Towards railway virtual coupling

Towards Railway Virtual Coupling
Francesco Flammini1, Stefano Marrone2, Roberto Nardone3, Alberto Petrillo3 , Stefania Santini3 and Valeria Vittorini3
1 Department of Computer Science and Media Technology, Linnaeus Univeristy, Växjö, Sweden, francesco.flammini@lnu.se
2Department of Mathematics and Physics, University of Campania Luigi Vanvitelli, Caserta, Italy, stefano.marrone@unicampania.it
3Department of Electrical Engineering and Information Technology, University of Naples Federico II, Napoli, Italy
{Roberto.Nardone,alberto.petrillo,stsantin,valeria.vittorini}@unina.it
7th-9th November 2018, Nottingham, United Kingdom

Cooperative Driving in Railways
• Cooperative driving is an essential component of future intelligent transportation
systems. It promises many advantages for ground transportation such as increased
safety (i.e. reduced accidents due to less drivers distraction) and improved
infrastructure utilization (i.e. increase of road capacity due to less traffic congestion).
• Improving infrastructure utilization and increasing line capacity are currently two
primary objectives for railways, addressed by the Shift2Rail.
• One of the proposed solutions by Shift2Rail to increase line capacity is the Virtual
Coupling that allows trains to virtually join by considerably reducing their headways.
• The idea is to transfer to the railways sector some
achievements and results of the automotive research
in vehicles platoon.

Railway Virtual Coupling
• Virtual Coupling adds to Automatic Train Control (ATC) systems the further functionality
of being able to virtually connect two or more trains, so drastically reducing their
headways and increasing line capacity.
• To reach the coupling, each train within the platoon exploits train-to-train
communication so to exchange information with neighbours and to receive the
reference signals coming from infrastructure.
• On the basis of the shared information the ATC is responsible of:
1. the safe tracking of a desired speed profile;
2. the maintenance of a desired inter-train spacing policy.

The problem of increasing the line capacity
• ERTMS/ETCS (European Railway Traffic Management System/European Train Control
System), the set of standards for management and interoperability of modern railways,
identifies three functional level of growing complexity.
• Nowadays, Level 2 is the most successful on high-speed lines and is based on the usage of
GSM-R (Global System for Mobile Communications Railway) for train-to-ground
communication and the use of fixed-block signalling with trackside train location
equipment.
• A fixed block is a portion of the trackside between two fixed points. The safety distance
among trains, computed by Radio Block Center (RBC) on the basis of the occupation
status of the portion of the track, is hence fixed and not adaptable to the train features.
• Fixed block strongly influences the capacity of the railway line.

How increasing line capacity?
• To increase line capacity we can exploit ERTMS Level 3 that introduces the concept of the
Moving Block signalling.
• In this context, each train reports its position and speed to RBC and on the basis these
information the RBC computes the safety distance among trains.
• To avoid collisions, RBC implements a supervision system and protects trains against
overrun of the safety distance.
• Since safety distance is not fixed, but depending on the actual train position and speed,
the capacity of the railway line increases.

Why Virtual Coupling?
• To maximize the probability of messages to be delivered, and hence the safety of trains,
the idea is to exploit the virtual coupling concept that introduces the train-to-train
communication paradigm.
• ERTMS Level 3 with moving blocks provide a management system that is fully
infrastructure based.
• In case of lack of information about train position:
1. the Level 3 could not work due to unsafe conditions;
2. the RBC will keep the last position report sent by the train.
• In case of track-to-train communication issues, a train could receive
information about its movement authorities from the preceding (and/or
the following) train.

Work Aims
• The aim of this work is:
i. to introduce Virtual Coupling in the context of ERTMS/ETCS (European Railway
Traffic Management System/ European Train Control System);
ii. to provide some preliminary hints, model and results disclosing its effectiveness in
augmenting railways line capacity.
• Contributions:
i. Providing a vision about the possibility to introduce Virtual Coupling extending the
current ERTMS (Level 3) standard, maintaining the backward compatibility with
existing infrastructures and on-board equipment.
ii. Proposing a simple stochastic model, as well as a preliminary numerical
simulation, to estimate the effects of introducing Virtual Coupling

Virtual Coupling concept: Our Vision (1/2)
• In case of train platoons, it is possible to imagine different communication topologies in
order to optimize bandwidth utilization and increase the reliability of message delivery.
• Possible topologies include:
1. Fully connected
2. Chain-like
4 3 2 15
4 3 2 15

Virtual Coupling concept: Our Vision (2/2)
• The RBC evaluates:
i. the safety distance within the train fleet on the basis of trains position;
ii. the common speed at which train fleet need to travel.
• Each train implements a local control algorithm that act on its acceleration so to:
i. track the desired speed profile as imposed by the RBC;
ii. to maintain the desired inter-train spacing policy, as imposed by RBC,
considering the position and speed information of the preceding train.
This guarantees an improved flexibility w.r.t. communication failures

Compatibility with ERTMS standard (1/2)
• Shif2Rail research area asks for “continuity and backward compatibility with the
current signalling and supervision systems through ERTMS standards”.
• Hence it is important to propose a Virtual Coupling concept that is compatible with
standards.
• Currently, ERTMS defines several Operating Modes for the on-board European Vital
Computer (EVC) according to different conditions affecting the status of the trackside
and of the train itself.
• The main operating mode are:
1. the Full Supervision (FS): all train and track data required for complete
supervision of the train are available on board.
2. Partial Supervision (PS): to manage degraded operational situations (e.g.,
Shunting, On-Sight, Staff Responsible, etc.).

Compatibility with ERTMS standard (2/2)
• To provide a concept that is compatible with standards, we propose a new supervision
architecture, where we add a new operating mode to the ones existing.
• It is called Full Supervision plus Virtual Coupling (FSVC) and it is reachable only by FS
operating mode.
• The switch from FS to FSVC is commanded through custom ERTMS “messages” sent by
the RBC that check that all necessary conditions are fulfilled, including:
i. sufficient Movement Authorities overlap among preceding and following train
to ensure the same routing in stations;
ii. successful establishing of train-to-train communication.
FS PS
FSVC
• When the RBC sends the “coupling” command to the trains, the
EVC switches from FS to FSVC; the “decoupling” command will
revert the EVC back in FS through an appropriate procedure.
• In case of train-to-train failures EVC switches from FSVC to FS.
• In case of train-to-track failures EVC switches from FS to PS.

Analysis of ERTMS Virtual Coupling Operating Mode: Capacity Model
• The switches between the ERTMS operating modes, including the additional FSVC, can
be modeled by the Stochastic Petri Net.
• T_FSVC and T_FS are transitions with a
deterministic firing rate as they
represent the frequency of trains
running at FSVC and FS operating point.
• Places P_T and P_R and the transition T
have been added to evaluate the trains
frequency on the track in function of
the sojourn times in FSVC, FS and PS.
• Specifically transition T will provide an
upper bound of the number of trains
that system could manage in the time
unit.
• The firing rate of transitions FS2PS and
FSVC2PS have been set to 1*10-6 which
the maximum failure rate for the RBC.
• FS2FSVC and FSVC2FS have a parametric firing rate. On the basis of the variation of these rates we evaluate
the Throuput of the line.

Capacity Model Simulation
• We simulate the model by considering the Rome-Naples high-speed track in Italy, where
trains run at 300 km/h with a safe distance of 15 km.
• In FS operating mode the maximum ideal train frequency on the track is 3 minutes (
15/300=0.05 h).
• At FSVC, we assume an inter-train distance of 3 km at 300 km/h, which correspond to a
frequency of 36 seconds (3/300 = 0.01 h).
• In our simulation we hence consider the firing rate of transitions T_FS=0.05 and
T_FSVC=0.01.
• Our aim is to evaluate the trains frequency in
an hour, by varying the firing rate of transition
FSVC2FS and FS2FSVC

Capacity Model Simulation: Results (1/2)
• If we had FS only (FS2FSVC is lower since it does not
work well), the T frequency would be 20 trains per
hour, i.e. one train each 3 minutes.
• On the opposite, if FSVC was implemented correctly
with a low rate of faulty conditions, the trains stay in
FSVC most of the time. The frequency would be one
train each 36 seconds, i.e. 100 trains per hour.
• If we had implemented a VC mechanism that is very
reliable with a very low rate for FSVC to FS, but if at
the same time the VC mechanism used to get to the
FSVC operating mode was very slow, we would a
frequency of about 60 trains per hour.
• On the opposite, if we had an efficient yet unstable VC
mechanism, with trains easily getting from FS to FSVC
but also leaving FSVC often due to e.g. communication
faults, we still would obtain a frequency of about 60
trains per hour.

Capacity Model Simulation: Results (2/2)
• For this analysis we fixed the FSVC2FS rate to the
value 0.5 and we varied the FS2FSVC from 0.1 to
0.9.
• By increasing the rate of the transition FS2FSVC,
the probability of being in the FSVC state (purple
bars) increases from 17% up to 64%, while the
the probability of being in the FS state (yellow
bars) decreases coherently.
• We finally plot the probability of the system to be in the FSVC and in the FS modes.

Virtual Coupling: Join Manoeuvre (1/2)
• Consider a fleet of 4 trains in FSVC mode traveling at the constant speed of 210 km/h with a relative
inter-train distance of 100 m when a new train in FS mode, labeled as train 5 and initially located at
500 m from the fourth vehicle of the fleet, performs a join maneuver.
• The fifth train, moving at 205 km/h, sends a request to join the fleet at t = 300 s. After the RBC
sends the coupling command to the train, the EVC of train 5 switches from FS to FSVC mode. In
doing so, the train 5 is now able to communicate with the first and fourth trains within the fleet.

Virtual Coupling: Join Manoeuvre (2/2)
• On the basis of the these shared information, the local control algorithm computes the desired
acceleration profile that has to be set on the fifth train’s dynamics in order to join the existing fleet.
Results show that, under the action of the local control strategy, the fifth train accelerates so to
achieve the required spacing policy of 100 [m] from train 4, and then it decelerates until it reaches the
constant speed of 210 km/h as set by the first vehicle of the fleet.

Conclusions and Future Work
• This work have addressed the problem of increasing the railway line capacity by proposing
the Virtual Coupling concept in the context of ERTMS/ETCS standard.
• We have provided a vision about the possibility to introduce Virtual Coupling extending
the current ERTMS (Level 3) standard, maintaining the backward compatibility with
existing infrastructures and on-board equipment.
• A preliminary analysis has been carried out by using stochastic Petri net to jointly model
performance and dependability (i.e. performability) of Virtual Coupling in normal as well
as in degraded modes of operation.
• We have performed a preliminary simulation Virtual Coupling control algorithm by
considering a fleet of 5 trains, where the fifth performs a join manoeuvre.
• As future work:
1. we will extend our performances and dependability analyses by considering also
the occurrence of more faulty conditions, e.g. communication time-varying delays;
2. we will address further analysis of coupling control algorithms by simulating new
scenarios accounting for topology variations and time-varying delays affecting the
Train-to-Train and Infrastructure-to-Train communication links.

Thank you
alberto.petrillo@unina.it

Towards railway virtual coupling

More Related Content

What's hot (20)

Similar to Towards railway virtual coupling (20)

More from Francesco Flammini (20)

Recently uploaded (20)

Towards railway virtual coupling