New Trends in Modern Electrical Traction

Profesor Maria PIETRZAK-DAVIDUniversité de Toulouse, INP-ENSEEIHT

2, rue Camichel , 31071 Toulouse Cedex 7

L a p l a c eLaboratoire Plasma et Conversion d’Energie – UMR N°5213

III KONGRES ELEKTRYKI POLSKIEJ2-3 Kwiecien 2019 Warszawa

Paris 1900

Much less electronics

but the basic idea was already there

Overview

Introduction

Power Electronics Topologies

Asynchronous Motor Control Strategies

Synchronous Motor Control Strategies

Centralized vs distributed propulsion

External and internal disturbances and system operation

Laplace test bench - reduced train bogie

Train or tramway without pantograph

Conclusion

Now the world is focused on CO2 emission reduction of the systems

Embedded Electric Traction Systems are very important for rail transport.

The design of these systems’ topology requires their principal components:

• power supplied from the electric traction line (catenry)

• pantograph,

• power-electronic transformer,

• DC filter,

• voltage inverter

• AC electric machine controlled according to the mission

Hard constrains have to be respected in these systems

Introduction

Mass

on the axleVibrations

Shocks

DC-AC

Multi-voltage

Supplies

Adherence

wheel-rail

Reduced

Volume

Sizing

Voltage

Variations

Loss of power Compatibility

to

Signalization

Constrains to respect

Topologies and Traction Motor Voltages

Onduleursde tension

L2

C2 ~ =

~ =

~ =

~ =

= ~ = ~ = ~ = ~

M1

M3

M2

M4

Redresseurs

C16mF

2.8kV

RFREINAGE

1.15

25kV 50Hz

Dis

jonc

teur Onduleurs

de tension

C04.2mF

= ~ = ~ = ~ = ~

M1

M3

M2

M4

L010mH

C8mF

2.8kV

RFREINAGE

2.3

2.8kV

2.3

L5mH

L5mH C

8mF

4 hacheurs enparallèle

3kV cc

Disj

onct

eur

RFREINAGE

Onduleursde tension

C016.8mF

= ~ = ~ = ~ = ~

M1

M3

M2

M4

L02.5mH

C8mF

2.8kV

RFREINAGE

2.3

2.8kV

2.3

L5mH

L5mH C

8mF

4 hacheurs élévateursen parallèle

1.5kV cc

Disj

onct

eur

RFREINAGE

AC: 25kV, 50Hz 3kV, DC

1,5kV, DC

DC Input filter sizing

Converter current harmonicsLine voltage harmonics

A good stability

has to be always ensured

L not too small to

guaratee a high

Impedance seen from

the line : ~ Z = L

A low impedance

to short-circuit

converter harmonics

then C high

Avoid

the resonance frequency

is too close to

the mechanical resonances

and the motor control

bandwith (0-10Hz)

LC

LCf

21

DC Motor

with colector

Synchronous Winding Motor

(TGV)

Squirel Cage

Asynchronous Motor

Permanent Magnet

Synchronous Motor

Railway Traction Motors

Two types of motor control in use

For Tram-Metro traction drives

Flux observer

Discret predictive model

Operates only with Space vector PWM

Does not allow max voltage

PNG: Pilotage New Generation

For EMU-locomotive traction drives

Flux estimator

Vector control and scalar control

Operates with any kind of PWM patterns

Allows the max available voltage

PLU: PiLotage Unified

Asynchronous Motor Controlwith a Voltage Source Inverter

PLU Motor Control PiLotage Unified -

Tsetting.LrP.Lm. r

Lm.Rr2 Lr

Isqr

cons =arctg(Vsq/Vsd)

Mod=6

Vsd + Vsq2

Vsd + Vsq2

Torque Tcontrol

Flux RControl

Vsq

Vsd

Phasecontrol

fs-MLI

fcorr

+

+

Demodulation

CalculationIsd, Isq, r

Is1, Is2

Uf

1, 2

Rotorspeed

Speed sensorfR fs

+

+

r Isq

+

r cons.t

r cons

T setting.

r

Isq cons.

cons

+

+

-

Modulation

PWM

Mod

Leg1

ON/OFFUf

UfU f

Decoupling

Isd

Leg2

Leg3

Uph 1,2,3fs

fR=p

FW-RE

PNG Motor Control Pilotage New Generation

A fact: Improve the motor torque control is an endless task

A dream-The perfect motor control:

accurate and fast torque control from 0 to max speed

robust against the variations of the motor parameters

easy to tune

no more sensors ( no speed sensor, no voltage sensor…)

PLUtoo many parameters to tune

weak decoupling (7Hz instability)

limited torque dynamic (scalar)

poor flux estimation at low speed

allows all PWM patterns

allows 100% output voltage

PNGeasier to tune-less parameters

high torque dynamic possible

good flux estimation by observer

slight static torque error

asynchronous Space vector only

need voltage margin: 95% output

12

Kalman Filter

inputu(t) y(t)

Observed outputyobs(t)

MODELstate xprédit (t)

K+

-

Observer

y(t)-yobs(t)

Minput

u(t)

Estimated outputyest(t)MODEL

state xest(t)

Estimator

Measured outputy(t)M

MOTORstate x(t)

PLU

ESTIMATION

PNG

OBSERVATION

Measured output

MOTORstate x(t)

No feed back

Errors due to measurement noise

and inaccurate model parameters

Feed back and correction

controlled by gain K

Convergence motor-model

Estimator

PWM Strategy

•The commutation frequency Fc of the inverter switches is limited by the silicone losses•The stator frequency Fs can reach several hundreds Hz at max speed (see chart)•The ratio Fc/Fs can become too low at a certain speed and generate low frequency harmonics within the motor ( beating, vibrations)•Generally Fc/Fs is maintained higher than 10 to avoid these low harmonics•If Fc/Fs can not be maintained above 10 the modulation type has to be changed •It is the PWM strategy*

Inverter type Fcmax

GTO 4,5kV 300Hz

IGBT 1,7kV 1,5 à 2,4kHz

Application

Locomotive

Tram-Metro

IGBT 3,3kV 600HzEMU Electric Multiple Units

IGBT 3,3kV 350/450HzLocomotive

Fsmax

160Hz

160Hz

230Hz

160Hz

YES

NO

YES

YES

SeveralPWM

*Only one PWM asynchronous pattern can be used up to Vmax but that demands the thermal dimensionningto be improved or to increase the IGBT number

PWM changing

U legUI phase

U leg0

U

I phase

Asynchronous PWM

Synchronous PWM

Calculatedangle PWM

Square or full wave

PWM Asynchronous(ASY)

PWM Synchronous(SYN)

PWM calculated angles(CALC)

Full wave(PO)

Fc: fixed frequencyFc > 10Fs

Intersective modeNeeded for Fs low

Fc = kFsk= multiple odd of 3Fixed phase Fc/FsIntersective mode

Neededto access to calculated angle PWM

N calculated anglesFc= (2N+1)Fs

1 angle to control fondamentalN-1 angles to

minimise harmonicsFront + ou -

Fc=FsMax motor voltage achievable

Reduced switching lossesModmax=1

Different types of PWM

785,04

785,04

907,032

Pure sine

907,032

Sine +1/6H3

Pure sine

Sine +1/6H3

)kFT21(Mod sminmax

)FT21(Mod cminmax

PWM type Characteristics Modmax(theo.) Modmax (real)

mins

n1n TF2

when

Modmax=1

mins

n1n TF2

Permanant Magnet Synchronous MotorPMSM

Stator– 3-phase windings– no major changes compared with as asynchronous motor

Rotor– Samarium-Cobalt magnets (Bmax: 1,1Tesla , T°max 350°C)– Fractionned and insulated magnets, glued on the rotor surface– Glass fiber bandage to fix the magnets– Rotor yoke partially laminated to reduce the losses ( eddy

currents)Totally closed motor– autoventilated or water cooled according to the power– to avoid the pollution by the magnetic metallic dust

Main applications– Trams, EMUs, AGV : tough space constraint– PMSM as complementary of the ASM

Signals to control PMSM

IRIS

Uf

Control

Uf IR ISTsetting

Inverter on/off

IGBT gate pulsesRotation direction

PMSM

Vector Control with « PLUS »

id Modul

Phase

Control

Iq

Decoupling

terms

Control

Id

phir

iq_cons

id_cons

uq_cor

ud_cor

uq_dec

ud_dec

vq

vd

fs_ind

modul+

+

++

UF_pilot

alpha_cons

iq

Quite similar to the vector control for the asynchronous motor

Torque is controlled by IqId is generally maintained at 0

« PLUS »: PiLotage Unified for Synchronous motor

iq

id_filtre

gamma_satiq_cons

PI-

+

LqphifiltreidLd 0_

Scalar control with « PLUS »

At full voltage (Modmax) the motor is controlled by its internal angle

Angle : angle between voltage and electromotive force ( no load)

The flux weakening can be obtained due to the defluxing

created by the the stator current

Permanent MagnetSynchronous Motor

Magnets mounting

Pro and consof the asynchronous motor

ASYNCHRONOUS

Cost

Low leakage inductanceVery high SC current and SC torque

High torque ripple

Large rotor Joule losses

Several motors in parallel/inverter

Robust structure

PMSM

Volume gain -25%Weight gain -25%

Low rotor lossesEfficiency gain +3%

Smooth rotor, less gap harmonicsClosed motor

Less noise -3dBAHigh internal inductanceLimited overtorque: 1,2 Tn

Only a single motor per inverterThe magnets impose a closed motor (pollution)

Heat extraction more difficult

Higher inverter switching frequencyMore complex inverter? ( +IGBT,3 levels ?…)

No possibility to reduce the flux if inverter failure (magnets)Inverter must be disconnected of the motor by a contactor

Large number of poles Thinner stator yokeLarger inner diameter

Possibility of higher torques

Overvoltage at high speed if the inverter blocksDC bus clamping device to implement

Overcost + 10% at 20%

Pro and consof the PMSM

PMSM manufacturing steps

Magnet mounting

Rotor bandage

Banded rotor

Stator frame

Stator windings

Stator and rotor

Synchronous/Asynchronous Motor comparison

0,00

0,50

1,00

1,50

2,00

2,50

3,00

3,50

4,00

4,50

0 100 200 300 400Bore diameter (mm)

Surf

ace

pres

sure

Ts

(N/c

m2)

Asynchronous

Synchronous

Ex: Freight locomotiveD=540mm, L=470mm

Tnom=7529 NmTs = 3,5 N/cm2

Comparison between TGV motors

535 1130

3000 4000

1560 1525

DC motor Synchronous

2,9 1,35

TGV-PSE

1020

4000

1260

Asynchronous

1,23

EurostarTGV-A

700

4500

700

PMSM

1,00

AGV

Power kW

V max t/mn

Weight Kg

Type

Ratio Kg/KW

Centralized ou Distributed Traction

Number of electrical equipments reduced cost of production and maintenance reduced

Isolation between passenger wagons and locomotive vibrations and noises reduced

Braking with multi-axis recovery;

Reduction of the maximum weight per axis;

Reduction of the adherence coefficient required per axis;

Increased passenger capacity.

Distributed traction prefered solution

Centralized traction

The disturbances of the railway traction system

External disturbancesPantograph detachmentLoss of adherenceStick- slip perturbationIrregular wear of the wheels

Internal disturbancesThe parameter variationsPower component degradations and dead timeSensor losses (current, voltage, speed)Traction power losses

Cooperative Control with Monitoring Automaton

New test-bench in LAPLACE laboratoryReduced Model of train bogie with 2 Asynchronous / Synchronous Motors, 2 PWM VSI,

Common DC Bus and Common Loads

Common loads – Emulation (real mechanical connexion is assured by PC programming

Generation of external and internal disturbances

Studied controls : several RFOC with observers to electrical and mechanical modes

Mean Control with observers (Luenberger, Kalman)

Wighted Mean Control (Luenberger, Kalman)

Mean Differencial Control (Luenberger, Kalman)

3 types of Sensorless Control (Luenberger, Kalman, OSV)

Service continuity and optimization of the system operation

Monitoring Automatum Management of analytical and structural redundancy

Cooperative control : better system performances and energy efficiency

Tolerant Fault Control better system availability and performannces

Ground Feeding of tramway in Bordeaux

No pantograph

Autonomous train with hydrogen fuel cells- Alstom realization

• Diesel elimination and hydrogen fuel

cells introduction

• Autonomous train Coralia iLint was

tested in Germany in Reichenhoffen

in september 2018

• Alstom prepares the dual mode

operation: « hydrogen-electricity»

and 1000 trains will be realized

in 2035-2040.

In France, in Occitania Region an autonomous train « REGIOLIS »

will be operational for the local railway communication

"Zero diesel trains" were announced

in « La vie du rail » 21 December 2018

No pantograph

Conclusion

Motor for modern railway propulsion, choice of motors without or with permanent magnets

Autonomous train with fuel cells, pantograph suppression, What kind of voltage source?

Normalization of voltages used as train feeding : AC, DC

Power Electronic Topology and new power components SIC, GAN: High Voltage or Multi-level solutions?

Hybrid structure, or electrical structure only for train propulsin

Centralized or distributed configuration of propulsion

Multi-Machines / Multi-Inverters / Multi-Loads system management

Sensor number optimization – maintenance cost minimization -sensorless control strategy

global CO2 emission but analysis with all component production

Acknowledge to

for different traction system data