PROBLEMY KOLEJNICTWA

RAILWAY REPORTS Zeszyt 180 (September 2018) ISSN 0552-2145 (print) ISSN 2544-9451 (on-line) 133

Optimization of the Efficiency of Braking Energy Recovery

in Rail Transport by Changing Arrival Time
Michał URBANIAK1, Ewa KARDAS-CINAL2

Summary
The article refers to the previous work of the authors, in which the model of traffic organization of cooperating trains in-
cluding the optimization of the use of energy returned to the catenary was presented. In the presented article, the model
was modified by changing the main control variable, which affects the efficient use of energy. Departure time was changed
for the arrival time of the train to the stop or station. The optimization is done by controlling the arrival time to the station
in the acceptable (scheduled) range while maintaining the scheduled departure time. This model assumed optimization
using the interval halving method (bisection) to achieve the optimal solution. The modified optimization method has been
implemented in the original model of railway traffic organization. It considers the optimal use of energy recovered during
electrodynamic braking using the energy transmission strategy to the catenary, assuming the cooperation of a train pair
and volume of all recovered energy and stop time at the station.

Keywords: rail transport, regenerative braking, optimization of energy recuperation, traffic organization

1. Introduction  the ability to use it directly in trains for non-trac-

tion needs, such as lighting, air conditioning, etc.,
The need to optimize the effectiveness of regenera-  storing it in stationary or onboard energy storage
tive braking in rail transportation stems from existing devices, and then using at time of increased de-
transportation problems such as: the need to reduce mand [3, 14],
transportation costs while maintaining acceptable  transmission of recovered energy back to the na-
standards, reduce pollutant emissions while main- tional power grid [1, 13, 25],
taining required capacity and increasing demand for  the transmission of recovered energy back to the
transportation services. In addition to the rationaliza- electricity transmission infrastructure, allowing
tion of running trains [12], it is necessary to introduce the possibility of its immediate use by another ve-
modern and ecological technologies as described in hicle in the acceleration phase [10, 15, 16] or pro-
documents published by competent national institu- viding acceptable voltage level.
tions and the European Union [19].
One of the ways to reduce the energy consumption Each of the above methods has advantages and dis-
of rail transportation, and therefore lower its costs advantages [8, 23]. The direct use of recovered energy,
(including environmental ones), is the use of regen- notwithstanding AC to DC conversion costs (the situ-
erative technology to recover some of the electricity ation presented in the Polish railway power system),
during electrodynamic braking. The energy recovered does not require incurring additional infrastructure
in this way can be re-used and thus help improve the modernization costs. The use of energy from regen-
energy balance of not only a single run, but also the erative breaking for non-traction purposes of the
entire rail transport system. vehicle or transfer of recovered energy back to the
Among the ways of using electricity derived from catenary, assuming the energy cooperation of several
regenerative braking, the following methods are dis- vehicles, can be called potentially cost-free methods.
tinguished [8, 23, 24]: The method of using energy for non-traction needs of

1
Ph.D. student; Warsaw University of Technology, Faculty of Transport; assistant, University of Technology Faculty of Civil and Envi-
ronmental Engineering; e-mail: [email protected].
2
Prof., Ph.D., D.Sc. Eng.; Warsaw University of Technology, Faculty of Transport.
 134 Urbaniak M., Kardas-Cinal E.

the subject train does not require any additional inter- a pair of vehicles, and is mainly possible within sta-
ference in the technical or organizational process it- tions and stops, where decelerating and accelerating
self. Transfer of recovered energy back to the catenary trains are the most frequent processes [9, 10]. The
for use by other trains depends on the coordination ideal situation would be if the acceleration and decel-
and train schedules and traffic. eration of a pair of vehicles commuting in opposite
directions took place at each station or stop of a given
railway line in a short period of time (Fig. 2). Unfor-
tunately, in practice this assumption is very demand-
2. Model of energy regeneration by train ing and can only be introduced in perfectly function-
cooperation ing (with minimal delays) metro lines. Therefore, this
example of a railway line analyzed here assumes the
The model described below assumes that by cre- possibility of cooperating only at a few stations and
ating an applicable timetable, it will be possible to stops.
transfer energy from regeneration of the braking train Although the basics of rail vehicle driving mod-
(train B) via the catenary to another vehicle leaving eling have been widely described in the literature by
the station (train A). Thus, it will reduce the demand Podoski et al. [17, 18], nevertheless the computational
for the energy in the acceleration phase transferred model proposed there is constantly being modified
from traction substation (Fig. 1). The amount of en- and refined by various scientists, depending on the
ergy recovered and used in this way will depend on goal that needs to be achieved. The model of the theo-
the time of train departure in the range allowed by the retical run itself largely comes down to the solution
detailed timetable. and development of the rail vehicle move equation
(Newton’s equation), where a train is treated as a ma-
terial point with mass m. It can be formulated as fol-
lows [20, 21, 22, 24]:

 dv
km dt  u  t   Rs  v   R g (x ),
 (1)
 dx  v(t ),
 dt

where force u(t) acting on the vehicle is traction force

u = F(t) or braking force u = –F(t), depending on the
Downloaded from mostwiedzy.pl

movement phase and k takes into account the mo-

ments of inertia of rotating masses. The purpose of
these calculations is to determine the parameters of
a moving train (e.g., the volume drawn from the cat-
Fig. 1. The use of recovered energy by using control of train ar- enary, power demand needed to cover the route, etc.),
rival time Source: own study based on [20, 21, 24] depending on time or distance traveled at the given
traction characteristic of the vehicle and with known
route geometric parameters.
The recovery and use of energy from electrody- Resistance of movement Rs(v) is mainly dependent
namic braking requires the cooperation of at least on the aerodynamic forces and wheel – rail interac-

Fig. 2. Scheme of perfect

coordination on suburban
railway lines or on the subway
between braking and starting
trains; Source: own study
based on [10]
 Optimization of the Efficiency of Braking Energy Recovery in Rail Transport by Changing Arrival Time 135

tions. Their dependence on the speed is usually de- In the presented optimization model [20, 21, 24],
scribed by a quadratic function [7, 20, 21]: only the actual energy consumption during the jour-
ney is minimized, what takes the form of:
Rs  v   k0  k1v  k2 v 2 , (2)
Ep = E – Er, (4)
where coefficients k0, k1, k2 are constants related to the
construction of the rolling stock: with mass and param- where E = Ea + Eb is the sum of the traction energy
eters describing the interaction of the wheel with the rail. consumed by vehicles A and B, and Er is the part of
The movement resistance Rg(x) = mgp(x) depends the energy recovered during braking of the vehicle B
on the inclination p(x) of the railway line, which and is expressed by:
changes with the current position of the train x along
the track. This inclination is defined as p = Δh / l T

where Δh is the difference in height between two Er   min{Pa  t  , Pb ,r  t , s }dt . (5)

track points that are separated by l distance, usually 0
expressed in promiles. Then, the equation takes the
form: Rg(x) = mgp(x) / 1000. For each of these vehicles, the energy consumed dur-
In addition, if at time t the power consumed by ing the journey in the time interval [0,T] is expressed by
the vehicles during the acceleration phase equals Pa(t) the integral of the power related to the traction force:
and the power generated during the electrodynamic
braking is indicated as Pb,r(t,s), and velocity of vehicles T T
u t   u t 
A and B are denoted as: va = va(t) and vb = vb(t), the En  max u  t  ,0 v  t  dt 
  v  t  dt (n = a,b).
magnitude of Pa = F(t)va(t) and Pb,r(t,s) = φ(s)B(t)vb(t) 0 0
2
can be determined directly from the equation of mo-
tion [24]: (6)

 1 2 1 2  The specificity of the optimization process takes into

 Pa  t    2 mva (t  t )  2 mva (t )  Eloss , a  / t , account the dependencies between the different depar-
  
 ture times from the station and the overlap of decelerat-
P  t , s    (s) mv 2 (t )  mv 2 (t  t )  E  / t ,
 1 1 ing and accelerating rail vehicles as shown in Fig. 3. In
 b ,r 2 b b loss , b 
 2  the cases considered below, one train (A) is in the ac-
celeration phase and the other train (B) in the braking
(3) phase. The start time and end time of the braking ( t 0b and
t1b ) of the train B are the same in all three cases, while the
where Δt is a short segment of time t, φ(s) is a param- starting time t 0a of the acceleration phase for the train
Downloaded from mostwiedzy.pl

eter with values ranging between 0 and 1, which de- A changes. It can be delayed up to the limit t 0agr (latest
pends on the efficiency of energy transfer depending t 0a ) to allow for a scheduled arrival at the next station.
on the distance s between the cooperating trains, and As the time of departure of train A( t 0a ) increas-
ΔEloss,a, ΔEloss,b are energy losses incurred to overcome ingly overlaps with the time of breaking of train B,
resistance to movement. the energy recovered during the braking of the train B

Fig. 3. The use of

recovered energy based on
control of train departure
time; Source: own study
based on [20, 21, 24]
 136 Urbaniak M., Kardas-Cinal E.

as used by the train A, can be shown in the following delay in the departure time of the train from the sta-
equations [20, 21, 24]: tion. This implies that the minimum actual traction
– in case 1.: energy consumption Ep occurs for the departure time
t 0a = t**. This time in turn can be determined using the
t 1a
equal division method (bisection) [4, 5] by solving the
Er  Pb ,r  t , s  dt , (7) non-linear equation:
t 0b

dE p / dt 0a  0 . (11)
– in case 2.:
The solution to this equation is based on the as-
t2 t1a

 
Er  Pa t , t 0a dt  Pb ,r  t , s  dt , (8)
sumption of the earliest and possibly latest time of de-
parture of the train from the station and determination
t 0b t2
of the gradient of actual traction energy consumption
for these two departure times. If both the solved values
– in case 3.: are positive, the earliest possible departure time from
t2 t1b
the station is the optimal solution. In any other case,
Er  Pa  t  t 0  dt  Pb ,r  t , s  dt . (9) we determine the optimal value t 0a by gradually nar-
rowing the range of train departure times and analogi-
t 0b t2
cally we look for the optimal solution using the bisec-
After analyzing how the use of regenerative energy Er tion method applied to the function dE p / dt 0a in which
changes depending on the departure time t 0a (increases in both gradients have positive values [24].
cases 1. and 2. or decreases in cases 2. and 3.), it is evident
that the model can be described by an unimodal function.
Further, it should be noted that according to the following 3. Modification of the model
relations as presented in the works [20, 21, 24]:
At the outset, it should be assumed that the trac-
tion network is supplied with constant voltage (e.g.
 dE p 3 kV DC, which is a typical supply voltage on Polish
 a  0, gdy t 0  t ,
a **

 0 dt railway lines), and the rolling stock and infrastructure

 dE p will make it possible to use regenerative braking tech-
 a  0, gdy t 0  t , nology with the transfer of recovered energy to the
a **
(10)
 dt 0 catenary [11]. In the following model, in contrast to
 dE p the original model described in the previous section,
 a  0, gdy t 0a  t ** , a variable controlling the theoretical run in the form
 dt 0
Downloaded from mostwiedzy.pl

B
of braking end time ( t KH ) was proposed. This value
is equivalent to the actual time of arriving train B at
The actual traction energy consumption Ep will
B
the station or stop i.e., t KH  TRPB . This situation is il-
decrease in the first stage, and then increase with the lustrated in Figure 4.

Fig. 4. Changes of
the use of recovered
energy by using
control of train
arrival time
[own study]
 Optimization of the Efficiency of Braking Energy Recovery in Rail Transport by Changing Arrival Time 137

Thus, we have a situation where the start and end Force F(t) for the train’s acceleration phase can be
times of the acceleration phase for vehicle A are con- determined from [17, 18]:
A
stant ( t PR  const . and t KR
A
 const . ), and that means
the characteristics of movement of A train does not FA1 = k·m·aA. (16)
change in terms of acceleration and the amount of en-
ergy required for the acceleration phase. On the other for acceleration with constant force, where k is the
hand, the start and stop times of the braking phase factor of rotating masses, m is weight of vehicle, aA is
B B
for the train B ( t PH and t KH ) shall be shifted, with the maximum acceleration, or for acceleration with con-
earliest braking instance of the train B being ear- stant maximum power:
lier than the starting time of train A ( t KH B
 t PR
A
) and
the latest time of braking of the train B cannot occur PE  E
later than the moment the train A is started ( t PHB
 t KR
A
). FA2  , (17)
Further, as stated before, the actual time of arrival of v A (t )
train B at the station TRPB must be within the range
between the earliest and the latest time of arrival of where PE is power of all vehicle engines and vA(t) is
a given train to a station pursuant to the official de- the characteristic speed calculated from the condition
tailed timetable: TRPB  TPPB ; TWP
B
. FA1 = FA2.
Further, other basic assumptions of the theoreti- Similarly F(t) for the breaking train can be de-
cal run model (1) – (3) are unchanged. Additionally, scribed as:
knowing that the traction network is supplied with
constant voltage, the DC power can be expressed by FB1 = k·m·aB , (18)
the product of the voltage U and the current I:
PE  E
P(t) = ηUI(t), (12) FB 2  , (19)
v B (t )
where η is an efficiency of engine (ηE) or efficiency or
regenerative braking (ηB) and I(t) is the current drawn where aB is maximum deceleration and vB(t) is the
or generated by train. characteristic speed corresponding to FB1 = FB2.
With this substitution we find: Since the highest demand for power and electric-
ity occurs during the acceleration phase, these needs

can be reduced by appropriate energy management
1 2 1 2 
 EUI A  t    mva (t  t )  mva (t )  Eloss , a  / t , from the regenerative braking of another train. The
 2 2  electricity demand in this case will be equal to the

 UI  t , s    (s)  1 mv 2 (t )  1 mv 2 (t  t )  E  / t , energy balance needed to accelerate less the energy
2 b loss , b 
Downloaded from mostwiedzy.pl

 B B  2
b
 recovered from electrodynamic braking [18, 19, 24] of
the breaking train. According to the proposed arriv-
(13) al time control approach, there is no need to optimize
the entire trip in order to achieve relatively measur-
where the energy necessary to overcome the resis- able benefits in the energy balance, but only the part
tance of movement is expressed by [22]: of the trip when the braking and acceleration take
place needs optimization. Therefore, the main com-
Eloss ,a   Rs , a (va )  Rg , a (xa ) xa , ponent of the objective function is proposed to be:
(14)
Eloss ,b   Rs , b (vb )  Rg , b (xb ) xb , EP  E  ERW  min , (20)

where EP is the actual energy value consumed dur-

where Δxa = vaΔt and Δxb = vbΔt are the lengths of the
ing the acceleration phase of the vehicle A, E is the
track sections passed by: vehicle A being in the accel-
all energy required to perform the acceleration, and
eration phase and vehicle B being in the braking phase.
ERW is the energy recovered during the braking of the
The current absorbed / induced can be determined
electrodynamic vehicle B and used in the cooperation
by knowing the force needed to overcome the resis-
process of both trains. Referring to (12), individual
tance of movement F(t), the current train speed v(t)
energy values can be determined as:
and the efficiency of the engine / generator η:
TKR
v(t )
I  F (t ) 
U 
. (15) E U
TPR
 I (t )dt , (21)
 138 Urbaniak M., Kardas-Cinal E.

TKH ergy volume possible to be used in trains energy co-

ERW  U  min{I A  t  , I B  t , s }dt , (22) operation presented in [2] or [6].
TPR
Table 1
Results of optimization using the firefly algorithm for
where IA(t) is the current drawn by the accelerating a train stop Gdańsk Żabianka AWF
train (A) and IB(t,s) is the current enerated by the
EP [kWh] ERO [kWh] ERW [kWh]
braking train (B).
It was also found that there are other criteria that 15.3999 9.7495 5.8728
should be considered using a multi-criteria optimiza- 15.9565 9.5664 6.9408
tion, including: 18.8695 10.0022 7.1727
15.3999 9.7495 6.3806
TKH

I
17.0331 9.2331 6.5425
ERO  U B (t )dt  max , (23)
TPH 15.4019 9.7509 6.3814
17.0616 9.3467 6.5860
where ERO is the entire energy recovered during the 15.3885 9.7153 6.3669
braking of train B that can be used in a different way
16.1888 10.5684 6.8135
other than direct transmission to the catenary (e.g.
additional energy storage) and: 15.4021 9.7543 6.3826
Total: 162.10 97.44 65.44
TP  TOB  TRP  min, (24) Source: own study

where TP is the stop time, which should be as mini-

B
mal as possible due to the station’s capacity, TO is the 4. Conclusions
scheduled time of departure of the train B, while TRP
is its actual arrival time. In conclusion, we note that there are some models that
Taking the above into consideration, we obtain the already outline ways to optimize the use of energy from
following global objective function: regenerative breaking using transmission to the catenary
and mutual cooperation of several vehicles – e.g. by con-
F (TOB , TRP )  w1 EP  w2 ERO  w3TP  min, (25) trolling the time of departure of the train. However, there
are still many other unrecognized possibilities to increase
where w1, w2, w3 are the validity of the individual the efficiency of regenerative braking, such as energy
member functions reflecting the importance of the optimization with the use of a reserve of passage time
Downloaded from mostwiedzy.pl

function of the criteria in the existing railway network included in the timetable by controlling the arrival time.
conditions. This function is minimized, therefore, These methods, although similar, differ in the way trains
partial functions that are also minimized are written run and consume energy. In the first case (control of train
with a positive sign (EP,TP) and the maximized partial departure time) along with its gradual departure delay, the
function (ERO) is written with a negative sign. energy necessary to pass the next section is increased, e.g.,
By using presented model optimized volume of since higher speeds need to be obtained. In the second
recuperated and used energy of cooperating trains case (control of time of arriving train at the station), with
for the Gdańsk Żabianka AWFiS railway station on the arrival delay, the energy demand decreases, e.g., due to
the railway line no. 250 Gdańsk Główny – Rumia extended run without power consumption, need for low-
(Tab. 1). As a result of optimization by Firefly Al- er speeds or possibility of maximum braking deceleration.
gorithm, it has been proved that within 24 hours, Existing models can still be modified and devel-
the volume of electric energy of the cooperating oped adapting them to the set of requirements, existing
train (at the analyzed station) can be recovered (changing) conditions and needs – e.g., by developing
ERO = 97.44 kWh and ERW = 65.44 kWh can be used and explaining the record, introducing the ability to
in the process of energy cooperation of trains. For edit basic data or giving up some of the calculations
total volume of electricity demand of cooperating by replacing them with other dependencies potentially
train EP = 162.10 kWh (on the analyzed section: shortening the whole calculation and analysis process.
Gdańsk Oliwa – Gdańsk Żabianka AWFiS – Sopot In the context of optimization, it should be em-
Wyścigi) energy used directly from recuperation ac- phasized that the organization of train traffic (or oth-
counts for over 40% volume of energy. That means er rail transport systems) cannot be modified only in
the real demand for electric traction energy volume terms of optimizing energy consumption or regenera-
is 96.66 kWh. This result is close to the levels of en- tive breaking efficiency. Due to the superior criteria,
 Optimization of the Efficiency of Braking Energy Recovery in Rail Transport by Changing Arrival Time 139

such as a specific travel time, stoppage time, capacity skiej – Transport, z. 115/2017, Warszawa, Oficyna
of railway lines and stations and demand side require- Wydawnicza Politechniki Warszawskiej, pp. 69–80.
ments, it is possible to reorganize the traffic but only 12. Kwaśnikowski J.: Elementy teorii ruchu i racjonalizacji
in a narrow range. This was reflected both in the as- prowadzenia pociągów, Wydawnictwo Naukowe In-

sumptions of the model TRPB  TPPB ; TWP B

and in the stytutu Technologii Eksploatacji – PIB, Radom 2013.
minimized global objective function. 13. Ortega J. M., Ibaiondo H.: Kinetic energy recovery
The use of the proposed model along with the opti- on railway systems with feedback to the grid, 9th
mization in the analysis of a single railway stop showed World Congress on Railway Research – CRR 2011,
the possibility of saving over 40 % of traction electricity. Lille, France.
14. Pawełczyk M.: Rozwój systemów wykorzystujących aku-
mulację energii w transporcie szynowym, TTS Technika
Literature Transportu Szynowego 2011, nr 10/2011, s. 4146.
15. Pazdro P.: Koncepcja ruchowej optymalizacji efek-
1. ABB, ENVILINE ERS – trakcyjny układ odzyski- tywności hamowania odzyskowego. TTS Technika
wania energii, Rynek Infrastruktury, http://www. Transportu Szynowego – Eksploatacja 2003, nr
rynekinfrastruktury.pl/wiadomosci/enviline-ers- 12/2003, s. 6264.
-trakcyjny-uklad-odzyskiwania-energii-13893. 16. Pena Alcaraz M. et.al.: Optimal underground time-
html [access 15.11.2017]. table design based on power flow maximizing the use
2. Açikbaş S., Söylemez M.T.: Parameters affecting of regenerative-braking Energy, Proceedings of the In-
braking energy recuperation rate in DC rail transit, stitution of Mechanical Engineers, Part F: Journal of
Joint Rail Conference, 2007, pp. 263–268. Rail and Rapid Transit, 2010, no. 226, pp. 397408.
3. Barrero R., Tackoen X., van Mirelo J.: Stationary or 17. Podoski, J., Masłek J., Kacprzak J.: Zasady trakcji
onboard energy storage systems for energy consump- elektrycznej, Wydawnictwo Komunikacji i Łącz-
tion reduction in metro network, Proceedings of the ności, Warszawa 1980.
Institution of Mechanical Engineers, Part F: Journal 18. Podoski J.: Zasady trakcji elektrycznej, Wydawnic-
of Rail and Rapid Transit, 2010, nr 224, pp. 207225. two Komunikacji i Łączności, Warszawa 1967.
4. Burden R.L., Faires J.D.: The bisection algorithm. 19. Roadmap to a Single European Transport Area
PWS Publishers, Boston 1985. – Towards a competitive and resource efficient
5. Corliss G.: Which root does the bisection algorithm transport system : http://eur-lex.europa.eu/legal-
find? SIAM 1977 2/19, p. 325327. content/PL/TXT/PDF/?uri=CELEX:52011DC014
6. Czucha J., Karwowski K., Mizan M., Pazdro P.: 4&from=EN [access 15.11.2017 r.].
Efektywność odzysku energii hamowania elektro- 20. Su S., Tang T., Roberts C.: A Cooperative Train
dynamicznego w komunikacji miejskiej, Przegląd Control Model for Energy Saving. IEEE Transac-
Elektrotechniczny, 10/2004, pp. 10161019. tions on Intelligent Transportation Systems, Volu-
Downloaded from mostwiedzy.pl

7. De Martinis V., Gallo M.: Models and methods to op- me: 16, issue: 2, pp. 622631.
timize train speed profiles with and without energy re- 21. Su S., Tao T., Wang Y.: Evaluation of Strategies to
covery systems: a suburban test case. Procedia – Social Reducing Traction Energy Consumption of Metro
and Behavioral Sciences, 2013, Vol. 87, pp. 222–233. Systems Using an Optimal Train Control Simula-
8. Gonzalez-Gil A., Palacin R., Batty P., Powell J.P.: tion Model, Energies 2016, 9(2), no. 105.
Energy-efficient urban rail systems: strategies for an 22. Tomczyk M., Pacholczyk M., Karwowski K.: Compre-
optimal manage-ment of regenerative braking En- hensive modelling of the cost effectivness of railway line
ergy, Transport Research Arena, Paris 2014. electrification, Zeszyty Naukowe Wydziału Elektro-
9. Jacyna M., Urbaniak M.: Organizacja ruchu pocią- techniki i Automatyki Politechniki Gdańskiej, 2016,
gów w obrębie stacji a oszczędność energii, Prace nr 51, Gdańsk 2016, pp. 197200.
Naukowe Politechniki Warszawskiej – Transport, 23. Urbaniak M., Jacyna M., Kardas-Cinal E.: Metody
z. 115/2017, Warszawa, Oficyna Wydawnicza Poli- wykorzystania rekuperacji w transporcie szynowym,
techniki Warszawskiej, pp. 181189. TTS Technika Transportu Szynowego, 12(273),
10. Jacyna M., Urbaniak M.: Wybrane zagadnienia 2016, pp. 355359.
optymalizacji organizacji ruchu kolejowego w celu 24. Urbaniak M., Kardas-Cinal E.: Modelowanie reorga-
minimalizacji kosztów zużycia energii, Prace Nauko- nizacji ruchu w transporcie szynowym zwiększającej
we Politechniki Warszawskiej – Transport, z. 109, efektywne wykorzystanie energii z hamowania odzy-
Oficyna Wydawnicza Politechniki Warszawskiej, skowego, Prace Naukowe Politechniki Warszawskiej
Warszawa 2016, pp. 4558. – Transport, z. 118, Warszawa 2017, pp. 325337.
11. Karkosińska-Brzozowska N., Urbaniak M.: Bariery 25. Warin Y., Lanselle R., Thiounn M.: Active sudsta-
stosowania technologii odzysku energii w transporcie tion, 9-ty World Congress on Railway Railway –
szynowym, Prace Naukowe Politechniki Warszaw- WCRR 2011, Lille, France.