"the roll got out of control when the engine shut down due to the fire. This is because there was no roll control with the engine shutdwon."<br /><br />I am not sure. There was another statement that the burning, pressurized fuel was generating about 50-foot-lbs of roll torque before the engine shut down. I a perverse way this was sort of like the turbopump exhaust roll control gone to the dark side. <br /><br />From another statement by Musk, it seems that a connector was undone after avionics work. There are two possibilities. It could be in the fuel system. However, the fuel line is a large element, and if it was disconnected, there wouldn't be a leak but a real fireball.<br /><br />Another is in the hydraulic controls, although I don't know what is being controlled. (I thought that the engine steering was purely electromechanical, but that could be in error.) I believe that the Falcon uses pressurized fuel for this function. Leaking hydraulic fuel s is a more likely possibility in that Musk says that the telemetry in review shows the leak four seconds before lift-off. My guess would be that this is when the hydraulics are pressurized, or at least the pressure is allowed to flow to the actuators. <br /><br />This goes against what backtothemoon said. If a "hydraulic" fuel line disconnected on the Falcon 5 or 9, it would spew fuel around the engines that were still lit, causing a similar disaster, like mlorrey said.<br /><br />And there has been no report of the rocket exploding on impact. There are statements about parts like the aft fuel tank dome surviving relatively intact. Recall that if the engine shuts off around 30 seconds, and the fuel shuts off, the engines stop as it continues to coast up for probably another ten seconds, and decends for 18 seconds. (I did a crude estimate that the rocket got to about 1.02 miles. ) There may have been no flames by the time it hit, and it reportedly did hit the water on "a dead reef".<br /><br />Kimbal Musk (in kwajro