Pump output can be greatly diminished by cavitation and would give the results seen in this case.
The HP gear pumps show evidence of cavitation. Two of the principle determinants for the point at which cavitation occurs are pump speed and fluid viscosity.
A mixture of hydrocarbons such as jet fuel can have a temperature dependent change of viscosity while in the liquid state.
Cold high viscosity fuel along with an increase in pump speed at thrust demand may have lead to the cavitation.
Fuel that was not frozen but chilled to a high viscosity would have flowed through the large surface area filters without producing an alarm. And the boost pumps may have handled it without a problem too.
But how and why did it happen on this flight and not so many other flights?
I look back to a post on the previous thread (pg 20, post 394, by Glueball)
http://www.pprune.org/forums/showthr...310013&page=20
"The temperature probes in the 747-400, 777, and MD-11 are located where the bulk of the fuel is coldest. However, some fuel may be
colder than the fuel measured by the probes, such as the fuel that is in contact with the lower wing skin. This creates a
temperature gradient in the fuel tank from the wing skin to the location of the probe.
As fuel travels to the boost pump inlets, the bottom, cold layer flows through small flapper valves located on solid tank ribs next to the bottom wing skin. These valves are used to control fuel slosh. Thus, the
cold fuel tends to flow toward the boost pump inlets."
Speculation ...
This flight started with fuel from a location that had cold weather. The initial FL resulted in a cold soak. The original flight plan was prudent for the conditions expected. The crew elected to honor a request from ATC to climb to FL348, ambient temperature approximately -65ºC. This was followed by a climb to FL380, -76C.
It seems to me that there was an extreme gradient between the coldest side of the tank and the fuel probe. While the the bulk of the fuel remained above the min. temp, I think that along the tank side, the fuel was chilled to the point that it congealed on the tank surface. Maybe it was still not frozen but perhaps so high in viscosity that it no longer moved with fuel currents to be remixed.
Once the tank surface warmed at arrival this high viscosity fuel would be released to flow toward the boost pump. Since the temp probe is not located at the pump inlet, it may have never seen this cold spot in the fuel.
Holes in the cheese ...
To reproduce this you have to cold soak the fuel. Expose the tank to extreme cold to get the required gradient, tank to probe, otherwise the crew will see the problem and correct for it. On arrival, a little more fuel in the tanks and the currents might not direct the cold spot to the pump or less fuel and the sloshing of flight maneuvers might remix the fuel. And the timing has to be just right. Early release of the cold spot gives a greater time for the remix of the fuel. A late release and the plane is on the ground before the cold spot gets to the pumps.
On this flight it all lined up. And it's going to be hell to prove. I bet it's a long time before we see a final report.