HomeMy WebLinkAbout20210318Avista to Staff PR 38 Attachment A.PDF Avista Contract R-40516
Prepared by
Boise State University
Boise, Idaho
September 2016
Staff_PR_38 Attachment A
i
PROJECT
RESIDENTIAL STATIC VAR
COMPENSATOR
Avista Contract R-40516
Final Report Draft version II, September 2016
Avista Project Managers: John Gibson
Randy Gnaedinger
Reuben Arts
Contractor: Boise State University
Staff_PR_38 Attachment A
ii
RESIDENTIAL STATIC VAR
COMPENSATOR
Avista Contract R-40516
Final Report, August 2016
Prepared by
Boise State University,
1910 University Drive,
Boise, ID 83725-1135
Principal Investigator
Said Ahmed-Zaid
Authors
Said Ahmed-Zaid
John Stubban
Andrés Valdepeña Delgado
Muhammad Kamran Latif
Prepared for
Avista Corporation,
141 I E. Mission Ave.,
Spokane, WA 99220
Avista Project Managers
John Gibson
Randy Gnaedinger
Reuben Arts
Staff_PR_38 Attachment A
iii
REPORT SUMMARY
The report provides the performance of a Residential Static VAR Compensator (RSVC) with a PWM-based
switching technique. The theoretical findings were verified by simulating and testing an RSVC prototype.
The proposed RSVC has several advantages compared to a conventional thyristor-fired SVC which include
an almost sinusoidal inductor current, sub-cycle reactive power controllability as opposed to half-cycle
controllability, lower footprint for reactive components, and the feasibility of building a single-phase
voltage regulation device. RSVC has a wide range of applications for utilities and customers. One such
application is Conservation by Voltage Reduction (CVR) which results in cost savings for both electric
utilities and customers during peak demand hours. This report also provides an analysis and evaluation of
the deployment of multiple RSVC in the Spokane downtown network to correct the power factor of the
four networks that form the downtown network. The methods of analysis include a local voltage control for
the RSVC and quasi-static time-series (QSTS) as well as traditional static analysis to perform the power
flow analysis in the distribution network. Results of the data analyzed show that it is possible to correct the
power factor in the network if proper settings are chosen for the load tap changer (LTC) transformer at the
head of the network. The network is a strong system where the voltage drop along the network is minimal
and it is difficult to effect the voltage using small reactive components (RSVC). The main objective of the
study was to correct the power factor in the four different networks in downtown Spokane. A higher power
factor reduces the feeder losses in the network by reducing magnitude of line currents. The main benefit of
deploying RSVCs in the downtown network is for the power factor correction. In all four network cases,
the power factor was corrected to almost unity. In some cases, the power factor of the network became
leading with the addition of RSVCs. The report also provides an analysis and evaluation of the benefits of
deploying multiple RSVCs in a distribution feeder. The feeder chosen for the study was SAG 741.
Extensive use of quasi-static time series (QSTS) determined the potential savings over a year in the SAG
741 feeder. The deployment of RSVCs can be an effective additional tool for the reduction of energy
consumption and peak demand via a strategy of conservation by voltage reduction (CVR). The deployment
of RSVCs throughout the feeders in conjunction with the optimization of the LTC, voltage regulator and
capacitor settings resulted in a reduction of the energy usage while keeping the voltage within the limits at
all times. The peak demand of the feeder was also reduced by applying CVR. The main benefit of deploying
RSVCs was the reduction in energy consumption. Other benefits of deploying the RSVC in the feeder were
peak reduction and voltage balancing of the phases. The cost- benefit analysis shows that it is beneficial to
install these devices in feeders with a relative weak source.
Staff_PR_38 Attachment A
iv
CONTENTS
INTRODUCTION ............................................................................ 7
Project Background ....................................................................................... 7
References ..................................................................................................... 8
VAR COMPENSATOR REVIEW ..................................................... 9
Principles of Var Compensation ................................................................... 9
Shunt Compensation ............................................................................................... 10
Series Compensation ............................................................................................... 11
Traditional Var Generators .......................................................................... 11
Thyristor Switched Capcitor (TSC) ............................................................ 12
Thyristor Controller Reactor (TCR) ....................................................................... 13
TCR with Fixed Capcitor (TCR-FC) ...................................................................... 15
Pulse Width Modulated (PWM) Switched Reactor ................................................ 15
References ................................................................................................... 18
BIDIRECTIONAL SWITCHES REVIEW ........................................ 19
Bi-directional Switch Topology .................................................................. 19
Commutation strategies for bi-directional switch ....................................... 22
Four-step commutation by measuring output load current ......................... 23
Four-step commutation by measuring input voltage sign ........................... 25
Commutation Methodology for Single Phase RSVC ................................. 26
References ................................................................................................... 27
Staff_PR_38 Attachment A
v
PROTOTYPE DESIGN AND SIMULATION RESULTS .................... 29
Distribution Network Modeling .................................................................. 29
Residential Loads Modeling ........................................................................ 31
Modeling Service Transformer Leakage Reactance ................................... 33
RSVC Reactive Component Sizing............................................................. 33
Summary for RSVC Reactive Requirements .............................................. 35
Simulation Results for RSVC ..................................................................... 36
Gate driving signals ................................................................................................ 36
Input and Output Voltage Waveforms .................................................................... 37
Inductor voltage and current waveforms ................................................................ 38
Current through bidirectional switches ................................................................... 39
RSVC output voltage for different duty cycles ....................................................... 40
Low-Voltage RSVC Testing ....................................................................... 41
Gate driving signals ................................................................................................ 41
Input and Output Voltage Waveforms .................................................................... 42
Inductor current waveform ..................................................................................... 43
Inductor voltage waveform ..................................................................................... 43
Current through top bidirectional switches ............................................................. 44
Hardware Prototype Board .......................................................................... 45
RSVC hardware prototype testing .......................................................................... 46
Problems encountered during RSVC hardware testing .......................................... 46
References ................................................................................................... 47
STUDY METHODOLOGY ............................................................. 48
Analysis Tools ............................................................................................. 48
OpenDSS................................................................................................................. 48
GridPV .................................................................................................................... 50
Staff_PR_38 Attachment A
vi
Data Conversion .......................................................................................... 50
PowerWorld Data Conversion ................................................................................ 50
SynerGI Data Conversion ....................................................................................... 52
Model Validation ......................................................................................... 52
Downtown Spokane Feeder Validation .................................................................. 52
Feeder SAG-741 Model Verification ...................................................................... 53
References ............................................................................................................... 55
STUDY RESULTS ......................................................................... 56
Spokane Downtown Feeder ........................................................................ 56
Spokane Feeder Description ................................................................................... 56
Power Factor Correction Utilizing RSVC .............................................................. 59
Future Work ............................................................................................................ 60
SAG-741 Feeder .......................................................................................... 60
Static Analysis ........................................................................................................ 62
Time-series Analysis ............................................................................................... 68
Future Work ............................................................................................................ 69
COST BENEFIT ANALYSIS .......................................................... 70
PATH TO MARKET ...................................................................... 72
APPENDIX .............................................................................................. 73
Staff_PR_38 Attachment A
7
INTRODUCTION
Project Background
Conservation by Voltage Reduction (CVR) is the implementation of a distribution voltage strategy whereby
all voltages are lowered to the minimum allowed by the equipment manufacturer. This is a consequence of
the observation that many loads consume less power when they are fed with a voltage lower than nominal.
In order to guarantee a good quality service, loads should not be supplied with a voltage higher or lower
than 5% of nominal. A range of standard service voltages used in the United States is specified by the
American National Standards Institute (ANSI) as 120 volts nominal, 114 volts minimum (120 V minus 5%)
and 126 volts maximum (120 V plus 5%).
Despite the regulatory history, electrical companies are forced to become more efficient and more
competitive by working to reduce costs. One such big cost is when a company buys costly energy from
another utility in the market when it cannot satisfy its own demand with its own installed capacity.
Furthermore, distribution companies, as well as final customers, must pay a higher price per kilowatt-hour
(kWh) during peak demand hours. The goal of our proposed residential CVR implementation is to reduce
power consumption during peak hours in order to save energy and costs.
Before applying CVR, power system operators and analysts must also understand the characteristics of their
loads. Even if all loads consumed less power with less voltage, which is generally not true, we would not
be saving energy in all cases. Some devices can give good service by working at a lower voltage. For
example, decreasing the voltage of a lightbulb will definitely yield energy savings. However, there are other
devices, such as air conditioners and ovens, which will have to work longer to give the same service. So in
the end, we may not be saving energy and, instead, it is possible to consume even more. Whereas lowering
the voltage may increase line current losses, the decrease in power consumption is expected to be bigger,
so that the overall balance will be positive [1-5].
Since the implementation of a conservation by voltage reduction (CVR) system is beyond the scope of this
project, we are proposing instead to develop a solution based on the concept of a Residential Static VAR
Compensator (RSVC) for regulating residential voltages, especially during peak demand hours, when the
benefits coincide best with the interests of customers and those of the electric companies. These RSVCs
will be an additional tool for smart demand-side management. By controlling remotely, the RSVC, a utility
can apply CVR at specified individual locations during specified periods. Our goal is to develop such an
RSVC prototype and we will leave it to the electric utility companies to develop strategies for conservation
Staff_PR_38 Attachment A
8
by voltage regulation. Our solution involves installing an individual apparatus which will decrease the
voltage before each customer’s service. This may not be cheap but many independent studies (with different
authors and procedures) have proved the great profit that can be achieved by working with CVR and these
additional costs can be justified over the long term [1-5]. In other words, the cost of implementing CVR
per kWh saved would be smaller than buying that amount of kWh in the market. A question remains as to
whether the payback for the initial cost investment will be in the range of three to five years.
Our previous experience with two senior design projects on CVR and where we tested the current and
power sensitivities of many common household appliances to voltage regulation has provided us with
general conclusions and guidance regarding the feasibility of this method. Another potential benefit of these
individual residential devices is that they can also be used by utilities with high penetrations of distributed
energy sources that would normally complicate the implementation of a global CVR system for energy
reduction.
References
[1] Kennedy, W. and R.H. Fletcher, “Conservation Voltage Reduction at Snohomish County PUD,” IEEE
Transactions on Power Systems, vol. 6, no. 3, pp. 986-998, August 1991.
[2] Erickson, J.C. and S.R. Gilligan, “The Effects of Voltage Reduction on Distribution Circuit Loads,”
IEEE Transactions on Power Apparatus and Systems, vol. PAS-101, no. 7, pp. 2014-2018, July 1982.
[3] Warnock, V.J. and T.L. Kirkpatrick, “Impact of Voltage Reduction on Energy and Demand: Phase II,”
IEEE Transactions on Power Systems, vol. PWRS-1, no. 2, pp. 92-95, May 1986.
[4] Fletcher, R.H. and A. Saeed, “Integrating Engineering and Economic Analysis for Conservation Voltage
Reduction,” IEEE 2002 Summer Meeting, 0-7803-7519-x/02, pp. 725-730.
[5] Lefebvre, S., G. Gaba, A.-O. Ba, D. Asber, A. Ricard, C. Perreault, and D. Chartrand, “Measuring the
Efficiency of Voltage Reduction at Hydro-Quebec Distribution,” PES General Meeting – Conversion and
Delivery of Electrical Energy in the 21st Century, pp. 1-7, Pittsburgh, PA, 20-24, July 2008, IEEE 2008.
Staff_PR_38 Attachment A
9
VAR COMPENSATOR REVIEW
Reactive power plays an important role to enhance the quality of power systems. In theory, reactive power
is defined as the ac component of the instantaneous power with double the system frequency and zero
average value. This means that the reactive power generated by the ac source can oscillate between the ac
source and reactive components (capacitors and reactors) at twice the rated frequency (60 Hz) without
circulating between the load and the source. This allows reactive power to contribute in the voltage stability
for the power systems. Reactive power also provides the power factor correction (PFC) for industrial plants
that present poor power factor. Many methods for PFC have been proposed in the literature but reactive
power method is the most commonly used technique for PFC [1-3].
A widely used methodology for providing Var compensation include shunt capacitors. But with largely
varying loads fixed shunt capacitors can often lead to either under-compensation or over-compensation.
Dynamic Var compensation can be achieved using switchable capacitor banks. Depending on the Var
requirements, these capacitor banks are switched in and out of the power system. However, they require
vast installation area and need a reactor to avoid series and parallel resonance between the capacitor and
source impedance [2].
Modern reactive compensation techniques use Flexible AC Transmission Systems (FACTS) devices that
provide dynamic reactive power to the power system. FACTS are high-speed power devices that combine
advanced control system techniques with the fast processing power of microprocessors to respond to
reactive power needs for the power system [4-5]. Commonly used FACTS devices are static Var
compensators (SVCs) and inverter-basted static synchronous compensators (STATCOMs). For utilities,
FACTS technology has become an essential tool to alleviate problems associated with reactive power to
get the most service from their transmission and distribution networks and enhance the grid reliability [6].
This chapter briefly describes the principles for series and shunt reactive power compensation. Also, an
overview for the commonly used FACTS (SVCs) devices are also presented. An improved SVC topology
is proposed that help mitigating the problems caused by the conventional SVC.
Principles of Var Compensation
Based on the type of reactive compensation required, it can be categorized as series Var compensation and
shunt Var compensation. Shunt compensation changes the effective equivalent impedance of the load
whereas series compensation modifies the series transmission or distribution parameters [1]. In both
Staff_PR_38 Attachment A
10
compensation techniques, reactive power flows through the system thus improving the performance of the
overall ac power system.
Shunt Compensation
In shunt compensation, the lagging current drawn due to inductive load is compensated by leading
current generated by the compensator device. The leading current can be achieved in three different ways
using a current source, a voltage source and a large single capacitor or capacitor banks. The reactive
source provides a part of the total reactive power requirement by injecting positive Vars into the power
system thus improving the voltage regulation at the receiving terminals. In
Figure 2-1 a typical power system is shown that draws a lagging current due to an inductive load which
results in reduced voltage levels at the receiving end. The power system is compensated by reducing the
lagging current supplied by the generator with the help of a reactive source. This helps to improve the power
factor as well as overall voltage profile for the power system.
Source
X RV1 V2
Load
V1
V2 I.R
I.Xδ
φ
I
I
Figure 2-1: Radial ac system without reactive compensation
V1
V2 I'.R
δ'
φ
Source
X RV1 V2
LoadQ
I'.X
I
Ic
Ic
I'
I'
Staff_PR_38 Attachment A
11
Figure 2-2: Radial ac system with reactive compensation
Series Compensation
Series compensation involves decreasing the inductive reactance of the power lines by installing the series
capacitors. As a result of series Var compensation, total reactance of the lines is reduced that helps in the
minimizing the voltage drops.
Figure 2-1 represents the phasor diagram of ac system without series reactive compensation. Figure 2-3
shows the effect of series compensation on an overall ac system. Decrease in the line reactance not only
reduces the voltage drop in the lines but also improves the voltage at the receiving end specifically for the
loads with low power factor.
Source
X RV1 V2'
Load
V1
V2'I.R
I.(XL-Xc)δ'
φ
V2
I
XL
Figure 2-3: Radial ac system with series reactive compensation
Traditional Var Generators
This section provides an overview of the shunt Var compensator that are discussed in literature and
commonly applied in electric utilities to mitigate effects of poor power factor and lower receiving voltages.
Common shunt Var compensators that are employed by the electric utilities include capacitor banks and
Static Var Compensator (SVC). SVC has various topologies depending upon the nature of the reactive
compensation required and reactive components used. Common SVC topologies include Thyristor
Controlled Reactor (TCR), Thyristor Switched Capacitor (TSC), and Thyristor Controlled Reactor with
Fixed Capcitor (TCR-FC).
Staff_PR_38 Attachment A
12
Thyristor Switched Capcitor (TSC)
Capacitor banks provide dynamic reactive power to the power system. In capacitor banks, individual
capacitors can be switched on or off in discrete steps depending on the reactive power needs. Figure 2-4
shows a capacitor banks where reactive components are sized to address the needs of varying reactive loads.
Each branch in Figure 2-4 consists of a capacitor C in series with two thyristor valves Sw1 and Sw2. This
assembly is connected in series with an inductor L to avoid the sudden increase in the rate of current through
the thyristors and to prevent the resonance with the power network. An individual branch in Figure 2-4 is
referred as Thyrisor Switched Capcitor.
AC Mains
C1 C2
SW1 SW1SW2 SW2
L1 L2
CN
SW2
LN
SW1
Figure 2-4: Switchable capacitor banks
It is important to provide the gating signals to the thyristors at the instant when system voltage is equal to
the capacitor voltage to allow the transient free switching. This is achieved by using a synchronization block
with a phase-locked loop. Also when the capacitors are turned on, their stored charge adds on to the network
voltage. Thyristors are exposed to the capacitor’s charge as well as the network voltage. Therefore, for safe
operation, thyristors should be rated at atleast twice as much as the peak of the system voltage.
TSC are not commonly used for reactive power compensation since their usage is practically limited by
number of disadvantages: the Var compensation is done in discrete steps, the thyristor switches are required
for each TSC branch resulting in the bulky structure, and power rating for the thyrisor switches should be
atleast twice the peak voltage.
Staff_PR_38 Attachment A
13
Thyristor Controller Reactor (TCR)
Figure 2-5(a) shows a thyristor controlled reactor used in a SVC. In a TCR, an inductor is connected with
bidirectional thyristor valve. The reactive power generated by a TCR is dynamically controlled from a
minimum value (zero) to a maximum value by controlling the current conduction through the thyristor
valves. The current injection is based on the firing angle (gating angle) of the thyristors. The maximum
current injected by TCR is obtained by a firing delay of 90°. Partial current contribution is obtained when
the firing angle is between 90° and 180°. By increasing the thyristor firing angle, the lagging current injected
into power system is reduced which in turn increases the inductance of the TCR and reduces the reactive
power level of the power system. The fundamental component of the instantaneous current supplied by the
TCR is given by
𝐼 𝑉
𝑋2𝜋 2α sin2α
Where:
α Firing angle of the thyristor
𝑋 𝜋𝜔𝐿 = minimum TCR reactance for α 90°
Staff_PR_38 Attachment A
14
Continuous Conduction
Partial Conduction
Minimum Conduction
V
V
V
I
I
I
(a) (b)
Figure 2-5: (a) Thyristor control reactor (b) Voltage and current waveforms in a TCR for different
thyristor gating angles
Figure 2-5(b) shows three cases for different firing angle for α. It can be seen that continuous current
conduction is obtained when thyristors are fired at 90° during each half cycle. At this configuration,
contribution of reactor is maximum in lowering the reactive power of the system. However, the change in
reactor current can only happen once during each half cycle. The lack of controllability for the reactive
power during the complete cycle is one of the major drawback for thyristor controlled reactor.
As with the case for TSC, the thyristor gating signals must stay in synchronization with the ac mains voltage
under all circumstances. This synchronization is required to properly fire the gating signals for controlling
the TCR. The synchronization is achieved by using a phase-locked loop that runs in synchronous with the
ac system voltage and produces thryistor gate firing sequence with respect to the peak of that voltage. The
control system responsible for phase-locked loop must be fast enough to respond during system faults and
voltage fluctuations. The use of synchronization block for producing the gate signals not only poses a
technical difficulty but also increases the total components required for constructing a SVC. Figure 2-5(b)
also shows the firing of thryristor at angles greater than 90°. These conditions are labelled as paritial
conduction and minimum conduction. At firing angles greater than 90°, the lagging current injected by the
reactor is non-sinusoidal. This produces low-order harmonics which requires filtering thus incurring
additional cost.
Staff_PR_38 Attachment A
15
Due to above mentioned difficulties, the use of conventional thyrisor-firing based reactive power control
for a reactor is inefficient and prone to harmonics. An advanced method for controlling the reactive power
for the reactor is presented in the final section of this chapter. This advanced method uses a pulse-width
modulated control of reactive power instead of discrete reactive power as in the case of TCR.
TCR with Fixed Capcitor (TCR-FC)
Figure 2-6(a) shows a thyristor controlled reactor with fixed capacitor (TCR-FC). This assembly uses the
combination of capacitor and reactor to provide adequate amount of reactive power support to the power
system. A distinguishing feature of TCR-FC is that the amount of the reactive power generated or absorbed
is a function of the input voltage that can be varied linearly depending upon the voltage. Figure 2-6(b)
shows the reactive power against the voltage (Q-V) characteristics of a fixed capacitor with a thyristor
controlled reactor. The graph depicts that the amount of Var exchanged with the power system depends
upon the input voltage. The linear region of operation is limited by the rated reactive power of the reactive
components. Beyond these points, the Q-V characteristics is no longer linear.
VREF
ICMAX ILMAX
C
L
SW1 SW2
v
i
AC Mains
(a) (b)
Figure 2-6: (a) Thyristor control reactor with fixed capacitor (TCR-FC) (b) QV characteristics for
Thyristor control reactor with fixed capacitor
Pulse Width Modulated (PWM) Switched Reactor
As described in previous sections, thyristor controlled reactor (TCR) has multiple problems associated with
it. A summary for the problems encountered with TCR can be summarized as follow:
1. Control System for TCR (as well as TSC) require synchronization with the AC mains for
generating the gate signals.
Staff_PR_38 Attachment A
16
2. TCR Var compensator produce low-order harmonics due to non-continuous current that
requires filtering.
3. Var compensation is not continuous. It can only occur once in each half cycle.
4. Huge reactive power components leads to bulky structure with high cost.
An alternating approach based on a pulse-width modulated (PWM) control of reactive power is proposed
for RSVC. The advantages of PWM-based Var compensators include a simpler control system with no
phase-locked loop, an uninterrupted reactor current, and a substantial reduction in the components size and
cost. The operating principle for proposed compensator is discussed in the next sub-section. Also, design
guideline of the designing RSVC system is proposed and experimental results are provided later in this
report to verify the proposed concept.
Operating Principle
Figure 2-7 shows a single phase, PWM-based switched reactor topology. A detailed analysis for this
topology can also be found in the literature [7-8]. The single-phase PWM-based switched reactor circuit
requires two bi-directional switches or four unidirectional switches.
Bidirectional
Switch
Bidirectional
Switch Inductor
SW1
SW2 VL
ZIN
IIN
Figure 2-7:Single phase PWM based switched inductor
In the Figure 2-7, switches SW1 and SW2, represents bidirectional switches with complementary gating
signals. When switch S1 is ON, the input current IIN is equal to the inductor current. When switch SW1 is
OFF, the inductor current is allowed to wheel through switch SW2 which is ON. The result is a fairly
sinusoidal inductor current whose higher-order harmonics are negligible.
Staff_PR_38 Attachment A
17
By using high frequency switching, the fundamental component of the inductor current can be controlled
[7]. Assuming the switches to ideal, meaning their switching and conduction losses are negligible, when
SW1 is closed (SW2 is open), input voltage 𝑣 appears across the inductor. This inductor voltage 𝑣 can
be expressed by the following expressions:
𝐼𝑛𝑑𝑢𝑐𝑡𝑜𝑟 𝑉𝑜𝑙𝑡𝑎𝑔𝑒 𝑣 𝑣𝐷
where D is the duty cycle ratio defined as the time interval when SW1 is conducting. Similarly, the current
can be expressed
𝐹𝑢𝑛𝑑𝑎𝑚𝑒𝑛𝑡𝑎𝑙 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝐶𝑜𝑚𝑝𝑜𝑛𝑒𝑛𝑡 𝑖
where D is the duty cycle for switch SW1. From the equations above, the equivalent input inductive
reactance (𝑋) can be found by using following expression:
|𝑋||𝑣|
|𝑖|
|𝑣/𝐷|
|𝑖𝐷|
|𝑋|
𝐷
In the equation above, 𝑋 can be controlled by the duty cycle D. This makes the fixed inductor reactance
to appear like variable reactance as a function of duty cycle. The dynamic reactive power for PWM switched
inductor can be expressed as
𝑄𝑣
𝑋.𝐷
It can be seen that by increasing the duty cycle, the output inductive reactance (𝑋) increases, while the
supplied reactive power decreases.
There is a noticeable improvement in the performance of PWM-based switched reactor compared to
conventional TCR. There is a freewheel path for the inductor current through switch SW2 which is helpful
in keeping the inductor current sinusoidal. This improvement eliminates the requirement for an additional
filtering circuit that is required for limiting the harmonics. The reactive power can be adjusted at any time
by varying the duty cycle D of switch SW1. This improvement eliminates the need of any synchronization
block that is required for providing proper gating signals to thyristors. Due to lower ratings for the reactive
components involved and reduction in the total components required to realize this device, it is
economically more feasible than convention SVC. Using this approach, utilities can benefit with the same
TCR-FC characteristics without polluting the ac system with low-order harmonics.
Staff_PR_38 Attachment A
18
References
[1] J. Dixon, L. Moran, J. Rodriguez and R. Domke, "Reactive Power Compensation Technologies: State-
of-the-Art Review," in Proceedings of the IEEE, vol. 93, no. 12, pp. 2144-2164, Dec. 2005.
[2] A. Prasai, J. Sastry and D. M. Divan, "Dynamic Capacitor (D-CAP): An Integrated Approach to
Reactive and Harmonic Compensation," in IEEE Transactions on Industry Applications, vol. 46, no. 6, pp.
2518-2525, Nov.-Dec. 2010.
[3] A. Prasai and D. M. Divan, "Control of Dynamic Capacitor," in IEEE Transactions on Industry
Applications, vol. 47, no. 1, pp. 161-168, Jan.-Feb. 2011.
[4] Narain G. Hingorani; Laszlo Gyugyi, "FACTS Concept and General System Considerations,"
in Understanding FACTS: Concepts and Technology of Flexible AC Transmission Systems, 1, Wiley-IEEE
Press, 2000, pp.1-35
[5] R. Mohan Mathur; Rajiv K. Varma, "Introduction," in Thyristor-Based FACTS Controllers for
Electrical Transmission Systems , 1, Wiley-IEEE Press, 2002, pp.1-15
[6] D. Coates, "FACTS: a transmission utility perspective," Flexible AC Transmission Systems - The
FACTS (Ref. No. 1998/500), IEE Colloquium, London, 1998, pp. 2/1-2/7.
[7] H. Jin, G. Goos, and L. Lopes, “An efficient switched-reactor-based static Var compensator,” IEEE
Trans. Ind. Appl., vol. 30, no. 4, pp. 998–1005, 1994
[8] Sanghun Kim, H. G. Kim and H. Cha, "Reactive power compensation using switching cell structured
direct PWM AC-AC converter," 2016 IEEE 8th International Power Electronics and Motion Control
Conference (IPEMC-ECCE Asia), Hefei, 2016, pp. 1338-1344.
Staff_PR_38 Attachment A
19
BIDIRECTIONAL SWITCHES REVIEW
A bi-directional switch (also known as four quadrant or 4Q-switch or AC switch) is an essential part of the
proposed residential static VAR compensator (RSVC). A bi-directional switch has the capability of
conducting current in both directions as well as blocking the voltages of both polarities. The realization of
a bi-directional switch is a technical challenge in the implementation of forced commutation techniques in
direct AC switching. Research has been underway to fabricate a bidirectional switch on a single silicon die
[1-2]. So far, very few bi-directional switches [3] are available in the power-electronics market. Therefore,
discrete unidirectional switches are used to realize a bi-directional switch.
This chapter provides a brief introduction to bi-directional topologies described in the literature. Also,
different current commutation strategies are reviewed based on the output current sign and input voltage
polarity across the bi-directional switches. Preference of one strategy over another depends upon a
particular application. For the RSVC application, a voltage based current commutation is chosen because
of ease to detect the voltage polarity across the bi-directional switches. For the sake for simplicity, “bi-
directional switches” and “switches” will be used interchangeably in the discussion. For reference to the
high-power transistors, “device” will be used throughout this chapter.
Bi-directional Switch Topology
Bi-directional switches are realized using discrete semiconductor devices. The diode bridge bi-directional
switch arrangement, shown in Figure 3-1, consists of an insulated gate bipolar transistor (IGBT) at the
center of a diode bridge arrangement [4]. The main advantage associated with a diode bridge topology is
that only one active driving circuitry is required to control the flow of current. However, the direction of
current cannot be controlled through the bi-directional switch. When the current changes sign, it is
commuted through the opposite conducting diodes. Moreover, there are three (3) devices involved in the
conduction of the current which contribute to greater conduction losses compared to other bi-directional
topologies. These disadvantages restrict the use of a diode bridge bi-directional switch for a limited number
of applications.
For a diode bridge bi-directional switch, it is not possible to define a safe commutation sequence. For safe
operation, the two switches should operate in a complementary manner. Since, the complementary
operation of the switches cannot be guaranteed, the only possible commutation strategies include “make-
before-break” and “break-before-make.” These commutation strategies lead to short circuiting of the input
source or opening an inductive current leading to destructive current and voltage spikes respectively.
Staff_PR_38 Attachment A
20
Figure 3-1: Diode bridge bi-directional switch topology
Consider the circuit shown in Figure 3-2. In this figure, the switches represent diode bridge bi-directional
switches. In case of “make-before-break” strategy, the on-coming switch SW1 turns on before the off-going
switch SW2 turns off. This causes a short circuit for the input source which leads to huge current spikes as
illustrated in Figure 3-3. These current spikes are destructive for the semiconductor devices and may lead
to permanent damage of the switches. In the circuit where “make-before-break” strategy is applied, reactors
are added to facilitate current transition during commutation. It then requires a snubber circuit to limit the
voltage transients [5].
Figure 3-2: Input phase with two bi-directional switches
For a “break-before-make” topology, the off-going switch SW1 is turned off before the on-coming switch
SW2 is turned on. This causes a breakage in the conduction of inductive load current which in turn produce
huge voltage spikes across the opened switches, as shown in Figure 3-4. These voltage spikes are destructive
for the switches and impose a threat to proper functioning of the switches. Circuits using “break-before-
make” strategy normally use voltage clamp circuit along with local snubber circuits to prevent the damaging
voltage spikes [6].
SW2
SW 1
CAPACITOR
INDUCTOR
LEAKAGE
REACTANCE
AC
MAINS
PWM BASED SWITCHED
INDUCTOR
Staff_PR_38 Attachment A
21
Figure 3-3: Short-circuiting the input phase using “make-before-break “strategy
Figure 3-4: Inductive current interruption using "break-before-make" strategy
Another proposed bi-directional switch topology includes two IGBTs with series diode connected in an
antiparallel configuration [7-8]. In this topology, a conduction path for the current exists in both directions.
The two IGBTs can be used to independently control the current paths. Also, the diodes provide the reverse
voltage blocking capability. By using proper commutation strategy, safe commutation of load current is
possible, thus eliminating the risk of short-circuit and over-voltage spikes. This topology has reduced
conduction losses as only two active devices are used to conduct current in either direction. These basic
features allow this topology to be superior to the one described in the previous example. Depending upon
the application, common-emitter or common-collector, shown in Figure 3-5 and Figure 3-6 respectively,
configurations can be used for constructing a bi-directional switch. A common-emitter configuration
requires an isolated power supply for gate driver circuitry for each bi-directional switch, but this
configuration is helpful in providing transients benefits during switching as well as reducing
electromagnetic interference (EMI) effects.
Figure 3-5: Common-emitter configuration for a bi-directional switch
SW 2
SW 1
CAPACITOR
INDUCTOR
LEAKAGE
REACTANCE
AC
MAINS
PWM BASED SWITCHED
INDUCTOR
SWITCH 2
SWITCH 1
CAPACITOR
INDUCTOR
LEAKAGE
REACTANCE
AC
MAINS
PWM BASED SWITCHED
INDUCTOR
Staff_PR_38 Attachment A
22
Figure 3-6: Common-collector configuration for a bi-directional switch
Commutation strategies for bi-directional switch
The commutation for a bi-directional switch, consisting of two IGBTs with series diodes connected in
antiparallel manner, can be based on measuring output load current or input voltage across the commutating
bi-directional switches. For the discussion to follow, refer to Figure 3-7. When the output phase needs to
be commutated from one input phase to other, it must satisfy following two rules
i) the commutation from one input phase to the other should not short circuit the two input phases.
The short circuit in the input phases leads to destructive current spikes which are fatal for the
switching devices;
ii) the commutation should not interrupt the output load current. Interrupting the flow of inductive
current leads to overvoltage spikes which are likely to destroy the IGBT devices.
In order to ensure the above conditions are met, it is required to know either the sign of the input voltage
across the switches or the direction of the output load current before a safe commutation sequence is
applied. [8-14].
Input phases Output phase
V1
V2
SW1F
SW1B
SW2F
SW2B
R
L
Io
Figure 3-7: Two-phase to one-phase current commutation using bi-directional switches
Staff_PR_38 Attachment A
23
Four-step commutation by measuring output load current
A reliable method for current commutation involves a four-step commutation strategy which can be used
to control the direction of current through the commutation switches [15]. The goal of this strategy is to
strictly follow the aforementioned two rules for safe commutation. In order to explain this strategy, consider
the two-phase circuit shown in Figure 3-7. In the circuit, the output load current Io is assumed to be positive,
if it flows from input to output phase. The subscript “F” represents the forward or positive direction of
current flow whereas, subscript “B’ means backward or negative direction of the current flow. It can be
seen that the forward current, from input phase to output phase, flows through switches named as SW1F
and SW2F, and the backward current, from output phase to input phase, flows through switches named as
SW1R and SW2R.
By looking at the circuit, the following combination of non-hazardous combinations of switching sequences
can be achieved.
Table 3-1: Non-hazardous devices combination for current commutation
State SW1F SW1B SW2F SW2B
Output
current (Io)
1 1 1 0 0 + or -
2 0 0 1 1 + or -
3 1 0 0 0 +
4 0 0 1 0 +
5 1 0 1 0 +
6 0 1 0 0 -
7 0 0 0 1 -
8 0 1 0 1 -
In Table 3-1, a logic-high or logic-one (1) means the IGBT is ON or conducting whereas, a logic-low or
logic-zero (0) means that the IGBT is OFF or open. Any other combination of switching signals will violate
either of the two rules for safe-commutation. States 1 and 2 are called Steady States and states 3 till 8 are
called Transitional States. The commutation of switches should start from a steady state and end in the other
steady state while going through three transitional states.
Suppose, in the circuit shown in Figure 3-7, switch SW1 is turning off and SW2 is coming on. Assume that
the output load current Io is flowing in the positive directions. The current commutation from the outgoing
Staff_PR_38 Attachment A
24
switch SW1 to the incoming switch, SW2, based on output current direction, involves the following four
steps:
i) the IGBT from the outgoing switch that is not conducting the output current is turned off. In
this case, SW1B is turned off;
ii) the IGBT from the incoming switch that will conduct the output current is turned on. In this
case, SW2F is turned on;
iii) the IGBT from the outgoing switch that is conducting the output current is turned on. In this
case, SW1F is turned off;
iv) the IGBT from the incoming switch that will not conduct the output current is turned on. In this
case, SW2B is turned on.
In case the output current in following in the opposite direction, i.e. from the output phase to the input
phase, the commutation from outgoing switch SW1 to the incoming switch SW2 is performed in the
following steps:
i) IGBT SW1F is turned off;
ii) IGBT SW2B is turned on;
iii) IGBT SW1B is turned off;
iv) IGBT SW2F is turned on.
The commutation sequence can be summarized in the state machine diagram shown in Figure 3-8. It should
be noted that there is a built-in time delay td in between each transition. This time delay should be greater
than the maximum propagation delays required by the IGBT gating signals.
Figure 3-8: Four-step switching diagram for two bi-directional switches based on output load current
Staff_PR_38 Attachment A
25
Four-step commutation by measuring input voltage sign
In the previous method, the output load current was used to properly commutate between two-input phases.
This section will describe another method for current commutation between the two input phases. This
commutation technique is based on the sign of the input voltage across the bi-directional switches which
are involved in the commutation [16].
As with the commutation using load current, this strategy assumes that when an output phase is connected
to the input phase, both the IGBT devices for the corresponding bi-directional switch are on. A general
strategy is to identify the freewheeling diodes between the two bi-directional switches. Freewheeling diode
paths are those that allow the current to follow from lower input phase to higher input phase voltage.
Consider the circuit shown in Figure 3-9. For the first case consider the voltage at SW1, i.e. V1 is at higher
potential than voltage across SW2, i.e. V2. The IGBTs that aid freewheeling diodes path for the current to
flow from V2 to V1 are SW2F and SW1B. The current commutation from SW1 to SW2 involves the
following steps:
i) IGBT of the incoming switch aiding the freewheeling diode, i.e. SW2F is turned ON;
ii) IGBT of the outgoing switch present in non-freewheeling current path, i.e. SW1F is turned OFF;
iii) IGBT of the incoming switch present in the non-freewheeling current path, i.e. SW2B is turned
ON;
iv) IGBT of the outgoing switch aiding the freewheeling diode, i.e. SW1B is turned OFF;
Input phases Output phase
V1
V2
SW1F
SW1B
SW2F
SW2B
R
L
Io
Freewheeling Diode current
Path
Figure 3-9: Freewheel diode current path when V1 > V2
In the case when V1 < V2, the current commutation from SW1 to SW2 involves the following steps:
i) IGBT of the incoming switch aiding the freewheeling diode, i.e. SW2B is turned ON;
Staff_PR_38 Attachment A
26
ii) IGBT of the outgoing switch present in non-freewheeling current path, i.e. SW1B is turned OFF;
iii) IGBT of the incoming switch present in the non-freewheeling current path, i.e. SW2F is turned
ON;
iv) IGBT of the outgoing switch aiding the freewheeling diode, i.e. SW1F is turned OFF;
This switching signal is presented in a state machine diagram in Figure 3-10. As in the case of load current
based current commutation, states where only both IGBTs for the bi-directional switch are on are called
steady states and remaining states are called transitional states. A small delay time between each transition
is required to account for the propagation delays in the gating signals for the IGBTs.
1 1 : 0 0
1 1:1 0
0 1:1 0
0 1:1 1
0 0 : 1 1
1 1:0 1
1 0:0 1
1 0:1 1
S1F S1B : S2F S2B
S1F S1B : S2F S2B
Step 1 – TD1
Step 2 – TD2
Step 3 – TD3
Step 4 – TD4
V1 > V2 V1 < V2
Transitional
states
Transitional
states
Steady states
Figure 3-10: Four-step switching state machine diagram for two bi-directional switches based on input
voltage sign
Commutation Methodology for Single Phase RSVC
For the development of the bidirectional switches in RSVC, voltage-based current commutation is
chosen over the output load current commutation. For single-phase voltage, the bi-directional switch
commutation become relatively simpler as it only involves detecting the sign of the main line with respect
to the neutral point. For a balanced three-phase system, neutral is often grounded (zero potential), thus V2
in the circuit shown in Figure 3-9 reduces to zero volts.
Detection of voltage sign is done using a high-accuracy analog-to-digital converter. An incorrect
measurement of the voltage sign during the commutation process will produce a short circuit path. For
RSVC, detection of voltage sign is achieved using a ADC from Maxim Integrated Santa FE
(MAXREFDES5#) [17]. This ADC is capable of 16-bit high-accuracy analog to digital conversion that
Staff_PR_38 Attachment A
27
accepts -10V to +10V analog signals. Digital output from ADC is then passed on to programmable devices
like micro-controller or FPGA for the execution of the proper state machine sequence.
References
[1] Shuming Xu, R. Plikat, R. Constapel, J. Korec and D. Silber, "Bidirectional LIGBT on SOI substrate
with high frequency and high temperature capability," Power Semiconductor Devices and IC's, 1997.
ISPSD '97, 1997 IEEE International Symposium on, Weimar, 1997, pp. 37-40.
[2] Ying-Keung Leung, A. K. Paul, J. D. Plummer and S. S. Wong, "Lateral IGBT in thin SOI for high
voltage, high speed power IC," in IEEE Transactions on Electron Devices, vol. 45, no. 10, pp. 2251-2254,
Oct 1998.
[3] IXYS (2003) Bidirectional Switch with NPT3 IGBT and fast Diode Bridge [Online]. Available:
http://www1.futureelectronics.com/doc/ixys/fio50-12bd.pdf
[4] S. Bernet, T. Matsuo and T. A. Lipo, "A matrix converter using reverse blocking NPT-IGBTs and
optimized pulse patterns," Power Electronics Specialists Conference, 1996. PESC '96 Record., 27th Annual
IEEE, Baveno, 1996, pp. 107-113 vol.1.
[5] P. D. Ziogas, S. I. Khan and M. H. Rashid, "Analysis and Design of Forced Commutated Cycloconverter
Structures with Improved Transfer Characteristics," in IEEE Transactions on Industrial Electronics, vol.
IE-33, no. 3, pp. 271-280, Aug. 1986.
[6] C. L. Neft and C. D. Schauder, "Theory and design of a 30-hp matrix converter," in IEEE Transactions
on Industry Applications, vol. 28, no. 3, pp. 546-551, May/Jun 1992.
[7] A. Alesina and M. G. B. Venturini, "Analysis and design of optimum-amplitude nine-switch direct AC-
AC converters," in IEEE Transactions on Power Electronics, vol. 4, no. 1, pp. 101-112, Jan 1989.
[8] C. Klumpner, P. Nielsen, I. Boldea and F. Blaabjerg, "New steps towards a low-cost power electronic
building block for matrix converters," Industry Applications Conference, 2000. Conference Record of the
2000 IEEE, Rome, 2000, pp. 1964-1971 vol.3.
[9] M. Ziegler and W. Hofmann, "Semi natural two steps commutation strategy for matrix
converters," Power Electronics Specialists Conference, 1998. PESC 98 Record. 29th Annual IEEE,
Fukuoka, 1998, pp. 727-731 vol.1.
Staff_PR_38 Attachment A
28
[10] L. Empringham, P. W. Wheeler and J. C. Clare, "Intelligent commutation of matrix converter bi-
directional switch cells using novel gate drive techniques," Power Electronics Specialists Conference, 1998.
PESC 98 Record. 29th Annual IEEE, Fukuoka, 1998, pp. 707-713 vol.1.
[11] B. H. Kwon, B. D. Min and J. H. Kim, "Novel commutation technique of AC-AC converters," in IEE
Proceedings - Electric Power Applications, vol. 145, no. 4, pp. 295-300, Jul 1998.
[12] C. Klumpner, P. Nielsen, I. Boldea and F. Blaabjerg, "A new matrix converter motor (MCM) for
industry applications," in IEEE Transactions on Industrial Electronics, vol. 49, no. 2, pp. 325-335, Apr
2002.
[13] J. L. Galvez, X. Jorda, M. Vellvehi, J. Millan, M. A. Jose-Prieto and J. Martin, "Intelligent bidirectional
power switch module for matrix converter applications," Power Electronics and Applications, 2007
European Conference on, Aalborg, 2007, pp. 1-9.
[14] P. W. Wheeler, J. Rodriguez, J. C. Clare, L. Empringham and A. Weinstein, "Matrix converters: a
technology review," in IEEE Transactions on Industrial Electronics, vol. 49, no. 2, pp. 276-288, Apr 2002.
[15] L. Empringham, P. Wheeler and J. Clare, "A matrix converter induction motor drive using intelligent
gate drive level current commutation techniques," Industry Applications Conference, 2000. Conference
Record of the 2000 IEEE, Rome, 2000, pp. 1936-1941 vol.3.
[16] J. Mahlein, J. Igney, J. Weigold, M. Braun and O. Simon, "Matrix converter commutation strategies
with and without explicit input voltage sign measurement," in IEEE Transactions on Industrial Electronics,
vol. 49, no. 2, pp. 407-414, Apr 2002.
[17] [Online] Available: https://www.maximintegrated.com/en/design/reference-design-center/system-
board/5561.html
Staff_PR_38 Attachment A
29
PROTOTYPE DESIGN AND SIMULATION
RESULTS
Residential Static VAr Compensator(RSVC) is conceived as device that can provide voltage stability at the
distribution level of the power system. SVCs exist at the transmission level of the power system, but they
are non-existence at the distribution level renders utilities to opt for solutions that lead to switching
transients and overvoltage problems [1-2]. The prototype design for RSVC is for a 25 kVA pole-mounted
transformer typically serving three residential homes. The concept of RSVC is an extension of already
developed SVC that are in-service on the transmission side of the power system. In contrary to SVC, RSVC
regulates residential voltages, especially during peak demand hours, when the benefits coincide best with
the interests of customers and those of the electric companies. These RSVCs provide an additional tool for
smart demand-side management. By remotely controlling RSVCs, utilities can apply CVR at specified
individual locations during peak demand hours. The device is developed with a goal to maximize VAr
compensation advantages and minimize the cost of components involved. It is the role of electric utility
companies to develop strategies for CVR. A proposed solution involves installing an RSVC device at the
pole mounted transformer which will decrease the voltage before each customer’s service main. RSVC
design can be easily scaled up for larger residential homes, buildings, and even neighborhoods with single-
phase or three-phase service transformers. The discussion in this chapter is based on designing a single
phase RSVC that will serve a typical 25-kVA residential load from a pole-mounted service transformer. A
complete open-loop design for RSVC along with reactive component sizing and bidirectional switches
topology is presented towards the end of this chapter.
Distribution Network Modeling
The prototype of a RSVC is designed to serve single-phase residential loads connected to a 25 kVA pole
mounted service transformer. For sizing RSVC reactive components, it is deemed necessary to analyze the
primary side of the pole mounted transformer so that the strength of substation network as seen from the
RSVC device is included in the simulation. In general, an ideal distribution network will appear as an
infinite bus with negligible reactance to an RSVC device.
To study the effects of distribution network on RSVC, a 5-mile-long distribution feeder serving five
uniformly distributed 1 MW loads was modelled using 397.5 MCM ACSR (Aluminum Conductor Steel
Staff_PR_38 Attachment A
30
Reinforced) conductor. The secondary sides of the service transformers are held at 0.95 per unit i.e., the
minimum allowable service voltage outlined in ANSI c84.1, by using reactive generators.
Figure 4-1: PowerWorld Model for a Distribution Network
Figure 4-1 shows a model of a distribution feeder that is being fed by a substation. The substation is modeled
as an ideal source that can server any power require by the distribution network. The distribution network
consists of five distributed loads of 1 MW at unity power factor. The parameters used to calculate the
Thevenin impedance equivalent of the distribution feeder are as follow; the power base is 100 MVA; the
voltage base is 240 V; the leakage reactance for the service transformer is 10% in a 2 MVA base. The
reactive generator maintains the voltage at the secondary of service transformer at 0.95 per unit by providing
the reactive power that is specified in Table 4-1. In Table 4-1, the reactive generator that is closer to the
substation is labeled as Q1 and so on. The total active and reactive power supplied by the substation is 4.86
MW and 1.09 MVAr respectively.
In order to calculate the strength of the distribution network, the Thevenin equivalent of the distribution
network is found at a suitable point from the substation. In this case, the Thevenin equivalent of the
distribution network is calculated at the third generation point. Figure 4-2 shows the impedance diagram
for the distribution network in Figure 4-1. The components value in impedance diagram are in per unit
system referred to the low side of service transformer.
Staff_PR_38 Attachment A
31
Table 4-1: Reactive power requirement for each generator in the Distribution Network
Substantion Generator
j755.07 408
LT1= j5
Thevenin Equivalent looking
into this point
R1 = 100L1 = j208.33
L
O
A
D
L
O
A
D
L2= j208.33 R2 = 100
L
O
A
D
L3 = j10000 R3= 100
LO
AD
C4 = -j1111.11 R4= 100
L
O
A
D
C5 = -j714.29 R5= 100
Transformer Leakage
LT2= j5
Transformer Leakage
LT3= j5
Transformer Leakage
LT4= j5
Transformer Leakage
LT5= j5
Transformer Leakage
Line Reactance
j755.07 408
Line Reactance
j755.07 408
Line Reactance
j755.07 408
Line Reactance
j755.07 408
Line Reactance
Figure 4-2: Equivalent impedance circuit for distribution network to calculate Thevenin impedacnce
Using a MATLAB script to solve the equations, the Thevenin impedance value (𝑍) comes out to be
0.0501 + j0.0086. The result shows that a distribution network will appear as an infinite bus to an RSVC
device.
Residential Loads Modeling
Residential loads vary enormously and involve sophisticated load modeling algorithms to predict the
residential load profiles [3-5]. The primary purpose for RSVC is to regulate residential loads for CVR
application. This voltage regulation is mainly dependent on the size of the reactive components involved.
Therefore, a general perception of the load profile is sufficient to model residential loads.
For RSVC, residential loads were modeled using load profiles provided by Avista for Spokane downtown
network for the year of 2013. The load profile data is available on per hour basis. In order to understand the
Reactive
Generator
Reactive
Power
(MVAr)
Reactance
(per unit)
Q1 -0.48 -208.33
Q2 -0.19 -526.32
Q3 -0.01 -10000
Q4 0.09 1111.11
Q5 0.14 714.29
Staff_PR_38 Attachment A
32
behavior of residential loads at the distribution transformer, the power factor of the load was calculated
using the load profiles. Based on power factor calculated, average power factors were determined for winter
peak month and summer peak month.
Figure 4-3 shows power factor variation during January 2013 and June 2013. The dashed line in Figure 4-
3 represents the average power factor calculated for each peak. The average power factor for winter peak
is found out to be 0.9446 (lagging) while the average power factor for summer peak is found out to be
0.8724 (lagging).
Figure 4-3: Power Factor variation during January’13 and June’13 for Post St. West 13521 Downtown
Spokane, WA
As stated earlier, the voltage regulation capability of RSVC is mainly dependent on the size of reactive
components involved. Therefore, a general idea regarding the residential loads is sufficient to properly
simulate the RSVC device with residential customer loads fed from a 25-kVA pole-mounted service
transformer. For the purpose of RSVC, the power factor for residential loads were considered to be 0.95
lagging. The resistive and inductive component of the load model were calculated to be 2.4253 Ω and
19.5726 mH respectively.
po
w
e
r
f
a
c
t
o
r
po
w
e
r
f
a
c
t
o
r
Staff_PR_38 Attachment A
33
Modeling Service Transformer Leakage Reactance
Leakage reactance plays an important role when considering the operation for an RSVC. The proposed
location for an RSVC installation will be in close proximity of the service transformer. Thus, it becomes
increasingly important to study the effects of leakage reactance for the service transformer.
In an ideal transformer, all the flux will link with both primary and secondary windings of the transformer.
However, in practice it is impossible to link all the flux with both primary and secondary of the transformer.
There will be a small amount of flux which will leak out of the either of the windings. This flux is called
leakage flux which will pass through the winding insulation thus contributing to the leakage reactance of
the transformer. While sizing the SVC components, the transformer leakage reactance was assumed to vary
between 10% to 20% of the rated transformer reactance 𝑋,. The parameters used to calculate the
transformer leakage reactance are similar to parameters used for modeling Thevenin equivalent of the
distribution network i.e. the power base is 100 MVA; and the voltage base is 240 V. Assuming the core
resistance is negligible, for 10% or 0.1 per unit of the rated service transformer reactance, the leakage
reactance is 0.2304 Ω whereas for 20% or 0.2 per unit the leakage reactance is 0.4608 Ω.
RSVC Reactive Component Sizing
The discussion in this section focuses on the RSVC circuit that is shown in Figure 4-4. As discussed in the
previous two sections, the leakage reactance for transformer is assumed to be 10 % of the rated transformer
impedance and residential loads at distribution transformer are modeled at 0.95 lagging power factor. It is
assumed that an addition 5 % reactance of cable wiring is also present and it is combined with the
transformer reactance.
The most important design parameter for an RSVC is its reactive components. The success of RSVC relies
on its ability to regulate the voltage at the service transformer. Reactive power needs are highly dependent
on the load profiles for specific service transformer but in order to generalize the RSVC design, worst-case
scenarios for voltage regulations were considered to size the reactive components.
A capacitor provides reactive support by providing VArs to the power system. Adding VArs to the power
system aid in boosting the receiving end voltage as discussed in the previous chapter. For residential
customers, the minimum permissible voltage is minus 5% of the ANSI c84.1 standard. Therefore, RSVC
capacitive power needs should be calculated in order to maintain the minimum nominal voltage at service
transformers. As reactor (LSVC) contributes in absorbing the VArs, it can be isolated from the proposed
RSVC design by opening switch SW1. The reactor is essentially cut off from the remaining circuit when
switch SW1 is opened and reactor current flows through switch SW2. The simulations performed resulted
Staff_PR_38 Attachment A
34
in a capacitive reactive power of 10.205 kVAr. Simulation results for capacitor modeling is shown in Figure
4-5.
Figure 4-4: RSVC circuit for calculating Fixed Capacitor
Figure 4-5: Simulation result for RSVC capacitor modeling when AC mains voltage is 228 V
The proposed RSVC design is based on Fixed-Capacitor with PWM switched inductor. Once the reactive
requirements for RSVC capacitors are determined, the reactor can be sized accordingly to meet the reactive
power needs. As with the case of sizing capacitor, reactor is modeled at the worst-case voltage condition
for CVR purpose. This condition will occur when the distribution transformer is operating at the nominal
voltage i.e. 240 V. In addition to nominal voltage at the service transformer, the effect for the RSVC fixed
RS
V
C
O
u
t
p
u
t
V
o
l
t
a
g
e
(
V
o
l
t
)
Staff_PR_38 Attachment A
35
capacitor must be included for determining the reactive power requirement for a reactor. In other words,
the reactor is sized to compensate the voltage at service transformer along with the VArs added by the
RSVC fixed capacitors. The simulations performed resulted in a reactor reactive power of 11.752 kVAr.
The simulation results for inductor modeling are shown in Figure 4-6.
Figure 4-6: Simulation result for RSVC inductor modeling with fixed capacitor and at AC mains voltage
of 240 V
Summary for RSVC Reactive Requirements
Table 4-2 shows a summary of the RSVC reactive requirements for different distribution transformer
reactance.
Table 4-2: Reactive requirements for RSVC
Transformer Reactance (including the
wire reactance)
(% of rated transformer reactance)
Load Reactive
Requirement
RL
(ohms)
XL
(ohms)
QC
(kVAr)
QL
(kVAr)
10 2.43 7.39 10.205 11.752
15 2.43 7.39 11.4 6.79
20 2.43 7.39 12.594 3.819
RS
V
C
O
u
t
p
u
t
V
o
l
t
a
g
e
(
V
o
l
t
)
Staff_PR_38 Attachment A
36
From the table above it is found that the reactive requirements for the SVC components are reasonable for
a reactance of 15%. This value of reactance is realistic for a 10% transformer reactance plus an additional
5% of cable wiring.
Simulation Results for RSVC
Simulation for RSVC was done using a fixed capacitor of 10 kVAr capacitor and inductor of 11.752 kVAr.
The leakage inductance of distribution transformer was modeled as 10% of the rated reactance. Figure 4-7
shows the simulation model for RSVC. The simulation results were obtained a switching frequency of 1
kHz and a fixed duty cycle of 0.5. The simulation results presented in this section are compiled by Simulink
and plotted in Matlab for readability.
Figure 4-7: RSVC Simulink model with fixed capacitor and switched inductor having bidirectional
switches
Gate driving signals
In the Simulink model, gate driving signals were generated using Xilinx System Generator which
incorporated the hardware descriptive language procedures to produce the logical conditions for
bidirectional switch commutation. Figure 4-8 shows the transitioning from top bidirectional switch to
bottom bidirectional switch and vice versa for both voltage polarities. It is important that commutation from
one switch to another switch is performed following the state machines described in Figure 3-10. An
Staff_PR_38 Attachment A
37
incorrect transition state will result in huge current and voltage spikes that will destroy the switching
devices.
Figure 4-8: RSVC commutation sequences for bidirectional switches
Input and Output Voltage Waveforms
As mentioned in the previous chapters, PWM-switched inductor with conventional relays poses two serious
hazards for the safe operation of RSVC. These threats include the shortening of the AC mains and opening
a continuous inductive current. To mitigate these hazards, a bidirectional switch with safe commutation
strategy is proposed. This commutation strategy ensures that the mains input is not shorted at any instant
during the operation of RSVC. Figure 4-9 shows the simulation result for the input and output voltage
which indicates that AC mains remains sinusoidal without shortening phase to neutral.
Lo
g
i
c
l
e
v
e
l
Lo
g
i
c
l
e
v
e
l
Staff_PR_38 Attachment A
38
Figure 4-9: Simulation results for RSVC input and output voltage
Inductor voltage and current waveforms
An important parameter of success for RSVC operation is that the inductor current remains sinusoidal which
results in fewer or no harmonics. Figure 4-10 shows the output for the RSVC inductor current and RSVC
inductor voltage. It can be seen that inductor voltage is chopped replica of the input AC mains waveform
for the time specified by the duty cycle of the top bidirectional switch. Also the inductor current, which
lags the inductor voltage by about 90°, remains sinusoidal throughout the operation of RSVC irrespective
of the duty cycle at which the inductor is switched.
AC
M
a
i
n
s
V
o
l
t
a
g
e
(
V
o
l
t
s
)
RS
V
C
V
o
l
t
a
g
e
(
V
o
l
t
s
)
Staff_PR_38 Attachment A
39
Figure 4-10: Simulation results for RSVC inductor voltage and RSVC inductor current
Current through bidirectional switches
Bidirectional switches have the capability to conduct current in both directions. Current through the top and
bottom switch combine to form an envelope of the total inductor current. Figure 4-11 shows the current
through top and bottom switches of the RSVC. There is a commutation period which prohibits the
commutation of the gate driver signals near zero crossing of the input voltage. In this way, the commutation
of gate signals is properly carried out and the possibility of detecting the wrong sign of input voltage near
zero crossings is eliminated.
In
d
u
c
t
o
r
V
o
l
t
a
g
e
(
V
o
l
t
)
In
d
u
c
t
o
r
C
u
r
r
e
n
t
(
A
m
p
)
Staff_PR_38 Attachment A
40
Figure 4-11: Simulation results for RSVC top and bottom bidirectional switches
RSVC output voltage for different duty cycles
The switched inductance value varies inversely with the square of the duty cycle amounting to a
continuously-variable inductor. This means that the reactive power absorbed by the inductor can be adjusted
by controlling the duty cycle of the top and bottom switches. The RSVC uses this dynamic reactive power
to regulate the residential voltages. In Figure 4-12, the duty cycle of the top switch was varied from 0.2 to
0.8. It can be seen that the RSVC output voltage reduces as the duty cycle increases. Thus by controlling
the duty cycle, it is possible to regulate the residential voltage.
Figure 4-12: Simulation results for RSVC output voltages at different duty cycles of bidirectional switch
To
p
S
w
i
t
c
h
C
u
r
r
e
n
t
(
A
m
p
)
Bo
t
t
o
m
S
w
i
t
c
h
C
u
r
r
e
n
t
(
A
m
p
)
9.95 9.955 9.96 9.965 9.97 9.975 9.98 9.985 9.99 9.995 10time(sec)
-400
-300
-200
-100
0
100
200
300
400
RS
V
C
O
u
t
p
u
t
V
o
l
t
a
g
e
(
V
o
l
t
)
Output Voltage at D = 0.2
Output RMS = 234.44 V
AC Mains
RSVC Output VoltageRMS RSVC Voltage
9.95 9.955 9.96 9.965 9.97 9.975 9.98 9.985 9.99 9.995 10time(sec)
-400
-300
-200
-100
0
100
200
300
400
RS
V
C
O
u
t
p
u
t
V
o
l
t
a
g
e
(
V
o
l
t
)
Output Voltage at D = 0.4
Output RMS = 229.47227.18 V
AC Mains
RSVC Output VoltageRMS RSVC Voltage
9.95 9.955 9.96 9.965 9.97 9.975 9.98 9.985 9.99 9.995 10
time(sec)
-400
-300
-200
-100
0
100
200
300
400
RS
V
C
O
u
t
p
u
t
V
o
l
t
a
g
e
(
V
o
l
t
)
Output Voltage at D = 0.6
Output RMS = 227.18 V
AC Mains
RSVC Output Voltage
RMS RSVC Voltage
9.95 9.955 9.96 9.965 9.97 9.975 9.98 9.985 9.99 9.995 10
time(sec)
-400
-300
-200
-100
0
100
200
300
400
RS
V
C
O
u
t
p
u
t
V
o
l
t
a
g
e
(
V
o
l
t
)
Output Voltage at D = 0.8
Output RMS = 225.23 V
AC Mains
RSVC Output Voltage
RMS RSVC Voltage
Staff_PR_38 Attachment A
41
Low-Voltage RSVC Testing
Lab testing of the RSVC at low voltage was performed using a fixed capacitor of 10 kVAr capacitor and a
low-rated laboratory inductor. The leakage inductance of distribution transformer was modeled as 10% of
the rated reactance. The switching frequency was set to 12 kHz with fixed 50 % duty cycle. Figure 4-13
shows the lab testing model of the RSVC. Initial testing of the RSVC model was performed without the
residential loads with AC mains voltage set at 30 Vrms.
Figure 4-13: Lab testing model for RSVC
Gate driving signals
Figure 4-14 shows the gate driving signals for the bidirectional switches. These signals were generated
using Spartan3E Field Programmable Gate Array (FPGA) board. In order to turn on the high power IGBTs,
opto-isolators are used to provide the required currents to drive the low power FPGA signals. Opto-isolators
provide a high electrical isolation between the high and low terminals allowing relatively small FPGA
digital signals to control much large AC mains voltages, currents and power.
SW1
SW2
LEAKAGE
REACTANCE
CAPACITOR
AC
MAINS
PWM BASED SWITCHED
INDUCTOR
INDUCTOR
SW1ASW1A
SW1B
SW2B
SW2A
V12
Staff_PR_38 Attachment A
42
Figure 4-14: Gate signals generated using FPGA
Input and Output Voltage Waveforms
As mentioned previously, an important criterion for measuring successful operation of RSVC device is
preventing short-circuiting of the input voltage.
Figure 4-15 shows the input and output voltages. The top waveform represents the AC mains and the
bottom chopped waveform shows the output of RSVC without loads.
Figure 4-15: Experimental result at low voltage for RSVC input voltage and output voltage
Staff_PR_38 Attachment A
43
Inductor current waveform
The first waveform in Figure 4-16 shows the input voltage; the second represents the output voltage and
the third shows the inductor current. The inductor current is scaled as 100 mV/A. The inductor current is
continuous and produce little harmonics. It is important to keep the prohibition period for commutation as
little as possible. Even though this period aids in detecting the correct sign of the input voltage but a large
prohibition period will cause the inductor to become non-sinusoidal.
Figure 4-16: Experimental result at low voltage for RSVC inductor current
Inductor voltage waveform
The output voltage across the inductor should be a chopped replica of the input AC mains. Figure 4-17
shows the AC mains voltage and chopped inductor voltage. For the sake of clarity, the switching frequency
was lowered to 1 kHz.
Staff_PR_38 Attachment A
44
Figure 4-17: Experimental result at low voltage for RSVC inductor voltage
Current through top bidirectional switches
The first waveform in Figure 4-18 shows the input voltage, the second represents the output voltage and
the third shows the current through the top bidirectional switch. This current waveform is a chopped
envelope for the inductor current. The current is scaled as 100 mV/A
Figure 4-18: Experimental result at low voltage for RSVC current through top switch
Staff_PR_38 Attachment A
45
Hardware Prototype Board
After the successful testing of RSVC operation at low voltage, an RSVC prototype was developed for
testing the operation of RSVC at rated voltage and current. Figure 4-19 and Figure 4-20 show the schematic
and layout of the RSVC prototype board respectively. The low-power electronics is electrically isolated
from the high-power IGBTs (or MOSFETs) using opto-isolators. These opto-isolators provide necessary
current to turn on the IGBTs when digital high is applied from the FPGA board.
Figure 4-19: RSVC prototype board schematics
Figure 4-20: RSVC prototype board layout
Staff_PR_38 Attachment A
46
RSVC hardware prototype testing
Figure 4-21 shows the RSVC hardware prototype testing setup. The switching board is mounted on heat
sinks to dissipate the heat generated by the bidirectional switching devices. Testing for RSVC prototype is
undergoing.
Figure 4-21: RSVC hardware prototype testing
Problems encountered during RSVC hardware testing
Initial testing of RSVC was performed with prototype switching board. Figure 4-22 shows the result of the
inductor voltage. The chopped inductor has some spikes whenever the top switch is turned on. Moreover,
an abnormal behavior was detected during the prohibition period for commutation.
Detailed testing for the RSVC prototype is left for the next phase. Results complied with the RSVC
prototype testing will be incorporated in the next phase report.
Staff_PR_38 Attachment A
47
Figure 4-22: Problems encountered during RSVC prototype testing
References
[1] T. E. Grebe, "Application of distribution system capacitor banks and their impact on power quality,"
in IEEE Transactions on Industry Applications, vol. 32, no. 3, pp. 714-719, May/Jun 1996.
[2] R. A. Adams, S. W. Middlekauff, E. H. Camm and J. A. McGee, "Solving customer power quality
problems due to voltage magnification," in IEEE Transactions on Power Delivery, vol. 13, no. 4, pp. 1515-
1520, Oct 1998.
[3] G. C. Giaconia, G. Fiscelli, F. L. Bue, A. Di Stefano, D. La Cascia and R. Miceli, "Integration of
distributed on site control actions via combined photovoltaic and solar panels system," Clean Electrical
Power, 2009 International Conference on, Capri, 2009, pp. 171-177.
[4] F. Y. Xu, X. Wang, L. L. Lai and C. S. Lai, "Agent-Based Modeling and Neural Network for Residential
Customer Demand Response," 2013 IEEE International Conference on Systems, Man, and Cybernetics,
Manchester, 2013, pp. 1312-1316.
[5] A. Capasso, W. Grattieri, R. Lamedica and A. Prudenzi, "A bottom-up approach to residential load
modeling," in IEEE Transactions on Power Systems, vol. 9, no. 2, pp. 957-964, May 1994.
Staff_PR_38 Attachment A
48
STUDY METHODOLOGY
Analysis Tools
OpenDSS
Electric Power Research Institute (EPRI) Open-Source Distribution System Simulator (OpenDSS) is a
comprehensive electrical system simulation tool developed to perform most analyses required by utilities
in distribution networks. OpenDSS can be implemented as a stand-alone executable program, or can be
driven by a variety of well-known software tools through a Component Object Model (COM) interface.
Two of the most popular software tools used to drive OpenDSS are MS Office Tool through Visual Basic
Applications (VBA) and MATLAB from Mathworks. The ability to be driven by a third-party analysis
program makes OpenDSS a powerful software tool.
OpenDSS can compile user-written models by using Dynamic Link Libraries (DLL), allowing the user to
focus on the model of the device and let OpenDSS handle the distribution system modeling. The DLL can
be written in most common programming languages.
Figure 5-1 shows the basic architecture of OpenDSS. The software accepts entries from different sources
such as user-written DLLs, COM interface, and its own script language. OpenDSS is able to display results
and also export them in the form of csv files for post-processing the data using a third-party software [1].
Figure 5-1: OpenDSS architecture
Staff_PR_38 Attachment A
49
The OpenDSS tool has been used for:
Distribution planning and analysis.
General multi-phase AC circuits analysis.
Analysis of distributed generation interconnections.
Annual load and generation simulations.
Wind plant simulations.
Analysis of unusual transformer configurations.
Harmonic and inter-harmonic analysis.
Neutral-to-earth voltage simulations.
Development of IEEE test feeder cases.
The program has several built-in solution modes, including:
Snapshot power flow
Daily power flow
Yearly power flow
Harmonics
Dynamics
Fault study
OpenDSS accepts data in different formats, minimizing the conversion effort by utilities and users.
OpenDSS allows data entry using tables, and is also script-based which provides the flexibility to model
almost any configuration.
In recent years, many of the new Distributed Generation (DG) projects have been connected to the sub-
transmission and distribution networks. Traditionally, distribution networks have been radial, and with the
increase of DG in distribution networks, a software simulator able to model distribution networks with
multiple sources was needed [2].
OpenDSS was developed to meet the need of distribution networks with multiple sources and power
flowing in both directions. OpenDSS is also able to capture the time and location dependence of DG, giving
a more accurate model of a distribution network with DG. OpenDSS has a frequency-domain simulation
engine with features to create models of electric power distribution systems and to perform any types of
analysis related to distribution planning and power quality.
Staff_PR_38 Attachment A
50
GridPV
GridPV is a toolbox for MATLAB® developed at Georgia Institute of Technology and Sandia National
Laboratories. It is a collection of functions that facilitate the use of OpenDSS via the COM interface.
GridPV is a well-documented tool for Matlab that can be used to build distribution grid performance models
using OpenDSS. Simulations with this tool can be used to evaluate the impact of solar energy on the
distribution system. The majority of the functions are useful for interfacing OpenDSS and Matlab, and they
are of generic use for commanding OpenDSS from Matlab and retrieving information from simulations [3].
Although GridPV was developed to study the integration of solar, the toolbox was of great use in this
project. The ability to extract data from the simulations and apply complex control algorithms to modify
the simulation parameters helped streamline the simulation process.
Data Conversion
The nature of the study requires the ability to perform time-series analysis also known as pseudo or quasi
steady-state time-series (QSTS). At the time this project was performed, PowerWorld (PW) and SynerGI
were in the early stages of developing time-series analysis and were not ready to perform the studies
required for this project. Due to the limitations of PW and SyngerGI, the use of OpenDSS was needed. The
data from both models was translated from PW and SynerGI to OpenDSS.
PowerWorld Data Conversion
The data provided by Avista for this section of project includes the following information for the downtown
Spokane power network.
Downtown network maps
PowerWorld Model (peak load)
Primary feeder maps
Secondary feeder maps
Hourly load data including kW, kVAR and PF from 2013-2015
The PowerWorld model was solved to ensure the case converged without any errors. Once the model
converged, the data was sent to excel in a .xlsx format by using the export tool inside PowerWorld. Once
the data was in excel, the data was separated into several files, one file for substation transformers, one for
lines, one for loads and finally one for service transformers.
Staff_PR_38 Attachment A
51
Using the spreadsheets, each of the files were manipulated to match the proper OpenDSS input format.
Each file was tested separately to facilitate the debugging process. A master file was created to test the
entire four networks at the same time.
OpenDSS can only have one circuit active at any given time, but it has the ability to create an unlimited
number of subnetworks. To overcome this limitation, an equivalent voltage source was created to represent
the 115 kV system, then four subnetworks were created, each subnetwork containing a slack bus and tied
to the main voltage source. A meter was placed at the head of every subnetwork to measure the active and
reactive power flowing to every network.
Figure 5-2: Data Conversion
Figure 5-3: Model Structure
Staff_PR_38 Attachment A
52
SynerGI Data Conversion
The data provided by Avista for this section of project includes the following information for the SAG-741
feeder.
Access data base (SynerGI model)
Measured load data
Measured voltage at regulator
The model was provided in an “.mdb” file (access data file). The conversion of the model was done by
creating separate files for the lines and the loads. The data was imported from access to excel and then
manipulated to match the input format required by OpenDSS. The conversion was done one file at the time,
performing testing and debugging at the end of every file. Finally, a master file was created to put all the
parts together and test the entire feeder.
Model Validation
Downtown Spokane Feeder Validation
The model built in OpenDSS was compared to the model provided in PW for verification purposes. Figure
5-4 shows the voltage at the four networks in OpenDSS and in PW. Table 5-1 shows the results of the two
different models for the downtown Spokane network.
Figure 5-4: Bus Voltage
Staff_PR_38 Attachment A
53
Table 5-1: Comparison of PowerWorld (PW) and OpenDSS (OD) Simulation Models for Peak Load
Active Power Reactive Power Power Factor
PW
(kW)
OD
(kW)
PW
(kVAr)
OD
(kVAr)
PW
(kVAr)
OD
(kVAr)
Network 1 9,058 9,038.5 5,257 5,285 0.864 0.863
Network 2 14,418 14,370 8,242 8,018 0.868 0.873
Network 3 12,019 11,979 5,704 5,546 0.903 0.907
Network 4 8,464 8,438 3,895 3,852 0.908 0.909
Feeder SAG-741 Model Verification
The model verification for the SAG-741 feeder was done in two steps. First, three cases were prepared for
a static analysis. One for winter peak, one for summer peak and one for a light load condition. The total
load was compared to the measured data provided by Avista. The second step of verification was performed
using the hourly measure data provided. The total energy consumed by the feeder was calculated by
integrating all the hourly power data over the year. Table 5-2 shows the three cases used for the model
validation.
Table 5-2: SAG 741 Power Flow Cases
Winter Peak Summer Peak Light Load
PTotal 3470 2041.7 695.6
QTotal 1254.9 392.3 -419.7
PLoss 111.4 36.8 4.6
QLoss -103.0 -193 -230.8
Table 5-3 shows the results of the measured data compared to the model over the course of a year. The load
allocation algorithm was modified to better fit the measured data.
Staff_PR_38 Attachment A
54
Table 5-3: Feeder SAG-741 Validation Results
Total Energy (kWh) Max kW
Measure Data (Avista) 13,727,435 3,492
OpenDSS Model 13,719,071 3,470
% Error 0.0609% 0.6300%
The voltage profile of the feeder at the three different loading conditions is shown in Figure 5-5.
Figure 5-5: Voltage Profile at three different loading conditions
Staff_PR_38 Attachment A
55
References
[1] Nie, Song, et al. "Analysis of the impact of DG on distribution network reconfiguration using OpenDSS,"
Innovative Smart Grid Technologies-Asia, 2012.
[2] Liu, Hao Jan, and Thomas J. Overbye. "Smart-grid-enabled distributed reactive power support with
Conservation Voltage Reduction," Power and Energy Conference at Illinois, 2014.
[3] M. J. Reno and K. Coogan, "Grid Integrated Distributed PV (GridPV) Version 2," Sandia National
Laboratories SAND2014-20141, 2014.
Staff_PR_38 Attachment A
56
STUDY RESULTS
The analysis of the RSVC consisted of simulating a model of the device with two different feeder
topologies. The first one, downtown Spokane, is a tightly coupled grid with a strong source. The main
objective of the downtown network study was to study the possibility of correcting the power factor rather
than to support the voltage. The downtown network is a short dense network that has little voltage drop and
hence no voltage problems.
The Spokane downtown network was initially model in PW as a balanced three-phase model. Distribution
systems are more unbalanced than transmission systems, assuming that the distribution network is a
balanced system can cause some inaccuracies.
The SAG-741 feeder was originally model in SynerGI, and it was modeled as an unbalanced three-phase
feeder. The unbalanced model is a more accurate representation of the feeder and the results obtained from
the SAG-741 study could be more accurate due to the modeling of the unbalance system.
The Spokane downtown network explicitly models the service transformer and the service line, while the
SAG741 does not. The modeling of service transformer and the service line provides a more accurate model.
It is a standard practice in utilities to only model the medium voltage buses and just assume a constant
voltage drop from the feeder level to the customer meter.
Spokane Downtown Feeder
The first study consists on deploying multiple RSVCs at the downtown network of Spokane, Washington
with the objective to correct the power factor. The deployment of the RSVCs was performed uniformly
throughout the downtown network, and the size of the RSVC was selected based on the size of the
transformer it was connected to.
The original RSVC was developed as a 15 kVA device to be connected to a 25 kVA service transformer.
The ratio from the RSVC to the transformer rated capacity is 60%. The same ratio was used in the sizing
of the RSVCs used in the Spokane Downtown network study.
Spokane Feeder Description
The downtown network is composed of four independent networks fed from the 115 kV system. Every
network has two parallel transformers that convert the voltage from 115 kV to 13.2 kV. The voltage is
distributed across the downtown network at the 13.2 kV voltage. Finally, the voltage is stepped down to
utilizable voltage by the customers, using several service transformers, to either 240 V or 480 V.
Staff_PR_38 Attachment A
57
The original model was built in PW, and consisted of four different networks. Each network is modeled
independently and has an equivalent generator, “slack” bus, which represents the connection to the 115 kV
grid. The equivalent generation provides the necessary active and reactive power for the networks. The
active and reactive power flow results of the four networks are shown in Table 6-1, along with the
corresponding power factor.
Table 6-1: Downtown Spokane Network Power Flow Results
Active Power
(kw)
Reactive Power
(kVAR)
Power Factor
Network 1(PST115) 9,058 5,257 0.864
Network 2 (PST115) 14,418 8,242 0.868
Network 3 (Metro115) 12,019 5,704 0.903
Network 4 (Metro115) 8,464 3,895 0.908
The voltage at each bus obtained from the PW model for the maximum loading condition is plotted in
Figure 6-1. There are three buses with a voltage value of zero. These buses correspond to the normal open
switch between the networks. Figure 6-1 shows both medium voltage buses (13.2 kV) and low voltage
buses (240 and 480 V), with the medium voltage buses closer to the 1.05 per unit and the low voltage buses
having different values for each network. The figure also shows that the medium voltage network has a
more constant voltage then the low voltage network, where the voltage can vary significantly.
Figure 6-1: Downtown Spokane Network Bus Voltage
Staff_PR_38 Attachment A
58
Figure 6-2 shows one of the network with the 115 kV system modeled with a slack bus, two transformers
that step-down the voltage from 115 kV to 13.2 kV.
Figure 6-2: PW Single-line of Downtown Spokane Network
The complete single line diagram of one of the networks is shown in Figure 6-3
Figure 6-3: Downtown Spokane Complete Single Line Diagram
Staff_PR_38 Attachment A
59
Power Factor Correction Utilizing RSVC
Two cases were developed to test the benefits of deploying RSVCs in the downtown network. In both cases,
the same number of RSVCs were used. Assumptions made for every case and results obtained are presented
in the following case scenarios.
Case 1
Case one was developed by connecting one RSVC equal to 60% of the transformer rated capacity. Each
RSVC was controlled by local voltage. The voltage reference for each RSVC was set to 1.03 per unit.
Table 6-2 shows the results for case 1. Simulation results show that networks 1 and 2 have a power factor
close to unity and networks 3 and 4 are operating with a considerable low power factor.
The RSVCs utilized in this project are only controlled by voltage, and do not have the ability to be controlled
by reactive flow. One option to increase the power factor in networks 3 and 4 is to raise the voltage reference
of the RSVCs to force them to operate in a capacity mode.
Table 6-2: Results Case 1
Active Power
(kW)
Reactive Power
(kVAR)
Power Factor
Network 1 (PST115) 9,028 446 0.998
Network 2 (PST115) 14,374 2,573 0.984
Network 3 (Metro115) 11,984 4,367 0.939
Network 4 (Metro115) 8,441 4,193 0.896
Case 2
Case two was developed after case one was not able to correct the power factor in all four networks. The
voltage at the secondary side of network 3 and network 4 is relatively high and not all the RSVCs injected
their maximum reactive capacity. The voltage reference of all the RSVCs was raised to 1.04 per unit to
force more reactive power and help compensate for the reactive power absorbed by the loads. Table 6-3
shows the results for case two.
Staff_PR_38 Attachment A
60
Table 6-3: Results Case 2
Active Power
(kW)
Reactive Power
(kVAR)
Power Factor
Network 1 (PST115) 9,027 -610 -0.997
Network 2 (PST115) 14,372 765 0.998
Network 3 (Metro115) 11,978 1,064 0.996
Network 4 (Metro115) 8,433 1,449 0.985
Setting all the RSVCs to 1.04 per unit resulted in a better power factor overall. Network 1 is operating with
a leading power factor but really close to unity while networks 2, 3 and 4 are operating with a lagging power
factor, but also close to unity.
Future Work
The RSVCs utilized in the study were only controlled by voltage, thus the set point had to be raised to force
the RSVCs to operate in a capacity mode. A possible future project is the implementation of VAR control
in the RSVC. In short dense feeders such as the downtown network the voltage drop could be really small,
this is where a VAR controller could be beneficial.
The voltage set point for all the RSVCs was the same, another potential future project could be the
implementation of a dynamic controller where the voltage set point of every RSVC or a group of RSVCs
change the set point according to the loading conditions on the feeder.
SAG-741 Feeder
The SAG-741 is a feeder operating at 20.8 kV fed from the 115 kV system. The feeder has one 300-kVAR
capacitor bank close to the substation and 5 sets of voltage regulators, three of which are three-phase, one
which is two-phase and one which is a single-phase regulator. The substation has a three phase Load Tap
Changer (LTC) transformer.
The model contains distribution lines at the 20.8 kV level and the loads are modeled at the 20.8 kV level
without the explicit model of the service transformer and the service line, a typical situation in utilities.
Figure 6-4 is a not to scale oneline of the SAG-741 feeder, with the substation indicated by the blue star.
Staff_PR_38 Attachment A
61
Figure 6-4: Single Line Diagram of the SAG-741 Feeder
The SAG-741 is a winter peak feeder with a peak load of 3,470 kW. Figure 6-5 shows the load profile of
the feeder for one year, in 1-hour intervals. The load profile shows that there are at least three events where
the feeder was out of service. The points when the load becomes zero were replaced with the previous non-
zero value in the data set to avoid converging problems with the software.
Figure 6-5: SAG-741 Load Data
Staff_PR_38 Attachment A
62
To find the location of the RSVCs, an iterative process was used. First, the power flow was solved, and the
bus with the lowest voltage was identified. On the bus with the lowest voltage, an RSVC was placed on
that bus and the power flow was solved again. The process was repeated until the addition of an RSVC did
not improve the bus voltages. After the algorithm was run, an inspection was performed to ensure that two
RSVCs were not connected to the same bus. If two RSVCs were connected to the same phase at the same
bus, one of them was removed. The assumption that the service transformers are 25 kVA was made, and
hence only one RSVC was placed for a single bus.
Static Analysis
Four different case scenarios were created to determine the best combination of in-line voltage regulators,
capacitor banks and RSVCs that produce the least amount of energy consumed.
Case 1
Case one consisted of zero in-line voltage regulators and the substation capacitor bank on (fixed). To
develop this case, all the in-line voltage regulators were removed from the circuit. The capacitor bank was
put in service without any type of controller and fixed in the ON position.
A total of 35 RSVCs were added to the circuit. The majority of the RSVC were located at the end of the
feeder where the voltage was at its lowest. All the RSVCs were set to 117 Volts in a 120-V base for voltage
control. The RSVC’s were distributed as follows:
Phase Number of RSVCs used
Phase A 8
Phase B 13
Phase C 14
The voltage profile at peak load is shown in Figure 6-6. The addition of the RSVCs helped to balance the
voltage profile for all three phases. The fact that the RSVCs are single phase allows for great flexibility to
compensate an unbalanced feeder due to the single phase loads.
Staff_PR_38 Attachment A
63
Figure 6-6: Peak Load Voltage Profile Profile
The location of the RSVCs is shown in Figure 6-7, highlighted as an orange splash symbol.
Figure 6-7: SAG 741 Feeder RSVC Locations
Table 6-4: Test Result Case 1
Real Power (kW) 3,490.7
Reactive Power (kVAR) 1,193.8
Staff_PR_38 Attachment A
64
Case 2
This case was developed with the capacitor banks and voltage regulators off for the entire simulation.
A total of 35 RSVCs were used for this case and their phase distributions were:
Phase Number of RSVCs used
Phase A 8
Phase B 13
Phase C 14
The voltage profile at peak load is shown in Figure 6-8.
Figure 6-8: Peak Load Voltage Profile
Staff_PR_38 Attachment A
65
The single line with the RSVC distribution is shown in Figure 6-9
Figure 6-9: SAG 741 Feeder RSVC Locations
The total amount of power at peak load is shown in Table 6-5
Table 6-5: Results Case 2
Real Power (kW) 3,490.6
Reactive Power (kVAR) 1,787.6
Case 3
The third case was developed with the in-line voltage regulators and capacitor bank remaining in service
for the simulation. The location algorithm was implemented with the same process as with the case without
regulators. The amount of RSVCs needed in this case is considerable less than in the previous two cases. A
total of 9 RSVCs were used in this case, with the distribution of the RSVC’s as follows:
Phase Number of RSVCs used
Phase A 0
Phase B 5
Phase C 4
The voltage profile at peak load is shown in Figure 6-10.
Staff_PR_38 Attachment A
66
Figure 6-10: Peak Load Voltage Profile
The single line with the location of the different devices is shown in Figure 6-11.
Figure 6-11: SAG 741 Feeder RSVC Locations
The total power at peak load is shown in Table 6-6.
Staff_PR_38 Attachment A
67
Table 6-6: Results Case 3
Real Power (kW) 3,468.4
Reactive Power (kVAR) 1,343.8
Case 4
The fourth case consists of the in-line voltage regulators placed in service, and the capacitor bank out of
service. The same location and number of RSVC’s as in case 3 were used for this case.
Figure 6-12: Voltage Profile
Figure 6-13: SAG 741 Feeder
Staff_PR_38 Attachment A
68
Table 6-7: Case 4 Results
Real Power (kW) 3472.1
Reactive Power (kVAR) 1935.6
Time-series Analysis
The time-series analysis is the backbone of this project, since it provides a more realistic approximation of
the real benefits of deploying RSVCs in a distribution feeder.
The first step to run a time-series simulation is to have a seasonal base case starting point. In this project,
the winter peak was used since it is the season with the highest load. Once the base case is solved, measured
data provided by Avista at the feeder head was fed into the model. The model used a load allocation
algorithm to allocate the load every hour based on topology of the base case.
Most utilities only have measured data at the feeder head and a load allocation is usually used to estimate
the load at different parts of the feeder. The load allocation algorithm included with OpenDSS was used
since it is fairly accurate and there is no need to develop a new one.
The four previous cases were simulated in a yearly mode, and the total energy and peak load were recorded.
During the simulations it was found that the capacitor played a big role in the amount of losses if it was
kept on or off.
A local voltage control was developed to determine when the capacitor was in or out of service. With the
capacitor controller in place, two cases were investigated; one with the in-line voltage regulators in the
circuit and the second one without the in-line voltage regulators in the circuit.
The three cases were simulated for an entire year using the amount of RSVCs previously calculated for the
case with the in-line voltage regulators in place and for the case without the in-line voltage regulators.
Results showed that some of the RSVCs were only “working” for a small periods of time during the peak
load. The number of RSVCs in the feeder was reduced to maximize the utilization of the devices. The
simulations were re-run with the new number of RSVC calculated. The final number of RSVC utilized for
every case is shown in Table 6-8.
Staff_PR_38 Attachment A
69
Table 6-8: RSVC per case
Case Number of RSVCs
In-line voltage regulator (case 1) 8
No voltage regulator (case 2) 22
The load allocations from the previous cases were adjusted to match the total energy measured in the model.
The adjustment was done for all three cases and Table 6-9 shows the end results.
Table 6-9: Summary Results
Total Energy (kWh) Max kW
Base Case 13,719,071 3,470
Voltage Regulator in place 13,446,160 3,392
No Voltage regulators in the feeder 13,381,926 3,386
Future Work
One of the possible benefits of the RSVC is to damp the effects of switching capacitor banks. The effect of
RSVC on existing capacitor switching will be worth studying.
The size of the RSVC used in this project was equal to the prototype being developed at Boise State
University (15 kVA). There may be cases where a different size will be a better fit, especially if the service
transformers are large, 100 kVA or more. An automatic algorithm that calculates the proper size will need
to be developed.
The algorithm used to place an RSVC is in its early stages and it could be improved. The use of an Optimal
Power Flow (OPF) algorithm could be a great improvement from the current algorithm used to place the
RSVCs.
The dynamic interaction between RSVCs and other voltage control devices was not studied in this project
and should be taken into account if a large enough number of units are used, especially if they are close
together to each other.
Staff_PR_38 Attachment A
70
COST BENEFIT ANALYSIS
The simulation results for the base case yielded a total energy consumption of 13,719,071 kWh per year.
The peak consumption was recorded to be 3,470 kW.
Table 7-1: Summary of Results
Energy Used
(kWh)
Peak Power
(kW)
Energy
Saved
(kWh)
Peak
Reduction
(kW)
Potential savings
($/year)
Base Case 13,719,071 3,470 NA NA NA
Case 1 13,446,160 3,392 272,911 78 $12,281
Case 2 13,381,926 3,386 337,145 84 $15,172
The break-even point for the 2 cases is shown in Table 7-2. The break-even point was calculated based on
a $45/MWh of avoided cost, and is shown in the following equation.
Break-even period𝒚𝒆𝒂𝒓𝒔 #𝑹𝑺𝑽𝑪𝒔∗$𝟏𝟎𝟎𝟎𝒆𝒂𝒄𝒉
𝒔𝒂𝒗𝒊𝒏𝒈𝒔/𝒚𝒆𝒂𝒓
Table 7-2: Break-even period for two cases
Break-even
period
Number of
RSVCs
Case 1 0.65 years 8
Case 2 1.45 years 22
The goal of this project is to keep the price of the RSVC under $1000, which is considerably lower than the
breakeven price for both cases. It is worth mentioning that in case 2 all the voltage regulators were taken
out of service. The savings associated with maintenance and possible replacement of these devices is not
Staff_PR_38 Attachment A
71
considered in the calculation, but it could be a significant part of the budget, especially if a voltage regulator
needs to be replaced.
Staff_PR_38 Attachment A
72
PATH TO MARKET
The potential market path of the residential static var compensator (RSVC) being built and simulated in
this project is similar to that of other shunt-connected, reactive-injection-based devices currently being
deployed by some utilities. An RSVC prototype is being tested in hardware at Boise State University for a
voltage control application (Conservation by Voltage Reduction) on the consumer side of the distribution
feeder. The single-phase RSVC device has the advantage over conventional shunt capacitors of being able
to operate in a capacitive or inductive mode without generating large undesirable harmonics. The RSVC’s
harmonic footprint is not typical of most thyristor-based SVCs currently deployed. The RSVC uses a novel
pulse width modulation (PWM) scheme to create the variable VAR compensation, which pushes the RSVC
harmonics into a higher frequency band. This smart device can be used in multiple applications such as
continuous voltage control at a load point, power factor control, and mitigation of power quality issues. The
benefits of mass deployment of RSVCs on the consumer side of distribution networks will be demonstrated
through a number of simulation studies proposed in this project. The new RSVC device has the potential to
disrupt other competitor’s devices on three fronts: cost, power quality, and smart-grid applicability or
compatibility.
Staff_PR_38 Attachment A
73
APPENDIX
Table A.1 OpenDSS Power Flow Sample Description Language
# Lines and Buses
New Line.Line1 Phases=3 Bus1=100 Bus2=200 r1=0.0017424 x1=0.0017424 c1=0 length=1 units=kft
enabled=FALSE
New Line.Line2 Phases=3 Bus1=100 Bus2=121 r1=0 x1=0.7997616 c1=0 length=1 units=kft
enabled=TRUE
New Line.Line3 Phases=3 Bus1=121 Bus2=131 r1=0.0052272 x1=0.008712 c1=0.34848 length=1
units=kft enabled=TRUE
New Line.Line4 Phases=3 Bus1=122 Bus2=132 r1=0.0069696 x1=0.0104544 c1=0.34848 length=1
units=kft enabled=TRUE
New Line.Line5 Phases=3 Bus1=123 Bus2=133 r1=0.0052272 x1=0.008712 c1=0.34848
# Loads
New Load.Load1 Phases=3 Bus1=1501 kV=0.21 kW=78 kVAR=39.96 model=1 enabled=TRUE
New Load.Load2 Phases=3 Bus1=1501 kV=0.21 kW=0 kVAR=0 model=1 enabled=FALSE
New Load.Load3 Phases=3 Bus1=1504 kV=0.21 kW=18.43 kVAR=9.44 model=1 enabled=TRUE
New Load.Load4 Phases=3 Bus1=1504 kV=0.21 kW=0 kVAR=0 model=1 enabled=FALSE
New Load.Load5 Phases=3 Bus1=1508 kV=0.21 kW=7.84 kVAR=4.01 model=1 enabled=TRUE
# Transformers
New Transformer.XMFR1 Phases=3 Buses=[10 100] kVs=[115 13.2] kVAs=[30000 30000]
xhl=5.7947720370809 taps=[1 1.05]
New Transformer.XMFR2 Phases=3 Buses=[10 100] kVs=[115 13.2] kVAs=[30000 30000]
xhl=5.7947720370809 taps=[1 1.05]
New Transformer.XMFR3 Phases=3 Buses=[20 200] kVs=[115 13.2] kVAs=[30000 30000]
xhl=5.7947720370809 taps=[1 1.05]
Staff_PR_38 Attachment A
74
New Transformer.XMFR4 Phases=3 Buses=[20 200] kVs=[115 13.2] kVAs=[30000 30000]
xhl=5.7947720370809 taps=[1 1.05]
New Transformer.XMFR5 Phases=3 Buses=[30 300] kVs=[115 13.2] kVAs=[99999 99999]
xhl=5.8846599166292 taps=[1 1.05]
Staff_PR_38 Attachment A