If all goes well, a massive fireball of hydrocarbons will ignite in the New Mexico desert some time in the next year. It will be part of a multiyear Department of Energy research project to understand whether the chemical composition of unconventional crude oils changes the risk they pose to the nation’s highways, pipelines, and railroads.

If all doesn’t go well, a similar massive fireball could go up over a derailed train, as happened in the town of Lac-Mégantic in July of 2013, killing 42 people, or outside Casselton, North Dakota, in December of that year, where somehow there were no reported injuries.

These explosions, and several other high-profile derailments and spills, called attention to the danger of transporting crude oil in unprecedented amounts on the North American rail system. As more pipelines have come into play, the rail-transport boom out of North Dakota has eased—approximately 150,000 barrels of oil travel the rails each day now, down from over 800,000 barrels at the peak in late 2014. Media attention has declined in lockstep, but there are still open questions about the chemical properties of these oils—many of which are produced from newly tapped shale formations around the country.

Researchers and oil industry workers know that what we call “oil” has undergone a massive change over the last 15 years. Oil is composed of many different hydrocarbons of different kinds. Some are very light, literally meaning there are fewer atoms of carbon per molecule—think propane—and others are heavy—think tar or the bitumen in asphalt. As more types of liquid fossil fuels are extracted in more varied ways, the actual chemical components that go into a barrel have become more variable.

David Lord and a team at Sandia National Laboratories have been working on finding and developing methods for analyzing oils from different fields to quickly assay their chemical constituents. Traditional techniques have been good at tracking the heavy stuff because those molecules are worth more; businesses measure it precisely because they need to. But the lighter substances have mostly been ignored by the industry’s standard procedures. These light components are more volatile, and it seems logical that oils with higher levels of volatility might be more dangerous to transport.

Over the last several years, Lord and the team have shown that they can accurately identify the components of different oils and their volatility. And this year, they will try to understand what those different solutions of hydrocarbons do in the real world.

“We’ve demonstrated that we have the means to delineate small differences in vapor oil pressures and composition,” Lord told me, “but the million-dollar question is: Does it result in any hazard consequence on the ground in real terms?”

* * *

Run the clock back to 1950 or 1980 or even 2000 and look inside a barrel of oil. It might vary in how much sulphur it has, making it sweet or sour, and it might vary in molecular weight, making it heavy or light. But those variations aside, most oils were very similar to one another.

That began to change with the rise in production throughout the shales of the Midwest, primarily the Bakken shale in North Dakota and the Permian and Eagle Ford formations in Texas, as well as the development of a wide variety of unconventional extraction techniques for deepwater drilling.

“There are different extraction techniques. Fracking, of course, but even fracking isn’t one category. You can inject fluids or you can acidize and melt rock away,” says Deborah Gordon, who began her career at Chevron and is now the director of the Energy and Climate Program at the Carnegie Endowment for International Peace. “Oil used to be one thing, but now oil is a much more complicated thing.”

For consumers of gasoline or diesel, of course, nothing has changed. We can still go to the pump and get exactly what we got before because refiners can take all these new and different components and transform them into the products that our cars and trucks need. But that’s getting harder, as both the extraction and refining steps become more complicated.

“Everything is changing but no one really is aware or cares,” Gordon says.

Her recent research unintentionally brought her into contact with this oil-composition problem. She started off looking at the climate. For the last several years, her team, including Joule A. Bergerson from the University of Calgary and Jonathan Koomey from Stanford University, has been trying to develop an Oil-Climate Index that looks at the life-cycle greenhouse-gas emissions from different oils. What they’ve found is remarkable: Some oils produce many times the total emissions of other oils. But all these substances go by the same name.

Some of this information is publicly available, especially in Canada, where a service called CrudeMonitor provides detailed chemicals assays of different oil streams. But as Gordon’s team searched for data on the components of oils, they kept finding that this information was generally hidden. Oil suppliers sometimes released assays of all the components of a crude as marketing material, but generally speaking, they were not publicly available. “There has never been a global inventory,” Gordon says.

It might seem that somebody must know what actual molecules are in these oils, but Lord’s team found that the traditional sampling methods had been missing the more volatile components known in the industry as “the light ends.” So, even the companies selling the stuff might not know everything that’s in it, though they certainly know more than regulators or the citizens of towns along the rail links or on top of the pipelines.

Koomey added, “The thing that struck us most forcefully as we studied oil more deeply is just how little safety and environmental regulators know about oil and oil products that are being shipped in large volumes around the world.”

This issue came to the fore again with the sinking of an Iranian freighter carrying condensate in the East China Sea. No one has ever seen an oil spill like that, especially the fire that engulfed the ship, making rescue of the crew impossible. For a time, there was a ring of fire in the ocean surrounding the burning ship.

The fire aboard the Iranian tanker, the Sanchi (China Daily / Reuters)

Condensate has no true definition, but it is a kind of very light oil that’s generally coproduced with natural gas and that contains a mixture of light hydrocarbons, from butane to octane. The problem is that in such a mixture, Gordon says, the different components “flash,” or go from liquid to gas, at different temperatures.

“That’s what makes them very dangerous. The butane component, which can be 5, 6, 7, 8 percent of the total, flashes at below freezing,” Gordon says. “If there is any weakness in the containment system, or an accident where you are having an impact, as soon as you have the butane exposed to the atmosphere, it is an explosion.”

If a tanker, or a whole 101-car line of tankers called a unit train, were traveling through a town, it makes sense that local authorities would want to know the very particular contents of those cars.

The railroad industry has made progress on this score in recent years. They designed an app called AskRail in collaboration with the International Association of Fire Chiefs to provide information to first responders in train accidents involving hazardous materials, according to a spokeswoman for the Association of American Railroads, an industry group.

But the app is only as good as the information that’s contained in it, which is provided by the shippers of the goods, which, in this case, would be the oil industry. They would probably use the codes developed by the Pipeline and Hazardous Materials Safety Administration, which provide several options for encoding what’s in a given tank car: “petroleum crude oil,” “petroleum distillates,” “petroleum products,” “oil, petroleum,” and “petroleum oil.”

All of these products call for the same safety protocol, which dictates isolating any “tank, rail car, or tank truck” that’s on fire for 800 meters (half a mile) in all directions with consideration of “initial evacuation for 800 meters (half a mile) in all directions.”

On the other hand, flammable gases, like butane, call for a different protocol, doubling the distance of isolation to 1,600 meters (a mile).

If a train full of condensate derailed, what would pop up on the app? Half a mile or a mile? What should pop up?

* * *

It’s possible that I’ve overstated the importance of the chemical composition of these crudes. A major National Academies of Sciences, Engineering, and Medicine report, which hit prepublication late last year, Safely Transporting Hazardous Liquids and Gases in a Changing U.S. Energy Landscape, recommended a variety of measures, and analyzing the oils themselves was not on the list. “The limited (less than a decade) experience with flammable-liquids unit trains suggests the high kinetic energy of the derailments has been a factor in the ignition of released product, more so than the specific volatility characteristics of the product being transported, such as its vapor pressure,” the report states. “However, there has been limited modeling of derailment kinetic energy that results in the conversion of thermal energy associated with multi-tank car derailments.”

In other words, if you have a train derailment of tank cars filled with flammable liquids and that stuff comes out, it’s gonna be bad, regardless of the specifics of the oil.

“There are issues with categorizing the crude oil, and there are definitely some differences between oils,” says Micah Himmel, a National Academies of Sciences, Engineering, and Medicine program officer who worked on the report. But when it comes to the safety of rail transport, the composition of the oil is not quite a red herring, he says, but it is a “pink herring.”

So, the obvious recommendations are to do better track maintenance to keep the trains on the tracks and to update the tank-car fleet to improve its crashworthiness. Unfortunately, very few new tank cars have actually been put into use in recent years, and even cheaper retrofits have been slow. The bulk of the fleet carrying flammable liquids are the same old tank cars that were identified as part of the problem in previous accidents.

On the other hand, the report does also recommend that authorities systematically model “the full array of factors that can give rise to and affect the severity of flammable-liquids train crashes” and derailments.

And that is the reason for all the burn tests in New Mexico.

The Sandia Laboratory team will use two facilities, a controlled indoor facility called the Thermal Test Complex and a larger outdoor site, the Lurance Canyon Burn Facility. They’ll run two types of tests: a pool fire, where a bunch of oil is sitting in a basin burning, and then the fireball, which Lord described as “a situation where a vapor plume has been cast up into the air because a containment vessel has failed and it creates a large fire suspended above the ground that’s fairly short-lived.”

The researchers have been systematically working down what the Department of Energy calls its Sampling, Analysis, and Experiment Plan. For the burn tests, the plan lays out the goals of identifying any oil properties that seem to affect the severity of a hazard, and then creating a list of ways that this analysis should change how crude oil is tested.

In any case, it doesn’t seem like having more information about all the crudes crisscrossing our nation can be a bad thing.

“In order for the regulators to understand the risks for different oil and gas products, they really need the assays, but they have been reluctant because of the power of the oil industry,” Stanford’s Koomey says. “If the regulators had the assays and the right people with the right knowledge, they could make more sensible judgments about shipments. But they don’t have it and they don’t ask for it.”