Let's put it this way;
It is well known that dumping water on landfill fires is not particularily effective.
It is known that it is nigh on impossible to stop oxygen from getting to the site of the burn in landfill fires..
It is also known that in numerous above ground fires that reports of molten steel/metal are common and in fact there are pictures of molten aluminum from automobile fires..
Physics tells us that in an insulated volume if the heat produced from combustion exceeds the heat dissipated by conduction (heat dissipation by convection and radiation is extremly small and zero respectively) the the temperature will rise.
Increased heat retention can be acheived also by using some heat to raise the temperature of the incoming air and the fuels yet to be consumed. In fact this would increase the efficiency of the burn and heat supplied by such things as plastics and auto tires.
The debris volume of the WTC towers had an underground oxygen supply, was insulated well and contained an enormous supply of various fuels.
Given all of the above there is no reason to suspect any self oxidizing incindiary had to be involved.
The WTC rubble was like a landfill in that it was an insulated pile of both combustibles and non-combustibles.
It was also like a landfill fire in that cutting off air and dumping water on it was largely ineffective.
It was unlike a landfill in that it was not packed down to the extent that a landfill is and it had an underground tunnel to feed it air.
It was unlike a landfill as well in that it was contained in a concrete basin with engineered drainage and surrounded by paved streets. ("so what!" tmd cries - it means that the initial condition of the rubble and soil is DRY and thus a better insulator than what would be found at a landfill)
If I had the time and inclination here is the experiment I would perform:
Using a 3-5 gallon steel can put 1 gallon of gasoline, one cup of motor oil, a cup of sawdust in the can (a mixed hydrocarbon fuel)
On open ground set the gas on fire and take temperature readings and log observations of significant events every 5 minutes until the gas burns out. ( This will take about an hour)
IMO what you will see is a smokey fire that flares up and subsides depending on the breeze. It will be quite hot but not close to the melting point of aluminum. .
Now redo the experiment but make a few changes.
First, fashion a 1/2 inch copper tube that runs , outside of the can, from just above the rim to a level just above the bottom of the can then back up and into the can to just above the level of the fuel. Do the same with 1/2 inch aluminum tube on the opposite side of the can.
Install a BBQ sparker in the can and run the wire up through one of the tubes.
Now, surround the can with at least a foot of dry sand up to the rim of the can. ( a plywood box at least 2 feet wider than the diameter of the can with the can, surrounded by dry sand, in the center)
Cap the can with a cone shaped lid with a 3/4 inch hole at the top and cover up to the rim of the hole with more dry sand. ( a large tin funnel would suffice )
Ignite the fuel with the BBQ sparker and take temp readings at the center of the smoke hole.
After the fuel goes out examine the aluminum and copper tubes for signs of melting.
IMO this is what will be observed:
The temps will greatly exceed the first run due to the pre-heating of the air supply, and the retention of heat by the insulating sand. The exit speed of the gasses will be quite a bit more than the first run and there will be less smoke due to more complete burning.
The aluminum pipe , and possibly the copper pipe will show signs of melting. In effect this will be a natural draft furnace.