A team of scientists at a federal lab in Beltsville, Maryland, is working to re-create the characteristics of the fire that burned the apartment house at 20-24 Noyes St. in Portland on Nov. 1, killing six people.

The Fire Research Laboratory of the federal Bureau of Alcohol, Tobacco, Firearms and Explosives is trying to replicate in a computer model, as precisely as possible, the exact conditions under which the fire occurred, including the location and type of furniture in the building, whether doors or windows were open, and the burn rates for the various materials used in construction of the two-story duplex, built in 1920. That could provide answers about how the fire started, why it spread the way it did, where smoke and superheated gases concentrated and, ultimately, why it was so deadly.

“A lot of what we’re hoping to get out of the lab results is to see if our hypothesis is accurate for the starting of the fire,” said state Fire Marshal Joe Thomas, a former Portland fire chief. “Is that what did it?”

Thomas wouldn’t elaborate on investigators’ theory of the fire other than to say it does not appear someone set it.

“Certainly we’re not looking at a whodunit here,” he said Monday.

Investigators had hoped the ATF would be done with its analysis by Thanksgiving, but by Monday they still had received no word on the results of the fire modeling.

“We’re building a house all over again inside the computer, which takes time,” said David Sheppard, senior technical adviser at the ATF’s research lab, billed as the largest and most advanced fire science lab in the world. “Once we’ve done that, and we’re running the model, those runs can take anywhere from days to months to complete.”

Sheppard said he could speak generically about the work of the lab but not about any specific case.

The ATF has posted a video online that shows the technology at work – a detailed investigation into a 2011 fire that killed a Baltimore firefighter. The model was able to shed light on the impact of different firefighting techniques and on the ramifications of residents’ behavior.

Most inquiries to the Fire Research Laboratory are not so complex. The lab helps investigators in the field determine whether a source, like a frayed wire or a light bulb, would be enough to ignite nearby combustibles. Investigators try to determine whether any accidental sources could be the cause of fire, because if there aren’t, then the fire is suspicious.

The lab runs tests to determine how fast, for instance, a given type of couch would burn and how much heat it would give off. Investigators often try to recover a sample of a couch or mattress or some other relevant material for the scientists to use in determining how fast the fire would move. They also try to mimic the qualities of the building materials as closely as possible, which is harder with a 94-year-old building than a newer structure.

“It’s not a quick process, unfortunately,” said Sgt. Joel Davis, supervisor for the southern division of the State Fire Marshal’s Office. “There’s a lot of legwork in finding like materials.”

In the meantime, the families and friends of those killed, neighbors and city officials can only speculate on how the Nov. 1 fire started and became the deadliest in 51 years in Maine’s largest city.

The death toll is one of the reasons the ATF offered the services of its forensic fire science lab. Investigators are able to determine the cause and origin of most fires, but each year there are a couple of fires in each state where the lab’s advanced technology can be helpful.

The ATF is the federal agency with jurisdiction over arson crimes, concentrating on those with serious loss of life or large property destruction. The lab has been open for 11 years and has performed more than 9,000 experiments, Sheppard said.

The lab includes a huge enclosed area, capable of holding a three-story building so scientists can replicate the conditions in a given fire. Engineers can rebuild a structure and burn it – sometimes multiple times – to assess how a fire would behave. The enclosure measures the heat emanating from different areas within the fire scene.

Information about burn rates of materials, environmental factors like temperature and wind, as well as observations by firefighters and other witnesses are input into a computer model that can show where and how quickly smoke and superheated gases moved through a structure. That Fire Dynamics Simulator program breaks down three-dimensional space inside a structure into cubes that can be as small as 4 inches square. It then calculates the rate of heat change in each square and how it affects the squares adjacent to it.

That can show how heat moves through a building, and when air next to surfaces like a ceiling or a wall gets hot enough to ignite them.

“If there’s fire in one room, and people are incapacitated in the other end of the building, we’ll use computer models to determine: Would products of combustion – toxic gases – have spread to the other regions of the house,” Sheppard said.

Figuring out when and if working smoke detectors would have been activated requires understanding relatively small movements of air near the ceiling or wherever the detector is located, Sheppard said.

That complexity is one reason that the results of fire modeling can take so long, Sheppard said. In the early years, the modeling could only be done by supercomputers. And the models often must be run repeatedly.

“What if there was a window that was open? What if somebody left the door closed? What if you had forced hot air heating and the furnace goes on? That would push air away from the smoke detector,” Sheppard said. Models have to be run on each variable, he said.

In the Baltimore case, investigators knew the cause of the fire – a pot of cooking oil on the stove had ignited – but sought to understand what about the fire’s behavior created the deadly conditions in the six-unit building. The model showed that a door left open in the apartment where the fire originated allowed gases heated to 600 degrees to drift into a stairway and up toward the second-floor apartments at about 6 mph. At the same time, flames from the first-floor apartment leaped through the open door to ignite a second-floor porch.

As firefighters rescued residents, the flames on the exterior porch broke through a glass window, igniting plastic slat curtains and then a pair of couches.

Couches burning on the second floor, and heat and air coming up through the stairwell, caused the second-floor apartment to “flash over” – an expression meaning that when the air temperature is so hot, all of the exposed combustibles suddenly ignite. The third floor, where firefighters were searching for victims, flashed over 30 seconds later, the model showed. Couches in a third-floor apartment ignited, blocking the retreat of the two-person search team, and the balcony had been ignited by the fire below, the model showed, blocking that exit.

One firefighter survived by climbing through a window and going head-first down a ladder. The other was unable to escape. Firefighters fought their way back in and retrieved him, but he died later at the hospital.

Apartment doors left open gave the fire ample oxygen to burn and spread quickly up the central stairway, the model showed. The model also was able to show that simply shutting the front door of the apartment where the fire started would have dramatically slowed the spread of the fire.

It is impossible to re-create all of the conditions exactly as they were at 20-24 Noyes St. on Nov. 1. But the computer simulation might help answer questions about why so many people were unable to escape and why one of those who did was burned so badly that he eventually succumbed to his injuries. It might also direct investigators to the point of origin in spite of the immense damage to the structure.

The heaviest fire damage in the 2½-story duplex was in the area of the front porch, but where the fire started precisely and whether it was just inside or out on the porch itself has yet to be determined definitively.

“Their analytic ability here is almost to be able to give us answers to the unknown,” Thomas said, referring to the scientists at the ATF lab.

“In this case here, we know where the fire was at different points of time,” Thomas said. “We at least know what some conditions were. What we want to be able to do is work backwards.”

If the fire did start on the porch, he said, investigators want to know “where would it have been inclined to breach the house, how it got into the second floor.”