This material is taken from a 1990 issue of the French Revue Generale
du Chemin de Fer (RGCF). It is loosely paraphrased and translated. If this
text is too technical for you, you can study up on railway suspension systems
at the Railway Technical Web Pages by Piers Connor, referenced in the
section under the "General" heading. The SR10 suspension described in this
article is now in use on all TGV trainsets in various forms. Even the
first-generation TGV Sud-Est trainsets were retrofitted with this design.
The articulated trainset architecture presents many advantages, but can
at first sight present some challenges in the design of a secondary suspension
that do not arise in a conventional architecture. These challenges include:
The lack of overhangs beyond the load bearing points of the secondary suspension
leads to a less favorable frequency response than for a conventional railway
car with two bogies. The end-supported car body in the articulated design
causes a natural excitation of the fundamental vibration mode, with the lowest
natural frequency and most noticeable discomfort to vehicle occupants.
The placement of the suspension points at the ends of the car body
leads to a higher hunting frequency (about the yaw axis) for a given lateral
suspension stiffness. This higher frequency body hunting mode increases the
cross-coupling to the bogie's own hunting mode, because of the lessened spacing
of the respective natural frequencies of car body and bogie. To avoid this
coupling, a stiffer lateral suspension can be used, but this reduces comfort
and causes its own problems at high lateral accelerations when running through
curves at 150 to 200 km/h on conventional lines. In the TGV architecture, the
stiffening of the lateral suspension by progressive elastomeric springs does
not solve the problem because it introduces non linear behaviors in the vertical
axis which adversely impact ride comfort.
The articulated nature of a TGV trainset where car bodies are
linked by a pin joint leads to coupling between car bodies. The chain formed
by consecutive cars thus has vibration modes of its own in addition to the
modes inherent in each car.
This abundance of vibration modes causes a more complex
and varied frequency response for a given dynamic input.
Despite these various handicaps, the performance obtained with the first
generation of TGV trainsets (the TGV Sud-Est) is quite acceptable, judging from
the response of passengers to the new service. The arrival of the second
generation TGV Atlantique and the increase of service speeds to 300 km/h warranted
an in-depth optimization study of the articulated suspension by SNCF.
Knowing the shortcomings of the articulated architecture, the designer can
be tempted to sidestep the problems altogether. One approach is to remove the
coupled modes of the car bodies by removing the pin joint between them and
instead suspending each separately on the bogie frame, with traction and buffing
forces transmitted by a conventional coupling. At constant suspension stiffness,
halving the suspended mass leads to a square root of two increase in the natural
frequency. Alternately a more flexible suspension can be developed to preserve
the modal characteristics, something which is technically difficult in the
lateral axis. This latter approach was tested by modifying the TGV 001 prototype
but did not prove entirely satisfactory.
Another research avenue is to create an efficient dynamic filter by making
the natural frequency of the secondary suspension very low, below 1 Hz. This has
the benefit of filtering out parasitic modes such as the vibration of the steel
spring used in the original TGV suspension, a large (300 kg each) spring with
very little internal damping and a natural frequency of about 20 Hz. The size
of the spring is all the larger because each bogie supports a load of
approximately 35 metric tons, much more than a conventional design. The large mass
leads to a lowered natural frequency more prone to coupling with other dynamic
From this stand point, a pneumatic suspension offers the advantages of very
low mass (only a few kilograms for the membrane), good internal damping, and a very
high flexibility made possible by an external air reservoir. By controlling
air pressure, a pneumatic suspension also affords a load leveling capability
which is made necessary by the increased flexibility. However, this type of
suspension presents two challenges:
If the pneumatic connection between the membrane and the air reservoir is not
carefully designed, the vertical flexibility can be frequency dependent. Above a certain
cutoff frequency the flexibility drops off and the reservoir no longer has an
effect, leaving the dynamics essentially determined by the membrane's flexibility.
This phenomenon is sometimes used to provide additional damping, but in this case
it can reduce the quality of the vibration isolation.
The membrane itself exhibits a lateral flexibility that is not constant with
displacement, unlike the linear behavior of a steel spring. On a stress-strain
diagram, a steel spring describes a straight line through the origin, while a
pneumatic membrane describes a hysteresis cycle. Strictly speaking, this makes
the simple notion of flexibility meaningless for a pneumatic membrane, unless
one defines it over a hysteresis cycle. The membrane has an effective transverse
flexibility that is the opposite of the desired behavior, exhibiting low
flexibility for small displacements (where high flexibility is desired to decouple
bogie hunting from the car body) and high flexibility for large displacements
(where more stiffness is preferable to maintain a better balance between
lateral displacement and passenger comfort).
The design finally selected for this suspension is composed of a pneumatic
membrane connected to a high capacity reservoir. The articulation of a TGV
trainset, with its inter-car support frames, affords ample space to accommodate
large reservoirs directly above, and in contact with the membrane, something
that would not be possible in a conventional architecture.
The high vertical flexibility obtained with this system requires the use of
a anti-roll member to prevent excessive leaning in curves while preserving
the desired dynamic properties. The natural frequency of the suspension
(in a loaded configuration) is on the order of 0.7 Hz, and appropriate design
of the membrane to reservoir connection keeps the vertical flexibility nearly
constant over the relevant frequency range.
The lateral suspension tuning was more difficult, in light of the membrane's
undesired lateral behavior. Through the research program a pneumatic membrane
was developed that offered minimal hysteresis and a high enough flexibility that
the car body's hunting frequency is on the order of 0.75 Hz (neglecting damping)
and thus well below the bogie hunting frequency. This specially designed membrane
incorporates a metal skirt with a geometry carefully determined in order to
reduce flexibility with increasing lateral displacement. This configuration does
not call for additional lateral bumpers, which as mentioned above adversely
affect the vertical axis. The lateral flexibility is reduced by a factor of 2
over a displacement from the neutral position to the bumper stops, which
themselves are still required to limit the lateral motion at maximum displacement.
Bogie-to-Body and Body-to-Body Connections
Any suspension design is composed of a restoring element (steel spring or as in
this case a pneumatic spring) that forces the vehicle back to its equilibrium
position, and a damping element (in general a hydraulic damper) which damps out the
body's oscillation by proper tuning to the spring's frequency. The dampers are
absolutely necessary and are generally mounted in parallel with the corresponding
restoring element; however, this placement causes problems. Above the damper's
natural frequency, the damper responds as a solid element and transmits parasite
vibrations from the bogie to the car body. Theoretical calculations as well as
experience have shown that removing the dampers, while unacceptable because car
body oscillations become poorly damped, significantly reduces the residual vibrations
inside the vehicle.
Bogie yaw dampers are less susceptible to this phenomenon because they are well
decoupled from vertical and transverse movements of the car body. Their function,
in a design where bogie hunting is already well decoupled from body hunting, is
to improve bogie stability. Their influence on vibration levels is almost
eliminated if the bogie design further benefits from a natural stability sufficient
for the desired speeds, which the long wheelbase design of the TGV bogies already
For the remaining dampers, the articulated architecture turns its liabilities
into assets. As any vertical or transverse movement of two car bodies articulated
on a common bogie induces a relative rotation between the car bodies, and hence a
relative displacement between the car ends, the damping in these axes can be done
entirely between car bodies by using longitudinal dampers at each corner of the car
end. This makes it possible to remove entirely the vertical and transverse dampers
that link the bogie directly to the body in a traditional architecture, thus
eliminating the parasitic vibrations transmitted by these dampers.
This arrangement provides excellent damping (all the more important because of
the very low natural frequencies obtained with the pneumatic suspension could cause
passenger discomfort) and maintains excellent dynamic filtering properties, resulting
in excellent ride quality.
Furthermore, the damper characteristics and their configuration affords a
significant improvement of the transient response of the suspension (switches/points,
track defects on curves taken at high cant deficiency) because the bogie's motion is
unconstrained by transverse dampers.
Finally, the excellent decoupling obtained between bogie hunting and car body
hunting led to a redefinition of the yaw damper, which remains the only damper linking
bogie and body directly, in the sole function of damping bogie hunting. When
decoupling is not so great, yaw dampers serve to damp both bogie and body hunting.
In this case, they are optimized for maximal effectiveness near the natural bogie
hunting frequency, which is about 4 Hz and much above the body hunting frequency.
An exceptional ride stability and level of comfort has been achieved through a
successful combination of computer modeling, subsystem testing and full scale line
testing on TGV Sud-Est trainset number 10.
Trainset 10 performed test runs in all configurations likely to be encountered
in revenue service (empty, loaded, with new wheels, with wheels at the wear limit
of 450000 km, on high speed lines, on standard lines, etc.) In addition, several
high speed test runs demonstrated that the SR10 suspension offers considerable
performance margin above commercial service speeds. The speed record runs of TGV Atlantique trainset 325 amply demonstrated
this. Trainset 325 was equipped with a stock SR10 suspension, the only modification
consisting of four yaw dampers rather than the usual two per bogie. A speed of 515
km/h was reached with an astonishing level of vertical and transverse comfort. At
400 km/h the dynamic environment inside the vehicle was equivalent to a standard
Corail coach at 160 km/h. Trainset 325 ran through curves at over 360 km/h with
cant deficiencies above 180 mm, as well as ran over switches/points at 500 km/h
without even coming close to the comfort guidelines set out by SNCF. This suspension
presents a significant advance:
In the vertical axis, the comfort figure of merit as calculated by SNCF was
multiplied by two. The effective vertical acceleration spectrum over a band from
4 Hz to 30 Hz was cut in half compared to the old suspension.
In the transverse axis, the comfort figure of merit was multiplied by 1.5.
The effective transverse acceleration spectrum over a band from 4 Hz to 30 Hz was
cut by a factor of 2 to 2.5 (depending on the point of measurement)
The transient response to track defects, while difficult to quantify,
was qualitatively greatly improved as the suspension swallowed up these
According to the ISO2631 standard defining effective acceleration measurements,
the mean vertical and transverse acceleration at 300 km/h on the LGV Sud-Est was just
For SNCF, the development of the SR10 suspension represents both a breakthrough
and a vindication.
The vindication, first: the articulated architecture with its many advantages
turned out to be an appropriate solution to the problem of passenger comfort at
high speeds, and turned its liabilities into assets. Articulation has important
advantages, such as a very low natural frequency for the suspension, the accommodation
of large air reservoirs between cars, and the reduction of parasitic vibration through
the elimination of dampers that mechanically link the bogie to the car body.
Finally, the SR10 suspension is a breakthrough that came at the end of a long
research program and allows SNCF to fit its TGV trains with a suspension worthy of
the speed, safety and comfort that customers can expect.
End RGCF article
Last Update: April 2000