You are on page 1of 6

Simulation of Nanjing Metro Line 1 Using Metro Simulator

Wei WANG, Ming CHENG

School of Electrical Engineering, Southeast University


Nanjing 210096, China (mcheng@ seu.edu.cn).

Abstract: Based on Nanjing Metro Line 1 (NJML1), this paper presents a metro simulator (MSim).
Firstly, NJML1 is analyzed and divided into four subsystems: traction power supply system, track, train,
and train controller. The modeling method for each subsystem is presented. Based on the proposed
models and Visual Basic 6.0, MSim is developed, which has two operation modes: single-train mode and
multi-train mode. Single-train mode is used to check train performances and generate train time
schedules. Multi-train mode aims to check and optimize system performances of metro lines. At last, the
NJML1 is taken as an example to verify the effectiveness of MSim.
Keywords: Modelling, trains, simulators, railway.

1. INTRODUCTION 2.1 System information


Due to the excessive release of CO2, greenhouse effect has
become critical and received worldwide attention. On the NJML1 is located in Nanjing of China, and the length is 14
other hand, traffic congestion is globally existed, which has 318 m. It has 12 stations and the dwell time at each station is
obviously increased the time cost during urban travel, listed in Table 1. It is supplied by 1500V DC with the
especially in major cities. Due to the advantages of high substation at each station. And the substation is irreversible.
speed, punctuality and environment protection, urban mass
transit, especially metro, is a viable solution to meet the two Table 1. Dwell times of MJML1
proposed world challenges above. Generally, metro is a Station Position (m) Dwell time (s)
multidisciplinary subject which covers broad and complex Maigao bridge 0 120
aspects. Considering complexity and cost, the performance of Hongshan zoo 1141.0 25
metro must be ensured before construction, while it is Nanjingzhan 2264.0 30
difficult and expensive to carry out real experiments.
Xingmofan road 3949.0 25
With the development of computer techniques, digital Xuanwumen 5010.0 25
computer simulation of rail transit system has been widely Gulou 6263.5 30
adopted as a standard computer-aided engineering tool during Zhujiang road 7126.0 25
the design and development stages of existing and new rail Xingjiekou 8273.0 35
transit systems (Goodman et al., 1998), and many simulation Zhangfuyuan 9399.0 25
methods of rail transit system have been presented. Generally, Sanshan street 10310.0 25
the objectives of these simulation methods can be classified Zhonghuamen 12226.0 25
into three main categories: simulation of train movement Andemen 14318.0 35
(Khmelnitsky, 2000; Ho et al., 2002; Li et al., 2007; Leva et
al. 2008; Allgre et al., 2010; Wang et al., 2010), calculation 2.2 Train
of traction power supply systems (Liu and Chen, 2005; Ho et
al., 2005) and the generation of train time schedules (Assis The structure of the train is shown in Fig. 1, in which Tc is
and Milani, 2004; Albrecht, 2009). However, most of the the trailer car; Mcp, Mc are the motor car with and without
existed simulation methods for rail transit systems only focus pantograph respectively. The traction motors are only located
on one objective, which makes the study of multi-objectives in motor cars.
difficult. Therefore, a multi-objectives simulator Metro
Simulator (MSim) is proposed in this paper, in which train Traction power supply system
movement, traction power supply system and train time
schedule are taken into account. Nanjing Metro Line 1
Tc Mcp Mc Mc Mcp Tc
(NJML1) is taken as an example to verify the effectiveness of
MSim. Track
2. NJML1 DESCRIPTION
Fig. 1. Structure of train.
2.3 Traction system Due to the fact that the departure interval of NJML1 is from 4
to 7 minutes, there is only one train running between every
The traction system only exists in motor cars. The topology two adjacent substations in one direction at any time. In this
of traction system in one motor car is illustrated in Fig. 2, paper, every substation is simplified as a constant voltage
which consists of one converter, four traction motors, four source. Therefore, every two adjacent substations can make
gears, four bogies. The traction motor is squirrel-cage up a dependent power supply section, named as sub-TPSS.
induction machine. The converter is controlled by the Because NJML1 has 12 substations, TPSS can be divided
strategy of variable voltage variable frequency (VVVF). The into 11 sub-TPSSs. A train is thought to be a moving load.
traction system can work in both traction and regenerative The structure of sub-TPSS is shown in Fig. 3. xA, xB and xTrain
braking states. are the position of Station A, Station B and the train
respectively. The line resistances between the train and the
Traction power supply system two stations can be given by
Pantograph
R1 = r ( xTrain x A )
DC (2)
AC R 2 = r ( x B xTrain )
M M M M Motor
Converter
where R1, R2 are the line resistances between the train and
G G G G Gear
Station A, Station B, respectively; r is line resistance ratio.
Wheel Udc is the DC voltage of substations, PTrain, uTrain, iTrain are the
Track power, voltage, and current of the train respectively. Among
these above mentioned variables, uTrain, iTrain are unknown
which can be given by (3), while others are known.
Fig. 2. Topology of traction system in a motor car.
Direction
xA xTrain xB
2.4 Braking system R1 R2

U dc PTrain U dc
Both electric and mechanical brakes are used, and electric (uTrain , iTrain )
brakes are preferable. Electric brakes include two types:
regenerative braking and rheostatic braking. The energy Station A Station B
(Substation) (Substation)
generated by regenerative braking is firstly fed back to the
traction power supply system, which can be reused by
adjacent trains. However, if there are no adjacent trains or the Fig. 3. Simplified structure of sub-TPSS.
adjacent trains cant absorb all the energy, the voltage of the
traction power supply system will increase. Once the supply
voltage increases over the upper voltage limit, the rheostatic U + U dc 2
4 R1R2 PTrain /( R1 + R2 )
braking will be immediately used to consume the excess uTrain = dc
2 (3)
energy in the form of heat energy. The braking resistors are
i
on board. Train = PTrain / uTrain

2.5 Speed limit 3.2 Track model


The maximum available train speed is 80 km/h. Additionally, NJML1 is mainly underground. Considering the complicated
the train speed is also limited by the radius of track curves. conditions of ground surface and cost saving, more restrictive
According to the practical experiences, the track curve limits design requirements of metro lines are usually demanded,
the train speed by which complicate the running condition of trains.
Consequently, an accurate track model is important.
v= R (1) Therefore, a database is set up to store the information of
stations, ramps and curves. Stored data types are illustrated in
where v, R are the train speed and the radius of present track
Fig. 4. When the train is running, the data stored in track
curve respectively.
model are looked up by the position x of trains.
3. MODELLING Database Storage format

Generally, NJML1 can be divided into four subsystems: Station Number Position
traction power supply system, track, train, and train controller.
Ramp Number End position Gradient
In this section, all these four subsystems will be modelled.
Curve Number End position Radius
3.1 Traction power supply system (TPSS) model

Fig. 4. Stored data types of track model.


3.3 Train model normal state, the motor torque must be limited in the shadow
area. Additionally, the velocity is assumed to be always non-
As is known, except traction system, there are also auxiliary negative in this paper.
systems, such as lighting system, air conditioner, and so on.
The energy absorbed from TPSS is distributed by traction Pe_motor can be given by
system and auxiliary system. The latter is only considered as
a constant load, which is denoted as Paux. The former is the Tmotorm / motor , Tmotor _ ref 0
core of a train, which has been shown in Fig. 2.
Pe _ motor = (7)
motorTmotormotor , Tmotor _ ref < 0
The converter is an interface between DC and AC side, which
can be modeled by where Tmotor, motor, motor, and Tmotor_ref are torque, rotational
speed, efficiency of motor, and reference value of Tmotor,
Pac _ conv /conv , Pac _ conv 0 respectively. If Tmotor_ref is located in the shadow area of Fig.
Pdc _ conv = (4) 5, Tmotor = Tmotor_ref; otherwise, Tmotor is the torque value which
conv Pac _ conv , Pac _ conv < 0 is closest to Tmotor_ref in the toque curve corresponding to the
same rotational speed.
where Pdc_conv and Pac_conv, are the power of the converter at
DC and AC side respectively, conv is the efficiency of the Gearbox is the interface between traction motors and wheels,
converter. When Pac_conv is negative, the train is braking; which can be modelled by
otherwise, the train is in the state of traction.
gear ngearTmotor , Tmotor 0
As is mentioned before, there are four converters in a train. In Tgear = (8)
this paper, their currents are supposed to be equal. Thats to ngearTmotor / gear , Tmotor < 0
say, the power PTraction absorbed from TPSS for the traction
purpose is equally distributed by four converters. Therefore, motor = wheel ngear (9)
PTraction can be given by
where Tgear, wheel, gear, and ngear are the torque and rotational
PTraction = nconv Pdc _ conv (5) speed of gear at wheel side, the efficiency and ratio of gear,
respectively.
where nconv is the number of converters.
Wheel is the interface between rotational and linear motion,
As is shown in Fig. 2, one converter controls four traction which works by
motors. Similarly, Pac_conv is also supposed to be equally
distributed by the four traction motors. Therefore, Pac_conv can Fwheel = Tgear /rwheel
be given by (10)
wheel = vTrain / rwheel
Pac _ conv = nmotor Pe _ motor (6)
where Fwheel, vTrain, and rwheel are the traction effort from one
where nmotor and Pe_motor are the number of traction motors in traction motor, train speed, and radius of wheel, respectively.
one motor car and the electric power of each traction motor, The total traction effort Fe can be given by
respectively.
Fe = nconv nmotor Fwheel (11)
Tmotor
When the train is running, the running resistance f from the
environment must be considered, which can be given by
Traction
0 f = f (vTrain , r , g ) (12)
motor
Braking
where g and r are the gradient and the radius of current track
respectively.
At the most time, the braking resistance Fm caused by
mechanical brakes is zero. However, when electric braking
Fig. 5. Torque curve of traction motor. can not provide enough braking effort, mechanical braking
system will be used, in which the acceleration of train is
The traction motor is represented by the torque curve as supposed to be known and constant. Therefore, Fm can be
shown in Fig. 5. The torque curve can be obtained through given by
some tests of the motor under specific conditions. And the
curve contains two parts: the positive part is the traction Fm = M Train aTrain Fe + f (13)
characteristic, and the negative part is the braking
characteristic. Therefore, there are two different torque values
corresponding to one rotational speed in the torque curve. In
Tmotor _ ref

Paux
+
Pe _ motor Pac _ conv Pdc _ conv PTraction PTrain
Motor nmotor Converter nconv
+
motor
Tmotor
Tgear Fwheel n Fe
motor nconv vTrain xTrain
Gear Wheel
wheel Bogie Environment
vTrain Fm
f < r , g >

Fig. 6. Block diagram of train model.

where aTrain and MTrain are the acceleration and the equivalent Based on Visual Basic 6.0 and previous models, MSim is
mass of train, respectively. developed, which has two modes: single-train mode and
multi-train mode. The human-computer interfaces of two
In normal situation, the acceleration of train can be given by modes are shown in Fig. 8. In single-train mode, only one
train is used and a round trip is completed. The train
aTrain = ( Fe f ) / M Train (13) performances such as speed, traction effort, DC voltage, and
so on, are computed, as shown in Fig.8 (a). Generally, single-
The integral of aTrain is vTrain, and the integral of vTrain is xTrain train mode is used to check train performances and generate
(the position of train). train time schedules. On the other hand, multi-train mode is
The integration of train model is shown in Fig. 6. to simulate the daily operation of metro lines. Some system
performances are computed. It aims to check and optimize
system performances of metro lines.
3.4 Train controller model

The train controller is used to generate Tmotor_ref. Its flowchart Speed - reference
is shown in Fig. 7. The speed-reference generator is to Generator
generate the target speed vTrain_ref, in which the controlled vTrain_ref
braking distance (CBD) is defined. CBD is the distance
between the train and the next station. If the distance between Yes
the train and the next station is longer than CBD, vTrain_ref is vTrain_ref = 0 ?
set to be the maximum velocity at current situation. If the
distance between the train and next station is shorter than No
CBD, vTrain_ref is set to be 0. If vTrain_ref is non-zero, the Acceleration - reference Acceleration - reference
acceleration reference aTrain_ref is generated by the Generator 1 Generator 2
acceleration-reference generator 1 to achieve vTrain_ref as soon
as possible, in which the position is not considered. If vTrain_ref
is zero, it means that the train will pull into the station or has aTrain_ref
stopped in the station. In order to minimize the error between
actual and desired stop position, the acceleration-reference Traction - effort - reference
generator 2 is developed to generate the acceleration referenc, Generator
in which the position is considered. In both acceleration- Fe_ref
reference generators, the jerk is limited. To acquire aTrain_ref, Motor - torque - reference
the corresponding traction effort reference Fe_ref are generated Generator
by the traction-effort-reference generator. The motor-torque-
reference generator is designed by the inversion-based model
of traction system. In normal situation, when the command of Tmotor_ref
Tmotor_ref is input, Te will be equal to Te_ref.

4. MSIM DEVELOPMENT Fig. 7. Flowchart of train controller model.


Track
DC voltage

Train Station
Height

Traction effort

Train speed System performances

Train performances

(a) (b)

Fig. 8. Human-computer interfaces of MSim: (a) single-train mode; (b) multi-train mode.

2.0 2.0

DC voltage (kV)
DC voltage (kV)

1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.0
400 400
Train speed (km/h) Traction effort (kN)

Train speed (km/h) Traction effort (kN)

200 200
0 0
-200 -200
-400 -400
120 120
80 80
40 40
0 0
Limited speed Actual speed Limited speed Actual speed
-40 -40
0 500 1000 1500 2000 2500 3000 340 360 380 400 420 440 460
Time (s) Time (s)
(a) (b)

Fig. 9. Simulated results of single-train simulation: (a) entire waveforms; (b) partial waveforms (Nanjing railway terminal
Xingmofan road).

speed follows the limited speed well. When the train is in the
5. SIMULATION
state of traction, DC voltage drops. And DC voltage rises
To verify the effectiveness of MSim, NJML1 is taken as an when the train is braking. During the whole simulation,
example. Both single-train and multi-train modes are mechanical brakes are not used. It means that electric brakes
executed. are effective enough to stop the train in time under normal
situation.
Firstly, single-train mode was executed and the computing
time is 424 s. The entire duration including dwell times is Secondly, multi-train mode is carried out to simulate the
2693.3s and the run times between every two adjacent daily operation of NJML1. The entire duration is 1 hour. The
stations are listed in Table 2, which can be used to generate departure interval is 4 minutes, and 12 trains are put into
time schedules for NJML1. In Table 2, the up and down time service. To evaluate energy consumption of NJML1, three
mean the duration of every adjacent stations in the up and cases with conditions listed in Table 3 are investigated.
down direction respectively. The simulated train Curves of the energy consumption under three cases are
performances are shown in Fig. 9. And Fig. 9(b) is the partial shown in Fig. 10. The total energy consumption in Case I is
waveforms of Fig. 9(a). As is seen in Fig. 9, the actual train 10.100 MWh, among which the electric braking energy
wasted by braking resistors is 4.077 MWh. Once energy- mainly used to check train performances and generate train
storage devices are used, the total energy consumption is time schedules, while multi-train mode is aimed to check and
significantly reduced. Under the assumption that all the optimize system performances. The NJML1 has been taken
electric braking energy is reused, the total energy as an example to verify the effectiveness of the MSim.
consumption in Case II and Case III are 6.023 MWh and Moreover, three cases have been investigated to evaluate
5.784 MWh respectively. energy consumption of NJML1 and the improvement on
energy efficiency has been suggested. The results have shown
Case I is actually used by NJML1. Obviously, Case I is not that the MSim can effectively simulate the operation
an efficient operation mode and Case II or Case III should be performance of metro system.
a better solution from the viewpoint of energy saving.
ACKNOWLEDGMENT
Table 2. Duration of every two adjacent stations
This work was supported by the Program Sponsored for
Start station End station Up time (s) Down time (s)
Scientific Innovation Research of College Graduate in
Maigao bridge Hongshan zoo 91.7 88.9
Jiangsu Province, China (Project: CXZZ_0149).
Hongshan zoo Nanjingzhan 78.3 87.8
Nanjingzhan Xingmofan road 113.6 114.3
Xingmofan road Xuanwumen 77.7 77.5
REFERENCES
Xuanwumen Gulou 87 87 Assis, W.O., and Milani, B.E.A. (2004). Generation of
Gulou Zhujiang road 68.5 69.3 optimal schedules for metro lines using model predictive
Zhujiang road Xingjiekou 78.7 82.4 control. Automatica, 40(8), 1397-1404.
Xingjiekou Zhangfuyuan 81.4 79.6 Albrecht, T. (2009). Automated timetable design for demand-
Zhangfuyuan Sanshan street 70.9 70.3 oriented service on suburban railways. Public Transport,
Sanshan street Zhonghuamen 117.3 118.8 1(1), 5-20.
Zhonghuamen Andemen 133.4 124.8 Allgre, A.-L., Bouscayrol A., Verhille, J.-N., Delarue, P.,
Chattot, and E., El-Fassi, S. (2010). Reduced-scale-
power hardware-in-the-loop simulation of an innovative
Table 3. Conditions of three cases subway. IEEE Transactions on Industrial Electronics,
57(4), 1175-1185.
Case Conditions Goodman, C.J., Siu, L.K., and Ho, T.K. (1998). A review of
The electric braking energy is firstly reused by simulation models for railway systems. IEE
adjacent trains. If the adjacent trains can not Developments in Mass Transit Systems Conference, 80-
Case I absorb all the electric braking energy, the 85.
energy will be consumed by braking resistor on Ho, T.K., Mao, B.H., Yuan, Z.Z., Liu, H.D., and Fung, Y.F.
board. (2002). Computer simulation and modeling in railway
The electric braking energy is firstly reused by applications. Computer Physics Communications, 143(1),
adjacent trains. If the adjacent trains can not 1-10.
Case II absorb all the electric braking energy, the Ho, T.K., Chi, Y.L., Wang, J., Leung, K.K., Siu, L.K., and
energy will be stored by energy-storage device Tse, C.T. (2005). Probabilistic load flow in AC
at stations. electrified railways. IEE Proceedings Electric Power
The electric braking energy is stored by energy- Applications, 152(4), 1003-1013.
Case III
storage device on board. Khmelnitsky, E. (2000). On an optimal control problem of
train operation. IEEE Transactions on Automatic Control,
12 45(7), 1257-1266.
Energy (MWh)

10 Case 1 Leva, S., Morando, A.P., and Colombaioni, P. (2008).


8 Case 2 Dynamic analysis of a high-speed train. IEEE
6 Case 3 Transactions on Vehicle Technology, 57(1), 107-119.
4 Li, K.P., Gao, Z.Y., and Mao, B.H. (2007). Energy-optimal
control model for train movements. Chinese Physics,
2 16(2), 359-364.
0 Liu, Y.C., and Chen, J.F. (2005). Control scheme for
0 600 1200 1800 2400 3000 3600
reducing rail potential and stray current in MRT systems.
Time (s) IEE Proceedings Electric Power Applications, 152(3),
612-618.
Fig. 10. Curves of energy consumption under three cases. Wang, W., Cheng, M., Hua, W., Zhao, W.X., Ding S.C., and
Zhu, Y. (2010). Hybrid modeling and applications of
virtual metro systems. IEEE Vehicle Power and
6. CONCLUSION
Propulsion Conference, 1-5.
To simulate multi-train metro system, a multi-train simulator,
MSim, has been developed, which has two work modes:
single-train mode and multi-train mode. Single-train mode is

You might also like