«Portable Plastic Gasoline Container Explosions And Their Prevention Glen Stevick, Ph.D., P.E., David Rondinone, Ph.D., P.E., Allan Sagle, Ph.D. of ...»
The maximum or worst case negative pressures from the pouring tests were used as inputs for the COMSOL finite element analysis. The results further confirmed that air entrainment can and does lower the percent hydrocarbon in a gasoline can vapor space during pouring. However, with all but the smallest amount of gasoline being stored in such a container, some combination of aging, temperature reduction and/or low volume amount in the can is still required for the vapor space to be in the explosive range. For example, at -9.1 oC, the percent hydrocarbon must be 13% or lower to produce an explosive mixture (percent hydrocarbon below the upper explosive limit) in the can due to air entrainment. This would have to be significantly aged gasoline as the percent hydrocarbon in the head space for fresh gasoline is approximately 35%. Air entrainment is a complex effect that we are continuing to study.
Modeling 30 seconds of pouring requires several days of computer time even with fast, multi-processor computers. Figure 5 below is a graphic representation from one modeling run illustrating the relationship between pouring and air entrainment.
Figure 5. Lines of flow (red streamlines) show the motion of vapor, and partial vapor pressure/gasoline vapor concentration (color contours) show the variation of gasoline vapor concentrations inside a 18.
9 liter (5 gallon) can during pouring with aged gasoline.
Field evidence and aging tests and calculations indicate that gasoline can explosions can and do occur under foreseeable conditions when their concentration is below the upper explosive limit UEL (7.6%) and above the lower explosive limit LEL (1.4%).13 However, given the skepticism as to the occurrence of such events in some literature, it was decided to conduct explosion testing to verify the reported events under reproducible conditions.
In these tests, measured amounts of gasoline samples were added to the container, and the container was inverted as in pouring, which mimics the most common scenario for incidents reported from the field. Inversion of the can was accomplished using a test jig in a reinforced test cell. The open spout 13 David R. Lide, ed., CRC Handbook of Chemistry and Physics, 90th ed. (Boca Raton, FL: CRC Press; 2009-2010); “Material Safety Data Sheet (MSDS),” 30 Oct. 2008, http://www.msdssearch.com.
was sited near an ignition port and a small flame was passed near the spout as shown in Figure 6.
The explosion tests were initially conducted using 18.93 liter commercially available gasoline cans with gasoline contents ranging from 20 to 1,750 ml in various degrees of aging. In addition, tests with 7.85 liter (2 gallon) containers were also conducted.
Most tests were run using cans modified to provide explosion relief using plastic plugs, as seen in figures 6 (below). This prevented the type of excessively violent explosions which had occurred in some of the cases studied, thus avoiding a safety hazard and/or excessive property damage.
Figure 6. Gasoline cans tested w/o flame arresters from 4 separate tests with gasoline in the explosive range.
Confirming the need for such explosion relief, tests were also performed with gasoline cans in the as-is state, without the pressure relief port. Typically, the
cans ruptured from the overpressure as shown in figure 7 below:
Figure 7. 18.
9 liter can (above) and 7.85 liter can (right) from unmodified gasoline can explosion tests.
The ignition process leading to the overpressure conditions in the portable plastic gasoline containers is also being studied. In this case, the travel of a flame front can be observed traveling up the plastic spout of the 7.85 liter can (figure 8) using high speed video recording. The published gasoline flame speed of 0.4 meters per second was confirmed.
Figure 8. Top row: stills from the video 7.
85 liter can tests showing the flame traveling up the translucent plastic spout (blue arrow). Bottom row: blowup of the plastic spout area corresponding to the frame above in the top row.
Over 50 gasoline can explosion tests have been conducted. Plotted relative to the percent hydrocarbon in the vapor space, the results confirm the published
upper and lower gasoline explosive limits (figure 9):
Figure 9. Explosion Go-NoGo tests as a function of percent hydrocarbon concentration.
The fresh gasoline data points in the explosive range were created by low temperature. Red and orange points indicate an explosion occurred in the test.
The test results include some No-Go or “not explosive” data points above the lower explosive limit (LEL). It should be noted that the concentration was measured without the spout, a few minutes prior to the time of ignition. During the low concentration tests, the vapor in the spout at the time of ignition may be leaner than the measured value. When the gasoline vapor concentration is low, the diffusion and gross motion of the gasoline vapor/air mix from the can to the spout is slower than for high concentrations. This can lead to a situation where the actual gasoline vapor concentration in the spout is leaner than the LEL, while the value measured is richer.
6. Flame Arresters
Flame arresters have been used industrially for almost two hundred years. Sir Humphry Davy14 first developed a flame arrester for coal miner lamps in 1815.
Patents were issued for flame arresters in the 1880’s for both chemical processes and flammable liquid containers (Allonas 1878).15 During the early 1930s, R.J. Anschicks, assignor to Protectoseal Company, developed and 14 Henry A. Pohs, The Miner’s Flame Light Book: The Story of Man’s Development of Underground Light (Denver: Flame Pub. Co., 1995).
15 J. Allonas, “Spark-Arrester,” US Patent No. 295716 (July 9, 1878).
patented a tank fitting that incorporated a flame arrester.16 All gasoline containers currently manufactured by the Protectoseal Safety Container Division have perforated metal flash arresters at each container opening.
A flame arrester works by removing heat from a flame and keeping the temperature of the fuel on the other side of the arrester below its ignition point.
The flame arrester mesh breaks the flame into many flamelets, and the heat of these flamelets is transferred to the walls of the mesh. There are two criteria for successful operation: the holes in the mesh must be less than the critical diameter, and the critical velocity must be higher than the flame speed. The
critical diameter dcr from Gossel17 is given by:
where α is the thermal diffusivity in air (1.0 m2/s), and Su is the fundamental burning velocity (0.4 m/s for gasoline), a property of the fuel.12 The critical velocity, a function of mesh geometry, is also given by Gossel:17
where a is the fractional free area of the arrester surface, y is the thickness (width) of the arrester elements (cm), and d is the diameter of the apertures (cm). Gossel presented two different equations for the critical velocity. We are using the more conservative equation (shown above) with a safety factor of 2.5.
We have evaluated these criteria for flame arresters in three gas cans using the
equations above. The results are given in Table 1 below:
16 Rudolph J. Anschicks, “Filling and Venting Device,” U.S. Patent No. 1,814,656, (Assignor to Protectoseal Company of America, July 14, 1931).
17 Stanley S. Grossel, Deflagration and Detonation Flame Arresters, (New York:
American Institute of Chemical Engineers, 2002).
For each of these arresters, the mesh hole size is less than the critical diameter, and the critical velocity is larger than the fundamental burning velocity for gasoline. All of these flame arresters have been successfully tested in gas cans. Blitz states that their screen is meant to be used as a filter, but we have found that it also works as a flame arrester. The flame arrester from
the JUSTRITE can is shown below:
Figure 10. JUSTRITE flame arrester; fits into the can at the spout base.
Fifteen tests were conducted with flame arresters with the same aged gasoline and under the same conditions as tests that produced explosions. Not once did an explosion occur. When measured, the temperature of the flame arrester (installed at the base of the spout and similar in size to the JUSTRITE device) was always found to be elevated near 100oC; the flame had reached the arrester, but could not pass through.
Video still pictures of the explosion tests with flame arrester are uneventful and shown below. Note the test flame at the end of the spout.
Figure 11. Gasoline cans tested with flame arresters exhibited no explosion.
7. Discussion and Conclusions The extensive testing and analysis performed in this study demonstrate that explosions of gasoline stored in commercially available portable plastic gasoline containers can and do occur. Preconditions for such events are the presence of an open flame or static ignition source and the presence of a vapor space where the percent hydrocarbon concentration is within the explosive range between the accepted upper and lower explosive limits for commercial gasoline blends.
Combinations of conditions leading to the vapor space being in the explosive range include aging of gasoline, low ambient temperature and/or a small amount of gasoline remaining in the can.
The testing and analysis also demonstrate that an inexpensive screen flame