Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
I. INTRODUCTION
Manuscript received July 22, 2005; revised November 30, 2005. Recommended by Associate Editor M. G. Simoes.
The authors are with the School of Electrical Engineering and Computer Science, Seoul National University, Seoul 151-744, Korea (e-mail:
gpkim74@eepel.snu.ac.kr; gpkim74@empal.com).
Digital Object Identifier 10.1109/TPEL.2006.880260
1421
Fig. 1. Hybrid RTGC system with a supercapacitor as the energy storage device.
trolley-left hoist-
a container is first moved from the fourth layer and sixth column
to the first layer and first column and then moved back to the
starting point. These are the simulation results of a real 4 1
RTGC, where the term 4 1 means that there are five layers
and the one additional layer above the first layer is for horizontal
motion, with a half load (a 25-ton container). It is assumed that
a 120-kW GenSet supplies the average power requirement. In
is the power needed (or regenerated) in the
Fig. 3(a),
is the ou3mtput power of the supercapacRTGC operation,
is the power supplied by the GenSet. The difitor, and
and
is supference between the two powers
plied (or stored) by the supercapacitor. The maximum power
demand reaches up to 290 kW in the simulation. The power demand in the constant speed region of the first hoist-up operation
is 240 kW. In this case, the energy needed from the supercapacitor is 2.9 MJ. The energy needed for the worst case with full
load is 3.4 MJ.
1422
B. System Specification
From the above simulation results, the number of supercapacitor cell is selected around five hundred (several hundreds of
series and two parallel), on the basis that the maximum voltage
of each supercapacitor cell is 2.3 V, and its minimum voltage
is 0.76 V. As a consequence, the capacitance of the supercapacitor bank used in this study is several tens of Farad and the
equivalent series resistance (ESR) is around 100 m . The maximum stored energy becomes around 4 MJ. The total weight of
the supercapacitor bank is about 400 kg. A third of the original
GenSet is used to supply the average power demand. The total
dc link capacitance normally used in the inverter power stacks
for hoist/trolley/gantry motors is no less than 23 mF.
C. Topology of Bidirectional DCDC Converter
There can be lots of converter topology for realizing a bidirectional dcdc converter. This application requires handling
several hundred amperes of load current. The high current requirement is best met by using parallel converter modules [8].
The three-phase inverter stack as shown in Fig. 4 has many advantages. First of all, it is commercially available. Second, the
converter can be operated in the interleaved manner, for it has
three buckboost converter modules. Therefore, the current is
distributed among three converter modules. With an appropriate
phase shift among the switching sequence of converter modules,
three switching ripples cancel one another greatly reducing the
input and output current ripples. Third, the modular approach
provides the system with fault tolerance against the failure of a
single module. Finally the whole stack size including three inductors is smaller than that of a single buckboost converter.
The three paralleled inductors are made from metal powder.
Their inductances vary from 220 H to 110 H as the current
increases to 300 A. It is designed that the converter always operates in continuous conduction mode (CCM).
The bidirectional dcdc converter in Fig. 4 is operated on the
boost mode in which electric power is supplied from the supercapacitor stage (low voltage stage) to the dc link capacitor stage
(high voltage stage), and on buck mode in which electric power
is absorbed from the dc link capacitor stage to the supercapacitor stage.
Fig. 5. Simulation results of dcdc converter for regulating the dc link voltage
in case of discharging the supercapacitor.
where
is the internal voltage of the supercapacitor,
is
the terminal voltage of the supercapacitor, ESR is the equivais the output
lent series resistance of the supercapacitor, and
current of the supercapacitor.
III. CONTROL STRATEGIES
A. Load Sharing Between GenSet and Supercapacitor
To optimize the fuel usage of the GenSet, it is required to
control load sharing between the GenSet and the supercapacitor. To accomplish the optimum load sharing, that is, to control
the output power of generator, it is essential to know the information of the generating power of GenSet. In this paper, instead
of direct measurement of the power, an indirect control method
using the speed-droop characteristics of the GenSet is presented
to reduce complexity and the components.
B. Speed-Droop Characteristics of GenSet
Basically, a GenSet has two control loops; governor and automatic voltage regulator (AVR). The governor regulates the
engine rotational speed by adjusting the throttle angle. Engine
1423
Fig. 6. Speed-droop characteristics according to the output power of the generator: (a) steady-state characteristics and (b) transfer function between power
and frequency.
speed means the frequency of the output voltage. The AVR regulates the terminal voltage of the generator by the field current
control. In our application, the generator output voltage was set
to 480 Vrms and the frequency to 60 Hz.
Normally, a governor has the speed-droop characteristics
[10]. It reduces the reference speed as the output power of the
generator is increased as shown in Fig. 6(a). If the frequency
is known, then the generating power can be estimated. Some
experiments were performed to get the speed droop characteristics according to the output power as shown in Fig. 6(b). And
it has been modeled as following equation.
where
is the droop frequency and
is the power output of the generator. From the experimental results, the inverse gain of the transfer function is calculated as
43.75 kW/Hz and the time constant is done as
0.5 s
for our 120-kW GenSet.
A single PLL method, which needs the information of a line to
line voltage, is applied to estimate the frequency of three-phase
voltages.
where
2
is the natural frequency of APF and
is
selected as 60 Hz.
are transformed into the
These two voltages
synchronously rotating -frame to get the synchronous -axis
using the estimated angle . The generator frevoltage
quency is estimated using a PI regulator and a feed-forward
term as follows:
where
is the feed-forward term and
is the estimated
generator frequency.
This paper proposes the frequency adjustment algorithm for
APF to take the frequency variation into account. In other words,
according to the frequency variation of the generator, the natural
frequency of APF is updated as follows:
, and
1424
where
is the no load frequency of the generator and
is the estimated generator frequency.
Using the dynamics of the dc link capacitor, a PI voltage controller is also designed as shown in Fig. 10.
1425
Fig. 12. Energy storage system with the supercapacitor for a RTGC.
=0
voltage by 690 V. But it goes down to 635 V, because the supercapacitor current is saturated by the maximum reference current of 500 A limiting the output power of the supercapacitor.
Therefore, the generator automatically supplies the difference
between the power demand and the output power of the supercapacitor. Its frequency goes down until the end of the hoist-up
operation. After 12 s, the supercapacitor voltage goes down to
350 V. This activates the frequency conless than
trol mode. At the end of this period, the supercapacitor voltage
goes down to 275 V. When the crane moves the container in
the horizontal direction (region II: trolley operation), the trolley
motor does not need much power. Because it is operated in the
frequency control mode, the large amount of power from the
generator flows to the supercapacitor. After the supercapacitor
440 V, the voltage control
is charged more than
mode is activated. When the crane hoist down the container (region III), the power regenerated from the hoist motor flows to
and charges the dc link capacitor. And the voltage controller
regulates the dc link voltage to 690 V so that this regenerated
power flows to and charges the supercapacitor. When the supercapacitor is charged up to 565 V, it stops charging. And thus the
1426
Fig. 15. Experimental waveforms of the hybrid RTGC with a 10 ton load container; hoist-up, trolley, and hoist-down operations.
Fig. 16. Experimental waveforms of the hybrid RTGC with a 40 ton load container; hoist-up, trolley, and hoist-down operations.
TABLE I
FUEL CONSUMPTION COMPARISON
the auxiliary load such as fans, hydraulic pumps and lights for
one hour. Normally, the auxiliary load is less than 20 kW. The
CASE 3 means that the 50-kW RL load draws the power from
the generator for one hour. The RL load consists of the several
inductors and resistors. The CASE 4 and CASE 5 mean
that the generator supplies the power required for the typical
one cycle of operations; the container is first moved from the
4th layer and 6th column to the 4th layer and 1st column and
then moved back to the starting point in a similar manner to that
shown in Fig. 2. While the CASE 4 shows the results with
10-ton container, the CASE 5 does the results with 40-ton
container. The CASE 6 and CASE 7 mean that the generator supplies the power required for one cycle of operations
with 10-ton container. The CASE 6 shows the results of the
best case; the container is first moved from the first layer and
sixth column to the fourth layer and first column and then moved
back to the starting point in a similar manner to that shown in
Fig. 2. The CASE 7 shows the results of the worst case; the
container is first moved from the fourth layer and sixth column
to the first layer and first column and then moved back to the
starting point in a similar manner to that shown in Fig. 2.
From CASE 4 to CASE 7, there are regeneration processes.
Therefore, it is the key issue in these cases how much energy can
be recuperated. If the RTGC was automated, it would be easy to
find the optimum energy saving strategy. But the crane is operated by a human operator; the next operation is not known to the
1427