Sei sulla pagina 1di 6

2013 3rd International Conference on Electric Power and Energy Conversion Systems, Yildiz Technical University, Istanbul, Turkey,

October 2-4, 2013

Thyristor Controlled Rectifiers (TCR) for traction


problems and solutions
Vitaly Gelman
VG Controls
Vemon, NJ USA
vgelman@vgcontrols.com

Abstract-The paper discusses specifications for the Thyristor


Controlled Rectifiers (TCR) in traction applications. Covered are
TCR voltage selection, output filter requirements, harmonics,
reactive power, power circuit configuration, power recuperation
and

cooling.

Compares

TCRs

and

diode

rectifier

usage

in

traction. It summarizes TCR requirements specific for traction.

Keywords - traction; Thyristor Controlled Rectijiers;TCR; DC


jiltering; AC harmonics; voltage selection; reactive power;
recuperation
I.

[NTRODUCTION

There is a significant interest to the TCRs from the


transportation authorities yet the majority of traction projects
use diode rectifiers. The Thyristor Controlled Rectifiers (TCR)
advantage compared to diode rectifier is higher train voltage,
but this comes at a cost: we need a controller to regulate the
voltage and an output filter to reduce the DC voltage ripples.
There are very few publications comparing diodes and TCRs
for traction and no standards or specifications for traction
TCRs; this paper attempts to fill the gap. We will show how
TCRs improve train voltage regulation, reduce system cost by
increasing traction substation spacing and provide energy
savings through the power recuperation. This paper presents
analysis of TCR' power circuit options and summarizes the
requirements for traction rectifiers.
[I.

TCR VS.

DIODE RECTIFIERS, OR WHY DO WE NEED

TCR

[nternational
Standard
IEC 1287- [
( 1995)
"Power
converters installed on board rolling stock"[2] specifies the
train equipment shall operate normally at [.2 nominal voltage.
US standard nominal train voltage is 750 VDC, such train shall
operate normally up to at [.2*750=900 VDC.
What is the maximum nominal voltage of diode rectifier to
operate 750 V trains?
( 1) Using standard US utility contract specifying 10%
incoming AC voltage regulation, the maximum no load
voltage should be 900/ l. l=818 VDC.
(2) Using 6% regulation (typical in traction) we get
nominal voltage of 8 18* 0.94=769 VDC.
(3) To allow for 2% ripples we get 769*0.98=754 VDC
In practice the nominal voltage for Diode rectifier is
specified as 750 V for 750 V trains. This voltage (together with

voltage regulation at higher current) limits Traction Power


Sub-Stations (TPSS) spacing to about 1 mile.
We have more leeway with selection of TCRs output
voltage because TCR regulates the DC voltage to compensate
for changes in incoming line voltage and load current.
US transportation authorities specifY TCR rated voltage
from 825 to 850 VDC; this voltage stays constant with load
current changing from 0% to an inflection point (selected
anywhere from [00% to 300% current). The lower voltage (e.g.
825 V) would leave more margins for regenerative braking
power transfer.
Assuming linear diode rectifier loading curve with 6%
regulation and 750 V nominal voltage and a TCR with 825
VDC nominal voltage and 150% inflection point, we arrive at
Figure 1. As you can see from Figure 1, there is TCR voltage
gain of [25 V for loads above 150% comparing to Diode
rectifier.
Rectifier Loading Curves
650
600

750

8,

J!!

. -r--

700

----

---

650

----

r--

---

r--

---

--

r--

600
550

-DiOde

-TCR

500

450

400

50

Figure

100

I.

150

200

Loadj%)

250

300

350

400

Voltage regulation of Diode rectifier and TCR

The higher TCR voltage allows us to increase TPSS


spacing and reduce their number with corresponding savings,
and improve train performance see [ 1], [4] and [5] for details.
Many new transportation systems use [500 V trains. Since
all voltages are proportional to the rated train voltage, for the
[500 (=750*2) trains we can use the rated TCR voltage in the
range of 1650 to 1700 V (double the 825 to 850 V).
Consequently for 1500 V trains with similar assumptions as for
we arrive at 250 V voltage gain for TCRs, leading to similar
reduction in TPSS quantity.

978-1-4799-0688-8/13/$31. 00 2013 IEEE

2013 3rd International Conference on Electric Power and Energy Conversion Systems, Yildiz Technical University, Istanbul, Turkey, October 2-4, 2013

Additional advantage is fault current limiting by the TCR


for both close and far faults to 300 - 450%, depending on
design. This reduces stresses on the transformer and the
breakers, extending service life. The separate issue of output
capacitor transient fault current is discussed below.
III.

BASIC TRACTION RECTIFIER REQUIREMENTS

From my experience the traction applications require from all


types rectifiers the following:

Ability to withstand short circuits.

Nominal Power is I to 5 MW. Typical nominal


rectifier power is 3 MW. For 850 V nominal voltage
this corresponds to rated current of 3.5 kA, for 1700 V
nominal voltage rated current is 1.75 kA.

Up to 300 to 400% overload. For 3MW rectifier 400%


current translates into 14 kA DC for 850 V nominal
voltage and 7 kA for 1700 V.

Nominal DC train voltage of 750 or 1500 VDC.

Without output filter the TCR peak voltage exceeds 938


VDC for 12 pulse mode and 1000 VDC for 6 pulse for typical
TCR parameters. Increasing incoming AC line voltage by even
5% will bring peak voltage to 985 Vpk ( 12 pulse) or 1050 Vpk
(6 pulse). This is well above allowable train voltage 900 Vpk.
Solution to the TCR peak voltage problem is installing
large output filter capacitor to eliminate output ripples and
satisfy IEC 1287 requirements. We have the following
requirements for selecting the output capacitance:

It is possible to use both electrolytic and film capacitors,


but at 850 V and above the film capacitors are more practical:

For diode rectifiers the nominal output voltage is 750 or


1500 V, respectively.

For TCRs the nominal output voltage is 825 to 850 V for


750 V systems and 1650 to 1700 V for 1500 V systems.
IV.

A.

OUTPUT RIPPLES AND AC INPUT HARMONICS

Do we need an output filter?

The Standard IEC 1287 -I ( 1995) "Power converters


installed on board rolling stock" specifies allowable train
overvoltages as a function of pulse duration (see Figure 2).

The COnYarttH :.nall QPrallB norm any wl'll!n!! SubjeCloBL1 ID D'U'tI(\rQltagos glv9t'l In CU1\'8 1 Qf
figUf6 -4

,,'

r--:

l
10"'

g. If, 10-

FligtJro 4

ill

11110-"

III

..

1G....z

S 1104
.}--...

DC supply 5.slcm OVCrvOIiDglii IIIVQls

Figure 2.

, ' , . . ,,'
-

Allowable train overvoltage

The peak overvoltage should be less than 1.2 nominal


voltage for pulses with duration longer than I ms.
The higher TCR voltage (see above) allows us to increase
TPSS spacing and reduce their number with corresponding
savings, but this brings increased peak voltage problem:
increasing TCR DC voltage requires higher rectifier
transformer secondary AC voltage, so TCR peak voltage is
higher comparing to diode rectifier. Also, because of the
delayed firing required for voltage regulation, TCR has higher
output ripples than diode rectifier.

The resonance frequency between the transformer


inductance and filter capacity should be well
below the six harmonic of the mains (e.g. 360 Hz
for 60 Hz system) to avoid parallel resonance in 6
pulse operating mode and control circuits
oscillations.
The capacitor current rating should be below
actual ripple current.

Film capacitors have high voltage rating;


electrolytic capacitors are limited to 450 V, so
they would require series connection for 850 V.
Film capacitors have lower conduction losses, thus
they have higher current rating.
Film capacitors have service life at least 10 times
longer
than
electrolytic
one.
Electrolytic
capacitors are limited to few thousand hours
service life at maximum temperature, to extend
service life we need to de-rate the current rating
and provide good cooling.
Film capacitors are available in large sizes;
electrolytic capacitors are not. To satisfy voltage
and current requirements with electrolytic
capacitors we will need both parallel and series
connection of many small units.

For the above reasons the film capacitors are preferable for
traction TCR output filters. For a typical 3 MW, 850 V TCR
we will need about 20 mF to avoid problems with resonance.
At this capacitance level and 1000 V voltage, the film capacitor
current rating exceeds 1000 A, few times higher the actual
capacitor ripple current. Also the output ripples are reduced
well below the diode rectifier ripples. Conclusion: to utilize
the TCR advantages of increasing TPSS spacing and
improved train performance without jeopardizing train's
equipment, we need to use output filter, preferable parallel
film capacitor.
B.

Is capacitive filter safefor the circuit?

Somebody might claim that using a capacitor filter is very


dangerous because a short circuit will lead to an extremely
large fault current and will destroy everything in its path. In
reality the peak fault current

Ipk

1pk

= -.!!....

2013 3rd International Conference on Electric Power and Energy Conversion Systems, Yildiz Technical University, Istanbul, Turkey, October 2-4, 2013

where:

Vd is DC voltage at the time of fault


p
is a characteristic impedance,

L and C are total loop inductance and filter


capacitance, respectively

TCR substation ( 12 pulse, parallel bridges, capacitive filter, no


IPT) with network short circuit power of 125, 190 and 360
MVA (denoted as xl of 1.53, I and 0.63, respectively). These
are the actual system parameters for one of the TCRs
installations in the US.
Voltago THO at dlneronl sourco Impodancos

If we have a fault just at the feeder breaker terminals outside


the substation, the loop inductance will be still few uHy. Even
assuming it to be just 2 uHy (very close positive and negative
buses inside the TCR and throughout the substation) and using
an output capacitor of 20 mF, we get
p

JT
=

20 00

0.01
=

Ipk =

V,I
P

850
0.0 1

1
=

2nJLC

TCR

output

capacitor

doesn't

create

by loop inductance, is zero crossing and should be cleared


by traction DC breaker with no problem.

How big are TCR AC harmonics?

TCRs have higher harmonics contents than diode rectifiers


due to delayed firing required for voltage regulation. However
the voltage distortion for 12 pulse TCR is well below IEEE5 19
standard requirements of voltage THD (Total Harmonic
Distortion) to be below 5% with individual harmonics below
3%. Figure 3 shows simulation results for THD for 2.5 MW

Voltage distortion of 3 MW TCR with different system

We can see that 12 pulse TCR harmonics are well below


IEEE5 19 requirements (3% THO vs. 5% limit), even for a
weak system with 125 MVA short circuit power. We can see
even 6-pulse operation yields voltage THO below IEEE5 19
requirements of 5%. The distortion for stronger system is much
lower.
To verify the simulation, we measured the actual 12 pulse
TCR voltage and current harmonics and compared them with
simulated; the results are shown in Figure 4. We can see very
good correspondence between measured and simulated values.
The only difference is high value for the 5th and ih voltage
harmonics. The last two columns show system impedance
calculated from the measured voltage and current (ZsysCalc)
and theoretical system impedance based on system short circuit
power (ZsysThr). We can see the good correspondence
between ZsysCalc and ZsysThr for high order harmonics but
large difference for the 5th and ih harmonics, indicative for the
th
resonance around 5 harmonic. Further investigation revealed a
capacitor bank connected in parallel to the 13 kV line without
series inductors, apparently it was creating a parallel resonance.
Calculated and measured harmonics in

dangerous transient fault currents. The current is limited

C.

,,""

impedances

= 7.2kJ

Traction DC breaker can dissipate over 100 times higher


energy when interrupting 100 kA DC fault.
Conclusion:

Figure 3.

Even if we assume the fault circuit has high Q and the fault
current will not decay substantially by the time the feeder
breaker contacts separate, the peak current is well within DC
breaker interrupting capability (over 120 kA per ANSI C37. 14
and C37. 16) and it is an AC current, it is easier to interrupt than
DC. So the typical feeder DC breaker will have no problem
interrupting the fault current right at the feeder breaker
terminals. In a more practical situation with the fault occurring
at the tracks, the loop inductance will be above 50 uHy;
inductance increase by a factor of 25 will decrease the
capacitor current and its frequency by a factor of 5. Thus the
capacitor fault current will be only 17 kA with 160 Hz
frequency; it will be even easier to interrupt. Additional
argument is a small amount of energy stored in the TCR output
filter. At 850 V a 20 mF capacitor has energy
2
3
20 1 0- .850

-.....

'00'

795 Hz

-----

V-

----

85 kA

This current is actually decaying AC current with the


frequency of
f

V
/'

Assuming the TCR voltage Vd is 850 V, we arrive at the


peak current

N hann

IiTAACI

Idb
13.1

41.99

0. 19

CiA

Ihcal

41.02

-70.0

-67.0

11

-46.5

13

-53.5

13

-62.0

15

-60.1

0.59

-71.5

0.16

0.11

35

37

47

49

Figure 4.

-73. 9

-71.4
-74.3

0.2 7
1.84

1.27

0.00
0 00
1.95
1.11

0,48

0.57

0.11

0.10

0.16

0.13

0.12

0.55

0.05

12 pulse mode 100% load simulated at 113 Vin


V

Vdb

-23.6

-73.1
-73,4

-68.0

VlVACI

3673.9

CiV

Vhcal

4100.00

13.3

0.00

12.6

000

239

20.31

-73.7

12.4

901

-77.0

8.5

6.67

-80.0

6.0

-87.1

1.6

-61.7

4.4

-852

-64.6

6.16
1.18

1 (Ohm)

l sys cal lsys Thr

40.4

1.0

4.9

4.4

27.6

1.6

5.6

5.2

8.3

10.0

7.3

9.2

3.3

4.79

12.6

14.0

11.9

14.8

3.5

1.65

17.1

3 .93

15.7

16.6

19.6

Comparison between measured and calculated TCR harmonics

Figure 5 shows TCR voltage and current oscillograms at


67% load, we can see the voltage distortion is low.

2013 3rd International Conference on Electric Power and Energy Conversion Systems, Yildiz Technical University, Istanbul, Turkey, October 2-4, 2013

Assuming + I 0% over-voltage and customary 100% of SCR


voltage margin, we arrive at required SCR voltage of 4400 V.
There are plenty of high current SCRs available at voltage
above 4400, standard devices go above 6500 V. Conclusion:

T8k Prevu

Refl CycRMS
'13 V
Ref2 CycRMS
1.39 A
Refl-tRef2 P'ha
55.700
,+

:(:

__

....,J:-M135r.

.J...._. c.-- ....

Figure 5.

AW2rg'r-0!i!YJ

12 pulse TCR primary voltage and current at 67% load

The TCRs had been successfully operated at numerous


locations causing no problems with either harmonics or power
factor (in the US it is Dallas and Phoenix).
V.

A.

POWER SECTION CONSIDERATIONS

Should TCR use

or

12

C.

TCRs.

Parallel or series bridge connection

12-pulse system is combined of two 6-pulse bridges that


can be connected either in parallel or in series. With series
connection we do not have to worry about current sharing
between the bridges but the SCR losses are almost twice as big
as for parallel connection. The DC current has to go through
four SCRs for series bridge connection versus two SCRs for
parallel connection. Parallel connection requires the SCR
withstanding voltage to be twice as big and their forward
voltage drop might be higher, but the difference in forward
voltage drop is below 15% (one of the offsetting factor is lower
current, under parallel connection each SCR is carrying half the
current). So from the point losses reduction the parallel bridge
connection reduces the losses by almost 50%.

Do we need an 1PT?

Inter Phase Transformer (IPT) is connected between the


parallel rectifier groups, in our case between two bridges, to
reduce circulating currents. While the IPT are necessary for
single-way rectifiers (e.g. 3 phase double WYE circuit 45), the
12 pulse double-way systems have the circulating current
limited by the rectifier transformer impedance. Loosely
coupled standard Wey-Delta rectifier transformer sufficiently
limits the circulating current to eliminate the need for an IPT.
Successful operation TCRs without IPT (e.g. DART) proves
the IPTs are not needed.
Figure 7 shows the power circuit of 12-pulse reversible
TCR with parallel bridges connection, output capacitive filter
and no IPT. Conclusion: we do not need an IPT.
VI.

pulse mode?

12 pulse rectifiers allow us to eliminate half of the


harmonics and reduce total hannonic distortion (THD) by
about 30 to 40%, as we showed in the previous section. Since
the required DC current detennines total number of SCRs, the
difference between 6 and 12 pulse systems boils down to
marginal cost increase of the control system and about 5%
increase in the rectifier transfonner cost. The cost of power
filters to remove additional harmonics present for the 6-pulse
system is much higher. Conclusion: we should use 12 pulse
B.

we should use the parallel connection of the bridges.

POWER FACTOR AND REACTIVE POWER

The reactive power is phenomenon related to a phase shift


between AC voltage and current. The current component
shifted 90 degrees from voltage transfers no power but it
creates losses and it is a major source of AC voltage change. It
is customary to use power factor (a cosine of angle between
voltage and current) as a measure of reactive current. Though it
is adequate for steady loads, it is very misleading for wide
changing loads such as traction where the substation operates
for 15 seconds at 250%, then 70 seconds at 100% and then
draws 10% power for the rest of 300 seconds cycle.
The real negative effects of the reactive current are voltage
sags (proportional to the current) and additional losses
(proportional to the current square). If we have poor power
factor at low power level both effects are negligible. It is more
practical to use the reactive current directly or its product with
voltage (called reactive power) to estimate the negative effects.

A.

Comparing reactive power for diode rectifier and TCR

Lets neglect harmonics and assume for simplicity that all


losses are located on the DC side. First lets consider diode
rectifier with a 6% voltage regulation (industry standard). Lets
further assume the resistive component provides 2% regulation
and commutation inductance provides 4% regulation. Basic
rectifier equations are:

Additional advantage of parallel connection is redundancy:


if one of the bridges is inoperable (cooling failure, multiple
SCRs failure, etc.) the rectifier could continue to operate with
single bridge at reduced current (and higher harmonics). In a
rectifier with series bridge connection the failure of one bridge
means failure of the whole rectifier.

VdO = --

One might ask if we have to connect more that one SCR in


series to achieve required voltage ratings. For a rectifier with
1700 VDC nominal output voltage we will need about 1400
VAC incoming voltage (margins for firing angle delay, etc.).

Vd
=_
VdO
r
Vdout = Vd - Jd; Pdout = Vdout Jd

3J2
n

coscp

2013 3rd International Conference on Electric Power and Energy Conversion Systems, Yildiz Technical University, Istanbul, Turkey, October 2-4, 2013

where:

Pd, V;, I), Vd and I d are primary power, DC

power before taking losses, RMS values of primary voltage and


current, and DC voltage before losses and DC current.
VdO' cp, Vdout and Pdout are no load DC voltage, angle
between voltage and current, output DC voltage after losses,
and output DC power after losses, respectively. x and r are
inductive and resistive impedances, 4% and 2%, respectively.
From these equations we can calculate output voltage Vdout
and cos <p for any current and fmally calculate reactive power
Q as a function of output power Pdout.

QI

J3 ll sin rp = J3 ll 1- cos2 rp

[t is important to calculate the reactive power as a function


of output power (not the current) to compare different rectifiers
because TCR has higher voltage at the same current and trains
require power, not current.
1.Em .---..,...-----....
.
--,...
.,.
---.
..
1.400 f--+------___+-----.lF-
1.axJ
1.ooo f--+----------.1-+_

a O.ax)

O.Em t---+---<--____t--t______i

0.400 f--+-I_--___+--+_

---____t--t______i
O.axJ t-"t-:.,..
0.000 ""'-_I_-----___I--""_----'
0.0

Figure 6.

0.5

1.0

1.5

2.0

2.5

3.0

Relative reactive po.wer Q* as afunctin of output power for diode


rectifier and TCR

We can write similar equations for the TCR, the only


difference will be the equation for Vd

Vd
where:

IO cosa -x1d

is the firing angle.

The TCR changes firing angle to keep the output voltage


constant while the current changes from zero to inflection
point. Above the inflection point the firing angle stays at
minimum value (S to 10 deg) to assure proper SCR firing.

Vdout

Figure 6 shows the relative reactive power Q* as a function


of relative DC power p* for diode rectifier and TCR with
ISO% current inflection point

= -- ;

i rated

'"

iolll
i rated

where Pdrated is the rated output power (after the losses).


We can see on figure 6 that the reactive power at low loads
(below inflection point of ISO%) is lower for the diode
rectifiers due to the TCR delayed firing. But once we are above
ISO% the difference almost disappears. [t is understandable
because above inflection point the TCR operates with
minimum firing angle, almost like a diode rectifier. The

negative effects of reactive current (voltage change and


additional losses) are relevant only at high current where both
diode rectifiers and TCRs behave similarly. Conclusion: TCR
has higher reactive current (lower power factor) than diode
rectifier but the difference is insignificant in the high
current region.

V[1.

A.

ENERGY RECUPERATION

Why do we need energy recuperation?

All modem trains have regenerating capability and can


supply mechanical energy to the DC bus during braking; we
might get up to 80% of mechanical energy converted back to
DC power at train's terminals.
[f there is another train accelerating nearby it can absorb
the braking energy and reduce the overall system power
consumption. Theoretically we could capture almost all
recuperated energy if we would always have another train
accelerated at the same location and at the same time our train
is braking, but it is impossible to achieve in practice.
With short headway of 2.S-3 min during peak hours the
trains are close to each other and there are many of them.
Consequently, the braking energy is transferred efficiently; the
lost energy is small, 8% of total. With long headway of 30 min
during off hours the lost energy could be up to 60% of the
traction power supplied by traction substation [6, TableS].
Both simulation and actual operating experience in [6] show
the total energy savings of about 8 to %.
To utilize braking energy we can add recuperating
converters to traction substations and transfer the recuperated
energy to the power grid. Essentially we use the power grid as
a huge energy storage system [7]. Since the power grid always
has loads, it can easily absorb the recuperated braking energy.
Many states now require utilities to accept the power from
customers and bill them on the net energy (net metering).
Acceptance of reversible TPSS by power utilities depends on
their acceptance of distributed power generation. The issue is a
matter of policy, not so much of technology.
Since the power demand is typically measured as a
running average over IS minutes interval, recuperating
substation should cut the demand charges as well, even for
traction substation with individual metering because the time
between train braking and acceleration is typically 30 sec, well
below IS min. The power demand reduction would be even
more apparent for a system with a combined metering where
AC distribution lines belong to the transportation authority
(e.g. BART).

11

B.

How expensive is adding energy recuperation to TCR?

The RTCR is actually two thyristor controlled rectifiers


connected anti-parallel. By firing either one of the bridges
(forward or reverse) we can transmit the energy either from
AC to DC or in the opposite direction. Since we can choose to
fire only one bridge at a time we can eliminate circulating
currents without adding the inductors. Comparing to the
regular TCR the modifications are adding reverse bridge; there
is minimal addition to the control system to add reverse bridge
functionality. Assuming we can use the same transformer for

2013 3rd International Conference on Electric Power and Energy Conversion Systems, Yildiz Technical University, Istanbul, Turkey, October 2-4, 2013

both forward and reverse thyristor bridges, we can estimate


the transformer power increased by about lS% for the
additional current due to reverse power flow. We can use many
common components (bus structure, cooling fans, etc.)
between forward and reverse bridges.
Though the number of semiconductor switches is
doubled, the overall cost increase to add energy recuperation
of about 3S%. We can estimate cost of 3MW TCR plus
rectifier transformer with 400% maximum current to be
$440k, while 3 MW reversible TCR with transformer is $600k
[7]. Figure 7 shows power circuit of reversible TCR.
The harmonics and power factor in the recuperating mode
are the same as in the rectification mode. If we use 12 pulse
bridges for both power flow directions we keep the harmonics
below IEEES 19 limits for forward operation even at 400%.
The reverse current is 100%, so the harmonics in reverse
operation are lower than in the forward direction.
Running the power in reverse direction should not cause
any problems because the current is much lower.

Conclusion: we need to use forced air cooling.


B.

Traction Substation Cooling

Since the TCRs require forced air-cooling the air going


through the TCR must be clean.
In urban environment the air quality is often poor with high
amount of dust and other pollutants. We can try to use air
filters at the TCR cabinet and/or at the substation level but then
air filter changes becomes a major maintenance problem; in
Dallas (DART) they had to change filters weekly. A better
approach is to use close loop system: have redundant air
conditioners to cool the whole substation. This eliminates filter
changing maintenance problem and has additional benefits of
running all substation electronics and other equipment in a
clean environment at low temperature (e.g. 30C). Conclusion:
we need to use close loop system with redundant air
conditioners to cool TCR substation.

Reverse
Bridge

IX.

'-----+--'-++J-'----+--....-----...

Figure 7.

TCRs use forced air-cooling with the fans driven by variable


frequency drives to keep their speed low at lower load. This
way both fan noise and their service life are improved.

Power circuit of 12 pulse reversible TCR

Conclusion: adding energy recuperating to the TCR is

CONCLUSIONS

The TCR for traction applications should be built as 12


pulse systems with parallel connection of two 6 pulse bridges
and no IPT. Output capacitive filter is necessary to keep peak
train voltage within standard IEC 1287- 1 requirements. Though
both AC harmonics and reactive power of the TCRs are higher
than of a diode rectifiers, they are within the IEEES 19 standard
requirements and do not represent the problems. Adding
energy recuperation capability to the TCR is inexpensive and
can be justified on the energy savings. The traction TCRs
should use forced air cooling; TCR traction substations should
use redundant air conditioners.

inexpensive and can be justified on the energy savings [7].

VIII.

A.

ACKNOWLEDGEMENTS

COOLING

TCR cooling: Forced air or convectional?

Modem diode traction rectifiers customarily use


convectional air-cooling. This is practical because the diodes
for traction rectifiers have high operating junction temperature
of 160C through 190C. Assuming maximum ambient
temperature of 40C and a safety margin of lSoC, we arrive at
the
difference
between
junction
and
ambient
of
190-40-1S= 13SoC for 190C diode, for 160C diode the
difference is 10SoC. The TCRs use SCRs for rectification, their
maximum junction temperature of 12SC. With the same
assumptions we arrive at the difference between junction and
ambient for SCRs as only 12S-40- 1S=70C.
The lower junction-to-ambient temperature difference
means the thermal impedance junction to ambient must be 30
to SO% lower for the SCRs than for diodes (additional factor
here might be S to lS% higher losses in the SCRs). To decrease
thermal impedance SO% while operating at lower temperature
we need to at least double the number of heat sinl(S and
rectifier footprint; this would be very costly. In practice the

The author wishes to thank Moustapha Ouattara from


WMATA for helpful discussion.
REFERENCES
[1]

V. Gelman, "Braking Energy Recuperation (Reversible Thyristor


Contrlled rectifiers)". September 2009,IEEE Vehicular Technology

Magazine
[2]

The Standard IEC1287-1 (1995) "Power converters installed on board


rolling stock"

[3]

IEEE-519: IEEE Recommended Practices and Requirements for


Harmonic Control in Electric Power Systems

[4]

V. Gelman,S. SagareJi "Implementation Of New Technologies In


Traction Power Systems". Proceedings ofJRC 2004: 2004 ASME/IEEE

Joint Rail Conferences


[5]

V. Gelman, "Thyristor rectifier for traction - practical at last'".


Proceedings of 2002 APTA conference.

[6]

.l.M. ORTEGA et at." Kinetic energy recovery on railway systems with


feedback to the grid". 9th World Congress on Railway Research,May
22-26,2011

[7]

V. Gelman, "Comparison between wayside storage and reversible


thyristor controlled rectifiers (rtcr) for heavy rail applications".
Proceedings ofJRC 2013: 2013 ASME/IEEEJoint Rail Conference

Potrebbero piacerti anche