IEEE Power Electronics Magazine - June 2017 - 65

capability is required. For this case, an overall cost comparison of the MV stage considering those two different
semiconductors is shown in Figure 6. Despite the high cooling system required by the Fuji Electric power modules
[see Figure 5(a)], the semiconductors' price is considerably
lower, compensating for the high investment in the cooling
system. Therefore, the MV stage design using Fuji Electric
is more economically viable.
Due to the higher power level in case C, the semiconductor is better used than in cases A and B. On the other
hand, the amount of cooling is much higher, since the
power dissipation is also higher. For the LV-side and dc/dc
stages, the designs for cases B and C are the same, since
the amount of processed power and voltage remains the
same. As already mentioned, in case C, the saving cost
from the LV side and dc/dc is invested on the MV stage.
Considering the results obtained (illustrated in Figure 5),
the cost savings are shared among the cooling system and
capacitors, once the semiconductors of the MV remain the
same. For this reason, a reduction of 20% of power in the
LV and dc/dc stages allows an increase in MVA power of
around 100% with respect to case A. Thus, for the assumed
parameters, the MV stage can provide 2 MVA of apparent
power, keeping the same overall system cost of case A;
see Figure 7.

B, the power processed by the ST is lower than the case A
applications. Thus, the amount of reactive power injected in
the MV grid is lower, and no voltage support can be given
(refer to the green line in Figure 8). With the proposed design
approach in case C, the ST can be undersized in the LV and
dc/dc stages, and the MV converter can be increased up to
2 MVA, without increasing the transformer costs. However,
the benefits for the MV grid are clear. With higher reactive
power capability, the ST can sustain the voltage profile

Total Normalized Cost-MV Stage
Infineon IGBT Module
Fuji IGBT Module

1 p.u.

Case A

Case C

FIG 6 A cost comparison of the MV stage for different semiconductors' power modules.

A Simulation Case Study of the Proposed GTDA
Depending on the grid needs, the ST can provide different
services: local voltage support, voltage control in a specific
bus, and power factor control at the HV/MV substation busbar. The simplest service it can provide is the operation
under unity power factor. The LV-side converter produces
the reactive power for the LV grid, and the MV converter
can absorb only active power, reducing the reactive power
burden of the MV grid. The ST injects reactive power to
control the voltage at its busbar, or at a specific busbar in
the grid. In a specific case, it can control the power factor
at the HV/MV substation busbar, avoiding low power factor
conditions (i.e., below 0.9 p.u.). However, the reactive
power injection depends on the MV converter size and the
active power request in the LV side (both ac and dc). The
LV active power can be only partially regulated [6], but it
affects the power quality in the grid. Instead, the size of the
MV-side converter can be tailored to the MV grid to have
better control margins.
Figure 8 depicts a practical example of an experiment
described previously. A load flow simulation has been
performed on a modified IEEE 34-bus test feeder. The grid
voltage adopted is 10 kV, to match with the ST considered in
this article. The ST absorbs 700 kW of active power, and the
LV loads work with a power factor of 0.9 p.u. The ST injects
reactive power to its maximum capability to support the
voltage profile. If the ST is sized following the conventional
transformer or SST design strategy (case A), the amount
of reactive power injected is limited to 714 kVAR, which is
not sufficient to keep the voltage at about 0.95 p.u. For case

Case B

Sload = 1 MVA
3

ac
dc

MVac

1 MVA/1 MW

1 MW
(a)

LVac

1 MVA

Sload = 0.8 MVA
3

ac
dc

MVac

0.8 MVA

0.8 MW
(b)

LVac

0.8 MVA

Sload = 0.8 MVA
3
MVac

ac
dc

2 MVA/800 kW

0.8 MW
(c)

LVac

0.8 MVA

Standard Power Processing Design
Lower Power Processing/Cost Saving
Higher Power Capability/GTDA
FIG 7 A block diagram of the ST considering the three different design approaches, highlighting the power processed by
each stage: (a) case A, the standard design approach; (b) case
B, the standard design plus the LV services to reduce the load
consumption; and (c) case C, the proposed GTDA.

June 2017

z IEEE PowEr ELECtronICS MagazInE

Table of Contents for the Digital Edition of IEEE Power Electronics Magazine - June 2017