Beruflich Dokumente
Kultur Dokumente
4, OCTOBER 2001
I. INTRODUCTION
(1)
where:
where
, , and preferred-source line voltages;
, , and components of the preferred-
Fig. 3. Block diagram of the voltage detection scheme. source voltages in the rotating frame;
angular frequency of the rotating frame;
initial value of .
To offer ride-through capability, the load must be transferred The peak value of and is calculated as:
within the shortest possible time. Therefore, the STS must meet
the following requirements: (2)
• It must detect voltage fluctuations in the system as fast as
possible. The output of the transformation block, i.e., , is compared to
• In case the preferred source fails, it must perform a fast a dc reference, i.e., . The error is passed through a first-
load transfer to the alternate source. order low-pass filter which attenuates impact of voltage tran-
• The gating strategy, which controls the transfer process, sients. The filter introduces a delay to the error signal which is
must prevent paralleling the two sources. determined by the filter cut-off frequency ( ). The filter output
• Detection and transfer logic must function properly for all is then compared to a voltage-change tolerance limit ( ).
possible operating conditions. Output of the comparator is a transfer signal which initiates a
• Detection scheme must not be sensitive to temporary transfer process if the preferred source fails.
voltage transients, e.g., capacitor switchings. 2) Gating Strategy: The gating strategy is composed of
three identical sets of logic for the three phases of the system.
III. STS-1 BENCHMARK SYSTEM It provides selective gating patterns to thyristor switches which
results in a fast load transfer process and prevents source paral-
Fig. 2 shows a schematic diagram of STS-1 benchmark
leling. The selective gating strategy is based on the direction of
system which is composed of a power circuit and a control
line current flow. Fig. 4 shows the gating logic structure which
logic.
is composed of the following blocks:
A. Power Circuit 1) Current direction and zero-crossing detection logic
is responsible for detecting the status of the thyristor
Power circuit of STS-1 benchmark system is composed of switches, i.e., on/off state, and selecting the right switch
two 12 kV distribution feeders as preferred (p) and alternate (a) to trigger to prevent source paralleling during the transfer
sources. The sources are represented by ideal voltage sources in process. To avoid multiple zero-crossing detection, line
series with lumped resistances and inductances. The combina- currents are passed through a smoothing first-order
tion of the load and load transformer is connected to the sources low-pass filter. Output of the filter is an input to a hys-
through thyristor blocks and . teresis comparator which detects the direction of current
flow. The hysteresis comparator is used to make sure that
B. Control Logic current direction can be detected and the transfer process
Control logic of STS-1 is composed of a voltage detection can begin for all possible loading conditions. The hys-
and a gating strategy section. teresis comparator compares its input to a zero-current
1) Voltage Detection: Fig. 3 depicts a block diagram of the threshold limit ( ) and preserves its output status as
voltage detection scheme. The logic is based on transforming ac long as the line current is within .
726 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 16, NO. 4, OCTOBER 2001
Fig. 7. RL load, single-phase-to-ground fault. Fig. 8. RL + motor load, 35% three-phase under voltage disturbance.
second cage unsaturated reactance 0.105 p.u. Case 1: RL Load, Three-Phase Under Voltage: Fig. 6 shows
mechanical damping 0.008 p.u. a case in which a 35% three-phase under voltage occurs in
• Control circuit parameters: the system at ms when phase-a voltage angle is
kV ( ). The disturbance is detected at
voltage-change tolerance limit % ms [Fig. 6(b)] which results in a detection time of
filter cut-off frequency Hz 1.39 ms. Commutation in phase-b and phase-c is completed
line current smoothing filter cut-off frequency 1 kHz with no delay [Fig. 6(d) and (e)] but fails in phase-a [Fig. 6(c)].
zero-current threshold limit A At , phase-a current direction is positive and, therefore, is
thyristor turn-off time 1 ms gated. However, the voltage drop across the incoming thyristor
sampling rate 6660 Hz. is negative and it cannot be turned on. The disturbance
instant is such that the detection occurs at the zero crossing
D. Case Studies of phase-a voltage ( ) which results in maximum transfer
time. The control logic allows phase-a current to cross zero
Figs. 6–11 show simulated performance of STS-1 benchmark by re-gating . After 1 ms, i.e., at 8.49 ms, is forward
system. The simulations are performed by PSCAD/EMTDC biased and gated to successfully transfer phase-a to the alternate
package. Under voltage disturbances are created by reducing source. The maximum transfer time, which occurs in phase-a, is
the amplitude of the preferred source voltage. Faults are created 3.05 ms. The total load-transfer time is 4.44 ms. Load voltages
at the preferred source terminals. Fault resistance of 0.01 are shown in Fig. 6(a).
is considered. Definitions of detection, transfer and total Case 2: RL Load, Single-Phase-To-Ground Fault: Case 2
load-transfer times are as follows: presents the simulation results when phase-a of the preferred
detection time: The difference between the time at which source is subjected to a single-phase-to-ground fault. The wave-
a disturbance occurs and the time it is detected. forms are shown in Fig. 7. The fault occurs at ms
transfer time: The difference between the time at which a ( ) and is detected after 4.38 ms, i.e., at ms
disturbance is detected and the time at which the last faulty [Fig. 7(c)]. At this time, phase-a line current is negative, and
phase is transferred. the incoming thyristor is positively biased. Therefore, the
total load-transfer time: The sum of detection time and transfer process in this faulty phase is completed with almost
transfer time. no delay [Fig. 7(d)]. The transfer time is only the commutation
728 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 16, NO. 4, OCTOBER 2001
Fig. 9. RL + motor load, 80% three-phase under voltage disturbance. Fig. 10. RL + motor load, phase-to-phase fault.
time which is negligible, and the total load-transfer time is equal cannot conduct. Commutation in phase-b fails, and the load
to the detection time, i.e., 4.38 ms. becomes regenerative feeding the preferred source through .
Case 3: RL Motor Load, 35% Three-Phase Under Commutation cannot begin until the incoming thyristor is
Voltage: In this case, the preferred source is subjected to a positively biased. This occurs at the zero crossing of phase-b
35% three-phase under voltage disturbance when the load is voltage. The transfer is completed at 11.41 ms. Therefore, the
a combination of an RL load and a motor load. The system maximum transfer time (8.23 ms) occurs in phase-b. The total
behavior is shown in Fig. 8. This disturbance scenario is load-transfer time is 8.69 ms.
similar to Case 1 where only an RL load was employed. The Case 5: RL Motor Load, Phase-To-Phase Fault: Fig. 10
disturbance occurs at ms ( ) and is detected at shows the STS performance for the same regenerative load
ms (detection time of 1.39 ms). Phase-b and phase-c when a phase-to-phase fault occurs at ms between
are transferred to the alternate source as soon as the fault is phase-a and phase-b of the preferred source. It is detected
detected. Phase-a has to wait for the current zero-crossing at ms. The detection time is 2.85 ms. Commu-
which is reached at ms (including thyristor turn-off tation in both faulty phases fails. The transfer is completed
time). The transfer time and the total load-transfer time are at ms when both incoming thyristors ( for
3.25 ms and 4.64 ms, respectively. phase-a and for phase-b) are positively biased. The transfer
Case 4: RL Motor Load, 80% Three-Phase Under time and total load-transfer time are 5.83 ms and 8.68 ms,
Voltage: Fig. 9 shows the system behavior for a condition respectively.
where the load becomes regenerative. The system is subjected Case 6: RL+Motor Load, Cross Current Phenomenon: The
to a disturbance which results in a three-phase 80% under cross current phenomenon may occur when current direction
voltage. The load is the same as that of Case 3. The distur- is not correctly detected. This is due to the inaccuracy of cur-
bance occurs at ms ( ) and is detected at rent measurement devices when line currents are within .
ms corresponding to a detection time of 0.46 ms. Fig. 11 shows the impact of the cross current phenomenon on
Commutation in phase-a and phase-c begins as soon as the the STS performance. The zero-current threshold limit is
fault is detected, and these two phases are transferred to the set to 1% of the load nominal current, i.e. 4.8 A. Therefore, if
alternate source with no delays. However, in phase-b, due to the line current is within 4.8 A, the detected current directions
the opposite signs of line current direction and phase voltage may not be correct. Fig. 11(a) shows a phase-to-phase fault be-
polarity, the incoming thyristor is negatively biased and tween phase-a and phase-b at ms. As Fig. 11(c) shows,
TASKFORCE ON SIMULATION OF FACTS AND CUSTOM POWER CONTROLLERS: BENCHMARK SYSTEMS FOR DIGITAL COMPUTER SIMULATION 729
Fig. 11. RL + motor load, cross current phenomenon. Fig. 13. 35% three-phase under voltage disturbance (STS-2).
TABLE I
B. Case Studies STUDY RESULTS SUMMARY—STS-1 BENCHMARK SYSTEM