Sandia Labs Crude Oil Fire Report – August 2019

Last month Sandia National Laboratories published a report on a series of experiments done looking at the pool fire characteristics of a variety of crude oil samples with a wide range of vapor pressures. At least one news organization has jumped on the conclusions explicated in the Abstract to express support for the contention that the regulation of crude oil shipping should not take into account differences in oil vapor pressure.

Study Conclusions

The lengthy and very technical report from Sandia, part of the ongoing DOE investigation of the hazards associated with crude oil shipments, makes the following statement in the Abstract (pg 3):

“The results indicate that all the oils tested here have comparable thermal hazard distances and the measured properties are consistent with other alkane-based hydrocarbon liquids. The similarity of pool fire and fireball burn characteristics pertinent to thermal hazard outcomes of the three oils studied indicate that vapor pressure is not a statistically significant factor in affecting these outcomes. Thus, the results from this work do not support creating a distinction for crude oils based on vapor pressure with regards to these combustion events.”

The key phrase in the above quote is found in the last sentence; “with regards to these combustion events.” Pool fires and fireballs were created and analyzed. The study assumed that in a crude oil derailment accident the conditions would exist to cause these two conditions. In the discussion portion of the report it is noted that (pg 75): “Based on the Phase I effort, the premise is that most train accidents

provide enough kinetic energy to exceed the parameter thresholds indicating flammability;

consequently, ignition is highly probable regardless of the crude oil type.”

Vapor Pressure Measurement

This study used an automated vapor pressure measurement system (ASTM D6377) at 100˚F. This is a different method from the ones that I have discussed previously in this blog (see here for instance), but the report authors include an important discussion (Section 1.1, pg 26) about the need for proper sampling techniques and storage of tested samples. Any discussion of vapor pressure testing needs to address these issues.

Fireball Testing

The fireball testing conducted in this study was designed to look at the effects of the ignition of vapor clouds over a derailment event. It is clear from the description of the test methodology (pg 253) that investigators were concerned about vapor releases from intact railcars that were subject to the intense heating associated with direct flame impingement from a pool fire caused by a release of crude oil (or other flammable liquid) from a nearby ruptured railcar.

The test tanks were heated to 300˚C and 280 psi and a rupture disk was then command released via explosives. To ensure ignition of the resulting vapor cloud, a second explosive device was then detonated.

Commentary

The test information presented in the report is very valuable for fire response planning. It is not really surprising that the test concluded that there is little effective difference in the thermal effects of a pool fire from crude oil with wide variations in vapor pressure. Those thermal effects are more closely related to the heat released in the combustion of hydrocarbons and that is directly related to the number of carbon atoms burned, not the physical state of the molecules within which they are contained. Similar masses of carbon atoms in linear chains will produce similar amounts of heat. This is chemistry 101.

The testing of the fireball, similarly restricts the evaluation to the heat effects and the size of the fireball. Again, this is useful information for fire response planning, but it does little to address the underlying concerns about the dangers associated with variations in crude oil vapor pressure; that is the likelihood of a vapor cloud forming in a given accident.

Since DOT mandates that the pressure relief valve (PRV) on crude oil railcars release vapors at 32 psi, the testing at 280 psi is of little value. What would have been more impressive would have been gradually heating the samples in a pressure vessel until a standard 32 psi PRV opened and then igniting the resulting vapor cloud after some preset time limit. The emissivity testing reported in this study would be done on the resulting fireball, but overpressure testing at set distances from the test to evaluate differences in the blast effects from the resulting fireball should also be required.

Vapor pressure testing of crude oil is going to be of only very limited usefulness. For relatively pure substances, calculating vapor pressures at varying temperatures from single temperature testing is a rather simple application of Boyles Law. For complex mixtures like crude oil this is not the case. Each of the hundreds of components of crude oil has its own boiling point, the temperature at which it begins to significantly contribute to the vapor pressure of the mixture.

To be a valuable predictor of fireball formation in a crude oil derailment, we need a new vapor pressure testing method. Instead of measuring vapor pressure at a fixed temperature, it would be more useful to regulators to have a test that measures the temperature at which we would expect safety devices to release a vapor cloud. For rail transportation that would be 32 psi. Unfortunately, such a test would present an interesting set of potential physical hazards in the testing facility. And that would significantly increase the cost of testing.

It would be helpful in Sandia did a test evaluating a similar variety of crude oils as seen in this testing to see what sort of temperature variations were seen in the temperature necessary to reach 32 psi vapor pressure and what variations were seen in the fireball testing conducted at those temperatures. Oh, yes, please include overpressure measurements during the fireball testing. If there were relatively little practical difference in the release temperature, emissivity and/or overpressure, then we could probably conclude that vapor pressure testing was a dead issue.