3 29.3 0.055 2.3 66.1 1093 7.8 4.4 17.5 0.2 1 18 16.0 799 9.5 4.9 15.8 0.37 22 58.
47、1 2650 3.1 1.8 8.4 0.35 34 58.1 2642 3.1 1.8 8.4 0.38 36 30.1 1427 5.6 3.0 12.4 0.44 37 44.1 204 1 4.0 2.0 10.0 0.37 37 72.1 3255 2.6 1.4 7.6 0.40 44 72.1 3264 2.6 1.5 7.8 0.42 45 84.2 3685 2.3 1.3 7.8 0.42 51 28.0 283 29.5 12.5 74 0.43 51 86.2 3881 2.2 1.2 7.4 0.42 52 42.1 1924 4.4 2.0 11.0 0.48 55
48、 28.1 2537 6.5 2.7 36 0.75 171 44.1 3766 7.7 3.0 1 O0 1 .o0 449 2.0 1323 29.5 4.0 75 2.91 685 26.0 1218 7.7 2.5 1 O0 1.55 80 1 tion is for the stoichiometric mixture with oxygen, the molar ratio is 1.43 for HFC-152a and 1.67 for HFC-143a. However, if it is the stoichiometric mixture with air, the ra
49、tios may reduce to 1.12 and 1.19forHFC-l52aandHFC-l43a, respec- tively. In addition, these numbers are for the extreme cases where the whole space in question is filled with the stoichio- metric mixtures and the explosion extends to the full space in a flash. Otherwise, the ratio may not become substantially different from unity. In the present study, we have just ignored it for the sake of simplicity. Eventually, we decided to employ the heat production rate per unit flame area as the intensity factor for a new version of RF-number, “RF2”. RF2 can be defined by the followi
