Sorry, if this was covered elsewhere in the 19 pages of thread & I missed it..but..
Am I right in thinking that the severity wouldn't have been quite as bad with a tgv, since the tighter coupling between carriages on a tgv would have meant that the train wouldn't have jack-knifed as badly before coming to a standstill & the carriages would be more likely to remain upright?
I believe that the tgv is safer in this regard, so I wonder why the shared bogie concept is unique to Alsthom's TGV/AGVs. Gauge clearance limitations? less operational flexibilty to switch carriages? patents?
Watch the crash video - the train did not jacknife. Something caused a ripple but the point at which this happens is obscured by an OHL mast in the foreground. At this point something sets up a 'ripple' through the train.
A Talgo car (about 4/5 in rake) lifts up and the following rake heads off tangentially to the curve. Heading miraculously up the slope and on to the street above.
Further down the train a second Talgo car lifts up and the light central section, with either the leading end or training end having no wheelset, heads off, whilst the trailing section with the weight of the generator car and rear loco, follows the curve around - possibly with the heavier vehicles remaining on the rails for some distance.
The front locomotive remains on the rails to a point well past the start of the retaining wall until the loose Talgo car flailing around at the end hooks on to the start of the retaining wall and jerks the train back pulling the vehicles in front off the rails. The rear section then hits the last vehicles of the front section.
Generally the stability of rail vehicles with no defects running on track with no defects is such that taking a curve at 200% of the posted speed limit will not result in a derailment. In tests for APT a Mk 1 carriage was sent down a steeply sloping branch line in Kent. It remained on the rails up to a point where there was over 20 degrees of cant deficiency for the speed it was travelling on the test curve.
Put simply the passengers inside a train would be slammed against the side of the carriage before the outer wheel climbed the rail or the carriage rolled over.
Two incidents in the UK are worth considering - in 1937 an LMS Coronation Pacific reached 112 mph at a point 3 miles South of Crewe, and then they realised that the tracks at the South end of Crewe station had a 20mph speed restriction - the train hit the first set of points at 3 times the posted speed limit and stayed on the track. The track was seriously overstressed (broken chairs etc) and those on board emerged ashen faced. Then in 1969 at Morpeth the Northbound sleeper train hit a 40mph curve at 80mph. The locomotive stayed on the rails but the next coach (a BG) apparently had the inside wheel lift and the high rail wheel climb over the railhead. The BG turns on to its side and was wiped out to a bare underframe, still coupled to the loco by the time the train came to a halt. The rest of the train split off and the leading vehicles headed out tangentially.
The 40mph limit applied to a 17 chain curve, with around 4" of superelevation, however the track had some 'bounce' through space under some of the sleepers such that they could move downwards but also, if the outer rail had a greater weight pressing down a local and transient twist would be delivered as the wheels passed over the 'soft' sleepers. Track twist is a serious defect and can often cause derailments. In addition a close measurement of the curve revealed a local reduction in radius from 17 chains to 15 chains, right at the point of derailment.
Fierce brake applications can send a ripple through a train, as each wheelset will slow down at a slightly different rate. With old loose coupled wagons the progressive build up of forces can rip drawgear off or conversely smash a wooden bodied truck to matchwood.
So returning to Santiago de Compostela - somehting happened at that point on the curve - we have very few pictures or details of the track at that point, but for so many derailments the point at which the train smashes up is generally well past the point at which the chain of events began, and a good investigator will be looking a long way back down the track