Firing Key
Gas Meter
Feed Gas
Rotameter
Transformer
Back Pressure Regulator
Feed Gas Inlet System Shut-off Valve e
Ignition Plug (Hot
Heat Exchanger IPA Feed Reservoir
Explosion Vessel
S. S . Cylinder I.D. 2” Length 4” Fitted with 2 stainless steel screens as spray
IPA Drain
baffles
Figure 1.
Ignition limits of the system were measured in a cylindrical, electrically heated steel
I
bomb
HARRY SELLOl
Shell Development Co., Emeryville, Calif.
Ignition Limits of the Gaseous System Isopropyl Alcohol-Oxygen-Nitrogen To minimize hazards in handling and processing isopropyl alcohol, operate outside ignition limits or reduce oxygen. Sources of ignition are always present. This report tells how to avoid explosion
A
COMBUSTIBLE gas or vapor may, under certain conditions, be ignited so that a rapid self-propagating combustion or deflagration occurs. For such a combustion, the combustible and the oxidant gas must be present in the proper concentrations, and a source of ignition or suitable energy input be available to initiate reaction. The regions of suitable concentrations for the combustion of a fuel and oxidant gas which delineate this region are known as the limits of flammability. T o minimize the hazard in handling and processing of a combustible, the flammability region must be scrupulously
1 Present address, Shockley Semiconductor Laboratory, Mountain View, Calif.
avoided and sources of ignition eliminated. The latter goal, however, is practically impossible, and ignition must be considered omnipresent. Thus, avoidance of hazard essentially is dependent upon a knowledge of the limits of flammability. Coward and Jones (7) describe “standardized” apparatus used to measure flammability limits of a number of fuels. It consists mainly of a glass cylinder 5 cm. in diameter and 150 cm. long, placed vertically and open at the bottom. The fuel and oxidant gas are mixed in the cylinder and ignited by an open flame or spark gap, If a flame travels the length of the tube, the mixture is self-propagating and is said to be flammable. This method is applicable only to measurements a t atmos-
pheric pressure. I n the work reported here, the limits are more properly classified as ignition limits, as differentiated from true flammability limits, because the method of measurement precluded observation of the extent of flame propagation. Both types of limits are strongly dependent on the nature of the equipment and type of ignition ( 4 ) . This dependence is reflected, for example, by the variation in the values reported for the upper limit of flammability for isopropyl alcohol (IPA) in air (7, 6)-i.e., 11.8 and 7.99%. Many other methods have been used to study gaseous explosions, depending mainly on the nature of the source of ignition and the type of explosion chamber. The ignition sources commonly used are intense sparks, open flames, or VOL. 50, NO. 10
OCTOBER 1958
1561
heated wires; the explosion chambers are predominantly cylindrical types of metal or glass.
Experimental Apparatus The explosion vessel was a cylindrical, electrically heated, stainless steel bomb, 2 inches in inside diameter and 4 inches long (Figure 1). The effect of increasing tube diameter above 2 inches appears to be negligible for most explosions ( 3 ) . The bomb was made pressure-tight by a threaded cap seated against a nickel gasket. A stainless steel helical coil (diameter 2 inches) made from 3/g-inch tubing served as the isopropyl alcohol saturator. Approximately 100 ml. of liquid were required to fill the saturator. The coil fitted into one of two parts in the bottom of the bomb; an inlet to admit saturated feed gas and an exit for the drainage of overflow liquid isopropyl alcohol. The entire saturator (coil and attached metal block) was immersed in an electrically heated oil bath. The interior of the bomb contained two stainless steel screens fastened to the thermocouple well, which served as spray baffles to trap droplets of liquid carried into the vessel from the saturator. Attached to the top of the explosion vessel were an ignition plug and an exit gas line. Exit gas from the bomb passed through a condenser (to remove isopropyl alcohol) to a system of pressure gages, back-pressure regulator, and gas meter. Three pressure gages (Bourdon
fl
\
type), 200: 600, and 3000 p.s.i.g., comprised the system, which was utilized to observe the pressure increases due to explosions. To protect the exit system from excessive pressure increases, a bursting disk set at 1000 p.s.i.g. was connected to the top of the bomb. Because of the relatively short path, the disk generally burst before the excessive pressure wave was transmitted through the system. The oxidant gas was supplied from gas cylinders and metered through a rotameter; isopropyl alcohol was pressured into the saturator coil from a reservoir of ca. 150-ml. volume. The gas inlet to the saturator coil (Figure 1) consisted of a length of hypodermic tubing extending through a metal block into the coil. This provided small bubbles of feed gas, which were saturated with isopropyl alcohol vapor upon rising through the coil. A thermocouple was inserted into the coil just below the inlet to the bomb.
Method of Ignition An electrically heated wire bridging the terminals of an ordinary automotive spark plug was used as the source of ignition. The wire was a 2-inch length of 26-gage Nichrome, having a resistance of about 0.9 ohm. By suitable transformers, the 110-volr supply (alternating current) was stepped down to a working voltage of about 3 volts, which resulted in a power output at the hot-wire ignition source of approximately 10 watts. This corresponded to an ignition energy well above the minimum ignition energy observed in other systems.
System CHa-air CaHa-air This investigation
Ignition Energy. Cal. 30-40 (min.) 140 50-75
STOICHIOMETRIC COMPOSITION
Figure 2. ignition limits of isopropyl a Icohol-o x y g en nitrogen system at various pressures. Less oxygen i s required for an explosion as total p r e s s u r e i s increased
-
P.S.I.G. 0
0 100 0 200 0 300
A 0
NITROGEN, %v
1562
INDUSTRIAL AND ENGINEERING CHEMISTRY
IPA, C v
The minimum ignition energy was extrapolated from data (8) obtained by the use of fine Nichrome wires. I t was sufficient to heat the wires to flame temperatures (ca. 1800' '2.). Data on propane-air (21,published subsequent to this investigation, showed that in comparing several types of ignition-hot wire, spark, and guncotton-hot wire gave the widest limits of flammabil. ity. and little effect was noted when ignition energies were increased above 140 calories (60 watts for 10 seconds).
Measurement of Limits For actual measurement of the ignition limits the oxidant gas was bubbled through the isopropyl alcohol saturator coil at a given temperature and operating pressure until equilibrium (constant temperature) was reached in the bomb. The desired operating pressure was preset on the system by means of the backpressure regulator. Approximately 0.3 cubic foot of feed gas was introduced in about an hour. When bomb temperature became relatively constant (=!=lo C. fluctuation), the system was closed by the oxidant gas inlet and system shutoff valves (Figure 1). The firing key was then depressed, energizing the ignition wire. If an explosion took place, a pressure increase was noted on the appropriate pressure gage. The extent of pressure rise depended upon the mixture in the bomb, but usually the 600 or 200 p.s.i.g. gage was used to detect the pressure change. For gas mixtures rich in oxygen, the pressure increases in most cases were large, so that 3000 p.s.i.g. gage was used. If no noticeable pressure increase occurred after the firing circuit was energized for 30 seconds, the mixture was considered nonflammable. Generally, for a flammable mixture, an explosion occurred after 5 to 10 seconds of firing; however, this interval varied between 1 and 30 seconds. During the firing period, current flowed through the wire at the rate of about 3 amperes. If no explosion occurred, bomb temperature was elevated in increments of about 5' until a flammable mixture was obtained. The flammable region was first located to the nearest 5' increment and then more accurately (1' changes) with a fresh charge of isopropyl alcohol. The bomb was thoroughly swept out after every try for explosion, whether or not an explosion had occurred. In most cases, an explosion was easily recognized, as pressure increased 5 p.s.i.g. or more. I n a few cases (upper limits for air at 0 and 100 p.s.i.g.), transition from the flammable to the nonflammable region was gradual, because of the low temperature of the gaseous mixture at the time of ignition. Although temperature increased simultaneously with pres-
IONITION LIMITS Nz %
0 2
%
90
\
\ /
\ \
FLAMMABLE REGION
\
\
\
\
\
I 10
-
Figure 4. Isopropyl alcohol vapor-liquid equilibrium temperatures at ignition limits
NONFLAMMABLE I
I
I
I
I
I I
At atmospheric pressure and room temperature the mixture (with air) is on the borderline of flammability
Figure 3. Isopropyl alcohol vapor-liquid equilibrium temperatures at ignition With air the flammable region lies between 40’ and 84’ C. Total preisure, 100 p.s.i.g.
sure, the former was not as valid a criterion for explosion because of the slow response of the temperature recorder. Temperature increases varied from 5’ to 4 3 O The lower-limit (lean isopropyl alcohol) explosions were generally accompanied by much stronger pressure increases than the upper-limit explosions and were more easily reproduced. At the lower limit, pulses of 1000 p.s.i. were frequent, while the upper-limit pulses were more often in the range of 10 to 150 p.s.i. T h e transition from the nonflammable to the flammable region was sharper in lower-limit mixtures. Thus, two or three positive results (explosions) usually defined the lower limit. The variation in the lower-limit compositions, in going from flammable to nonflammable mixtures, was 10 to 15% of the experimental values. In the case of the upper-limit mixtures, because of the gradual transition, a number of explosions (3 to 6) were required to define limits clearly, particularly at lower temperature and pressures. This number was dependent upon the particular mixture. The experimental error, however, was smaller for the upper limits-about 5% of the reported value.
c.
NONFLAMMABLE REGION
Flammable regions or envelopes are delineated by nonflammable compositions (Figure 2 ) . The points plotted represent the nonflammable member of the pair of compositions requ’ired to define a given limit. Thus, a composition lying on the curve is nonflammable. The nearest flammable composition may be two percentage units (in terms of isopropyl alcohol) within the confines of the curve. Thus a safety factor roughly equal in magnitude to the experimental error is introduced. The composition of the vapor in the bomb was calculated from vapor pressure data for isopropyl alcohol, assuming an ideal system. Actually this is not so, largely because of presence of oxygenparticularly a t high partial pressures and low temperatures. However, the correction for nonideality is relatively small a t these pressures and temperatures when compared to the experimental variation in the ignition limits. *
Results
Effect of Nitrogen. The ignition limits for isopropyl alcohol and all mixtures of oxygen and nitrogen are shown in Figure 2. The flammable
region for isopropyl alcohol-oxygen at atmospheric pressure was between 2 and 35% isopropyl alcohol (lower and upper limit). As nitrogen is added to the mixture, the flammable region narrows -Le., the upper and lower limits approach each other. Finally, a composition is reached in which the nitrogen concentration is too high (or the oxygen too dilute) for the mixture to sustain an explosion. This is shown in Table I in terms of the volumes of oxygen required per volume of isopropyl alcohol a t the upper limit. The limiting ratio of oxygen to isopropyl alcohol at each pressure approaches the stoichiometric value for complete combustion to carbon dioxide and water. For each pressure, the “nose” or minimum of the curve lies near the stoichiometric ratio. The nose value also represents the maximum allowable or limiting oxygen content which can be tolerated in a given mixture before flammability is reached.
Limiting Oxygen Contents Total Pressure on System, Limiting 01 P.S.I.G. Concn., Vol. ?!& 0 100 200
300
VOL. 50, NO. 10
17 14 13 11.3
OCTOBER 1958
1563
Effect of Total Pressure; The ignition limits of this system were determined at total pressures of 0, 100, 200, and 300 p.s.i.g. For each pressure, a different flammable region (or envelope) was determined (Figure 2). As total pressure was increased, the upper ignition limit increased markedly, while the lower limit was marginally lowered. As total pressure is lowered, the upper limit decreases sharply and approaches the lower limit. Thus, for each oxidant gas there is a limiting or minimum pressure below which combustion cannot be maintained: about 90 p.s.i.g. for 14.6% oxygen-85.4% nitrogen. For richer oxygen mixtures, the minimum pressure decreases well below atmospheric pressure. In methaneair mixtures, for example, the minimum pressure is about 0.2 atm. (7). If flame propagation proceeds by burning successive layers of gas, increasing total pressure (and concentrating reactants) should facilitate propagation. Proceeding along lines of constant nitrogen composition (Figure 2), less oxvgen is required for an explosion as mtnl pressure is increased. A mixture containing 25 volume % nitrogen, 20’7’ oxvgen, and 55% isopropyl alcohol, while flammable a t 300 p.s.i.g., can be rendered nonflammable by decreasing total pressure to 200 p.s.i.g. At the upper limit, the volumes of oxygen required for flammability per volume of isopropyl alcohol decreased as the total pressure increased. For the mixture with 25 volume % nitrogen (Table I), roughly one sixth as much oxygen is required at 300 p.s.i.g. than a t atmospheric pressure; as pressure is increased, changes in the relative importance of the various chemical reactions determine the course of the burning processes. For
example, both cracking and oxidation occur at the lower pressures. Increase in pressure, however, diminishes cracking and favors oxidation reactions. Effect of Temperature. The method of measuring the ignition limits in This investigation (by establishing vaporliquid equilibria) introduced a variation in temperature. The range of temperatures at which upper-limit mixtures were obtained, when air or oxygen was employed as the oxidant gas, is: 300
Oxidant Gas
Atm.
P.S.I.G.
Air Oxygen
24.0
135.5
57.5
179.0
The ignition limits defined in terms of vapor-liquid equilibrium temperatures are shown in Figures 3 and 4. The broken lines in Figure 3 represent the mixtures which will result when a given oxidant gas is saturated with isopropyl alcohol at various temperatures and a total pressure of 100 p.s.i.g. For example, if 1 0 0 ~ ooxygen (upper dashed line) is bubbled through liquid isopropyl alcohol at temperatures below about 40’ C., the gaseous mixture will be below the lower ignition limit. Mixtures formed by saturating oxygen between 40’ and 131’ C. will be flammable; above 131’ C. (upper limit): the compositions are too rich in fuel (isopropyl alcohol) to be flammable. O n diluting the oxidant gas with nitrogen, a narrower range of flammable compositions results. ’Thus with air, the flammable region shown (Figure 3) lies between about 40’ and 84’ C. The use of an oxidant gas containing less than 15% oxygen will avoid this region. The complete range of isopropyl alcohol vapor-liquid equilibrium temperatures at the ignition limits is shown in Figure 4. t\s the total pressure upon the system is increased, a higher partial pressure of isopropyl alcohol and a higher temperature, are required to render a given mixture flammable. The flammable regions are thus transposed toward the higher temperatures. The normal equilibrium vaporization of isopropyl alcohol in air at room temperature and atmospheric pressure
Table I. Ratio of Oxygen to Isopropyl Alcohol at Upper Flammability Limit As Nitrogen Is Added, the Flammable Region Narrows
(Ratios calculated from values in Figure 2) N2 in
Mixture, ~701. % 0 25 50 75 a0
84 85 86
1564
Oz-IPA Volume Ratio at Upper Limit
.
Atm. press.
100 p.s.i.g.
200 p.s.i.g.
1.9
0.47 0.47 0.64 2.10 3.08 Nonflam. Nonflam. Nonflam.
0.42 0.42
2.1
2.4 4.0 Nonflam. region Nonflam. Nonflam. Nonflam.
INDUSTRIAL AND ENGINEER!NG CHEMISTRY
0.55 1.45 2.17
4.02 Nonflam. Nonflam.
300 p.5.i.g. 0.33 0.34 0.47 0.92
1.30 2.27 2.80
Nonflam.
results in a mixture just on the borderline of being flammable. A decrease in temperature to the region of 12’ to 25’ C. would yield hazardous mixtures. In general, increasing temperature widens ignition limits. Thus, a correction for temperature would have the greatest effect at the lower isopropyl alcohol concentrations where the temperatures were the lowest. However, the correction, which is roughly linear, would not have much effect upon the limits-. For example, at atmospheric pressure, the upper flammability limit for methane-air ( I ) increased by about 10.570 of its value and the lower limit decreased by about 8y0 for an increase in temperature of 200’ C. Such a correction would not be important in this investigation, because the ignition limits are already conservative-the curves are drawn through the nonflammable members of the pairs of compositions that define the limit. Sources of Error
I n addition to the effect of temperature variation upon the ignition limits, the reproducibility of the measurements was influenced by changes in energy of the source of ignition, oxidation of the hot wire, and “cool flame” phenomena. The formation of nickel oxide films has been observed upon the surface of nickel bars used in the ignition of natural gas-air mixtures (5). These oxide films exerted a passivating effect upon the metal and decreased the ease of initiation of explosions. This factor may have influenced the reproducibility of the ignition limits measured in this work. Some precaution was taken by periodically replacing the Nichrome ignition wire. Care was also taken to flush the system after each ignition. as carbon was observed to deposit upon the walls of the vessel. This could lead to some error, as excessive amounts of carbon could catalyze reactions at lower temperatures and lead to “preiknition” effects. Literature Cited (1) Coward, 1-1. F., Jones, G. W.:Bur. Mines Bull. 503, 97 (1952). (2) Di Piazza, J. T., others, IND. ENG. CHEM. 43,2721 (1951). (3) Jones, G. W., Chem. Revs. 22, 2 (1938). (4) Jost. W.. “Explosion and Combustion Processes i n Gases,” 1st ed., p. 128, McGraw-Hill, New York, 1946, 15) Lewis, B., Von Elbe, G., “Combustions, Flames and Explosions of Gases,” 1st ed., pp. 334, 447, Academic, New York, 1951. 16) Louis, M., Entezam, L., Ann. combustibles liquides 14, 21 (1939). (7) Mason, W., Wheeler, R. U., J . Chem. SOC.113, 45 (1918). (8) Stout, H. P., Jones, E., ‘Third Symposium on Combustion, Flame and Explosion Phenomena,” p. 329, Williams & Wilkins Co., Baltimore, 1949. RECEIVEDfor review October 14, 195ACCEPTEDJune 5 , 1958