Sunteți pe pagina 1din 8

Topic #8: Synchrophasors, applications and benefits

Synchrophasor Technology for Cascading Failure Effect Prevention


(1)

(2)

Jabil Circuit, Inc., St. Petersburg, FL, USA,

(2)

(3)

(3)

Zoran Mihailovic , Lakshan Piyasinghe , Yin Li , Lingling Fan , Zhixin Miao


(1)

(3)

Aclara Technologies, Hazelwood, MO, USA,

(3)

University of South Florida, Tampa, FL, USA


zoran_mihailovic@jabil.com

Abstract
Power generation, transportation and distribution (GT&D) community interest in synchrophasors has
emerged after several spectacular blackouts in the USA during the first decade of the new century.
Since then, syncrophasor technology, which core consists of a system of time-synchronized phasor
measurement units (PMUs), is one of the main features of utilities SCADA control modernization.
FPGA-based digital control, reliable global positioning system (GPS) and new communication
protocols enable fast execution of complex algorithms and fast transfer of synchronized data over
large distances. On the other side, modern protection systems help power consumers protect their
assets from irregular power grid activities by fast Zone 3 relay disconnects as the last line of defense.
These two systems work seemingly separately today, but their interaction cannot be neglected,
especially power grid dynamics after large power load disconnections. It is noticeable that most of the
time a local power grid problem exponentially grows to a large area blackout as a result of an
uncontrollable cascading avalanche effect caused by power grid instability and subsequent islanding,
augmented heavily by fast Zone 3-type protection relay disconnects of heavy power loads. This work
explains how synchrophasor technology could be used for smooth, preventive, control of dynamic
power load shedding. It is based on the idea that proper warnings, in the form of scalable digital
signals, and adequate time to react, could be given to consumers to gradually reduce their power
consumption ahead of time, when power grid problems are still to be maintained at a relatively safe
distance. The goal is to prevent both, Zone 3-type relay disconnects, and utility-forced power
shedding and subsequent time lost on gradual power reconnection. A simplified model of cascading
power grid failure pattern and predictive control-based algorithm have been used to present how
control of consumers power electronics (PE) and intelligent electronic devices (IEDs) could be used
to prevent these failures based on information obtained from a PMU network. Control dynamics and
stability concerns related to interactions between utility and consumer controls and two separate
consumer controls coupled through the grid nodes are addressed using master-slave communication
logic and inner-outer control loop configuration. PMU/PDC/Control Room/PE real-time communication
network has been established using IEEE C37.118 protocol. The focus at this stage is on a proof of
concept using simulated network combined with several geographically remote (within 20 mile radius)
power electronics/motor drive loads and PMUs. The following steps would be to adjust the control
algorithm to medium power level (50kVA-1MVA) loads and DERs, which would be simulated using
HIL testers and power grid simulators.

Introduction
Power blackouts cost millions of dollars in damaged equipment, power restoration and lost industry
operation time worldwide, every year. One of the most memorable and most expensive was the USCanada Northeast blackout in August 2003. NERC Commission forensic investigation report, [1], was
a turning point for US Department of Energy (DOE) to launch several programs and substantially
increase funding for power grid modernization research. Beside SCADA system automation,
protection and communication protocol breakthroughs, [2] and [3], non-traditional power grid
management using synchrophasor technology, [4] and [5], took the spotlight, due to its over ten-year
long history of successful deployment in wide area monitoring system (WAMS) data collection R&D
efforts, used mostly for system modeling and forensic analysis. Advances in microprocessor speed,
performance and memory size, as well as Ethernet and GPS data communications, were the
backbone for both of these developments. Other industrialized countries on all continents, but mostly
in Europe, Latin America, India and China, have followed the path [6].

Many control algorithms for power grid oscillation prevention have been developed and analyzed in
various power system disturbance situations. High impact of the nature of distributed energy
resources (DER) and their dynamics to power grid was also one of major R&D topics. In the looming
era of Big Data, enormous data storage centers for future IoT, sophisticated power supply units
(PSU) for medium and high power, HVAC and other large distributed electric motor loads, and
distributed energy management systems (DEMS), among others, synchrophasor technology has an
opportunity to become a key contributor for faster and more robust power grid system control due to
its measurement data accuracy, fast data transfer and time-synchronization core, [4] and [5], capable
to adapt to the challenges of new bi-directional power distribution.
Well-established power system control [7], featuring system oscillation prevention and reactive power
control using flexible AC transmission system (FACTS) devices, such as STATCOM, SVC, PSS and
LVR [8], as well as undervoltage load shedding techniques [9], has the opportunity to be augmented
with new control design options coming from synchrophasor implementation in new SCADA systems,
such as remote feedback load control analyzed in [10].
Simultaneously, power consumption control, from mass-production manufacturing to large motor drive
facilities or data centers, has evolved into complex electric power demand-response (DR) and
intelligent motor control centers (IMCC) using advanced industrial SCADA and energy consumption
control software to optimize their performances and energy consumption [11]. Voltage stability
demands on the distribution side of the substation transformers, so-called grid edge, become more
and more stringent, requiring distributed deployment of controllable active power filters, some of them
listed as FACTS devices above only at a distribution level voltage - static synchronous compensators
(STATCOMs), unified power flow controllers (UPFCs) and line voltage regulators (LVRs) [12].
One of the above-mentioned control design options is model based predictive control (MPC),
relatively recently introduced to the field of electrical engineering from the field of process control [13].
Afore-mentioned fast microprocessors and well-established knowledge of power supply and motor
drive systems, allow calculation-loaded model-based predictive control (MPC) algorithms to expand
from its relatively slow (time constants in seconds, minutes and above) process control environment
to fast (time constants in milliseconds and below) power electronics and motor drive control arena,
challenging conventional field-oriented PID PWM/SVM controls in very competitive applications [14].

Motivation
Nevertheless power utility companies took the responsibility for the 2003 US-Canada Northeast
Blackout [1], a stark dissonance between utilities and large industrial consumer controls was evident
before and during the sweeping cascading power failure avalanche throughout the region. It is worth
to notice that in the critical time these two protection systems have worked practically in opposition to
each other. While utilities have tried to navigate energy flow, control power generation and enforce
selective load shedding in order to stabilize power grid, large industrial customers continuous
operation, unaware of regional power grid behavior, have caused two unfavorable developments:
overloading surviving power lines and Zone-3 relay under-voltage protection tripping. Both types of
events brought more instability to the grid due to further load and power line disconnection transients.
Lack of communication and situational awareness has caused most of the systems respond only to
their imminent threats, except for attempts of manual control in RTO control centers. As a
consequence, it took only eight minutes for the cascading failure avalanche to spread over the entire
Northeast from a local, but highly interconnected power network around Akron, OH.
There is still a communication and control automation gap between these two systems despite system
protection, automation and communication breakthrough like generic object oriented substation event
(GOOSE), system integrity protection schemes (SIPS), and distribution automation (DA).
On one side, for power generation, transmission and distribution (GT&D) companies, so-called
utilities, power grid is regarded as a stiff load, which demands have to be satisfied and dynamics
suppressed by controlling power generation and energy flow through the system. Power consumer
network has been used as ballast for selective load shedding in critical situations. On the other side,
for power consumers, the power grid has been considered a stiff source, which eventual out-of-limit
perturbations should be prevented from damaging consumers equipment. Consequently, this
prevention would eventually trigger the same Zone 1, 2 or 3 on-off relays used for load shedding.
Although they can turn on and off relatively fast on a triggering command, the automatic reclosing

dynamics of the reengage-capable relays is slow in comparison to power electronics semiconductor


switches due to damaging effect of multiple on-off transients in relays. Unifying these two power
control systems through communication and system awareness would mean a dynamic
synchronization of power demand and distribution on a much faster scale.
A linking chain could be PMU-data loaded MPC applied in consumers power electronics load control.
Several applications have been reported in [13], [14] and [15] Finite control set MPC (FCS-MPC),
Figure 1, consists of a relatively accurate system prediction model with non-linearity handling at each
sampling interval, and constrained cost function minimization and switch sequence selection using
power converter switch discrete on-off state look-up table [15].

Figure 1: Model-based predictive control functional block diagram


Numerical optimization is the core of the FCS-MPC. It is a natural control environment for
synchrophasor technology implementation. Similar to FCS-MPC, the heart of the PMU-based control
is system monitoring, event or signal pattern recognition, using accurate system models (system
dynamics, event maps and look-up tables), and subsequent prediction of the power system behavior.
Both, PMU-based control and FCS-MPC have discrete number of solutions, which selection depends
on system modeling accuracy, sampling resolution and weighing factors in cost function minimization
at every sampling interval. They both count on positive definite values PMU on time-stamped
positive sequence sampling and FCS-MPC on least-square minimization of discrete cost function.

Principle of Operation
The goal of this work is to provide a concept of a dynamic load shedding as a customers response to
remote phasor measurement unit (PMU) data, received directly from the field or from a utility control
center, in order to alleviate stress on power grid and prevent eventual power system instability and
subsequent power shutdown. Dynamic partial load shedding could be possible using FCS-MPC or
other optimization techniques in conjunction with imported data from PMU-based power grid
monitoring. A concept of such system is shown in Figure 2.

Figure 2: PMU data injected load control system

Power units represent power electronics controlled loads, connected to a power substation
transformer. Dynamics of each of these units have been controlled locally, while central control from
the Distributed Load Control Room is responsible for the situational awareness monitoring, load
sharing and scheduled maintenance [2]. A new, here proposed, addition to this system is a PMUnetwork, which opens a window to external power grid situational awareness. This awareness
enables the load control outer loop to make a decision about load behavior based on power grid
information. Predictive control algorithms, such as FCS-MPC could weigh such decision against other
control objectives. Finally, there is a power system control from the regional transmission officers
(RTO) control room, which has the right to ultimate decision about the system control. It also provides
Distributed Load Control Room with the information about control decisions, including control
demands and recommendations in order to avoid eventual power shut-down.
Stability of this system could be the major concern due to apparent sharing of power grid control
between utilities and consumers, which should be addressed by master-slave type prioritization. As
the consumer has only an option to manipulate its own load system, it is obvious that utility control
prevails as it can still apply traditional load shedding, regardless consumers decision. Later, the same
control approach could be applied for bi-directional power flow, when a customer would not only be
energy consumer, but also energy provider. Finally, dynamics of each outer loop is slower than the
one of the next inner loop.

Simulation Modeling and Analysis


Simulation model of the power grid, which is used for the proof of concept, is an extension of the
model presented in [10]. It is still a simplified model of two-node four-generator power system with a
power line failure mechanism, Figure 3. An FCS-MPC regulated three-phase AC-DC-AC converter
driving a permanent magnet synchronous motor (PMSM) is added in parallel to load L2 through a
programmable circuit breaker. Two node systems, Area 1 (to the left of Bus 2) and Area 2 (to the right
of Bus 4), are connected by four parallel transmission lines of different lengths. The PMSM motor
drive represents multiple lower power motor drives in an IMCC with a total load of 100 MVA.

Figure 3: Two-node four-machine power system simulation model with power line fault insertion
Another model augmentation is a detailed PMU model with line voltage and current signal magnitude
and phase extraction from the signal positive sequence. Load FCS-MPC has a PMU-based control
signal incorporated in its cost function. In the PMU model, the phasor values of the input voltage and
current are derived from time domain values. Phasor magnitude and phase values have been
extracted from positive sequence time-domain values. Discrete Fourier analysis has been applied to
obtain first harmonic phasor values for each phase.

Simulation Results
A repetitive fault is simulated at Bus 3 of the power system, Figure 4. At the fault, the protection
circuits will be activated and faulted line would be disconnected. When the fault is cleared the system
will go back to its original operation. The figure shows the main display of the PMU system test bed,
consisting of each bus PMUs, RT Lab
control center and an interface of phasor concentrator units
(PDUs). Each PMU and the control center user interface signals are presented in the lists on the left
side of three out of four screens. The lists include magnitude, phase, first harmonic frequency and
frequency rate of change of each measured signal. Two displays to the left verify signal consistency
at different points in the PMU-PDC-control center communication chain. System voltage transients on
Buses 1 and 5 have been observed during the fault. The fourth screen shows the phase vector
diagram of the bus 1 and 5 voltages after the fault, where Bus 1 voltage (horizontal phasor line, phase
0
angle 0 ) is taken as the reference bus. Phasor magnitudes and angles displayed in time domain are
shown to the left of the phasor diagram.

Figure 4: Test bed for power fault simulation on Bus 3


Next simulation has implemented a cascade fault, triggered by 2 events. First fault occurs on Bus 3 at
20 seconds mark, causing line 1 breaker to operate and disconnect the line. Still, line 2, 3 and 4 will
transfer the power from area 1 to area 2 without any problems. At 40 s mark, Area 2 load increases by
10 %, causing an increase of power transfer form Area 1 to Area 2. Due to the increase of the
currents in each line, impedance relays of Line 2, 3 and 4 will operate, creating a cascade fault in the
system. Impedance relay implementation is shown in Table 1. Zone 1 and Zone 3 impedance relays
are implemented for line 1, 2, 3 and line 4. The voltage and current of the each line is measured and
fed in to the impedance calculation model. The calculated impedances are fed in to the Zone 1 and
Zone 3 relay models. The breaker for Zone 1 setting will trigger immediately and breaker for Zone 3
setting will be operated with 1 second delay.
Table 1: Impedance relay setting

Figure 5 shows the impedance of each line and the relay output signal based on zones 1 and 3
settings. It can be seen that the measured impedance of each line will increase after Line 1
disconnects. This is due to the increasing of currents in lines 2, 3 and 4, which means power transfer
increase on each line.

Figure 5: Impedance measurements of each line with cascade fault


When the load increase occurs at 40 second mark, the measured impedance of each line goes further
down as the transferred power increases. This causes Line 4 measured impedance go below the
Zone 3 settings, triggering the 1 second timer on the relay. Figure 6 shows how Zone 3 relay of Line 4
triggers after 1 second delay. It can be seen that Line 4 gets disconnected by this relay signal. Line 2
and 3 impedances will go further down to reach Zone 1 setting before reaching Zone 3, one second
triggering point, and will be disconnected immediately, causing the whole system to collapse.

Figure 6: Impedance measurements of each line after load increase


Figure 7 presents the voltage phasor measurements at buses 1 and 5. As before, Bus 1 voltage is
used as a reference. A significant deviation can be seen on Bus 5 voltage after Line 1 fault and load
increase. The phase angle difference has been used to develop a load control algorithm to avoid a
similar power system cascade failure event.

Figure 7: Voltage magnitude and phase angle of bus 1 and 5

Conclusion
Costly power blackouts have forced governments to heavily fund vivid research and development in
the area of power grid protection, automation and control. Advances in microprocessor and
communication technologies expanded research horizons to the areas of high computation burden
and real-time data transfer over large distances. Synchrophasors emerge as a reliable system
monitoring option for overall power system improvement. In the same time, practical developments in
power electronics digital control expands beyond limits of classical linear system control theory,
embracing MPC-based optimization approaches among other advanced control techniques. PMU
technology and FCS-MPC share many common characteristics and it looks natural to merge them
together into a powerful hybrid control system, where customer central control would become active
segment of much larger power system control.
This was a motivation for a research on the implementation of this idea on a power system simulation
model, which includes cascading power failure mechanism based on single power line failure followed
by a load increase. FCS-MPC algorithm has been implemented on a PMSM drive, representing a
cumulative load of an IMCC facility. Synchrophasor measurements, magnitude and phase angle
differences (errors), have been used as weighed variables for cost function minimization.
References
[1] NERC Steering Group: Technical Analysis of the August 14, 2003, Blackout: What
Happened, Why, and What Did We Learn? Report to the NERC Board of Trustees, July 2004
[2] S.E. Rudd et al.:Intelligent Monitoring of Distribution Automation, PAC World Magazine
March 2012
[3] J. Sykes and Y. Hu: Design of System Integrity Protection Schemes,, PAC World Magazine
March 2014
[4] A.G. Phadke and J.S. Thorp: Synchronized Phasor Measurements and Their Applications,
Springer, Sc.+ Business Media, New York 2008
[5] Jaime De La Ree, Virgilio Centeno, James S. Thorp and A. G. Phadke: Synchronized Phasor
Measurement Applications in Power Systems, IEEE Transactions On Smart Grid, Vol. 1, No.
1, June 2010, pp. 20-27

[6] Multiple authors: Monitoring the Health of the System Synchrophasor Technology, IEEE
Power & Energy Magazine, Vol. 13, No. 5, Sept./October 2015
[7] Kundur, Prabha: Power system stability and control, Eds. Neal J. Balu, and Mark G. Lauby.
Vol. 7. New York: McGraw-Hill, 1994.
[8] N. Mithulananthan et al.: Comparison of PSS, SVC, and STATCOM Controllers for Damping
Power System Oscillations, IEEE Transactions on Power Systems, Vol. 18, No. 2, May 2003
[9] C. Mozina: Undervoltage Load Shedding, Power Systems Conference: Advanced Metering,
Protection, Control, Communication, And Distributed Resources, Clemson, NC, March 13-16,
2007, PSC 2007, pp. 39-54
[10] Z. Mihailovic, L. Piyasinghe, L. Fan, Z. Miao and Y. Ma: Synchrophasor Technology for
Large Power Consumer Network Optimization, PAC World Americas Conference 2015,
Raleigh, NC, 1-3 September 2015
[11] D.C. Mazur, J.A. Kay and K.D. Mazur: Intelligent Motor Control, IEEE Industry Applications
Magazine, Vol. 21, No. 2, March/April 2015, pp. 30-37
[12] D. Divan, R. Moghe and A. Prasai: Power Electronics at the Grid Edge, IEEE Power
Electronics Magazine, Vol. 1, No. 4, December 2014, pp. 16-22
[13] S. Vazquez et al.: Model Predictive Control A review of Its Applications in Power
Electronics, IEEE Industrial Electronics Magazine, Vol. 8, No. 1, March 2014
[14] S. Kouro et al.: Model Predictive Control MPCs Role in the Evolution of Power
Electronics, IEEE Industrial Electronics Magazine, Vol. 9, No. 4, December 2015
[15] J. Rodriguez et al.: State of the Art of Finite Control Set Model Predictive Control in Power
Electronics, IEEE Transactions on Industrial Informatics, Vol. 9, No. 2, May 2013
[16] L. Wang, S. Chai, D. Yoo, L. Gan and K. Ng: PID and Predictive Control of Electrical Drives
and Power Converters using MATLAB /Simulink , John Wiley & Sons, Singapore 2015

S-ar putea să vă placă și