You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224588497

Bidirectional, High-Power DC Transformer

Article  in  IEEE Transactions on Power Delivery · November 2009


DOI: 10.1109/TPWRD.2009.2028600 · Source: IEEE Xplore

CITATIONS READS

190 1,168

1 author:

D. Jovcic
University of Aberdeen
122 PUBLICATIONS   4,028 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

PROMOTioN - Progress on Meshed HVDC Offshore Transmission Networks View project

All content following this page was uploaded by D. Jovcic on 26 January 2015.

The user has requested enhancement of the downloaded file.


1

Bi-directional, high-power DC transformer


D. Jovcic, Senior Member, IEEE

significant incentive to develop technology to provide


Abstract—This paper studies a bidirectional DC-DC converter additional access points to the existing HVDC lines. A suitable
concept which is capable of achieving very high stepping ratios MW size DC-DC converter would enable tapping on HVDC
with MW level power transfers. The converter can find potential lines and it would aid in development of multiterminal HVDC.
application in connecting high power DC sources, interfacing to In particular, the recent development of offshore renewable
HVDC (High Voltage DC) transmission or to FACTS (Flexible sources creates scenario of distributed DC sources and
AC transmission) elements. The converter is based on two
favorable conditions for submarine DC transmission, which
resonant circuits which share common AC capacitor. The
topology is simple and utilises thyristors with potentially all soft require DC voltage stepping at MW power levels. A high-
switchings. A complete analytical modelling is presented which power DC transformer would also aid in development of
enables systematic design procedure for the converter. The FACTS (Flexible AC transmission systems) technology, by
detailed digital simulation on PSCAD platform confirms enabling connection to a wide range of DC sources.
satisfactory operation on a 5MW test system, which connects 4kV The DC-DC converters have been extensively utilized at
DC source to 80kV high voltage DC grid. The converter shows low power levels and myriad of topologies exist. However,
good responses to rapid changes in power magnitude/direction most of these technologies are not suitable for scaling up to
and it is concluded that robustness to terminal voltage MW power levels. The limitations are linked to the nature of
disturbances is excellent. Since the highest-power phase-control
thyristors are employed, the converter can potentially be used at
the switches at highest power, the operating frequencies,
much higher power levels. The passive components are of efficiency, switch utilization and others.
reasonable size. The conventional, unidirectional boost converters [5] can
Index Terms—DC-DC power conversion, thyristor converters, not achieve gains larger than 2-4 or higher powers because of
High Voltage DC transmission. difficulties with the output diode. There have been attempts to
develop the flyback and forward converters [5-7] at higher
I. INTRODUCTION power levels; nevertheless some serious inherent limitation in
In the power range of tens and hundreds of MW there has terms of stepping ratios and power levels have been
not been much use of DC-DC converters because of the demonstrated [8]. Reference [1] studies scaling up to 5kW with
insufficient market need and the lack of suitable technology. stepping ratio of 5 and [2] describes a 100kW, 14kV forward
However, the market demand for DC-DC connection has converter. However these converters utilize MOSFETs as
significantly increased in recent years considering proliferation switches with around 10kHz frequency, which gives low
of power sources that generate DC [1,2]. The DC power prospect for further increasing to MW power levels.
sources that are approaching power levels of multiple MWs The switched capacitor converters have been proposed as a
include: fuel cells, photovoltaics, batteries and redox flow. method of achieving high DC boost without transformers or
Additionally, all variable speed machines (like permanent inductors [9]. On the downside, each module increases output
magnet wind generators or small hydro generators) may be voltage only by the value of the input voltage. To achieve
viewed as DC sources if the last converter stage is removed stepping ratio of say 10, 9 modules are needed and over 18
[3]. Furthermore, most of electrical storage and load leveling switches, which implies significant losses and complexity.
devices use a DC storage media (batteries, supercapacitors, The family of resonant converters has the potential for high
capacitors, superconducting magnetic energy storage etc). power development since some are capable of operating with
Many of these DC sources utilise very low voltage basic cells, thyristor switches [10]. Also, parallel resonant converters can
or require wide variation of DC voltage, and their integration achieve high step-up gains. The main limitations of these
into the power grid has traditionally been difficult. topologies are caused by increased switching losses, poor
The rapid development of HVDC (High Voltage DC) power quality, difficulties with power direction reversal, and
transmission technologies is also driving demand for DC-DC control difficulties.
converters. The recently developed HVDC light (HVDC with A new step up DC-DC converter has been proposed recently
Voltage Source Converters) [4] has already been implemented [11,12]. It has been demonstrated that this converter can
in dozen interconnections and it is being promoted as a very achieve very high step-up gains with a MW range test systems.
suitable solution for integration of renewable power sources. It however has no ability of step-down operation.
Virtually all current HVDC schemes (both line commutated A step-down DC power transfer and a power reversal may
and VSC HVDC) operate as two-terminal systems, but there is be required in many utility applications when connecting to a
high voltage DC grid and in particular with energy storage
Dragan Jovcic is with the School of Engineering, University of Aberdeen, applications. Various options for power reversal are applicable
Aberdeen, AB24 3UE, UK:d.jovcic@abdn.ac.uk
2

considering that DC-DC converters may be connected to Low voltage circuit High voltage circuit

voltage-source or current source systems. T1 T2 D5 D6


I1 I2
This paper develops further the converter concept from [11]
to enable step down operation and fast reversal in power L1
direction. It also studies different topologies where interface Vcr1 Ic Vc Vcr2L2
with DC terminals requires voltage or current reversal.
V1 Cr V2
II. UNIDIRECTIONAL HIGH POWER STEP UP CONVERTER T2 T1 D6 D5
This section gives only a summary of the basic step-up
Figure 1. Unidirectional step up converter (V1<V2).
converter concept from [11,12]. Figure 1 shows the topology
with four diodes in the high-voltage circuit. All switches
should have reverse blocking capability, but turn-off capability III. BIDIRECTIONAL CONVERTER WITH VOLTAGE POLARITY
CHANGE
is not required, and therefore thyristors are suitable. The
inductor L2 is not essential for operation however a small
A. Operating principles
inductor will reduce harmonics on I2 and reduce current
derivatives in the diodes. The capacitor Cr is rotated by Figure 2 shows the proposed bi-directional converter where
sequentially firing T1 and T2 pairs at 0.5 duty ratio. V1<V2. It consists of the low voltage circuit and the high-
The nominal switching frequency fs (fs=1/Ts) is firstly voltage circuit sharing a common energy storage capacitor Cr,
determined considering that the maximum switching frequency which is centrally grounded. The low voltage circuit includes a
fsmax is dependent on the nature of the switches: full thyristor bridge connected to the capacitor Cr on AC side,
and a resonant inductor L1 on the DC side. The high-voltage
circuit has a thyristor bridge with bidirectional valves (to
f s max = 1 /( 2Toff ) (1)
enable current direction change) and a resonant inductor L2.
All the converter voltages are symmetrical around central
where Toff is the turn-off (recovery) time for the thyristors. The point and therefore the rotating capacitor and each terminal are
main converter design equation is given as [11]: grounded at central point. This topology is convenient for
connecting to typical (VSC or LCC) converters and to a
I 2 (V2 − V1 ) bipolar HVDC. If other grounding arrangements are used than
= 2Cr f s , (2) insulated platforms may be required at some terminals.
V1V2
The low voltage circuit components, including the rotating
capacitor Cr, are selected using (1)-(3) [11]. A designer should
The above equation indicates controllability through frequency observe that (1)-(3) are derived for the converter in Figure 1
fs, and applies to both: continuous and discontinuous modes. assuming that the peak capacitor voltage is Vcp≈V2
Assuming that fs and the power/voltage levels are known (consequence of diode use and a very small inductor L2). With
(I2,V2,V1) then Cr can be determined from (2). The operation in a bidirectional converter one can presume higher value for V2
discontinuous mode, close to the border with continuous mode in (2) (20-30% higher is a good initial value), since larger L2
may offer overall advantages. With Cr known, the value for the is needed and this will imply higher capacitor peak voltages.
resonant inductor at the border of discontinuous mode (L1cr) is: This point is further discussed in section V.B. The low voltage

( )
circuit defines the operating frequency according to (1), which
L1cr = 1 / π 2 f s2Cr L1 ≤ L1cr (3) is common for the high-voltage circuit. The low voltage circuit
can reverse power direction if voltage V1 changes polarity.
The main advantages of the converter are: The high-voltage converter also operates as a resonant
converter with L1-Cr–L2 resonance. In step up mode, the
• The equation (2) shows very weak dependence on the gain
converter operates as the converter in Figure 1, and thyristors
(V2/V1), and the converter can achieve very high gains.
T5-T6 may be permanently gated. In this mode the inductor L2
• The high voltage circuit operates with all zero current
does not have much influence, but larger L2 reduces I2s
switchings and therefore losses/stresses are minimal.
derivatives, and increases peak capacitor voltage.
• It is suitable for MW-range power transfer because of the
The primary function of the inductor L2 is to provide
use of thyristors, and low switch stresses.
resonant turn off of T7-T8 in step-down operation, as it will be
• In discontinuous mode, all switchings are at zero current discussed in section III.B. The current Is2 in the high-voltage
implying minimal switching losses. The reverse recovery circuit is smaller than in the low voltage circuit, and this
losses are minimal and there is no need for snubbers. implies shorter conduction intervals. Consequently, there is a
Therefore only the conduction losses are applicable. freedom in choosing firing angle in the high-voltage circuit.
• In general, it is not vulnerable to commutation failure. The reversal of power is therefore achieved by changing
On the down side, the cost of silicon may be high since all current direction on high voltage side (I2) and changing
switches are rated for high voltage, and the harmonics will voltage polarity on the low voltage side (V1). Such converter
increase at lower powers, because of frequency control. The may find application as an interface between a current source
reliability may be lower than with AC transformers, but it will AC-DC converter (connecting to V1) and a high-voltage DC
be comparable with conventional HVDC converters.
3

grid (V2 connection), like with tapping on HVDC lines using circuits are active, and the corresponding current path is shown
line-commutated converters. in boldface in Figure 2. The firing instant αB (α2down) and the
The filters in Figure 2 are not essential and they will depend size of inductor L2 are the two design parameters. They are
on the power quality requirements. determined considering following:
Figure 3 shows the control system. It is based on the o Too early firing (at low αB) will imply low VcB and this
frequency regulation principle where the firing frequency may creates large potential across the inductor L2 causing increase
be synchronized with the capacitor voltage Vc using a Phase in dI2s/dt and peak value for capacitor voltage (Vcp),
Locked Loop (PLL). A single-phase PLL is required with o Too late firing will reduce the extinction time on T7 and T8.
magnitude compensation and the design in [13] may be (Toff7 in Figure 4). At point E switch T7 becomes forward
suitable. The PLL is not essential for the operation, but it may biased, and therefore it should regain blocking state in Toff7
improve stability at low operating frequency, when intervals interval. It should be ensured that Toff7>Toff, where Toff is the
between rotations are long, i.e., during a start up. The PI turn off time for thyristors in the high voltage converter
controller regulates current I1 (or some other variable) in (assumed same as for low voltage circuit).
feedback manner. The controller shown in Figure 3 also o Too small inductor L2 will imply large current derivatives
applies for converter in Figure 7, and therefore signals T3, T4 and current I2s peaks in the high-voltage circuit. The current
are neglected when considering converter in Figure 2 (signals derivatives should not exceed the maximum allowed switch
T1 and T2 are connected for both up and down). The gains kp values and also high peak currents imply high harmonics.
and ki can be determined using optimization on the simulator. o Too large inductor L2 will lead to long conduction intervals
T7 implying reduced extinction time Toff7 and large Vcp.
Lf1/2 L1/2 Low voltage circuit High voltage circuit L2/2 Lf2/2 The optimum parameters for the high voltage circuit are
I1 T1 T2 T5 T7 T6 T8 I calculated by developing a suitable analytical model with
I1s 2s I2
reference to Figures 2 and 4. For simplicity, the filters are
2Cr 2Cf2
2Cf1 neglected, and therefore I1s=I1, Vcf1=V1, I2s=I2, and Vcf2=V2.
Ic Also, assuming discontinuous mode, results in I1A=0.
V1 Vcr1 Vc Vcr2 V2
In the interval A-B, only the low voltage converter is
2Cf1 2Cf2 operating and there are two circuit equations:
2Cr
Lf1/2 L1/2 T2 T1 T6 T8 T5 T7 L2/2 Lf2/2 Vcp C D D’
filter fs fs α2up α2down filter
100 V2
E
controller B
α'D Vc
Figure 2. Bidirectional converter with V1 voltage polarity change (V1<V2).
Voltage Vc [kV]

50
αeC
- I1 PI controller αB=α2down F
I1ref
kp+ki/s up/down 0
+ V1 Vcr2
θ Vc, Vcr2 T7
integrator 360 T1 Toff7
+ 0 = T3
Vc
kpll + ω 1/s 0 comparator -50 Toff1
x T1
harmonic T2 T2
reset 180 = T4 Ts
elimination
= 360 comparator -100 A -3
x sin comparator x 10
Magnitude θ
Vcm 0 0.2 0.4 0.6 0.8 1 1.2
compensation T5 3
= T7
Single phase PLL
up/down comparator
α2up T6
2.5
+ = T8
α2down comparator
+180
Figure 3. Controller for the Bidirectional converter. 2
I1s I1s
Current I1 [kA]

B. Design of high-voltage circuit


1.5
In the step-up mode, the high-voltage thyristors T5 and T6
can be permanently fired to operate as diodes. In the step down
mode T5 and T6 are off and the reverse thyristors T7 and T8 are 1
fired on the rising slope of the voltage Vc. Figure 4 illustrates I2s
the converter key variables in step down operation in 0.5
discontinuous mode, for the test system given in the Appendix.
-3
The inductor L2 creates a resonant circuit with Cr, (L1-Cr-L2 x 10
0
resonance) and this resonance enables zero current turn off of 0 0.2 0.4 0.6 0.8 1 1.2
T7 and T8. This soft switching is fundamental advantage over t [s]
conventional boost/buck converters. In the interval B-C both Figure 4. Typical converter voltage and current waveforms.
4

dI1 and the constants for the joint circuit (B-C interval) are:
Lr = V1 − Vc (4)
dt
dV Ze = L2 ( L1 + L2 ) /(L1Cr ) , ωe = ( L1 + L2 ) /(L1L2Cr )
C r c = I1 (5) α eC = ωe (tC − tB ) (17)
dt
At the end of interval A-B the voltage and current are: In a practical converter there will be a small interval D-D’,
of zero current on the low voltage side (VcD=VcD’, I1D=I1D’),
VcB = −V1 − ( −V1 − VcA ) cos(α B ) (6) which is in the order of several microseconds. This interval is
desired since it will prevent simultaneous reverse recovery
I1B = (−V1 − VcA ) / Z1 sin(α B ) (7) current and high reverse voltage in thyristors and therefore
reverse recovery losses will be reduced. It is also mentioned
In the interval B-C, both circuits are conducting and three that point C may occur after point D with a large L2, but this
dynamic equations apply. case can be readily studied in the same manner.
The above static equations (6)-(7), (11)-(13) and (14)-(15)
dI1 can be used to design the converter. In these 7 equations, the
L1 = V1 − Vc (8) following 8 variables are unknown: VcA, VcB, I1B, I1C, VcC,, αB,
dt
αeC and α’D. For a given value of inductor L2, and the assumed
dV
Cr cr = I1 − I 2 (9) value of firing angle αB the remaining 7 variables can be
dt determined. The equations are however highly nonlinear with
dI trigonometric terms, and explicit solution is very challenging,
L2 2 = Vc − V2 (10) but numerical iterative methods can be readily used.
dt In a practical converter there will be an additional constraint
related to the turn off time for thyristors T7 and T8 (Toff7):
The initial values in point B are calculated in (6)-(7). The
point C is defined by the zero crossing of current I2. At the end
t E − tC > Toff (18)
of interval B-C:

The value of Toff=400µs (Silicon Power C784) is considered in


L1V2 − L2V1  L V − L1V2 
VcC = +  VcB + 2 1  cos( α eC ) + the test system as representative for highest power thyristors.
L1 + L2  L1 + L2  (11) The variables at point E can be determined using (4)-(5).
L1 I1 B Figure 5 shows the selection of L2, according to the above
+ Z e sin( α eC ) model, for the test system given in the Appendix. In this
L1 + L2 figure, the values for αB, VcA and dI2/dt are shown as the
L1I1B LI V +V α function of inductor L2 size and design trade-offs are evident.
I1C = + 2 1B cos(α eC ) − 1 2 eC −
L1 + L2 L1 + L2 L1 + L2 ωe If L2=0.3mH is selected, the maximum firing angle that will
(12) not violate (18) will be αB=127deg, the expected peak Vc
 L2V1 − L1V2  L voltage is Vcp=VcA=114kV, (V2=80kV) and the calculated I2
 − VcB  2 sin(α eC ) current derivative is dI2/dtmax=60A/µs (limit for C784
 L1 + L2  L1Z e thyristors is 100A/µs). In practice, some operating margin is
V +V α  L (V + V )  sin(αeC ) required and simulations show that the angle α2down=120deg is
0 = 1 2 eC +  − 2 1 2 + V2 −VcB  − appropriate. This firing angle (α2down) is considered constant in
L1 + L2 ωe  L1 + L2  Ze the proposed controller; however it can be manipulated for
(13)
LI LI further optimization of the converter performance.
− 1 1B + 1 1B cos(αeC )
L1 + L2 L1 + L2 C. Switch turn off study
The low voltage circuit thyristors T1 and T2, are turned off
In the interval C-D only the low voltage converter is either by I1 zero crossing (in discontinuous mode) or by firing
operating and same equations (4)-(5) apply. Point D is defined opposite thyristors (in continuous mode). It can be shown that
by I1 zero crossing and D’ is the firing of T2 at the beginning of there is no danger of shoot through, unless faulted controller
next cycle (VcD’= VcD= -VcA). In point D’, the equations are: issues asymmetrical firings [11]. The interval Toff1 (Toff1= tF-tD)
in Figure 4, should be larger than switch turn off time (Toff).
− VcA = −V1 − (−V1 − VcC ) cos(α 'D ) + Z1I1C sin(α 'D ) (14) This interval is fairly robust and constant, since it is primarily
dependent on the L1-Cr natural frequency: Toff1≈0.5π/ω1.
0 = I1C cos(α 'D ) + (−V1 − VcC ) / Z1 sin(α 'D ) (15) The high voltage thyristors (T7 and T8) turn off by I2 zero
crossing and they are more complex for analysis since they
In the above equations, the low voltage circuit constants are: operate in L1-Cr-L2 circuit. In interval B-C the time domain
equation for I2 is:
Z1 = L1 / Cr , ω1 = 1 /(L1C1 ) , α B = ω1t B , α 'D = ω1 (t D − tC ) (16)
5

(V2 + V1 ) V2 L1 − V1 L2  sin(ωet ) frequencies (ωe extremely small). In Figure 6, the shape of I2
I2 = t+ − VcB  for 50% reduced voltages V1 and V2 (terminal voltage
L1 + L2  L1 + L2  Z1 disturbances) are also shown, and these curves confirm that
(19)
 L (cos(ωet ) − 1)  switch turn off is not in danger.
+ I1 B  1  = I2x + I2 y + I2z
 L1 + L2  IV. BI-DIRECTIONAL CONVERTER WITH CURRENT POLARITY
CHANGE
The above three components of I2 are illustrated in Figure 6. In the applications where both connections have firm
It is evident that there is a linear component (I2x) which could voltages (V1 and V2) the power reversal requires current
threaten I2 zero crossings, however this is very unlikely in direction change on both sides. An example could be a voltage
practice. The studies of the parameters in (19) show that I2y source converter (VSC) of low DC voltage connecting to a
and I2z dominate for all reasonable parameters, and problems high-power DC network. To enable such interconnection, bi-
may arise only in the cases of extremely low characteristic directional switches are needed on both high-voltage and low-
voltage circuits. Figure 7 shows such converter.
140
The low voltage circuit operates as in Figure 1 and 2. With
120 positive current I1, T1 and T2 are alternatively fired, whereas
100
when negative current is required, T3 and T4 are fired. The
dI2/dt [A/us]

controller is shown in Figure 3. The high voltage circuit is


80 identical to that in Figure 2. The converter in Figure 7 can also
60 provide 4-quadrant operation at either terminal, since voltage
polarity can change in addition to current direction.
40

20 Lf1/2 L1/2 Low voltage circuit High voltage circuit L2/2 Lf2/2
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
119 I1 T3 T1 T4 T2 T5 T7 T6 T8 I
I1s 2s I2
118
2Cr 2Cf2
2Cf1
117 Ic
116 V1 Vcr1 Vc Vcr2 V2
VcA [kV]

115 2Cf1 2Cf2


2Cr
114
113
Lf1/2 L1/2 T4 T2 T3 T1 T6 T8 T5 T7 L2/2 Lf2/2
112 filter fs fs α2up α2down filter
111 controller
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
132 Figure 7. Bidirectional converter with current polarity change (V1<V2).
130
V. CONVERTER SIMULATION RESULTS
128
alpha [deg]

126 A. Test system


124
A 5MW converter connecting 4kV DC voltage to 80kV
Selected L2 transmission grid is considered as a test system. It may
122 represent an interface for a wind generator generating DC
120 (using an AC-DC converter) which is connecting to 80kV DC
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 collection network. The converter parameters are calculated as
L2 [mH]
Figure 5. Selection of L2 and high-voltage firing angle αB. described above and all values are given in Appendix, in table
A.1. A detailed model is developed on PSCAD/EMTDC [13].
3 The standard 4.4kV line-commutation thyristors with Toff=
I2y 400µs are employed, to illustrate possible application at
2 highest power levels. The fast turn off thyristors (Toff=100µs)
I2
Current I2 [kA]

I2(0.5V2) are available with lower ratings of up to 2.8kV and may also be
1 suitable for the studied 5MW converters. A detailed PSCAD
I2x model for the switches is used, (based on the Silicon Power
0 C784 thyristor), and the parameters are given in Table A.2.
B I2(0.5V1) C
The theoretical maximum operating frequency with the
-1 above thyristors is fsmax=1/(2Toff)=1250Hz, and consequently
I2z to allow some margin, the operating frequency of fs=1000Hz is
-2 selected. The operation in discontinuous mode is chosen.
-4
x 10 Because of practically zero switching losses, higher switching
-3
0 0.2 0.4 0.6 0.8 1 1.2 frequencies are possible, which bring advantages of smaller
time [s]
Figure 6. Components of current I2, and high voltage switch turn off study.
passive components (L1, Cr and L2). All the thyristors will have
6

some reverse recovery current which will be dependent on the voltage on the high-voltage terminals (V2). Therefore, in (2), it
slope of current decline at the zero crossings. However, the can be assumed that V2 is 20-30% above the actual value.
reverse recovery current will cause small losses since voltage
across thyristors is zero during reverse recovery [11]. 1.6 0.24
The low-voltage inductor L1 is crucial for the proposed 1.4 I1 0.2
converter in terms of impact on losses and weight. A

Current I1 [kA]

Current I2 [kA]
1.2 0.16
preliminary design of L1/2 and L2/2 inductors has been 1 0.12
performed and it is summarised in table A.3 and Figure A.1. It Iref
0.8 0.08
is concluded that a practical inductor is feasible and the 0.6 I2 0.04
expected L1/2 weight is around 1.8ton with total resistance of 0.4 0
around 4.7mΩ. The flux density can be further reduced by 0.2 -0.04
increasing air gap. An air-core design might be more 0 -0.08
appropriate at higher frequencies but it will have higher 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
number of turns and higher resistance. Since the capacitor Cr is time [s]
6 1600
the main energy storage component, the inductors in the 5 1400

Frequency fs [Hz]
4
proposed converters only provide resonant conditions and

Power P1 [MW]
3 P1 P2 1200
therefore they are smaller. The inductors L1 and L2 are 2 1000
1 fs
considerably smaller than a comparable (5MW, 50Hz, 0 800
80kV/4kV) AC transformer. These DC inductors will be of a -1 600
-2
similar design as smoothing reactors with HVDC systems. -3 400
-4 200
The filters (Lf1, Cf1, Lf2 and Cf2) are designed to allow -5
maximum 5% current ripple (on I1 and I2) and values are given -6 0
in Table A.1. Their size will depend on the application. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
time [s]
B. Converter with voltage polarity change 5 100
4
Voltage V1 [kV]

Voltage V2 [kV]
Figure 8 shows simulation responses of converter in Figure 3 80
2, which is subject to a series of excitation inputs. It is 2
1 V2 60
concluded that the converter:
0
• Follows very well the current reference (at 0.2s and 0.3s), -1 V1 40
• Responds well to significant disturbance on V1 (at 0.8s and -2
-3 20
0.9s) and V2 voltage (at 0.6s and 0.7s), -4
• Enables rapid change in power direction (at 0.4s and 0.5s) -5 0
During the power reversal (0.4s-0.5s) voltage V1 and current 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
I2 change polarity. It is evident that rapid reversal of power is time [s]
possible without any significant disturbance on other variables.
This might have importance for applications with energy Figure 8. Simulation results with bi-directional converter from Figure 2.
storage. It is seen that the converter readily starts in un-
energized state (Vc=0) and voltage ramping is simple. The 100 Vcr2
converter can also start in step-down mode. 80
60 Vcr1 V2
Voltage [kV]

Figure 9 shows detailed traces for the case of step-down 40


operation of this converter. The traces are in good agreement 20
with the theoretical results in Figures 4 and 5. Note that the 0
-20 V1
peak capacitor voltage is Vcp≈100kV, which is somewhat lower -40
than the value predicted in Figure 5. This lowering of Vcp -60
-80 Vc
results from the fact that the low voltage circuit is designed -100
using (1)-(3) which assume that VcA≈V2. In step-down mode 0.02 0.021 0.022 0.023 0.024 0.025 0.026
however, because of early firing of T7-T8, the following always
2.5 timeI1s[s]
applies: VcA>V2, and this implies that I1 increases beyond the
I1
nominal values considered in (1)-(3). Since I1 is regulated in 1.5
Current [kA]

feedback loop in the PSCAD model, the controller reduces


0.5 Ic
operating frequency as I1 increases, resulting in longer zero-
current intervals as seen in Figure 9. As the end result less -0.5 I2
energy is transferred to Cr leading to lower capacitor peak I2s
-1.5
voltage. To fully optimize converter at the final design stage, Ic
the low voltage circuit should be redesigned taking a higher -2.5
value of voltage V2 in (2) (V2≈100kV) and this would reduce 0.02 0.021 0.022 0.023 0.024 0.025 0.026
zero-current intervals. Further simulation test with different time [s]
converter ratings lead to the conclusion that the capacitor Figure 9. Detailed simulation results of bi-directional converter with voltage
voltage will typically have peak values of 20-30% higher than polarity change (step down mode).
7

C. Converter with current polarity change switches have V2 and I1 rating. Therefore the low voltage
Figure 10 shows simulation responses of converter in Figure converter needs higher rating of (V2/V1)xPrated where Prated is
7. It is concluded that this converter: the power transfer level. This increased rating will have cost
• Follows very well the current reference (at 0.2s and 0.3s), penalties, however since the lowest-cost switches (thyristors)
• Responds robustly to V1 and V2 voltage disturbance (at 0.6s, are employed the cost implications should be carefully studied.
0.7s, 0.8s and 0.9s) and D. Fault simulation
• Enables rapid change in power direction (at 0.4s and 0.5s), Figure 12 shows simulation of a positive V1 pole to neutral
by changing current I1 and I2 polarity. fault in step-down mode. All variables assume same labels but
It should be observed that the converters successfully subscript “p” is added for positive pole and “n” for negative
operate for significant reductions on terminal voltages (over pole. This fault results in power reduction to 50%, since the
20% sags on V1 or V2) as seen in Figures 10 and 8. This is current I1 is regulated to nominal value. One can observe
major advantage and indicates that the proposed converters are excellent tolerance to severe V1 faults and there are no
not vulnerable to commutation failure, despite the use of transient overvoltages/overcurrents. These simulations confirm
switches with no turn-off capability. In the proposed conclusions from III.C that the converter is not vulnerable to
converters, the commutation from one converter branch to commutation failures.
another is primarily dependent on the shape of the capacitor Figure 12 also indicates that symmetry of converter
voltage Vc, which has an AC waveform. This capacitor voltage variables is maintained despite unsymmetrical fault to ground.
is in the middle of the converter, and it is shielded from The converter possesses inherent feedback regulation which
external disturbances by an inductor on each side, (L1 and L2). reduces power input as V1 is depressed, as seen by the rapid
Figure 11 shows the efficiency obtained using PSCAD reduction in I2s, after the fault.
simulation, which includes the filter losses. The full-load
efficiency of over 95% can be expected. It is noted however
that PSCAD usually gives conservative values for efficiency.
Considering the measured switch stresses in Table A.4 in
the Appendix, it is concluded that the switches in high-voltage
circuit have V2 and I2 rating, whereas the low-voltage circuit

1.5 I1 Iref 0.15


1 0.1
Current I1 [kA]

Current I2 [kA]

0.5 0.05
Figure 11. Simulated efficiency. Step-up mode.

0 I2 0
880
-0.5 -0.05 V1p
V1 [kV], P1 [MW]

Frequency [Hz]
-1 -0.1 fs
0 870
-1.5 -0.15 V1n
-2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 860
P1
P1 time [s] -4
6 P2 1600
5 1400 -6 850
Frequency fs [Hz]

4
Power P1 [MW]

3 1200 0.035 0.036 0.037 0.038 0.039 0.04 0.041 0.042 0.043 0.044 0.045
2 1000 Vcp
1 60
Vcn
0 fs 800
-1 40
Voltage [kV]

-2 600
20
-3 400 Vcr2p
-4 200 0
-5 Vcr2n
-6 0 -20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -40

time [s] -60


5 100 0.035 0.036 0.037 0.038 0.039 0.04 0.041 0.042 0.043 0.044 0.045
90
4 80
Voltage V1 [kV]

Voltage V2 [kV]

0
70
Current [kA]

-0.5
3 V2 60 I2s
-1
V1 50 -1.5
2 40 -2
30 -2.5 I1s
1 20 -3
10 0.035 0.036 0.037 0.038 0.039 0.04 0.041 0.042 0.043 0.044 0.045
0 0
time [s]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
time [s] Figure 12. Simulation of a line-neutral fault on V1.

Figure 10. Simulation results with bi-directional converter from figure 7.


8

VI. CONCLUSIONS TABLE A.4 MEASURED SWITCH STRESSES FOR THE TEST SYSTEM
Thyristors Thyristors Max.values C784
This paper presents a bidirectional DC-DC converter which T1-T4 T5-T8 (Silicon Power)
can achieve very high stepping ratios and it is suitable for MW Vpeak [kV] 96 87 27x4.4=119
power transfers. Two converter topologies are studied: The Irms [kA] 1.25 0.063 1.65
topology with voltage polarity reversal would be suitable for dV/dt [V/µs] 10 4.4 1000
interfacing current source converters, and the topology with dI/dt [A/µs] 7 60 100
current direction reversal which would be suitable for
connecting voltage source converters. VIII. REFERENCES
The analytical study of step-down operation indicates that [1] D.K.Choi, at all. “A novel power conversion circuit for cost effective
the converter design is a trade off between high peak capacitor battery fuel cell hybrid system” Elsevier Journal of Power Sources,
Vol 152, (2005), pp 245-255.
voltages and high current derivatives. In a typical design the [2] L.Heinemann “Analysis and design of a modular, high power
resonant capacitor will have a 20-30% higher voltage than the converter with high efficiency for electrical power distribution
high voltage terminal. systems” IEEE PESC 2002,Volume: 2, Pages:713 - 718.
[3] D.Jovcic “Off Shore Wind Farm with a Series Multiterminal CSI
The detailed PSCAD digital simulations confirm successful HVDC” Electric Power Systems Research, Elsevier, Vol 78, issue 4,
operation with a 5MW, 4kV/80kV bi-directional DC-DC 2008, pp 747-755.
converter. The converter rapidly responds to power direction [4] Kjell Ericsson "Operational Experience of HVDC Light" Seventh
International Conference on AC-DC Power Transmission. IEE. 2001,
change and shows good robustness to terminal voltage pp.205-210. London, UK.
variations and faults. [5] N.Mohan, TM.Undeland, WP.Robbins, “Power Electronics
Converters, Applications and Design,” John Wiley & Sons, 1995
VII. APPENDIX TEST SYSTEM [6] RJ.Wai, RY Duan, “High step-up converter with coupled inductor”
IEEE Transactions on Power Electronics, vol 20, no 5, September
TABLE A.1 TEST SYSTEM PARAMETERS 2005, pp 1025-1035.
parameter value parameter value [7] Q.Zhao,F.C.Lee “High Efficiency, high step up DC-DC converters”
fs [Hz] 1000 L2 [mH] 0.3 IEEE Transactions on Power Electronics, Vol 18, no 1, Jan 2003, pp
V1 [kV] 4 Cf1 [µF] 200 65-73.
[8] K.Hirachi et al. “Circuit configuration of bi-directional DC/DC
V2 [kV] 80 Lf1 [mH] 5
converter specific for small scale load leveling system” Proc. IEE
I1av [A] 1,290 Cf2 [µF] 100 Power conversion conference, 2002, pp 603-609
I2av [A] 62 Lf2 [mH] 25 [9] O. Abutbul, et al “Step-up Switching Mode Converter With High
Cr [µF] 7.9 α2up [deg] 90 Voltage Gain Using a Switched-Capacitor Circuit” IEEE Transactions
Vcmax [kV] 110 α2down [deg] 112 On Circuit and Systems-I Vol. 50, no 8. August 2003, pp1098-2002.
L1 [mH] 12 [10] V.Ranganathan, P.D.Ziogas and V.Stefanovic “A regulated DC-DC
Voltage source converter using a high frequency link” IEEE
TABLE A.2 SWITCH DATA IN THE PSCAD MODEL (C784) Transactions on Industry applications, Vol 18, no3, May/June 1982,
pp 279-287.
T1-T8
[11] D.Jovcic “Step up DC-DC converter for MW-size applications” in
On resistance [mΩ] 27x0.37=10 print, IET Power Electronics, PEL-2008-0101, April 2008.
Off resistance [MΩ] 27x0.023=0.4 [12] D.Jovcic, “High gain DC transformer” UK patent office, PCT Patent
Voltage drop [V] 27x1.1=30 application no GB 0724369.4, December 2007.
Extinction time [µs] 400 [13] D.Jovcic, “Phase Locked Loop System for FACTS” IEEE
Forward/reverse voltage [kV] 119 Transactions on Power Syst, Vol 18, no 3, August 2003, pp 1116-24.
[14] Manitoba HVDC Research Center “PSCAD/EMTDC users’ manual,”
TABLE A.3 INDUCTOR DATA Winnipeg 2003.
L1/2 L2/2
Inductance L [mH] 6 0.15 IX. BIOGRAPHY
C Core size, length P [m] 0.6 0.22 Dragan Jovcic (SM’06, M’00, S’97) obtained a Diploma Engineer degree
C Core size, width E [m] 0.26 0.011 in Control Engineering from the University of Belgrade, Yugoslavia in 1993
Air gap 2xG [m] 2x0.1 2x0.01 and a Ph.D. degree in Electrical Engineering from the University of
Auckland, New Zealand in 1999.
Number of turns 5x122 2x84
He is currently a lecturer with the University of Aberdeen, Scotland where
Peak current [A] 2060 700 he has been since 2004. He also worked as a lecturer with University of
Steel mass [kg] 1107 14 Ulster, in period 2000-2004 and as a design Engineer in the New Zealand
Copper mass [kg] 702 72 power industry in period 1999-2000. His research interests lie in the areas of
Total resistance [mΩ] 4.7 0.48 FACTS, HVDC, control engineering and power electronics.
P
E
E

Core:
Material - laminated steel
G Relative permeability - 4000
P

G Maximum flux density - 1.5T


Coil:
Wire - 4 x AWG4/0
Fullness factor - 0.6

Fig. A.1 Inductor core dimensions.

View publication stats

You might also like