The value of this discussion is that it highlights the rational basis for the determination of the runway width standard. Sometimes with the passage of time, the rational basis gets a bit lost, and it becomes harder to see how the standard fits to the context of new developments and new aircraft [what a neat way of bringing the discussion round to the original basis of this post].
For runway width, I don't think requirements for each class of aircraft have changed since DC-3s were still in airline service. It could be simplistically argued that the now exclusive use of tricycle undercarriage in RPT aircraft means an increase in controllability and so a decrease in runway width requirement. Or it could be argued that the use of autoland means more accurate landings and a decrease in runway width requirement. These are empirical arguments which seem superficially attractive.
They miss the rational basis for the standard that JT outlined eloquently above, part of which is the takeoff case. And his point that if the critical engine fails at Vef just above Vmcg, and the runway is wet (with poor friction), and there is a cross-wind blowing which significantly increases the real Vmcg (to above Vef), then the nosewheel steering and rudder authority can be insufficient to control the aircraft and a rapid sideways departure can occur. Quite a believable scenario. All the requirements for runway width (and shoulder and runway strip) could come into play.
What was interesting in the certification assumptions, and missed by myself initially not being a performance engineer [thanks Santa for my copy of Swatton], was the detail of the definitions:
Vef is the speed at which, for the purposes of performance calculations, the critical engine is assumed to fail; it is never less than Vmcg.
Vmcg is the minimum control speed on the ground at maximum takeoff power such that, if the critical power unit becomes inoperative, it is possible by aerodynamic means alone without the use of nosewheel steering, using normal piloting skills to maintain a parallel path not more than 30 ft (9m) laterally from the original path.
This started me thinking about the original question by Prof2MDA on runway width. Taking the design? normal? engine failure case on takeoff where Vef is just above true Vmcg. The aircraft will probably have wandered slightly off centreline during takeoff anyway; I don't have the relevant research to hand, but from memory a normal aircraft wander could be up to say 5m off centreline. Then given a wheeltrack (distance between the outer tyres of the outer main gears of approximately 12.6m for a 747-400, this places the outer tyres at (12.6/2 + 5=) 11.3m off centreline. Then have an engine failure just above Vmcg and experience a parallel path departure of 9m allowed in the regs, and the outer wheels are now (11.3 + 9 =) 20.3m off centreline. If the minimum clearance between the outer tyre and the edge of the runway was set to be 1.5m (my estimate to allow for load spreading in the lower pavement), then that is equivalent to a minimum runway width of (20.3+1.5)*2 = 43.6m. Hmm - doesn’t leave much fat in a 45m wide runway. And it doesn't leave anything for JTs scenario with a crosswind. I don’t have the A380 gear width information to hand to redo the calculation for that aircraft - but I also wonder how THAT fits runway width.