Train tilting technology: How it works and why it makes sense
Rationale: more speed in curves
The limiting factor for train speeds on auxiliary lines is the high number of curves. A train or MU entering a curve from a stretch of straight track will usually have to brake and reaccelerate on leaving the curve. This slows the train and consumes current or fuel (usually the latter as auxiliary lines tend not to be electrified).
All this speed-limiting business is mainly done not to keep the rolling stock from flying off the track due to centrifugal pseudoforce, but to keep the passengers from falling off their seats or having to puke because of rapidly changing, intense transversal forces. Thus, if a way could be found to protect the passengers from these forces, the trains could go a lot faster on the existing track.
Tilt for less transversal force
As you might guess, tilting is this remedy. By tilting the vehicle and thus approaching the "up-down" axis of passengers' bodies with the direction of virtual gravity that results from the combination of gravity and centrifugal pseudoforce in a curve, the transversal forces acting on the passengers can be reduced or even eliminated. Ideally, this means that the train can go through a curve where it used to have to brake and reaccelerate without changing its speed. On some lines, scheduled running times have been reduced by 30% by using tilting trains or MU consists.
Engineers have worked on tilting train technology ever since the 1960s and 1970s. Then-modern "Trains of the future" such as the British APT or Deutsche Bundesbahn's equally ill-fated ET 403 featured tilting, yet only in the 1980s did regular service with tilting trains become a reality with Fiat's Pendolinos.
The seemingly simplest way of making a train tilt is simply suspending the superstructure of a car or MU in points higher than its centre of gravity. When going through a curve, the superstructure will tilt outwards below this axis and inwards above it.
It's not all that easy in reality, as it requires some kind of 'pillar' at each end of the vehicle and the arrangement of springs and/or shock absorbers becomes rather unusual. However, the Spanish manufacturer Talgo has been making passively tilting express cars since 1980 under the Talgo Pendular brand. (And in fact, those cars have their suspensions under the roof. Very strange.)
Passive tilting is low-tech, low-maintenance, does not consume power and does not require any on-track infrastructure. However, the achievable tilt angles aren't all that great: the most modern Talgo Pendular car, Talgo XXI, can tilt up to 3.5°. The transition process from untilted into tilted position and back is not controlled.
With active tilting, the superstructure of the car 'floats' over its bogies on hydraulic dampers connected an arrangement of levers. When transversal forces are detected by an inertial platform or the beginning of a curve is signalled to the vehicle's computer by either on-board navigation such as GPS or a trackside transponder, the car is actively tilted into the tilt angle necessary for compensation; at the end of the curve, it is tilted back again.
This is clever: active tilting allows for high tilt angles (around 8°) and it can anticipate curves so the passengers won't feel much. With the most refined tilting systems (those on the tilting diesel ICEs), even the inevitable sideways shaking-around of the superstructure on the dampers is compensated pneumatically so the cab will not hit the rubber bumpers limiting the tilt angle when reaching full tilt, eliminating one further source of uncomfortable transversal force and vibration.
However, active tilting requires lots of power, which is why it is usually only installed on MUs; it also requires lots of brains, i.e. computers, and it's notoriously high-maintenance, especially the German systems.
Active tilting was pioneered by Fiat in their Pendolino and Cisalpino trains. Their robust tilting technology is based on oil-hydraulic actuators.
To avoid Fiat patents and (purportedly) to save space, the German rolling stock industry has developed electric tilting systems based on linear motors. At first, they were woefully failure-prone, but today they do kind of a good job, for example in the 612 DMU.
I've heard Japanese tilting trains use a semipassive tilting system where the tilting is initiated, stabilised and finalised by an active system, yet the actual tilt angle is determined by the freely-suspended cab. Sounds like a good compromise, but I haven't got much information on this.