• unlimited access with print and download
    $ 37 00
  • read full document, no print or download, expires after 72 hours
    $ 4 99
More info
Unlimited access including download and printing, plus availability for reading and annotating in your in your Udini library.
  • Access to this article in your Udini library for 72 hours from purchase.
  • The article will not be available for download or print.
  • Upgrade to the full version of this document at a reduced price.
  • Your trial access payment is credited when purchasing the full version.
Buy
Continue searching

Flammability limits, flash points, and their consanguinity: Critical analysis, experimental exploration, and prediction

Dissertation
Author: Jef Rowley
Abstract:
Accurate flash point and flammability limit data are needed to design safe chemical processes. Unfortunately, improper data storage and reporting policies that disregard the temperature dependence of the flammability limit and the fundamental relationship between the flash point and the lower flammability limit have resulted in compilations filled with erroneous values. To establish a database of consistent flammability data, critical analysis of reported data, experimental investigation of the temperature dependence of the lower flammability limit, and theoretical and empirical exploration of the relationship between flash points and temperature limits are undertaken. Lower flammability limit measurements in a 12-L ASHRAE style apparatus were performed at temperatures between 300 K and 500 K. Analysis of these measurements showed that the adiabatic flame temperature at the lower flammability limit is not constant as previously thought, rather decreases with increasing temperature. Consequently the well-known modified Burgess-Wheeler law underestimates the effect of initial temperature on the lower flammability limit. Flash point and lower temperature limit measurements indicate that the flash point is greater than the lower temperature limit, the difference increasing with increasing lower temperature limit. Flash point values determined in a Pensky-Martens apparatus typically exceed values determined using a small-scale apparatus above 350 K. Data stored in the DIPPR® 801 database and more than 3600 points found in the literature were critically reviewed and the most probable value recommended, creating a database of consistent flammability data. This dataset was then used to develop a method of estimating the lower flammability limit, including dependence on initial temperature, and the upper flammability limit. Three methods of estimating the flash point, with one based entirely on structural contributions, were also developed. The proposed lower flammability limit and flash point methods appear to predict close to, if not within, experimental error. Keywords: flammability limit, flash point, DIPPR, flame propagation

TABLE OF CONTENTS

Chapter 1.   Introduction ............................................................................................................. 1   Chapter 2.   Flammability Limits ............................................................................................... 3   2.1   Experimental Determination ........................................................................................... 3   2.1.1   Bureau of Mines Tube Method ................................................................................... 3   2.1.2   ASHRAE Method ....................................................................................................... 4   2.1.3   EN 1839 Tube Method ................................................................................................ 5   2.1.4   Differences in Measurement Methods ........................................................................ 5   2.2   Temperature Dependence of the Flammability Limit ..................................................... 6   2.3   Flammability Limit Estimation Methods ........................................................................ 8   2.3.1   Chemical Equilibrium Methods .................................................................................. 8   2.3.2   Empirical Correlations .............................................................................................. 10   Chapter 3.   Temperature Limits and Flash Points ................................................................ 15   3.1   Flammability Temperature Limits ................................................................................ 15   3.1.1   Flammability Temperature Limits: Experimental Determination ............................ 15   3.1.2   Flammability Temperature Limits: Estimation ......................................................... 16   3.2   Flash Point .................................................................................................................... 16   3.2.1   Flash Point: Experimental Determination ................................................................. 16   3.2.2   Flash Point: Estimation ............................................................................................. 18   3.3   Interrelation of Fire-Hazard Properties ......................................................................... 21   Chapter 4.   Experimental Method ........................................................................................... 25   4.1   Selection of Measurement Compounds ........................................................................ 25   4.2   Lower Flammability Limit ............................................................................................ 26   4.3   Lower Temperature Limit ............................................................................................. 29  

i

4.4   Flash Point .................................................................................................................... 30   4.4.1   Pensky-Martens Procedure ....................................................................................... 30   4.4.2   Small-Scale Procedure .............................................................................................. 30   4.4.3   Additional Measurements ......................................................................................... 31   Chapter 5.   Experimental Results ............................................................................................ 32   5.1   Lower Flammability Limit as a Function of Temperature ........................................... 32   5.1.1   Experimental Data .................................................................................................... 32   5.1.2   Adiabatic Flame Temperature Analysis .................................................................... 34   5.1.3   Flame Temperatures and the Theory of Flammability Limits .................................. 38   5.1.4   Comparison with Reported Data ............................................................................... 43   5.2   Lower Temperature Limit and Flash Point ................................................................... 48   5.2.1   Calculated vs. Experimental Lower Temperature Limit ........................................... 48   5.2.2   Flash Point ................................................................................................................ 51   5.2.3   Effect of Measurement Parameters on the Flash Point ............................................. 55   5.2.4   Flash Point vs. Lower Temperature Limit ................................................................ 57   5.2.5   Interrelationship of Flash Point and Lower Flammability Limit .............................. 60   Chapter 6.   Critical Review of Previously Reported Data .................................................... 62   6.1   Flammability Data Compilations .................................................................................. 62   6.2   DIPPR ® 801 Database .................................................................................................. 65   6.3   Review Methodology .................................................................................................... 66   6.3.1   Measurement Method ............................................................................................... 67   6.3.2   Interrelation of Properties ......................................................................................... 68   6.3.3   Chemical Series Trends ............................................................................................ 71   6.3.4   Example Data Evaluations ........................................................................................ 74   6.3.5   Special Considerations .............................................................................................. 78  

ii

6.4   Database Statistics ........................................................................................................ 78   Chapter 7.   Property Estimation.............................................................................................. 86   7.1   Lower Flammability Limit ............................................................................................ 86   7.1.1   Magnitude Estimation ............................................................................................... 87   7.1.2   Estimation of Temperature Dependence ................................................................... 99   7.1.3   Overall Estimation .................................................................................................. 104   7.2   Upper Flammability Limit .......................................................................................... 106   7.3   Flash Point .................................................................................................................. 112   7.3.1   Property Correlations .............................................................................................. 112   7.3.2   Structural Contribution Method .............................................................................. 119   Chapter 8.   Summary and Recommendations ...................................................................... 124   8.1   Summary ..................................................................................................................... 124   8.2   Recommendations ....................................................................................................... 126   References .................................................................................................................................. 127   Appendix A.   Variables Affecting Flammability Limits ..................................................... 145   Appendix B.   Summary of Published Estimation Methods ................................................ 151   Flammability Limit Estimation Methods ................................................................................ 151   Flash Point Estimation Methods ............................................................................................. 169   Appendix C.   Apparatus Specifications ................................................................................ 182   ASHRAE Flammability Apparatus ........................................................................................ 182   Pressure Correction ................................................................................................................. 188   Flash Point Apparatuses .......................................................................................................... 188   Appendix D.   Experimental Results ...................................................................................... 191   Raw Data ................................................................................................................................. 191   Confidence Regions from Lower Flammability Limit Regression ........................................ 207  

iii

Appendix E.   Regression Datasets ......................................................................................... 213  

iv

LIST OF TABLES

Table 2.1. Comparison of three widely-used standardized flammability apparatuses ................... 5   Table 2.2. Group contributions for Zatsepin, Sorokin, and Stepachev [30] estimation method for lower flammability limits .................................................................................................... 9   Table 2.3. Single point prediction methods for lower and upper flammability limits .................. 12   Table 3.1. Five main flash point apparatuses currently used ........................................................ 17   Table 3.2. ASTM standardized method of measuring the flash point .......................................... 17   Table 3.3. Published methods of estimating the flash point ......................................................... 19   Table 4.1. Compounds for which experimental work was performed .......................................... 26   Table 5.1. Experimental lower flammability limit data and 95 % confidence intervals at temperatures between 300 K and 500 K ................................................................................. 33   Table 5.2. Comparison of reported calculated adiabatic flame temperatures (K) with values found using CEA, a chemical equilibrium calculator, and Equation 5.1 ................................ 35   Table 5.3a. The effect of initial temperature, T, on calculated adiabatic flame temperatures, T ad , for reported lower flammability limit data ....................................................................... 37   Table 5.4. Definitions of flame propagation listed in Figure 5.5 .................................................. 42   Table 5.5. Regressed slopes and intercepts from experimental lower flammability limit data .... 44   Table 5.6. Comparison of experimental lower flammability limit results with single-point literature values ....................................................................................................................... 47   Table 5.7. Lower temperature limits, T L ....................................................................................... 51   Table 5.8. Pensky-Martens (PM) and small-scale (SS) flash point measurements; p-value based on two-sided t-test for statistical difference between experimental values .................. 52   Table 5.9. Comparison of experimental flash points with literature values ................................. 52   Table 5.10. Effect of ramping rate, wait-time, and sample injection volume on the experimental flash point .......................................................................................................... 56   Table 5.11. Evaluation statistics of published relationships between the flash point and the lower flammability limit ......................................................................................................... 60   Table 5.12. Evaluation statistics of published relationships between the flash point and the lower flammability limit when parameters have been fit using the experimental data of this work.................................................................................................................................. 61  

v

Table 6.1. Reported flammability data for butanol ....................................................................... 74   Table 6.2. Flash point data for 1,6-hexanediol ............................................................................. 76   Table 6.3. Absolute changes in the recommended value greater than 100 % .............................. 84   Table 6.4. Average changes made to recommended values in the DIPPR ® 801 database ........... 85   Table 7.1. Average absolute deviation (AAD) of published lower flammability methods for data from the Bureau of Mines apparatus (BoM), the European standardized method (EN), the ASHRAE apparatus, and recommended lower flammability limit values ............. 88   Table 7.2. Average absolute deviation (AAD) of published lower flammability limit methods that do not apply to general organic compounds .................................................................... 89   Table 7.3. Structural contributions for Equation 7.5, regressed using lower flammability limit data from the Bureau of Mines apparatus (BoM), the European standardized method (EN), the ASHRAE method, and the recommended values ................................................... 93   Table 7.4. Average absolute deviations for values predicted using Equation 7.6 ........................ 95   Table 7.5. Errors exceeding 30 % when the lower flammability limit is predicted using Equation 7.6 ............................................................................................................................ 98   Table 7.6. Average absolute deviation (AAD) of published upper flammability methods for data from the Bureau of Mines apparatus (BoM), the European standardized method (EN), the ASHRAE apparatus, and recommended upper flammability limit data ............... 106   Table 7.7. Average absolute deviation (AAD) of published upper flammability limit methods that do not apply to general organic compounds .................................................................. 107   Table 7.8. Structural contributions for the prediction of the upper flammability limit using Equation 7.15 ........................................................................................................................ 108   Table 7.9. Average absolute deviations for prediction of the upper flammability limit using Equation 7.15 ........................................................................................................................ 110   Table 7.10. Errors exceeding 50 % when Equation 7.15 is used to estimate the upper flammability limit ................................................................................................................. 111   Table 7.11. Structural groups that determine the parameter k for Equation 7.16 ....................... 113   Table 7.12. Sample calculations for Equation 7.16 using groups from Table 7.11 .................... 115   Table 7.13. Average absolute deviation by estimation method for 1062 organic compounds in DIPPR ® 801 database ........................................................................................................... 115   Table 7.14. Reported flash points that differ significantly from the values predicted by the presented methods ................................................................................................................. 119  

vi

Table 7.15. Parameters and structural contributions for Equation 7.23 ...................................... 121  

vii

LIST OF FIGURES

Figure 2.1. ASHRAE 90 o flame propagation criterion ................................................................... 4   Figure 3.1. Vapor pressure at the flash point against the inverse of the moles of oxygen required for stoichiometric combustion. The solid line is the fit proposed by Leslie and Geniesse .................................................................................................................................. 20   Figure 3.2. Flammability diagram illustrating the relationship between the flammability limits (LFL/UFL), flash point (FP), and temperature limits (T L /T U ) ...................................... 22   Figure 4.1. Schematic for 12 L flammability apparatus; A. Oven, B. Magnetic stir mechanism, C. Electrode, D. RTD, E. Silicon stopper, F. Spring-loaded cover clamp, G. Air/fuel inlet ............................................................................................................................ 27   Figure 5.1. Lower flammability limit data versus temperature for: 2-nonanone (●); decyl acetate (□); isopropyl myristate (◊). Linear fits are given by dashed lines .......................... 34   Figure 5.2. Calculated adiabatic flame temperature, T ad , from experimental lower limit data as a function of initial mixture temperature: methanol (●); butanol (◊); 4-methyl-2- pentanol (+); 1-octanol (ⅹ); 2-methyl-1,3-propanediol (□) ................................................... 36   Figure 5.3. The effect of temperature on the lower flammability limit calculated by the modified Burgess-Wheeler law ( - - -), compared with experimental data for: 4-methyl-2- pentanol (●); butanol (□); 1-octanol (∆) ................................................................................. 39   Figure 5.4. Adiabatic flame temperatures assuming CO formation (T ad, CO ) at the lower temperature limit; OH termination becomes dominant over branching below about 1000 K, shown by the dashed line ................................................................................................... 40   Figure 5.5. Ratio of CO and CO 2 concentrations from product gases of flames that propagated to different extents: methanol at 100 o C (△) and 150 o C (▲), butanol at 150 o C (◊), 1-octanol at 100 o C (□) and 150 o C (■), decyl acetate at 150 o C (○) and 200 o C (●), dibutyl amine at 100 o C (ⅹ),and diisobutyl phthalate at 200 o C (+); 90 o ASHRAE propagation criterion is given by dashed line, and definitions of the flame propagation are given in Table 5.4 ................................................................................................................... 42   Figure 5.6. The slope of the adiabatic flame temperature with respect to initial mixture temperature assuming CO 2 (□) and CO (●) as the main carbon product ................................ 42   Figure 5.7. Confidence regions for the slope (a) and intercept (b) of the lower flammability limit as a function of temperature for ethyl lactate ................................................................. 44   Figure 5.8. Experimental lower flammability limit data for methanol; [117] (○); [118, 125] (□); [119] (∆); calculated from a single data point from ref [70] and reported slope from

viii

ref [21] (■- - -); data point from [32] with slope calculated using the modified Burgess- Wheeler law (♦- - -); this work (●) ......................................................................................... 45   Figure 5.9. Experimental lower limit data for butanol (●■♦+) and 1-octanol (○□); calculated from a single data point from [70] and reported slope from [21] (■/□- - -); data point from [32] and slope calculated using the modified Burgess-Wheeler law (♦- - -); [118, 125] (+); this work (● ○) ................................................................................................................. 46   Figure 5.10. Calculation of lower temperature limit by finding the intersection of the vapor pressure curve (—) and the regressed lower flammability limit curve (- - -) from experimental data (+), illustrated here with the experimental lower temperature limit data (●) for phenetole ..................................................................................................................... 48   Figure 5.11. Visual F-test for statistical difference between calculated and experimental lower temperature limits; confidence intervals for experimental data are shown on the zero line ................................................................................................................................... 49   Figure 5.12. Deviations between calculated and experimental lower temperature limits following correction of vapor pressure curves ........................................................................ 50   Figure 5.13. Pensky-Martens (PM) and small-scale (SS) flash point values; adamantane (□) appears to be an outlier ........................................................................................................... 54   Figure 5.14. 95 % confidence region for the coefficients of Equation 5.9 ................................... 54   Figure 5.15. Calculated PM values using Equation 5.9 compared to experimental data; previously reported values of isopropyl myristate (□) and anthraquinone (◊) were affected by extending the waiting time between injection and ignition in the small-scale apparatus, and decreasing the thermal ramping rate of the Pensky-Martens appartus ............................ 56   Figure 5.16. Flash point (FP) vs. lower temperature limit (T L ) for the Pensky-Martens (□) and small-scale (+) apparatuses .............................................................................................. 58   Figure 5.17. Confidence regions of the coefficients in Equations 5.12a (dashed line) and 5.12b (solid line) compared with the coefficients recommended by Evlanov (□) .................. 59   Figure 6.1. Illustration of documented circular referencing among compilations ........................ 63   Figure 6.2. Trend of experimental flash point data (●) with carbon number for n-alcohols. Plots like this can be used to quickly spot erroneous data (□) and predict missing values (+) ............................................................................................................................................ 72   Figure 6.3. Typical trends for flammability and temperature limits against carbon number, shown here for n-alcohols: lower flammability limit (top left), lower temperature limit (top right), upper flammability limit (bottom left), and upper temperature limit (bottom right). Predicted flammability limit data are depicted by (+) ................................................ 72  

ix

Figure 6.4. General flash point trends, shown here for alkenes and alkynes: FP vs. normal boiling point (T b ) (top left), FP vs. ΔH c (top right), FP vs. ΔH vp at the flash point (bottom left), and P*(FP) vs. ΔH c -1 (bottom right). Possible outliers are depicted by (□) ................. 73   Figure 6.5. Structure of 1,6-hexanediol (top) and selected comparison compounds (left to right): 1,3-propylene glycol, 1,4-butanediol, and 1,5-pentanediol ......................................... 76   Figure 6.6. Increase in the number of experimental data points for each property added to the DIPPR database ...................................................................................................................... 79   Figure 6.7. Number of compounds with at least one experimental data point in the previous DIPPR ® 801 database and the reviewed database .................................................................. 80   Figure 6.8. Number of accepted points in the previous DIPPR ® 801 database compared with the reviewed DIPPR database; points are divided into experimental (hashed bars) and other ........................................................................................................................................ 80   Figure 6.9. Histogram of changes in recommended values for flash point .................................. 81   Figure 6.10. Histograms of changes in recommended values for lower flammability limit (left) and upper flammability limit ......................................................................................... 82   Figure 6.11. Previously recommended flash points vs. values recommended following the critical review .......................................................................................................................... 82   Figure 6.12. Previously recommended lower flammability limits (left) and upper flammability limits vs. values recommended following the critical review ........................... 83   Figure 7.1. Adiabatic flame temperatures calculated assuming heat capacities independent of temperature are related linearly to flame temperatures calculated from Equation 5.1 ........... 91   Figure 7.2. Distributions of errors from using Equation 7.6 to predict the lower flammability limit for data sets of (top left) Bureau of Mines, (top right) European, (bottom left) ASHRAE, and (bottom right) recommended values .............................................................. 96   Figure 7.3. Predicted vs. experimental lower flammability limits from the (top left) Bureau of Mines, (top right) European, and (bottom left) ASHRAE apparatuses, and the (bottom right) recommended data set ................................................................................................... 97   Figure 7.4. Slope of the adiabatic flame temperature, γ, as a function of carbon number, n C . Equation 7.11 is shown by the solid line .............................................................................. 102   Figure 7.5. Evaluation of the temperature dependence of the lower flammability limit, as estimated by (left) Britton and Frurip (□), the modified Burgess-Wheeler law (+), and (right) Equation 7.13 ............................................................................................................. 103   Figure 7.6. Predicted lower flammability limits using Equations 7.6 and 7.13 vs. experimental points in Table 5.1. Data for methanol are shown with solid circles ............. 104  

x

xi Figure 7.7. Predicted lower flammability limits using the methods of the Bureau of Mines (top left), Catoire and Naudet (top right), and Britton and Frurip (bottom) vs. experimental points from Table 5.1 ...................................................................................... 105   Figure 7.8. Predicted upper flammability limits from Equation 7.15 vs. the recommended data values (left) and data determined using the European standardized method ................ 110   Figure 7.9. Histogram of errors between experimental flash points and values predicted using Equation 7.16 (left) with groups to determine k, and the method of Leslie and Geniesse ... 116   Figure 7.10. Experimental flash points vs. values predicted using Equation 7.16 (left) with groups to determine k, and the method of Leslie and Geniesse ............................................ 117   Figure 7.11. Histogram of errors and plot of predicted vs. experimental flash points for Equation 7.18 using the combined training and test sets ...................................................... 118   Figure 7.12 Experimental flash points vs. values predicted using Equation 7.23, shown here for the training set (left) and the test set ............................................................................... 123  

CHAPTER 1. INTRODUCTION Knowledge of the combustion potential of a chemical is crucial when designing safe chemical processes. To prevent explosion, it is often simplest to keep a chemical outside of its flammable concentration range, as described by the flammability limits and flash point. The flash point is an approximation of the lower temperature limit, the temperature at which a chemical evolves enough vapors to support combustion. Agencies such as the U.S. Department of Transportation, the National Fire Protection Agency, and the Occupational Safety and Health Administration classify flammable liquids for regulations and guidelines based on the flash point [1]. Consequently, the flash points of common chemicals are widely reported. Flammability limits represent the concentrations of fuel in air that will just support flame propagation. The limits are better descriptors of a chemical’s flammability, and more useful for safe process design because they are applicable to solids, liquids and gases. A substantial amount of flammability limit data has been published, though the temperature-dependence of the limits is nearly always neglected. For gases, data are frequently reported at 298 K. Flammability limit data for liquids and solids, however, are often reported at a single arbitrary temperature. Differences in apparatuses and experimental methods can influence the measured flash point and flammability limits significantly. A common practice of reporting the widest range of flammability, i.e., the lowest flash point, has apparently been adopted by many compilations, as

1

2 the values they report are often outliers and inconsistent with data for other related properties. These inconsistent data are then frequently used to regress parameters for estimation methods. The end result of reporting the widest flammability range instead of the most probable is unnecessary and costly restraints on chemical processes and inaccuracy in prediction methods developed from the reported data. This dissertation describes a critical evaluation of published flammability data, undertaken to provide a database of recommended values. In addition, flammability data are determined experimentally for 29 organic compounds, chosen for measurment to further understanding of the interrelationship of flammability properties, quantify the effect of standardized experimental methods and apparatuses on flammability data, explore the effect of initial temperature on the lower flammability limit, and to supplement previously reported experimental data. Finally, this dissertation presents flammability limit and flash point estimation methods for the critically reviewed compounds.

CHAPTER 2. 2.1 FLAMMABILITY LIMITS ASTM defines the upper/lower flammability limits as “the [maximum/minimum] concentration of a combustible substance that is capable of propagating flame in a homogenous mixture of the combustible and a gaseous oxidizer under specified conditions of test” [2]. Many investigators have theorized why such limits exist [3-27], but currently the dominant view is that flame propagation fails when the heat loss rate exceeds the rate of enthalpy generation during the combustion reaction. Experimental Determination Though many methods of measuring the flammability limit have been developed [2, 28-39], only the three standardized methods for which a significant amount of data have been reported are considered here. 2.1.1 Bureau of Mines Tube Method Data from the tube method developed by the U.S. Bureau of Mines were long considered the standard for flammability limits. Tests were performed in narrow tubes, 2 cm to 7.5 cm in diameter and at least 1 m high. A fuel-oxidizer mixture was considered flammable only if it

3

could, in theory, support flame propagation along an infinite tube. Thus, flame propagation to the top of the tube was required for a mixture to be called flammable [3-4]. 2.1.2 ASHRAE Method The ASHRAE (American Society of Heating, Refrigerating and Air-Conditioning Engineers) method [2, 40-42] was developed specifically to accommodate halogenated compounds that may be difficult to ignite in smaller vessels. Though measurements are made in a spherical 12-L flask, the flame propagation criterion for this method was empirically designed to reproduce data measured in jumbo tubes with full flame propagation [43-45]. A mixture is considered flammable when a flame forms a continuous arc subtended by a 90 o angle, measured from the ignition source to the vessel walls (Figure 2.1).

Figure 2.1. ASHRAE 90 o flame propagation criterion

4

2.1.3 EN 1839 Tube Method Originally called DIN 51649, this European standard was developed with the reasoning that a fuel-oxidizer mixture supporting any ignition could result in potentially dangerous situations whether propagation is possible or not [5-6]. When the method was renamed EN 1839, the flammability criterion was changed to require flame detachment and at least 10 cm of flame propagation to account for localized heating introduced by the ignition source [7]. The standard test vessel is a vertical glass tube 150 cm long with a diameter of 5 cm. 2.1.4 Differences in Measurement Methods Table 2.1 details the differences in the three flammability limit measurement methods described in this chapter. Several comparative studies have been published for these and other flammability apparatuses [44-48]. In general, it has been shown that data determined in the EN 1839 apparatus correspond to the widest fuel-concentration range over which a fuel-air mixture is considered flammable. The Bureau of Mines tube typically yields the narrowest range of fuel concentration, though data from the ASHRAE method are similar.

Table 2.1. Comparison of three widely-used standardized flammability apparatuses Bureau of Mines ASHRAE EN 1839

Vessel Shape Vertical glass tube Spherical glass flask Vertical glass tube Vessel Size 5 cm x 150 cm 12 L 8 cm x 30 cm Spark or flame, bottom of tube Spark, below center of sphere Ignition Source Spark, bottom of tube Continuous flame arc subtending 90 o angle from ignition source Propagation Criterion Flame detachment and 10 cm of propagation Full propagation to top of tube Mean value of last ignition and non-ignition points Mean value of last ignition and non-ignition points Definition of Limit Last non-ignition point

5

Many investigators have pointed to the discrepancies between flammability data determined using different methods as evidence that fundamental flammability limits may not exist [8-14]. On the other hand, much of the error between measurements may be a simple function of the flame propagation criteria and definition of the limit utilized in each study. However, the numerous other variables that affect the measured flammability limit make it difficult to show experimentally whether or not a fundamental flammability limit exists. A summary of these variables is provided in Appendix A. Because of the pertinence to this work, studies on the effect of temperature on the flammability limits will be summarized here. 2.2 Temperature Dependence of the Flammability Limit Burgess and Wheeler [15] showed that the heat liberated by a mole of a lower limit mixture at ambient temperature and pressure is approximately constant for many compounds (Burgess-Wheeler law): kHLFL = Δ − ⋅ )( c , (2.1) where LFL is the lower flammability limit and ΔH c is the heat of combustion of the fuel. Based on the findings of a constant adiabatic flame temperature with respect to the initial mixture temperature of the lower flammability limit, Zabetakis, Lambiris, and Scott [16-17] attempted to extend the law of Burgess and Wheeler to account for the temperature dependence of the lower flammability limit by adding the enthalpy required to raise a limit mixture from ambient temperature to the initial test temperature:

6

kTCHLFL p =Δ⋅+Δ−⋅ −airfuel,c )( , (2.2) airfuel,−p C where is the total specific heat of the fuel-air mixture, found by air,f,airfuel, )100( ppp CLFLCLFLC ⋅−+⋅= − , (2.3) and C p, f and C p, air are the molar heat capcities of the fuel and air, respectively. When the lower flammability limit is known at a given temperature, T 0 , Equation 2.2 may be rewritten as ( ) ( ) 0 c0 airfuel, 0 )( 100 1 )( )( TT HTLFL C TLFL TLFL p − Δ− ⋅ −= − . (2.4)

For many hydrocarbons, this approximately corresponds to a 7 % decrease in the lower flammability limit per 100 K, relative to the value at 293 K. Equation 2.4 is often expressed in the more general form ( ) 01 0 1 100 1 )( )( TT c TLFL TLFL −−= . (2.5) The parameter c represents the decrease relative to the lower flammability limit at T 0 , typically 293 K, per 100 K increase, i.e., c ~ 0.07 K -1 according to the modified Burgess-Wheeler law (Equation 2.4). Using a Bureau of Mines style apparatus with a 10 cm diameter, Hustad and Sønju [18] found c to be approximately 0.085 K -1 for hydrocarbons. Gibbon, Wainwright, and Rogers [19] showed c varied between 0.11 K -1 and 0.18 K -1

for common solvents in a 13-L closed sphere . Goethals et al. [20] and Brandes, Mitu, and Pawel [21] found c values between 0.13 K -1 and 0.23 K -1 in the DIN/EN tube apparatus for a wide range of compounds. A summary

7

of studies on the effect of temperature on the upper flammability limit is included in Appendix A. 2.3 Flammability Limit Estimation Methods Methods of predicting the lower flammability limit may be divided into two classifications: chemical equilibrium methods and empirical correlations. 2.3.1 Chemical Equilibrium Methods Based on the findings of Zabetakis, Lambiris and Scott [17], chemical equilibrium methods assume that the adiabatic flame temperature at the flammability limit is approximately constant among different fuels. The adiabatic flame temperature (T ad ) is the theoretical temperature of the flame assuming no heat loss: ∑ ∑ = reactants o products ad )()( THTH ii , (2.6)

where T o is the initial mixture temperature and H is the enthalpy of species i. The combustion products are typically estimated using a chemical equilibrium calculator. Mashuga and Crowl [22] estimated the entire flammability envelope for methane and ethylene with satisfactory results by assuming a T ad of 1200 K. Ervin et al. [23] also used an adiabatic flame temperature of 1200 K to predict the flammability limits of alkanes, carboxylic acids, and acetates. Shebeko et al. [24] and Vidal et al. [25] believed the temperature to be about 1600 K, while Melhelm [26] selected the conservative value of 1000 K.

8

Others have noted that the flame temperature is not constant at all, but rather it increases with the molecular weight of the fuel [27-29]. Zatsepin, Sorokin and Stepachev [30] developed a method to estimate the lower flammability limit from a T ad calculated using structural ontributions: c

∑ = ii xTT ad , i (2.7a) here T i is the contribution of bond-type i (Table 2.2), x s found by w i i

∑ j jj i ngk (2.7b) = ii ngk x, i is the multiplication factor of bond i (1.5 for aromatic bonds, 1 for all others), and ng i is the

Full document contains 262 pages
Abstract: Accurate flash point and flammability limit data are needed to design safe chemical processes. Unfortunately, improper data storage and reporting policies that disregard the temperature dependence of the flammability limit and the fundamental relationship between the flash point and the lower flammability limit have resulted in compilations filled with erroneous values. To establish a database of consistent flammability data, critical analysis of reported data, experimental investigation of the temperature dependence of the lower flammability limit, and theoretical and empirical exploration of the relationship between flash points and temperature limits are undertaken. Lower flammability limit measurements in a 12-L ASHRAE style apparatus were performed at temperatures between 300 K and 500 K. Analysis of these measurements showed that the adiabatic flame temperature at the lower flammability limit is not constant as previously thought, rather decreases with increasing temperature. Consequently the well-known modified Burgess-Wheeler law underestimates the effect of initial temperature on the lower flammability limit. Flash point and lower temperature limit measurements indicate that the flash point is greater than the lower temperature limit, the difference increasing with increasing lower temperature limit. Flash point values determined in a Pensky-Martens apparatus typically exceed values determined using a small-scale apparatus above 350 K. Data stored in the DIPPR® 801 database and more than 3600 points found in the literature were critically reviewed and the most probable value recommended, creating a database of consistent flammability data. This dataset was then used to develop a method of estimating the lower flammability limit, including dependence on initial temperature, and the upper flammability limit. Three methods of estimating the flash point, with one based entirely on structural contributions, were also developed. The proposed lower flammability limit and flash point methods appear to predict close to, if not within, experimental error. Keywords: flammability limit, flash point, DIPPR, flame propagation