Professional Documents
Culture Documents
Voltage Stability of Converter-Interfaced Energy Storage Systems
Voltage Stability of Converter-Interfaced Energy Storage Systems
PCC vac
LPF2 vmac
Figure 4 shows a modified version of the well-known
Grid − WSCC 9-bus, 3-machine test system originally presented
+ in [Sauer and Pai (1998)].
y ref
vac 9
iac,abc
md
+ VSC VSC Control
− mq
y ref 3
− + ref
vdc vmdc vdc
LPF1 S 8
− + W
6
idc
u
Storage Control Storage Device
G G
2
uref 1
7
Fig. 1. Overall scheme of a converter-interfaced reactive 5 4
power source connected to the AC grid. Light gray:
STATCOM. Dark gray: CI-ESS. Fig. 4. Modified WSCC 9-bus, 3-machine system.
246
2019 IFAC CSGRES
224 Korea, June 10-12, 2019
Jeju, Álvaro Ortega et al. / IFAC PapersOnLine 52-4 (2019) 222–227
vbus 5 [pu(kV)]
vbus 6 [pu(kV)]
stability, this network is considered in this case study, 0.8 0.8
as the main transient effect of the BES is its ability to 0.6 0.6
exchange active power with the grid. As any other network,
however, the WSCC 9-bus system shows a maximum load- 0.4 No STATCOM 0.4 No STATCOM
With STATCOM With STATCOM
ing condition and, consequently, can face voltage collapse. 0.2 0.2
This network is thus as good as any other system to carry 1 1.5 2 2.5 3 3.5 4 1 1.5 2 2.5 3 3.5 4
out voltage stability analysis. psys/pbc psys/pbc
(a) Load at bus 5 (b) Load at bus 6
Since the original WSCC system is fairly symmetrical,
some modifications to the base-case data are introduced
with the aim of creating a weak area, i.e., the bus where 1
vbus 8 [pu(kV)]
the wind turbine is installed and neighbouring buses. The 0.8
modifications are as follows.
0.6
• The power base of the system has been reduced by 3
times, i.e., sn = 33.3 MVA. 0.4 No STATCOM
With STATCOM
• The√voltage of all transmission lines has been reduced 0.2
by 3 times. 1 1.5 2 2.5 3 3.5 4
psys/pbc
• The synchronous machine at bus 3 has been replaced
by a wind power plant of the same capacity. The wind (c) Load at bus 8
power plant is modelled as an aggregation of 18 fifth-
Fig. 5. pv curves at the load buses of the modified WSCC
order doubly-fed induction generators with optimal
9-bus system with and without a STATCOM. The x-
cubic MPPT approximation, first-order primary volt-
axis represents the ratio between the loading level of
age regulation and static turbine governor.
the system, psys , and the base-case condition, pbc .
• In time domain simulations, the wind follows a
stochastic, exponentially autocorrelated Weibull-dis-
tributed process modelled as a set of stochastic differ-
4.1 Long-Term Voltage Stability
ential equations [Zárate Miñano and Milano (2016)].
• Consistently with time scales and conventional as-
sumptions, loads are modelled as constant PQ for The pv curves of the three loads of the modified WSCC
the continuation power flow analysis and as constant system of Fig. 4 with and without a 20 MVAr STATCOM
impedances for time domain simulations. at bus 8 are shown in Fig. 5. Reactive power limits of the
• The voltage at the synchronous machine buses is generators are of ±42 and ±55 MVAr for the synchronous
reduced by 2% with respect to the base case. machines at buses 1 and 2, respectively, and of ±10 MVAr
• The system includes a secondary frequency controller for the wind power plant at bus 3.
(AGC) with gain Ko = 2.
In steady state, the active power output of a CI-ESS is
• When considered, the CI-ESS and the STATCOM
null, as generation and demand are balanced. Therefore,
device are connected to bus 8 through a 133/21 kV
the static models of a STATCOM device and a CI-ESS
transformer.
coincide. In particular, the STATCOM is modelled as
• The CI-ESS is a Battery Energy Storage (BES) mod-
a constant active power, with pac = 0, and constant
elled using the well-known Sheperd’s model [Shepherd
voltage if the reactive power is within the limits, and as
(1965); Ortega and Milano (2016)].
a constant active and reactive power with pac = 0 and
• The active power rating of the BES is 3.2 MW.
qac = ±20 MVAr, otherwise.
• The CI-ESS active power control is designed to reg-
ulate the frequency at the point of common coupling Figure 5 shows that the STATCOM allows increasing the
(PCC) with the grid, i.e., the frequency at bus 8. The loading level of the system up to 3.7 times the base-case
frequency is estimated by means of a phase-locked conditions, as opposed to maximum ratio of 3.47 of the
loop (PLL) device [Cole (2010)]. case without STATCOM, i.e., a 6.56% relative increase.
The STATCOM also maintains the voltage of the load at
Two main scenarios are considered. First, Subsection 4.1
the PCC (i.e., bus 8) almost at a constant value up to near
discusses the long-term voltage stability of the WSCC
the point of collapse.
system using both static and dynamic models of the
power system devices and controllers. Then, the short- The impact of the STATCOM on the voltage stability
term voltage stability of the dynamic model of the WSCC is more relevant in the event of the outage of one of
system is studied through time domain simulations in the transmission lines, as represented in Fig. 6, where
Subsection 4.2. the pv curves are computed considering that the line
connecting buses 4 and 5 has been disconnected. After the
All simulations included in the paper are obtained using
line outage, the maximum ratio psys /pbc decreases to 2.64
the Python-based software tool Dome [Milano (2013)]. The
and 2.86 without and with the STATCOM, respectively,
Dome version utilised is based on Ubuntu Linux 18.04,
i.e., a 8.24% relative increase. The outage of the line
Python 3.6.7, cvxopt 1.2.2, klu 1.3.8, and magma 2.2.0.
implies approximately a 24% reduction with respect to the
fully connected system. Similar conclusions can be drawn
considering voltage-dependent load models.
247
2019 IFAC CSGRES
Jeju, Korea, June 10-12, 2019 Álvaro Ortega et al. / IFAC PapersOnLine 52-4 (2019) 222–227 225
1 1 5 5
vbus 5 [pu(kV)]
vbus 6 [pu(kV)]
0.8 0.8
0.6 0.6
Im
Im
0 0
0.4 No STATCOM 0.4 No STATCOM
With STATCOM With STATCOM
0.2 0.2 −5 −5
1 1.5 2 2.5 3 1 1.5 2 2.5 3 −0.2 −0.1 0 −0.2 −0.1 0
psys/pbc psys/pbc Re Re
(a) Load at bus 5 (b) Load at bus 6 (a) WSCC system with (b) WSCC system with
BES. psys /pbc = 1.673 BES. psys /pbc = 2.5
1
vbus 8 [pu(kV)]
0.8 5
0.6
Im
0.4 No STATCOM 0
With STATCOM
0.2
1 1.5 2 2.5 3
psys/pbc −5
−0.2 −0.1 0
(c) Load at bus 8 Re
(c) WSCC system with
Fig. 6. pv curves at the three load buses of the modified STATCOM. psys /pbc = 2.5
WSCC system with and without a STATCOM. The
line between buses 4 and 5 is disconnected. Fig. 8. Eigenvalue loci of the rightmost eigenvalues of the
WSCC system with a BES and a STATCOM. Dash-
5 5 dot lines indicate 5% damping.
Im
0 0
20 MVAr.
In Fig. 8(a), the inclusion of a BES system shifts the
−5 −5
−0.2 −0.1 0 −0.2 −0.1 0 eigenvalues that caused the Hopf bifurcation in the system
Re Re to a value of −0.396 ± j0.924. The 6 rightmost eigenvalues
(a) WSCC system without (b) WSCC system without are in this case within the 5% damping thresholds, being
BES. psys /pbc = 1.0 BES. psys /pbc = 1.673 the eigenvalue #7, which is related to the rotor speeds of
the two synchronous machines, the first one that shows
Fig. 7. Eigenvalue loci of the rightmost eigenvalues of the poorly damped behaviour (−0.288 ± j10.59). This result
WSCC system. Dash-dot lines indicate 5% damping. holds even for larger loading levels, as shown in Fig. 8(b)
A second long-term, small-signal voltage stability analysis for psys /pbc = 2.5. In this case, the aforementioned poorly
is presented below which considers dynamic models of the damped eigenvalues are −0.249 ± j10.76. The real part of
power system devices and controllers of the WSCC system. such eigenvalues is still about an order of magnitude larger
For this study, the increasing loading level of the system than the dominant modes of the system.
is compensated by the two synchronous machines of the Finally, from the voltage stability point of view, no rel-
system, as it is assumed that no more active and reactive evant differences can be observed between installing a
power can be extracted from the wind power plant. BES or a STATCOM device, as observed from comparing
Figure 7 shows the rightmost eigenvalues of the WSCC Figs. 8(b) and 8(c). This is an expected result due to
system for two loading levels: (i) the base-case condition, the similar configuration of the reactive power support
and (ii) 1.673 times pbc . It can be seen that, for psys /pbc ≥ of both devices. This result also indicates that the active
1.673, a complex pair of eigenvalues, which is related to power support of the BES does not compromise the voltage
the states er,q and ve of the synchronous machine at stability of the overall system for relatively large loading
bus 1, crosses the imaginary axis, thus becoming unstable levels.
modes. The Hopf bifurcation Seydel (2010) occurring when
psys /pbc = 1.673 prevents the operation of the system at 4.2 Short-Term Voltage Stability
higher loading levels which, from the static analysis above,
could be up to 3.47.
This section studies, through time domain simulations,
The installation of additional sources of reactive power the short-term dynamic response of the modified WSCC
such as STATCOM and BES devices allows increasing the system with inclusion of CI-ESSs. Two scenarios are pre-
loading level of the system beyond the critical value of sented to analyse both small- and large-disturbance volt-
1.673 times the base-case condition, as these devices shift age stabilities. The reactive power rate of the CI-ESS and
to the left the pair of eigenvalues that cause the Hopf the STATCOM device used in the comparison is 3.2 MVAr.
bifurcation. This is shown in Fig. 8 for the scenarios where This value is chosen with the aim of analysing the impact
248
2019 IFAC CSGRES
226 Korea, June 10-12, 2019
Jeju, Álvaro Ortega et al. / IFAC PapersOnLine 52-4 (2019) 222–227
vbus 8 [pu(kV)]
WSCC system + STATCOM
Small disturbances are modelled as stochastic wind pertur- 1.058
bations that follow a Weibull distribution with exponential 1.056
autocorrelation. The response of the WSCC system fol- 1.054
lowing such perturbations is shown in Fig. 9. Figure 9(a) 1.052
depicts the voltage at bus 8, i.e., the PCC with the CI-ESS 1.05
or the STATCOM device. Wind perturbations create sus-
tained voltage oscillations which, despite having relatively 0 20 40 60 80 100 120 140
Time [s]
small magnitude, can deteriorate certain components of
the load connected at the bus, and/or compromise their (a) Voltage at the PCC
operation.
1.066
WSCC system
The average value of the voltage varies substantially with 1.064 WSCC system + BES
respect to the operating condition during the simulation.
vbus 6 [pu(kV)]
WSCC system + STATCOM
The inclusion of a CI-ESS or a STATCOM in the system 1.062
allows reducing to a great extent the voltage oscillations, 1.06
and keeps the average voltage close to its reference value.
While both the CI-ESS and the STATCOM device show 1.058
similar performance, the ability of the CI-ESS to also 1.056
provide active power control leads to a more efficient 0 20 40 60 80 100 120 140
voltage regulation with respect to the STATCOM, as it Time [s]
requires less reactive power support and proved better (b) Voltage at the load bus 6
voltage response (see Fig. 9(c)).
Finally, despite the local nature of the voltage control BES
from both devices, their effect can be observed also at 0.01 STATCOM
qvsc [pu(MVAr)]
249
2019 IFAC CSGRES
Jeju, Korea, June 10-12, 2019 Álvaro Ortega et al. / IFAC PapersOnLine 52-4 (2019) 222–227 227
0.02
0.01 power system analysis. In Proceedings of the IEEE PES
0 General Meeting.
−0.01 Ortega, Á. and Milano, F. (2015). Design of a control
BES limiter to improve the dynamic response of energy
−0.02
STATCOM storage systems. In Proceedings of the IEEE PES
−0.03 General Meeting.
0 2.5 5 7.5 10 12.5 15 17.5 20
Time [s] Ortega, Á. and Milano, F. (2016). Generalized model of
(c) Reactive power output of the VSC VSC-based energy storage systems for transient stability
analysis. IEEE Trans. on Power Systems, 31(5), 3369–
Fig. 10. Response of the WSCC system following a line 3380.
outage. Sauer, P.W. and Pai, M.A. (1998). Power System Dynam-
Future work will focus on the study of the stability of CI- ics and Stability. Prentice Hall, Upper Saddle River,
ESSs included in microgrids. Low X/R ratios of microgrids NJ.
feeder lines imply strong couplings between voltage and Seydel, R. (2010). Practical Bifurcation and Stability
frequency. The capability of CI-ESSs to simultaneously Analysis. Springer Science & Business Media, New York,
regulate the frequency and voltage at the PCC in a very US, third edition.
short time frame makes these devices excellent candidates Shepherd, C.M. (1965). Design of primary and secondary
to cope with such couplings. cells II. An equation describing battery discharge. Jour-
nal of the Electrochemical Society, 112(7), 657–664.
Western Power Reserve Work Group (RRWG) (1998).
REFERENCES
Voltage stability criteria, undervoltage load shedding
Arabi, S. and Kundur, P. (1996). A versatile FACTS device strategy and reactive power reserve monitoring method-
model for powerflow and stability simulations. IEEE ology. Technical report, Western Electricity Coordinat-
Transactions on Power Systems, 11(4), 1944–1950. ing Council (WECC).
Cañizares, C., Corsi, S., and Pozzi, M. (1999). Modeling Xin, H., Huang, L., Zhang, L., Wang, Z., and Hu, J.
and implementation of TCR and VSI based FACTS (2016). Synchronous instability mechanism of P-f droop-
controllers. Internal report, ENEL and Politecnico di controlled voltage source converter caused by current
Milano. saturation. IEEE Transactions on Power Systems,
Cañizares, C.A. (2000). Power flow and transient stabil- 31(6), 5206–5207.
ity models of FACTS controllers for voltage and angle Yazdani, A. and Iravani, R. (2010). Voltage-Sourced
stability studies. In 2000 IEEE Power Engineering Soci- Converters in Power Systems – Modeling, Control and
ety Winter Meeting. Conference Proceedings, volume 2, Applications. Wiley-IEEE Press.
1447–1454 vol.2. Zárate Miñano, R. and Milano, F. (2016). Construction
Chakravorty, M. and Patra, S. (2016). Voltage stability of SDE-based wind speed models with exponentially
analysis using conventional methods. In 2016 Inter- decaying autocorrelation. Renewable Energy, 94, 186
national Conference on Signal Processing, Communica- – 196.
tion, Power and Embedded System (SCOPES), 496–501.
250