A BNSF train pulled out of a terminal in North Dakota's Bakken Shale region last month, hauling 106 tank cars filled with crude oil on a westward journey to a refinery near Ferndale, Wash.

It didn't get far. Just 30 miles into its thousand-mile-plus trek to the Pacific Coast, the train derailed July 16 on a straightaway outside Culbertson, Mont. Twenty-two tank cars left the rails. Five were breached, BNSF says, spilling 35,000 gallons of Bakken crude. In the pile-up, a live power line was knocked down.

Then something curious happened, different from 10 other oil-train crashes since 2013: The oil did not ignite.

"You could smell it from over a mile away," said Chief Deputy Corey Reum, of the Roosevelt County Sheriff's Department. "With the crush of the tank cars and a live wire, it did not catch fire."

No explosion. No drama. Scant national attention.

Untangling this Mystery of the Unexploded Train Wreck may yield important lessons for those seeking solutions to the fiery rail accidents that have beset North America since the domestic oil boom took off six years ago.

Without sufficient pipelines in place, about 60 percent of North Dakota's production moves by rail. Much of it ends up at Philadelphia-area refineries.

Philadelphia Energy Solutions' refinery complex in South Philadelphia alone processed 13.9 percent of last year's total Bakken output, according to regulatory filings. All of it came by rail.

The oil spilled in the latest accident had been conditioned to reduce its volatility to comply with a North Dakota order that went into effect in April. Some are suggesting that the reduction in vapor pressure may have reduced the chance of a fire.

"All we can tell you is that the [oil producer] was in compliance with our order," said Alison Ritter, spokeswoman for the North Dakota Department of Mineral Resources. "The reasons as to why it did not ignite would have to come out of the investigation."

The flammability of North Dakota's crude oil came under suspicion after a runaway train derailed in downtown Lac-Mégantic, Quebec, in 2013, setting off a conflagration that killed 47 people.

Last year, the U.S. Department of Transportation's Pipeline and Hazardous Materials Safety Administration declared that Bakken crude has a higher gas content, higher vapor pressure, lower flash point, and lower boiling point, "and thus a higher degree of volatility than most other crudes in the U.S., which correlates to increased ignitability and flammability."

The oil industry said that statement was not based on solid science. The American Petroleum Institute said that North Dakota petroleum was not substantially different from other light crudes and that it met standards for products that can be moved by rail.

The Energy Department's Sandia National Laboratory, in a March review of current scientific studies, seemed to support the industry. Sandia said the relationship between crude-oil properties and the probability or severity of a fire in a rail accident had not been established.

"No single parameter defines the degree of flammability of a fuel; rather, multiple parameters are relevant," Sandia said.

Nevertheless, several members of Congress have introduced bills to mitigate the volatility of oil shipped by rail. In April, U.S Energy Secretary Ernest Moniz announced a two-year study into how crude-oil properties affect its combustibility in rail accidents.

North Dakota regulators, seeking to address public concerns, approved in November new requirements that crude oil not exceed a vapor pressure of 13.7 pounds per square inch (psi), a measure of its volatility. The limit was based on national standards for stable crude.

The new rules require producers to warm the crude to 110 degrees, which separates some gaseous compounds, such as butane and propane. Vapors are flared off at the well site before the oil is transported to markets.

Five weeks after the rules went into effect, a BNSF train derailed May 6 in Heimdal, N.D. The oil ignited, but there was no massive explosion or fireball of the sort that has characterized other oil-train accidents. The state's mineral resources director said the Heimdal experience was evidence that the new rules were working.

Yet some oil-industry experts are skeptical that vapor pressure is the deciding factor if crude oil ignites in a train crash.

The oil in May's Heimdal accident tested at 10.8 psi, much lower than the new North Dakota standard. But the crude oil in a 2013 BNSF accident in Casselton, N.D., was also under 11 psi. That accident forced the evacuation of 1,400 people and emitted a series of blasts and a massive fireball.

A spokesman for Statoil, the Norwegian oil producer that shipped the crude involved in the most recent derailment, declined to disclose vapor-pressure data but said the crude complied with the new regulations.

Some experts doubt whether the North Dakota rule will make a difference. They say most Bakken crude already meets the new vapor-pressure standards.

"Very little of it actually needs to be conditioned to meet the guidelines," said Kathy Neset, a geologist and oil consultant in Tioga, N.D., who has worked in the region for decades.

Her assessment was echoed by Philip L. Rinaldi, chief executive of Philadelphia Energy Solutions, which operates the South Philadelphia refinery complex formerly owned by Sunoco. The refiner's strategy, outlined in filings with the Securities and Exchange Commission, is to source up to 80 percent of its 335,000-barrel-a-day capacity from the Bakken Shale. That's nearly four oil trains a day.

Since PES began shifting to Bakken crude-oil supplies in 2013, Rinaldi said, the oil has not exceeded the 13.7-psi standard.

"We have yet to get a shipment that has a vapor pressure in excess of that," Rinaldi said in a May interview. "This regulatory modification ain't going to make a difference because we already get stuff below that level."

Rinaldi says he believes new rail-car standards adopted by the Department of Transportation in May are more likely to reduce catastrophic accidents. Those standards require thicker shells, more puncture resistance, and thermal protection to reduce the chances that undamaged tank cars will overheat and explode if engulfed in fire.

The circumstances of the July 16 accident in Montana are still under investigation by the Federal Railroad Administration. The train was traveling at 44 m.p.h. when it derailed - the speed limit was 45 m.p.h. A spokesman for the agency said investigators took away some portions of broken rail for examination, suggesting that the track may be suspected. Most derailments are caused by failures of track or cars.

The BNSF train included CPC-1232 tank cars, an industry design adopted in 2011 that will be replaced by the newer standards the government announced in May.

BNSF spokesman Mike Trevino said the five tank cars that leaked suffered either punctures or valve failures. The 35,000 gallons that spilled was a little more than is contained in one full tank car.

"It's not a foregone conclusion that a derailment of an oil train will result in a fire," Trevino said.

Fires start when the right combination of flammable vapor and air comes in contact with an ignition source, said George Bibel, a professor of mechanical engineering at the University of North Dakota and author of the book Train Wreck: The Forensics of Rail Disasters.

Typically, a shower of sparks is created by the tremendous amount of energy given off by rail cars crashing against steel rails - each car weighs up to 286,000 pounds.

A downed power line might also provide an ignition source. But the sparks need to make contact with the fuel.

"It's a matter of how many sparks you make, which is kind of a random event," said Bibel. "And then it's a matter of the sparks finding vapor. So it's multiple random events."