Available online at www.sciencedirect.

com

ScienceDirect
Procedia - Social and Behavioral Sciences 87 (2013) 222 233

SIDT Scientific Seminar 2012

Models and methods to optimise train speed profiles with and

without energy recovery systems: a suburban test case
Valerio De Martinisa,*, Mariano Gallob
a

Federico II University of Naples, Department of Civil, Architectural and Environmental Engineering, via Claudio 21, Naples 80125, Italy
b
University of Sannio, Department of Engineering, piazza Roma 21, Benevento 82100, Italy

Abstract
This paper proposes models and methods to optimise train speed profiles in suburban areas in order to minimise total energy
consumption. In order to achieve this target, an optimisation framework is developed for designing driving strategies, which
requires a railway simulation model as a subroutine. In this paper we focus on the train optimisation model to define energyefficient speed profiles and the relative energy consumption, assuming train motion parameters as control variables. We
consider two cases: with and without energy recovery systems. Proposed models and methods are tested on a real suburban
case. Initial results show that significant energy savings may be obtained.
2013
Published
by Elsevier
Ltd. Open

2013The
TheAuthors.
Authors.
Published
by Elsevier
Ltd.access under CC BY-NC-ND license.
Selection and
under
responsibility
of SIDT2012
ScientificScientific
Committee.
Selection
andpeer-review
peer-review
under
responsibility
of SIDT2012
Committee.
Keywords: energy-efficient drive; speed profiles; energy recovery systems

1. Introduction
Developing energy-efficient transport systems is a major challenge for society in order to improve sustainable
mobility mainly in urban and metropolitan areas. The total energy spent on transportation needs amounts to over
33% of the entire world energy production, with the percentage accounted for by fossil fuels being around 85%.
In this scenario, rail systems represent one of the most interesting solutions for satisfying high travel demand
with low energy consumption: railways may have a significantly lower cost per pass-km than road systems (for
instance in the case of the same degree of occupancy) and allow high capacities to be reached without significant
loss of service quality. Indeed, although in the case of a rail system operational costs are about five times higher
with respect to road systems, vehicle capacity may be as much as ten times higher. Moreover, energy

* Corresponding author. Tel.: +39 081 7683770.

E-mail address: vdemartinis@unina.it

1877-0428 2013 The Authors. Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
Selection and peer-review under responsibility of SIDT2012 Scientific Committee.
doi:10.1016/j.sbspro.2013.10.606

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233

consumption and consequent greenhouse gases and air pollutant emissions of rail systems are also lower in
comparison to other transportation systems.
The literature on railway energy efficiency contains several contributions from different fields, in keeping
with its multidisciplinary nature, that propose different approaches to optimise energy consumption. Albrecht
(2010) and Hansen and Pachl (2008) have recognised the important impact of train operation on energy
consumption; many improvements in operation performance can be obtained both by means of driving assistance
systems (DAS) and communication technologies (DAriano & Albrecht, 2006; Beugin & Marais, 2012) and by
applying specific strategies and approaches to optimise the railway system (Bavafa-Toosi et al., 2008;
Bocharnikov et al., 2007; Liu & Golovitcher, 2003). In this respect, many studies have considered automatic train
operation (ATO) systems, especially in urban and suburban contexts, by defining optimal speed profiles in terms
of running time or energy spent (Dominguez et al., 2012). In other words, predefined speed profiles are uploaded
into the system and, according to the departure time and minimum running time required, the system selects the
speed profile that best fits the strategy requirements (Miyatake & Ko, 2010). Significant findings on specific
railway systems were reported, for example, by Lukaszewicz (2004) on freight train operations, by Ke and Chen
(2004) on the planning of mass rapid transit systems, and by Gu et al. (2011) on moving block signalling
systems.
Moreover, significant improvements on energy recovery from electric vehicles during braking regimes have
been developed in the last decade, insofar as energy consumption minimisation may be considered a balance
between the energy consumed and the energy recovered. Timetable rescheduling procedures are often used to
reallocate train departures or acceleration regimes to coincide with the braking regime of other trains within the
same powered section (Pena-Alcaraz et al., 2012; Albrecht, 2004) in order to maximise the use of recovered
energy which would otherwise be dissipated in heating resistors. The use of supercapacitors as storage devices on
board or at substations is one of the most promising solutions, since they can reduce the power peaks and
stabilise the voltage along the line, according to the planned timetable and the levels of service required
(Malavasi et al., 2011; Gunselmann, 2005). Feeding back the regenerative energy to storage devices at
substations leads to transmission losses. These losses can be avoided by placing the device on board the vehicles;
in this case, the train mass is increased and some space is consumed on the vehicle (Frasca et al., 2012; Steiner et
al., 2007). The trade-off between efficient speed profiles and regenerative energy maximisation is a complex
problem because higher speeds consume more energy and, at the same time, allow higher regeneration
capabilities. This problem can be solved first by optimising the speed profiles of each train and, secondly,
increasing network receptivity (Bocharnikov et al., 2010).
As a consequence, thanks to the use of optimisation procedures, railway simulation models and algorithms
(Quaglietta et al., 2011) have been developed to estimate the effects of the proposed speed profiles on quality of
service and travel demand costs (DAcierno et al. 2012; Gallo et al., 2011) where, following a what if approach,
the best solution is found by simulating different scenarios and choosing the one which best meets the proposed
requirements. Finally, recently, Corapi et al. (2013) and De Martinis et al. (2013) proposed to adopt a
microscopic approach for analysing effects of different drive strategies in terms of energy consumption.
In this paper we propose a general optimisation framework for defining the energy-efficient speed profiles for
urban and suburban lines. The paper is organised as follows. In Section 2, the problem is defined and analysed in
terms of energy-efficient strategies and energy consumption models. In Section 3 the optimisation model for
designing the speed profiles is formulated. In Section 4 numerical results on a real-scale case are reported.
Section 5 concludes the paper and identifies research directions.
2. Problem definition
Railway systems are widely studied both in the academic and industrial sectors, and are recognised as a
multidisciplinary field, involving aspects of mechanical and electrical engineering, information technology and

223

224

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233

transportation engineering. As regards energy efficiency, the literature classifies energy efficiency strategies in
two groups: energy saving strategies and energy recovery strategies.
Energy saving strategies seek to reduce energy consumption by optimising train speed profile parameters,
such as speed, acceleration, deceleration, and the coasting phase, and/or by proposing less aggressive driving
behaviour; one of the parameters that may measure driving behaviour are jerks (derivative of acceleration).
Energy recovery strategies consider the use of regenerative braking systems on electric vehicles, usually
coupled with energy storage equipments with specific characteristics on storage capacity and energy return speed,
as supercapacitors, that can be provided either on board or at electric reversible substations. The latter solution
allows the stored energy to be used also by other vehicles.
Albeit for different reasons, both these strategies require extra travel time to be available on the rail section
(recovery time) for their implementation; the first uses the recovery time for the introduction of the coasting
regime, while the second uses the recovery time to better synchronise the braking and acceleration phases of
different vehicles and to avoid voltage peaks caused by two or more vehicles in the same regime (acceleration or
deceleration).
In the literature, recovery times are usually considered as extra time vis--vis the minimum travel time
available for implementing energy efficiency strategies (Hansen & Pachl, 2008). Recovery times can be
classified in time for small delays (running time reserve, dwell time reserve) and time for avoiding delay
propagation (buffer time). In this paper we refer only to running time reserve, which is commonly assumed to be
about 3-8% of the minimum running time between two stops/stations.
2.1. The energy consumption model
Energy consumption for train operations is mainly comprises two elements: the energy required by traction
units and the energy required for auxiliary systems. For our purposes only traction units are considered in the
following.
Moreover, the energy consumption model with recovery systems usually implies an energy consumption
model of the entire line. The purpose of this paper is to optimise speed profiles for train operations in order to
minimise the energy balance, respecting the level of service of the line. This means that the single train operation
is constrained to its timetable, in the case of existing services, and to the forecasted travel demand and headways,
in the case of planned services. In this paper we refer to single train energy consumption, assuming line
constraints as external constraints to the optimisation problem and adopting a microsimulation approach.
The energy required for train running along a given track with given motion parameters can be expressed as
the integral of the corresponding mechanical power over time. The mechanical power is the power measured at
the wheel-rail interface and can be computed as the product of the tractive effort F and speed V:
E=

Pmech (t )dt = V F (V , t )dt

tT

(1)

tT

where the tractive effort F is defined in T, that is travel time on the track (or part of it), and can be computed by
solving the differential equation derived from Newtons theory, also known as the general motion equation, with
the finite difference method. Given a generic time step i of 1 second, we may arrive at the following expression:
F (Vi ) = M f p

V
+ R(Vi , S i )
(t i +1 t i )

(2)

where M is the train mass, fp is the mass factor of the rotating parts, R(Vi, Si) is the sum of vehicle resistances that
depend on speed (Vi) and line resistances that depend on slopes, curves and tunnels (Si).

225

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233

The energy computation model can be thus defined with a discrete approach for energy saving and energy
recovery strategies. As regards energy saving, the single train energy consumption can be computed as follows:
E=

Vi Fi

(3)

(Vi , t i )

i =1...T

The energy consumption refers to the positive values of the efforts applied at the wheels, i.e. tractive efforts
during acceleration and cruising.
In the case of energy recovery strategies, train energy consumption can be expressed as the sum of the whole
efforts applied at the wheels, as follows:
E=

(4)

Vi Fi + (Vi , ti ) + Vi Fi EL (Vi , ti )

i =1...T

i =1...T

where the Fi EL term is negative and indicates the part of braking effort generated by the electric braking
system. In general, the total braking effort during braking is partially supported by the electric engine until a low
speed is reached (generally 10-15 km/h). For lower speeds only the mechanical brake is activated.
In Figure 1, brief data on the trends of tractive (T) and braking (Btot) efforts are represented, in terms of kN per
tons of train mass, together with electrical braking (EB), which is part of the whole braking effort, without
considering the efficiency of electrical and mechanical parts.

Fig. 1. Tractive and braking efforts (courtesy of AnsaldoBreda)

3. Optimisation models
The proposed optimisation models are formulated considering motion parameters as control variables for
energy consumption and assuming the availability of recovery times suitable for implementing energy efficiency
strategies. The optimisation model for solving the energy saving problem is formulated as follows:

[a ,V
*

*
max

*
,d * ,TiC* ,T fC
= arg min E a ,Vmax ,d ,TiC ,T fC
a ,Vmax ,d ,TiC ,T fC

(5)

226

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233

subject to the following constraints:

J 1 s < a amax

(6)

Vmin < Vmax Vallow

J 1 s < d d max

(7)
(8)

*
TiC* < T fC

(9)

( ( )) T

*
fC

T + Tdec V T

*
fC

max

S acc + Scruise + S coast + S dec = Dist

where:
a
Vmax
d
TiC
TfC
E(.)
Vmin
Vallow
J 1s
amax
dmax
Tdec
Tmax
Sacc
Scruise
Scoast
Sdec
Dist

(10)
(11)

is the target acceleration (a* is its optimal value);

is the target speed (V*max is its optimal value);
is the target deceleration (d* is its optimal value);
is the starting time of coasting (T*iC is its optimal value);
is the ending time of coasting (T*fC is its optimal value);
is the total mechanical energy spent;
is the minimum value of the target speed that respects the scheduled arrival time, without coasting;
is the maximum speed on the section allowed by speed limits;
is the acceleration at 1 second, obtained by multiplying the jerk value by 1 second;
is the maximum acceleration compatible with passenger comfort;
is the maximum deceleration compatible with passenger comfort;
is the time needed to decelerate from a certain speed;
is the maximum travel time compatible with respect for the timetable (it is the sum of the minimum
running time and the running time reserve);
is the space covered during acceleration;
is the space covered during cruising;
is the space covered during coasting;
is the space covered during deceleration;
is the total distance to be covered.

The optimisation model for solving the energy recovery problem is formulated as follows:

[ , a ,V
*

*
max

, d * = arg min E (, a ,Vmax , d )

(12)

, a ,Vmax ,d

subject to constraints (6)-(8), (11) without the Scoast and to the following constraint:

0 TRTR
+ TER Tmax

(13)
(14)

where:

is the additional time spent by the train at the station before running and after closing the doors, waiting
for availability of energy recovered from other trains that are in the braking phase;
TRTR
is the running time reserve for the specific section.
is the running time in energy recovery driving, which includes the running time reserve not used by .
TER

227

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233

The optimisation model, specified for the single strategy adopted, allows for respect of the scheduled running
time as a constraint vis--vis the level of service for the single line.
Under constraints (6) and (8) passenger comfort and the variation in acceleration needed to reach the target
acceleration can be respected. Under constraint (7) the maximum allowable speed on the track given by track
limits can be respected; it does not consider speeds that are lower than the minimum speed for arriving at the
scheduled running time with the maximum acceleration and without coasting. Constraint (9) is an operative need
for the algorithm. Under constraints (10), (13) and (14) the scheduled running time when the running time reserve
is used can be respected. Under constraint (11) the travel distance to be covered can be respected.
Regarding the specific strategy, energy saving will use the extra time availability for the coasting phase, such
that the optimised speed profile has to respect the scheduled running time and the distance to cover, i.e.
constraints (9), (10) and (11). Energy recovery, instead, will use a classic speed profile, and will be optimised on
its acceleration and braking phases, while the travel time will start with the energy recovery availability with
respect to the scheduled running time and the distance to cover, i.e. constraints (13), (14) and (11).
3.1. Energy-efficient speed profile
The energy-efficient speed profile for a single train can be defined by eqn. (2). More specifically, we can
assume that vehicle resistances can be computed as follows:

RV (Vi ) = K1 M + K 2 M Vi + K3 Vi 2

(15)

where Ki are parameters to calibrate, M is the mass in tons of the single vehicle and Vi is the speed at time step i.
The line resistances depend on train position at time step i (Si) and can be computed with the formula of Roeckl
(16), as regards the resistances due to curves, and with the weight force component (17), for the resistances due
to the slopes:

6.3

Rr (Vi , Si ) = V 55 M
i

4.91
Rr (Vi , Si ) =
M
Vi 30

Ri (Si ) = M g

r 300 m
(16)

r < 300 m

i
1000

(17)

Finally, R(Vi, Si) can be defined as the sum of (15), (16) and (17). In other words:

R(Vi , Si ) = RV (Vi ) + Rr (Vi , Si ) + Ri (Si )

(18)

Acceleration can be computed with the following formula:

V
= ai = ai 1 J (t i t i 1 )
(t i +1 t i )

(19)

a + J (t i ti 1 )
ai = i 1
ai 1 J (ti t i 1 )

(20)

where:
(a)
(b)

228

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233

is the acceleration value during the transitional phases (approach to next motion phase), considering both the
approach to the target acceleration (a) and the approach to the target speed (b). The same conditions can be
assumed for the deceleration values. The jerk value, J, represents the variation in acceleration during the
acceleration phase, and can be optimised like the other speed profile parameters. However, in this paper we did
not consider it as a decision variable since it depends on driver behaviour. In future research this assumption will
be removed. Optimisation of this parameter usually makes sense only in the case of fully automated systems
(driverless).
The model for optimising speed profiles produces energy-efficient results; for energy saving strategies, it
defines the starting and ending points of the coasting phase, TiC and TfC. For a given coasting strategy, the speed
profile optimisation model verifies the consistency of the profile in terms of travel time available on a given track
and the distance covered, i.e. constraints (9), (10) and (11), using the motion parameters generated by the
optimisation algorithm. In practice, the starting and ending points of the coasting regime are defined a priori by a
coasting strategy; the driver has a planned coasting regime at a given track point. In this paper we use the strategy
ASAP (As Soon As Possible), which means that the driver starts coasting as soon as he/she can. This strategy
presupposes a driving assistance system.

Fig. 2. Optimisation framework with the railway simulation subroutine

In Figure 2 the proposed optimisation framework is reported. Given a set of target parameters of motion (a,
Vmax, d), the model for defining the speed profiles calculates the relative speed profile at each one-second time
step. The starting time of the coasting phase, TiC, is sought at each step with a parallel algorithm that runs
equation (21), where tractive effort F(Vi) is not applied and the variation of speed and the related resistances at
each step have to be computed. In other terms:

R (Vi , Si )
V
=
M fp
(ti+1 ti )

(21)

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233

and the speed profile with the coasting phase is accepted if the following two conditions, given by constraints
(10) and (11), are respected:
1.

T + Tdec(d,V(t)) = Tmax

2.

Space covered at time Tmax = space to be covered

These conditions mean that the whole running time reserve has to be used. The first condition requires
compliance with the maximum time available, Tmax, which must be achieved by allowing for the fact that at time t
we must add the time Tdec(d,V(t)) required to brake from speed V at time t with a deceleration d. The second
condition requires that the whole distance be covered.

Fig. 3. Speed profile definition model

In Figure 3 the speed profile definition model is represented. Though already sufficient to calculate the energy
consumed, the model does not contemplate the randomness of events on the railway network, such as interaction
between vehicles. Therefore, from this point of view, its use could be evaluated with the presence of driver
assistance systems, driverless trains or simple networks such as urban and suburban lines.
4. Numerical results
The proposed models were tested on a real-scale case, namely a rail track in Naples. In the coming months,
this test case will be monitored and will allow calibration of the simulation model. The test case is a double track

229

230

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233

of about 1800 m with two stations, respectively at the beginning and end of the track with no signalling systems.
The track is at ground level, and there are no slopes or curves. Given the characteristics of the track, this
preliminary test can be viewed as similar to a generic station-to-station urban line.
For our purposes two different scenarios were considered. The first scenario assumes the entire length of the
track as a station-to-station section, while the second considers a hypothetical third station in the middle of the
track, hence two sections of about 900 m.

Fig. 4. First scenario speed profile and total energy consumption

In Figure 4 speed profiles and the related total energy consumption are reported. The speed profile TO refers
to a time-optimal strategy to minimise travel time, with a total running time of 114 seconds which is the
minimum running time for this section. The recovery time was computed by allowing for an additional 15% of
the minimum running time to spend on energy-efficient strategies. The energy recovery (ER) speed profile
considers a required time of 5 seconds for energy recovery use from other trains while the other 10 seconds are
used for speed profile optimisation, with due consideration of the optimal energy balance. The energy saving
(ES) speed profile uses the whole recovery time to optimise energy consumption, thus considering consumption
only during acceleration and cruising regimes. An additional strategy is considered, assuming possible energy
recovery also in the case of energy saving speed profile optimisation (ES+ER). In this case it is possible to
consider an energy balance as in energy recovery strategies. Numerical results are summarised in Table 1, values
with (*) refer to the additional strategy (ES+ER).
In Figure 5 the time-optimal (TO) speed profile is reported; it allows the single section length to be covered in
70 seconds (minimum travel time), and also in this case an additional recovery time of 15% was considered for
energy-efficient speed profile optimisation. The same considerations as the first scenario regarding ER, ES and
ES+ER results can be made for this second scenario. In Table 2 the numerical results are summarised while
Figure 6 reports the space-time diagram of the train on the whole track length.
Table 1. First scenario numerical results

TO
ER
ES

J
[m/s3]
0.25
0.3
0.18

a
[m/s2]
1
0.9
0.9

V
[m/s]
20
18.4
19.1

D
[m/s2]
1
0.8
0.9

[s]

Tic
[s]

Tfc
[s]

31

112

Ttot
[s]
114
125
129

resTime
[s]
10
15

E spent
[kWh]
14.12
11.85
9.37

E rec
[kWh]
5.46
4.20
2.41(*)

E rec
[%]
38.70
35.47
25.77(*)

231

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233

Fig. 5. Second scenario speed profile and total energy consumption (1 section)
Table 2. Second scenario numerical results (1 section)

TO
ER
ES

J
[m/s3]
0.2
0.2
0.2

a
[m/s2]
1
0.8
0.8

V
[m/s]
20
18
15.4

D
[m/s2]
1
0.8
0.8

[s]

Tic
[s]

Tfc
[s]

22

64

Ttot
[s]
70
82
82

resTime
[s]
7
12

E spent
[kWh]
11.64
9.34
5.71

E rec
[kWh]
5.64
4.46
2.13 (*)

E rec
[%]
48.47
47.75
37.46 (*)

Fig. 6. Second scenario space-time diagram and total energy consumption (whole track)

5. Conclusion and research prospects

This paper proposed two optimisation models for minimising energy consumption by defining optimal speed
profiles; one model concerns energy saving strategies and the other energy recovery strategies. For both models a
solution procedure was proposed and tested on a real-scale case involving a suburban line. Initial results showed

232

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233

a promising reduction in energy consumption with the optimal energy-efficient speed profile for a given running
time reserve. Even if the energy balance supports energy recovery strategies, the technology in question results in
extra expenditure in terms of both investment and operations, which should be computed with reference to the
specific cases. Finally, numerical results show the applicability of the proposed procedures and the benefits in
terms of energy consumption.
Further research will be devoted to calibrating the model on real data. On this aspect, the rail track used in the
tests will be monitored in terms of motion parameters and energy consumption with and without energy recovery
systems.
Acknowledgements
Partially supported under research project PON - SFERE grant no. PON01_00595.
References
Albrecht, T. (2004). Reducing power peaks and energy consumption in rail transit systems by simultaneous train running time control.
Advances in Transport, 15, 885 - 894.
Albrecht, T. (2010). Dealing with operational constraints in energy efficient driving. IET Conference on Railway Traction Systems (RTS
2010), 22 - 22.
Bavafa-Toosi, Y., Blendinger, C., Mehrmann, V., Steinbrecher, A., & Unger, R. (2008). A new methodology for modeling, analysis,
synthesis, and simulation of time-optimal train traffic in large networks, IEEE Transaction on Automation Science and Engineering, 5, 43 52.
Beugin, J., & Marais, J. (2012). Simulation-based evaluation of dependability and safety properties of satellite technologies for railway
localization. Transportation Research Part C, 22, 42 - 57.
Bocharnikov, Y.V., Tobias, A.M., & Roberts, C. (2010). Reduction of train and net energy consumption using genetic algorithms for
trajectory optimization. IET Conference on Railway Traction Systems (RTS 2010), Birmingham, U.K., 1 5.
Bocharnikov, Y.V., Tobias, A.M., Roberts, C., Hillmansen, S., & Goodman, C.J. (2007). Optimal driving strategy for traction energy saving
on DC suburban railways, IET Electric Power Application, 1, 5, 675 - 675.
Corapi, G., Sanzari, D., De Martinis, V., DAcierno, L., & Montella, B. (2013). A simulation-based approach for evaluating train operating
costs under different signalling systems. WIT Transactions on the Built Environment, 130, 149 - 161.
DAcierno, L., Gallo, M., Montella, B. & Placido, A. (2012). Analysis of the interaction between travel demand and rail capacity constraints.
WIT Transactions on the Built Environment, 128, 197207.
DAriano, A. & Albrecht, T. (2006). Running time re-optimization during real-time timetable perturbations. WIT Transactions on the Built
Environment, 88, 531 - 540.
De Martinis, V., Gallo, M., & DAcierno, L. (2013). Estimating the benefits of energy-efficient train driving strategies: a model calibration
with real data. WIT Transactions on the Built Environment, 130, 201 - 211.
Domnguez, M., Fernndez-Cardador, A., Cucala, A.P., & Pecharromn, R.R. (2012). Energy Savings in Metropolitan Railway Substations
Through Regenerative Energy Recovery and Optimal Design of ATO Speed Profiles. IEEE Transactions on Automation Science and
Engineering, 9, 496 - 504.
Frasca, R., Andreozzi, F., Cascone, B., Della Valle, D., Fratelli, L., Giannini, G., Improta, V., Nappo, C., & Porzio, M. (2012).
Supercapacitors system for balancing power-supply gap in railway electric vehicles. International Symposium on Power Electronics,
Electrical Drives, Automation and Motion (SPEEDAM), 754-758.
Gallo, M., Montella, B., & DAcierno, L. (2011). The transit network design problem with elastic demand and internalisation of external
costs: An application to rail frequency optimisation. Transportation Research Part C, 19, 1276 - 1305.
Gu, Q., Lu, X.-Y., & Tang, T. (2011). Energy saving for automatic train control in moving block signalling system. Proceedings of the 14th
IEEE Conference on Intelligent Transportation Systems (ITSC), 1305 - 1310.
Gunselmann, W. (2005). Technologies for increased energy efficiency in railway systems. European Conference on Power Electronics and
Applications, 1 - 10.
Hansen, I.A., & Pachl, J. (2008). Railway timetable and traffic: analysis, modelling, simulation, Eurailpress: Hamburg, Germany, 2008.
Ke, B.R., & Chen, N. (2005). Signalling block layout and strategy of train operation for saving energy in mass rapid transit systems. IEEE
Proceedings: Electric Power Applications, 152, 129 -140.
Liu, R., & Golovitcher, I.N. (2003). Energy-efficient operation of rail vehicles. Transportation Research Part A, 37, 917 - 932.

Valerio De Martinis and Mariano Gallo / Procedia - Social and Behavioral Sciences 87 (2013) 222 233
Lukaszewicz, P. (2004). Energy saving driving methods for freight trains. Advances in Transport, 15, 885894.
Malavasi, G., Palleschi, P., & Ricci, S. (2011). Driving and operation strategies for traction-energy saving in mass rapid transit systems.
Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit, 225 (5), 475 - 482.
Miyatake, M., & Ko, H. (2010). Optimization of train speed profile for minimum energy consumption. IEEJ Transactions on Electrical and
Electronic Engineering, 5 (3), 263 - 269.
Pena-Alcaraz, M., Fernndez, A., Cucala, P., Ramos, A., & Pecharromn R.R. (2012). Optimal underground timetable design based on power
flow for maximizing the use of regenerative-braking energy. Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail
and Rapid Transit, 226 (4), 397-408.
Quaglietta, E., DAcierno, L., Punzo, V., Nardone, R., & Mazzocca, N. (2011). A simulation framework for supporting design and real-time
decisional phases in railway systems. 14th International IEEE Conference on Intelligent Transportation Systems (ITSC), art. no. 6082913,
846851.
Steiner, D.M., Klohr, M., & Pagiela, S. (2007). Energy storage system with ultracaps on board of railway vehicles. 2007 European
Conference on Power Electronics and Applications, 1-10.

233