20170727Angell Direct.pdf

ir '=iil 1l:i'l1..,+,;,.-l'rru
: L. '-',:1 Pl'i 3' 3l
I
BEEORE THE IDAHO PUBLIC UTILITIES COMMISSION
IN THE MATTER OE IDAHO POWER
COMPANY' S APPL]CATION FOR
AUTHORITY TO ESTABLTSH NEW
SCHEDULES EOR RESIDENTIAL AND
SMALL GENERAL SERVICE CUSTOMERS
WITH ON-SITE GENERATION
IDAHO POWER COMPANY
DIRECT TESTIMONY
DAVID M. ANGELL
CASE NO. IPC-E_17 _1.3
OF
)
)
)
)
)
)
1 Q. Please state your name and business address.
2 A. My name is Dave Ange11. My business address
3 is L22L West Idaho Street, Boise, Idaho.
4 Q. By whom are you employed and in what capacity?
5 A. I am employed by Idaho Power Company ("Idaho
6 Power" or "Company") as the Transmission and Distribution
7 Planning Manager.
8 Q. Pl-ease describe your educational background.
9 A. I graduated in 1984 and 1986 from the
10 University of Idaho, Moscow, Idaho, receiving a Bachel-or of
11 Science Degree and Master of Engineerj-ng Degree in
!2 El-ectrical Engineering, respectively. I have provided
13 electrical engineering instruction for both the University
1,4 of Idaho and Boi-se State University. Most recently I
15 instructed power system anal-ysis at Boise State University
16 during the 2009 spring semester. I am a licensed
Ll professional engineer in the State of fdaho and a senior
18 member of the Institute of Electrical- and Electronic
1"9 Engineers.
20 O. Please describe your work experience with
2L Idaho Power.
22 A. From 1986 to 7996, I was employed by Idaho
23 Power as an engineer in both communications and protection
24 systems. In L996, T became the Engineering Leader of
25 System Protection and Communications. I held this posi-tion
ANGELL, D]
Idaho Power
1
Company
1
2
3
4
5
6
'7
8
9
until- 2004, when I transferred to Transmission and
Distribution Planning. During the faIl of 2006, I accepted
the positions of System Planning Leader and Manager of
Delivery Planning. I have been managing Idaho Power's
interconnected-transmissj-on system, sub-transmission, and
distribution planning since 2006.
O. What is the purpose of your testimony in this
proceeding?
A I will- provide an explanation of the
and how the Company's residential- and small
(*R&SGS") customers with on-sj-te generation
10 electrical grid
11
t2
13
L4
15
76
71
18
19
20
2t
22
general
utilize the distribution system. I wil-l- then address the
questlon of whether increasing levels of distributed energy
resources ("DER") will contribute to the deferral- of future
investment in distribution infrastructure. FinalJ-y, I will
describe how smart inverters provide functional-ity that is
necessary to support the ongoing stabil-ity and reliability
of the dlstribution system and explain the Company's
request rel-ative to a smart inverter requirement in
Schedule 12 for customers who interconnect privately-owned
DER to Idaho Power's system.
I. THE GRID
o What is meant
servr_ce
23
24 A. The grid,
25 power system including
by the term "the grid"?
this context, is the el-ectrlc1n
the conversion, transformation,
ANGELL, DI 2
Idaho Power Company
1
2
3
4
5
6
1
I
9
transmission, distribution, and delivery of energy in the
form of electricity to customers.
The conversion of energy contained in reservoirs,
fossil- fuels, wind, geothermal well-sr or solar rays to
electricity power occurs at generation stations. Many of
the generation stations are located remote from the
customers' point of use. Therefore, the electricity is
transformed to extremely high voltages in order to reduce
the electrical losses when transmitting the el-ectricity on
transmission lines for long distances. Transformers are
used throughout the grid to change the electric voltage
level- to match utilization and reduce electric losses.
Once the electrici-ty is
transformed to a lower
delivered to communities, it is
voltage
10
11
72
13
t4
15 distribution.
71
The electricity
on distribution
at substations for l-ocal
is distributed through the
Iines where transformers16 l-ocaf community
18
are used to tap the line and deliver customers electricity
at a reduced voltage to match their intended use.
O. What klnds of services does the grid offer
Idaho Power customers?
A. The grid offers rel-iable electricity delivery,
in the context of dependability and balance of supply,
across large regions in the amount and at the instant of
customers' demand. The grid also provides fl-exibil-ity by
allowing the utility access to a diverse portfolio of
L9
20
2l
22
23
24
25
ANGELL, DI 3
Idaho Power Company
I
2
3
4
5
6
7
8
9
resources for power generation, even 1f those resources are
located far from where the power is needed.
O. What functions does Idaho Power perform in
order to maintain a safe and reliabl-e dlstributj-on system
and grid?
A. In order to provide safe and reliable energy
on demand, fdaho Power must perform the following
functions: voltage control, system protection, scheduling,
dispatching, and load balancing. These functions are
commonly referred to and col-lectively known as ancillary
services.
O. How does Idaho Power control voltage to
maintain a safe and reliable distribution system and grid?
A. Voltage control is achieved by managing the
voltage throughout the grid at the generator, transmission,
and distribution systems. Automatic voltage regulating
devices control- the voltage output of the generators to
match the grid operators set voltage. At the substations,
grid operators also remotely switch substation capacitors
and inductors to raise and lower the transmission voltage,
respectiveJ-y. Automatic voltage management occurs at the
distribution substation transformers with voltage control
based on 1oad, known as J-oad tap changers. Additional
automatic control signals are sent to swltched distribution
circuit capacitors based on substation transformer loading.
ANGELL, DI
Idaho Power
10
11
!2
13
t4
15
16
71
18
19
20
2t
22
23
24
4
Company
25
1 Fina11y, voltage control occurs at substations that service
2 large commercial and industrial customers.
3 Q. What is system protection?
4 A. System protection is the detection and
5 isolation of both short circuits and system operation that
6 may damage generation, transmissj-on, substation, and
7 distribution faci-Iities. ldaho Power coordinates the
8 protection equi-pment to isolate only the failed component
9 and al1ow the remaining grid to continue to supply energy.
10 O. How do scheduling, dispatchi.g, and load
11 balancing help Idaho Power maintain a safe and reliable
L2 distribution system and grid?
13 A. The Idaho Power-owned generation stations are
L4 controlled by grid operations personnel. These personnel
15 schedul-e a generator's electrical output ahead of time
76 based on the l-oad forecast and its optimal use in
Ll consideration of energy market economics. Duri-ng each hour
18 of the duy, the operators efficiently dispatch the
19 generation fl-eet to maintain the balance between production
20 and forecasted use. They operate the generation stations
27 within a set of operational, environmental, and economic
22 constraints to maximize customer val-ue. Some of these
23 generation stations are also configured with the nearly
24 instantaneous ability to automatically adjust the electric
25 output to balance the generated electricity with the actual
ANGELL, D]
Idaho Power
5
Company
1 use, known as automatic generation control. Additlonal
2 generation capability is held in reserve and is available
3 for dispatch if the actua1 load exceeds the forecast or if
4 some of the dispatched generation is forced out of service
5 unexpectedly.
6 Q. How do wind and solar resources i-mpact this
7 scheduli-ng, dispatchj-ng, and load balancing?
I A. Independently-owned wind and solar generation
9 resources differ from fdaho Power-owned and operated
10 generation stations because their production is difficul-t
11 to forecast and they cannot be dlspatched by Idaho Power's
72 grid operators. Because the actual output from these
13 independently-owned resources typically varies from the
74 forecast, they place increased demands on the dispatch and
15 utilization of the automatic generation control and reserve
16 generation.
17 O. Does DER located on the customer side of the
18 meter increase the complexity of forecasting?
19 A. Yes. While on an individual- basis a small
20 independently-owned on-site generation system (capacity of
22 automatic generation control, the aggregate amount of DER
23 installed across Idaho Power's system is noticeable and
24 does increase the complexity of forecasting. As of June
25 30, 20L1, Idaho Power's net metering service had a
ANGELL, DI
Idaho Power
6
Company
1
2
3
4
5
6
1
8
9
cumulative nameplate capacity of 11 megawatts including
customers who had submitted applications for net metering
servj-ce. Because the net metered systems are installed on
the customer side of the meter, Idaho Power is not abl-e to
detect the amount of DER at any given moment, whi-ch
increases complexity of both forecasting and l-oad
following.
O. You have described the overall servj-ces that
the grid provides to Idaho Power's customers and the
functions that Idaho Power performs to maintain a reliable
grld. Does the grid provide other services that are
specific to a person with privately-owned generation?
A. Yes. The grid provides the following services
that customers with privately-owned generation require:
inverter operatj-on, motor starting, energy balancing, and
standby service.
O. What is an inverter?
A. fnverters convert direct current ("DC")
electricity into alternating current (*AC") electricity.
Inverters are used in both off-grid and on-grid
applications. An j-nverter is required for customers who
install a photovol-taic ("PV") generation system because
solar panels produce DC electricity and the home appliances
requj-re AC power supplied by the inverter.
ANGELL, DI
Idaho Power
1
Company
10
t-1
L2
1_3
L4
15
L6
t7
18
79
20
27
22
23
24
25
1 Q. What is the difference between an "off-grid"
2 application and an "on-grid" application?
3 A. An off-grid generation system is one that is
4 not interconnected to the electric grid; the off-grid
5 system provides all electric needs of the owner they are
6 independent of the utility. In the case of an off-grid
7 solar PV system, the DC el-ectricity generated by the PV
I system is used to charge a battery bank connected to the
9 customer's equipment through an off-grid inverter, which
10 does not require the grid to operate.
11 On the other hand, an on-grld generatj-on system is
12 one that is interconnected to the electric grid. For on-
13 grid systems, the DC el-ectricity generated by the PV system
14 is sent directly to an on-grid inverter which converts the
15 electriclty to AC for use by the DER customer or other
16 customers through the grid.
77 O. How does the grid provide services for on-grid
18 system inverter operation?
79 A. Without the grid, the customer's generation
20 system woul-d not operate because these line commutating
21, inverters wou1d not be able to develop voltage or del-iver
22 energy. In other words, the grid must be present for
23 customers with on-grid inverters to operate their
24 generation system.
25
ANGELL, DI
Idaho Power
6o
Company
1
2
3
4
5
6
7
8
9
Eor the remainder of my testimony, al-I discussions
in regard to
inverters.
o.
generation to
A.
i-nverters wiIl be specific to on-grld
How does the grid enable a customer with self-
Electro-mechanical devices such as generators
and motors transfer energy via the interaction of magnetic
in additionfields. These magnetic fields
to the current associated with
additional current is known as
requJ-re current
the energy transfer. This
interconnected with the Idaho Power
start a motor?
10 reactive current. The
11 induction motor, the most widely used motor, is constructed
12 with an electro-magnet which relies on a power source to
13 develop a magnetic field. When energi-zed, the motor has no
!4 magnetic fiel-d to impede the current flow from the power
15 source. Therefore, during motor starting, a current draw
L6 of about six times the fu]l load value occurs. Most
L] inverters currently
not able1B
19
20
2t
22
23
24
system are
continuous
to supply
reactive currents.
and capacitors, supplies the
reactive current. fn other
generation woul-d not be able
these high starting and
The grid, via its generators
motor starting and continuous
words, a customer with on-site
to turn on certain equipment
and household motors without
ANGELL, DI 9
Idaho Power Company
Iike air conditioners,
being connected to the
pumps,
grid.
25
I
2
3
4
5
6
7
8
9
O. How is standby service provided by the grid
beneficial to a customer with self-generation?
A. When a customer's self-generati-on system is
not able to meet their demand, that customer must rely on
power from the grld. Al-so, when a customer's system is not
generating because of weather conditions, time of day or
operational- malfunction, the customer relj-es on power from
the grid to meet their el-ectricity demands.
a. Is it a requj-rement for someone with
privately-owned generation to be connected to the grid?
A. No. A person with privately-owned generati-on
is not requj-red to be connected to the grid. However, most
customers voluntarily choose to connect to the grid in
order to receive the services that the grid provides as
described above.
II. USE OF THE GRID BT STAT{DARD SERVICE CUSTOMERS AND
CUSTOMERS WITH ON-SITE GEIIERLTION
10
11
12
13
74
15
l6
11
18
t9
20 use
27
ZZ
o.
the grid
A.
How do R&SGS customers with on-site generation
compared to R&SGS standard service customers?
The prlmary difference is that the R&SGS
23
customer with on-site generation uses the grid in a bi-
directional manner by both consuming energy from the grid
and delivering excess net energy to the grid when not
consuming aII generation on-site. The standard service
residential- customer only consumes energy from the grid.
24
25
ANGELL, DI
Idaho Power
10
Company
26
1 Furthermore, while the daily demand requirements of the two
2 customers may be sj-miIar, the net monthly energy may not
3 reflect the utilization of the grid by the on-site
4 generation customer.
5 Q. What is meant by the term "net zero" customer?
6 A. A net zero customer i-s one that, over the
7 course of a year, generates as much or more energy
I (kilowatt-hours ('kV0h") ) than they consume. That is,
9 during certain hours of the year, the customer is a net
10 exporter of energy to the grid, and during other hours of
11 the year, the customer is a net consumer of energy from the
t2 grid.
13 O. Does the net zero customer util-ize the
t4 distrlbution system less than the standard service
15 residential customer?
L6 A. No. A net zero customer utilizes all aspects
Ll of Idaho Power's grid during the hours they are consuming
18 energy (including the generation, transmission, and
19 distribution systems) and utillzes the distribution system
20 during the hours they are exporting energy to the grid.
2l To illustrate this, the Company selected a single
22 residential net metering customer who netted their usage to
23 zero during 2016. Figure 1 demonstrates the hourly usage
24 of that residential- net zero net metering customer on the
25 Company's 20tG adjusted system peak day (June 29) and
ANGELL, DI
Idaho Power
11
Company
1
2
3
compares that customer's hourly usage to a standard servj-ce
customer's usage whose home is nearby the net metering
customer.
4 Figure 1. Residential. Net ![etering Customer vs. Standard
5 Service Residential Customer (June 29, 2016)
a Would you characterize these customers' usage
as simil-ar?
A. No. While the daily absolute demand
requirements of the two customers are similar, the net
monthly energy consumed by the net metering customer j-s not
representative of their usage of the grid.
O. Why is the net monthly energy not
representative of the customer's use of the grid?
A. When a net metering customer exports excess
net energy to the grid, their retail- meter "spj-ns
backwardsr " or in the case of a modern meter, subtracts
usage electronically. Later, during other hours of the day
or month when the customer is consumlng energy from the
6
1
I
9
10
11
12
13
L4
15
16
71
18
ANGELL, DI 72
Idaho Power Company
8
6
4
2
(21
(4)
(6)
t 2 3 4 5 6 7 8 9 1011t2L3L4L5t6t7 18192021222324
Hour Ending
r "Net Zero" Net Metering Customer r Standard Service Residential Customer
3.Y
19
1
2
3
4
5
6
7
8
9
grid, the meter "spins forwardr " and adds usage
electronically. On a monthly basis, the net metering
customer is using the grid, every hour, every day, but
because usage is measured for bil-ling purposes on a monthly
basis, that net metering customer appears to have "zero
usage" for the month.
To understand the extent to which the net metering
customer uses the grid, you can use the absolute value of
the energy being transacted to and from the net metering
customer. Figure 2 represents the same day as Eigure L,
but shows all of the energy as positive, that is, the total
amount of energy that is being transacted between the net
metering customer and the Company's grid regardless of
which direction the energy is flowing.
Fignrre 2. UtiJ.ization of the Distribution System by
Residential Net l[etering Customer vs. Standard Service
Residential Customer (ilune 29 20t51
10
11
72
13
L4
15
1,6
l'7
ANGELL, DI
Idaho Power
13
Company
8
7
6
5
4
3
2
1
L 2 3 4 5 6 7 I 9 1011L2L314L5t617 18192021222324
Hour Ending
r Net Metering E Exported Net Excess r Standard Service
EI
18
1
2
3
4
5
6
7
8
9
The sum of the hourly consumed
standard service residential customer
11.1 kwh, and the sum
hourly energy for the
energy for the
on June 29, 20L6, was
of the absolute val-ue of the net
residential net metering customer was
the standard service residential
55.4 kwh. Yet, when looking at thelr meter reads from that
10
day, it would appear
customer transacted 71.1 kWh of energy, while the
residential net metering customer's meter would register
that customer transacted 11.28 kwh.
When looking at the sum of all of the individual
hours wi-thin the month of June for those same two
customers, the sum of the hourly consumed energy for the
standard servj-ce residential customer was t,480 kwh, and
ANGELL, DI
Idaho Power
11
72
13
1,4 the sum of the absol-ute val-ue of the net hourly energy for
15 the residential net meterj-ng customer was t,323 kwh. Yet,
16 when looking at their meter reads for the month, it would
L7 appear the standard service residential customer transacted
18 7,480 kwh of energy, while the residential net metering
19 customer's meter woul-d register that customer exported
20 excess net energy of 440 kwh, to be carried forward to
2L offset consumption in a future month, and that customer
22 would be billed for zero kwh.
23 This demonstrates how the net monthly energy as a
24 basis for billing does not reflect a net metering
25 customer's util-ization of the grid
L4
Company
1
2
3
4
5
6
1
I
9
rII. LOCAL DISTRIBTITION II{I\IESIIIEIiIT
O. Has Idaho Power studied the relationship
between distributed, rooftop solar PV and its distribution
system operations?
A. Yes. In 2015, the Company performed a study
comparing solar intensity variations and distribution
circuit demand. This study is attached as Exhibit No. 74.
A portion of the study sought to determine if there was a
relationship between solar i-ntensity and distribution
cj-rcuit loading. The Company's system peak load is largely
driven by the Treasure Valley residentj-a1 and commercial
1oads. Therefore, weather stations with irradiance sensors
were installed on a Treasure Va11ey distribution circuit
that supplied primarily residential and some commercial
customers. Three irradi-ance sensor orientations at three
locations were used. The orientations were southerly facing
at a 35o til-t, horizontal- and westerly facing at a 53o
tiIt. The study demonstrated that there was a significant
time delay between peak solar j-ntensity and peak
distributlon circuit demand.
O. What did the study conclude?
A. The study demonstrated that a southerl-y facing
sensor peaked approximately four hours prior to the
distribution circuit peak load and a westerly faclng sensor
peaked approximately two hours prj-or to the distrlbution
ANGELL, DI 15
Idaho Power Company
10
11
72
13
t4
15
16
L1
18
19
20
21,
22
23
24
25
1 circuit peak l-oad. In both cases, similarly orj-ented PV
2 systems will not significantly reduce a distribution circuit
3 peak load. Additionally, the measurements demonstrate that
4 the western facing PV system will create a steep decline in
5 production at the end of the day that will result in a rapj-d
6 change in circui-t voltage and require enhanced voltage
7 regulating abil-ities to respond to this rapid decline.
8 As the study concluded, even a system oriented west
9 will not peak during peak load hours of the suflrmer peak on
10 a specific distribution circuit or substation. The
11 distribution circuit peak load occurred from 17:00 through
L2 19:00 hours. At 19:00 hours, the contribution from a PV
13 array wil-I be about 20 percent if faced southerl-y and about
14 55 percent if faced westerly. PV generation would only
15 shift the peak load to 20:00 hours and decrease it by 10
16 percent. To illustrate this, Figure 3 presents a typical
71 nominalizedl load shape with the irradiance shape for the
18 southerl-y-configured sensor and the resulting l-oad shape
79 less the solar irradiance.
20
2L
22
1 To find relati-onships between so1ar intensity and l-oad, the data
was nominafized so that each varj-able ranged between 1 and 0. That
al-1owed the two correlated time-series to be more easily refatable when
graphed.
ANGELL, DI 76
Idaho Power Company
1 Figrure 3. Southerly SoJ.ar Irradiance Shape vs. Load Shape
1.0
0.8
o?-c
=El1|N;
c,E
C,(ga
0.6
0.4
0.2
0.0
0:00 2:00 4:00 6:00 8:00 10:00 12:00 1 4:00 16:00 18:00 20:00 ?2:00 24:00
Hour
Load (Std) S lnadiance (Sld)
-
Load'lrradiance
2
3
4
5
6
7
I
9
Si-mi1arly, Fj-gure 4 presents a typical
shape with the irradiance shape for the
nominalized load
sensor and the resulting load shape l-ess
irradiance.
westerly-configured
the solar
10
11
72
ANGELL, DI
Idaho Power
L7
Company
13
1 Figrure 4. Westerly Solar Irradiance Shape vs. Load Shape
1.0
0.8
0.6
0.4
0.2
0.0
0:00 2:00 4:00 6:00 8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 24:00
Hour
Load (Std) W lrradiance (Sld)
-
Load-lrradiance
O. Why is the planning horlzon fj-ve years?
A. Idaho Power is able to forecast di-stribution
circuit and substation capacity requirements with some
certainty five years into the future. This planning
horizon peri-od all-ows the Company to investigate options to
avoid facility overloads, select more cost-effective
options, and design and construct improvements to meet the
identified overl-oads.
O. Can Idaho Power forecast on-site generatj-on
installations by distribution circuit and substation?
gt.=
=E(lI
.trE(!E
C,G'
(f)
2
3
4
5
6
1
8
9
10
11
72
ANGELL, DI 18
Idaho Power Company
,l
)I \l
t
I
I
I
I
\
III
13
1 A. No. A customer's interest and abil-ity to
2 invest in an on-site generation system is based on many
3 factors such as their ability to flnance a system, risk
4 tolerance, the local economy, and electricity prlces. The
5 last two factors are incorporated into the forecasts used
6 in the Integrated Resource Plan analysis for on-site
7 generation additions for the service area. However, it is
I extremely difficult to build a reasonable forecast with
9 those same assumptions for specific circuits and
10 substations.
11 O. Can increased levels of rooftop PV reduce
12 local distribution infrastructure investment?
13 A. Only
has infrastructure in place to serve all customers during
and undertakepeak load hours. Idaho Power must plan
distrlbution system investments in order
in limited circumstances. Idaho Power
t4
15
t6
77 reliable service. In order to reduce the
to provide this
infrastructure
18
79
20
investment, sufficient PV additions must occur on the
distribution circuit during Idaho Power's five-year
2L
planning
o
horizon.
Can you provide an example where
be reduced?22 infrastructure mlght
Z3 A An example of a cj-rcumstance where an
24 investment coul-d be deferred is a remote section of a
25 distribution circuj-t where the load peak occurs during
ANGELL, DI L9
Idaho Power Company
1
2
3
4
5
6
7
I
Y
daylight hours, customers are increasing at a slow rate,
and the voltage is decllning. A tradltional solution,
e.9., installation of a capacitor, might be reduced by 20
to 50 kW of PV generation.
O. Has this occurred on the Idaho Power system?
A. Yes. Idaho Power engineers reviewed the
distribution system and found one location that met the
criteria above. A pilot PV project was installed in 201,6
and the engineers are presently monitoring the performance
for evaluation this fa11.
O. Is the distributj-on system capable of handling
increasing levels of DER without any modification?
A. No. High DER penetration amounts create
distribution circuj-t operatJ-on challenges, such as voltage
management, short circuit detection, and islanding.
Islanding occurs when a customer's generation is capabJ-e of
supporting the load of other customers physically located
near the customer's generator when that section of the
electrical circuit is isolated from the Idaho Power system.
O. Why does the presence of DER impact the
distributlon cj-rcuit voltage?
A. When DER is contributing power to the circuit,
it changes the power requirement from the distrj-bution
substation transformer as shown previously. This change j-n
ANGELL, DI 20
Idaho Power Company
10
11
72
13
74
15
t6
l1
1B
L9
20
2t
22
23
24
25
1
2
3
4
5
6
7
8
9
power flow causes the typical circuit voltage drop to
change.
O. What is meant by typical cj-rcuit voltage drop?
A. Voltage drop (l-oss) occurs any time power
flows through a conductor. Eor a typical circult, the
voltage is highest at the substation (the power source) and
drops to the lowest point at the end of the circuit. The
rate of drop is based on the amount of current flow and
conductor resistance.
O. How j-s the distribution circuit voltage10
11 managed?
t2 A. The distribution circuit voltage is typj-caIly
13 automatically controlled by three components: (1) the
74 substation distribution transformer load tap changer
15 ("LTC"), (2) regulators located along the circuit, and (3)
16 shunt connected capacitors. The LTC automatically adjusts
Ll the substation bus voltage based on the power fl-ow through
18 the transformer. The LTC and regulators are mechanical
L9 devj-ces that slowly wear with each change of tap.
20 Therefore, the controls on these devices are set with
2L sufficient bandwidth and time delay to avoid excessive wear
22 and maintain the voltage within a range for the customer's
23 equipment to function properly. The capacitor controls are
24 set to manage the reactive power flow while keeping the
25 circuit voltage within the range described above.
ANGELL, DI 2I
Idaho Power Company
1
2
3
4
5
6
7
I
9
O. May the controls be set to accommodate DER
penetration in excess of 15 percent of peak load?
A. No. The traditional LTC and regulator
controls were designed assuming the cj-rcuit power would
only flow from the substation to the loads on the circuit.
During conditions of DER power
load, these controls will sense
output that exceed the l-ocal-
the
10 latest models that are
to increase the voltage
power flow and adjust
when it is not desired.
to replace the controll-ers with the
abl-e to detect a reverse power
the voltage
Idaho Power would need
11
t2
13
t4
15
76
77
18
t9
20
2L
22
condj-tj-on and adjust accordinqly.
This level of PV penetratj-on can require additional-
voltage adjustment (tap change) cycles in addition to the
daily load cycle adjustments. Adjustments down are needed
as the voltage rises with PV output peaking with the solar
peak and adjustments up when the voltage decreases during
the load peak while the sol-ar output wanes.
Finally, the PV output will change rapidly when
broken clouds pass over the PV systems. Rapid changes to
output result in rapid voltage fl-uctuations that cannot be
regulated by the
o. Are
time-del-ayed regulating devices.
there operational practices or equipment
ANGELL, DI 22
Idaho Power Company
23 avail-able to reduce these operational chal-lenges?
24 A. Yes. Idaho Power first replaces the
25 controllers and optimj-zes their settings for reduction of
1
2
3
4
5
6
7
8
9
voltage deviation wlthout substantially increasing the
devlce wear. Beyond this, there are two options. The first
option is to reduce the remaining voJ-tage deviation by
decreasing the circuit lmpedance through fuII conductor
replacement on a given feeder. This option is not
practical given that a full conductor replacement would not
be cost-effective. The second option is to require voltage
regulation from the DER.
rV. INVERTER EEJATT'RES AI{D EI'NCTION
O. How can on-sj-te generation provide regulation?
A. An on-sj-te generati-on system j-nterconnected to
the grid through a smart inverter can regulate voltage if
its voltage control- function is enabled.
O. What is a smart inverter?
A. A smart inverter provides configurable
functions beyond the conversion of DC to AC. A few of the
features are: voltage/reactive power control, anti-
islanding, monj-toring, and remote communication.
O. Have these regulation functions been
demonstrated?
A. Yes, multiple studies and experience from
10
11
72
13
t4
15
L6
L7
1B
79
20
2t
22
23
Germany, California,
depJ-oyment of smart
and Hawaii have shown that the
inverters can reduce the
24 of on-site generation. The industry adoption
voltage impact
of smart
25 j-nverter requirements will help to mitigate circuit voltage
ANGELL, DI 23
. Idaho Power Company
1
2
3
4
5
6
1
8
9
devj-ation. States like Cal-ifornia and Hawaii have already
started requiring smart inverters in residential
installations. Germany, the global leader in PV, has
required smart inverters for the last few years.
O. Is there a cost differential- between a smart
inverter and a standard inverter?
A. Yes, however, smart inverter costs are
decreasing and with the adoption of smart inverter
requi-rements in Cal-ifornia and HawaiJ-, these costs will
decrease even more rapidly due to scales of production. A
standard inverter costs approximately $0.23 per watt
whereas a premium j-nverter with the smart inverter
functionality buil-t in costs around $0.35 per watt. For
example, j-f a customer were to install a 6,000 watt system,
the price difference between a smart inverter and a
standard inverter for this system wou1d be roughly $120.
Most inverter manufacturers already provide smart
inverter function capabilities in their devices; it is just
a matter of upgradlng the software to enabl-e the smart
inverter functionality. The manufacturers that do not
currently provide smart inverter capabilities in their
products are generally lower cost but they will have to
offer smart inverter capabilities in the future to remain
competitive in the market. The Company recognizes that
ANGELL, DI
Idaho Power
10
11
L2
13
l4
15
L6
L1
1B
t9
20
2L
22
23
24
24
Company
25
1 there is a potential cost difference between lower cost
2 Legacy inverters and smart inverters.
3 Q. Does the Company currentl-y require that
4 customers who install privately-owned generation,
5 interconnect to the grid using a smart inverter(s)?
6 A. No. The current Schedule 72, fnterconnection
7 to Non-Utility Generation, requires that grid
8 interconnected inverters have either a certification with
9 Standard for Inverters, Converters , Control-l-ers and
10 Interconnection System Equipment for Use with Distributed
11 Energy Resources UL l7 47, Institute of El-ectrical- and
L2 Electronic Engineers Interconnecting Distributed Resources
13 with Electric Power Systems Standard 1547 (*IEEE 7541") or
t4 be subject to thj-rd-party testj-ng performed at the
15 customer's expense. None of the standards that are in
L6 effect today include a mandate for the use of a smart
L7 inverter.
18 O. Does the Company believe that it would be
L9 beneficial to require Idaho Power customers with privately-
20 owned generation to interconnect to the grid through a
2L smart i-nverter?
22 A Yes.As descrj-bed previously, the Company
benefits have been demonstrated23
24
believes that many
associated with the use of smart inverters. In fact, the
25 Instltute of Electrical and Efectronic Engineers are in the
ANGELL, DI
Idaho Power
25
Company
I
2
3
4
5
6
1
I
9
process of revising the IEEE 1547 to adopt the standards
around what constitutes a smart inverter.
O. When are the revisions to fEEE L547 and 7547.L
anticipated to be approved?
A. The IEEE 1547 and 1547.L standards are
presently being balIoted. It is anticipated that IEEE 1541
could be approved as early as the end of 2077 and IEEE
154'7 .7 could be approved by mid-2018.
O. What is the Company recoflrmending in this
filing regarding smart inverters?
A. Idaho Power requests that the Idaho Public
Utilities Commission acknowledge that smart inverters
provide functionality that is necessary to support the
ongoing stability and reliability of the distribution
system by ordering the Company to submit a compliance
filing in the form of a tariff advice within 60 days of the
adoption of the revised IEEE standards, or 60 days of the
conclusion of this case, whichever occurs l-ater. This
tariff advice will seek to modlfy Section 2 of Schedule 72
to require that customers with on-site generation install a
smart inverter that meets the requirements defined in the
revised IEEE standards.
0. Why would reductj-on of the voJ-tage deviation,
a power quality issue, be the responsibility of the DER?
ANGELL, DT
Idaho Power
26
Company
10
11
72
13
!4
15
76
77
18
19
20
27
22
23
24
25
1 A. It is the DER that creates the voltage
2 deviation it j-s also the DER that can cost-effectively
3 mitigate the deviation through the install-atj-on of a smart
4 j-nverter. Establishing this as a requirement in Schedule
5 12 is simil-ar to the requirement for customers to comply
6 with Practices and Requirements for Harmonic ControL in
7 ELectric Power Systems as set forth in the current IEEE
8 Standard 519 in Rul-e K-
9 Q. Does this conclude your testimony?
10 A. Yes.
11
72
13
t4
15
T6
71
18
T9
20
21,
22
23
24
25
ANGELL, DI 21
Idaho Power Company
1
2
3
4
5
6
1
I
9
ATTESTATION OF TESTIMONT
STATE OF IDAHO
SS.
County of Ada
T, David M. Ange11, having been duly sworn to
testify truthfully, and based upon my personal knowledge,
state the following:
I am employed by Idaho Power Company as the Planning
Manager in the Customer Operatl-ons Engineerj-ng and
Construction Department and am competent to be a wj-tness in
this proceeding.
f declare under penalty of perjury of the laws of
the state of Idaho that the foregoing pre-filed testimony
and exhibit are true and correct to the best of my
information and belief.
DATED this 27th day of JuIy, 2077.
David M. Ange I
SUBSCRIBED AND SWORN to before me this 21th day of
Ju1y, 20L7.
0Y(
10
11
L2
13
74
15
!6
L7
l_8
19
20
2t
22
23
24
25
26
21
28
29
30
31
No
Re
ry Ii
siding at l_
or Idaho
se, Idaho
7
My commission expires:t2/20/2020
ANGELL, DI 28
Idaho Power Company
frO?rf .?l
e(rlutc
rDa-)
OF TD
)
)
)
aa,Ii,
,lY't
-l
(,t :l(:
t+FIfiI?**e
)t.lilrlA
\
BEFORE THE
IDAHO PUBLIG UTILITIES COMMISSION
GASE NO. IPC-E-17-13
IDAHO POWER COMPANY
ANGELL, DI
TESTIMONY
EXHIBIT NO.14
A Method for Determining the Relationship between
Solar Irradiance and Distribution Feeder Peak Loading
Phillip Anderson
COEC Research, Development & Deployment
Idaho Power Company
Boise, Idaho USA
[e-mail]
Abstruct-L method for determining the relationship between
solar irradiance and distribution feeder peak loading is
presented. Two research questions were posed: (l) is there a
statistictl relationship between solar intensity and load and (2)
what other relationships might there be between load and
meteorological parameters. A distribution feeder was chosen
where data from three equidistant solar irradiance weather
stations (SIWSs) was coUected between June 21, and September
21,2013. The data was analyzed using a combination of cross-
correlation and maximum cross-correlation / kernel density. The
results of the study show that ttere geems to be a correlation
between solar intensity and load, but that correlation is
optimally significant when the variables are time-shifted.
Index Terms-Autocorrelation, photovoltaic systems, power
distribution, solar energy
I. I}NRODUCTION
As elecfical power utility customers install their own roof-
top photovoltaic (PV) systems, the question arises whether the
utility itself could benefit from PV systems that it owns and
operates. The effect ofcloud cover on PV operability could be
less than expected because cloud cover might also have an
effect on distribution feeder loading; in other words, as the
solar intensity decreases, so also might the loading on an
accompanying distribution feeder decrease, or vice versa.
A benefit to a utility when its residential customers begin
interconnecting PV systems onto its distribution system might
be to balance increasing loads, thus allowing the utility to
postpone future feeder upgrades necessitated by load growth.
Because ofthat benefit, a research project designed to collect
solar intensity and feeder loading data simultaneously during
the summer of 2013 was initiated.
II. A BREF OVERVIEW OF EXISTNG RESEARCH
The existing research of PV system effects on the
operations of their distribution feeders yielded three broad
categories:
l. Research that addresses distribution feeder
simulation that includes PV system interconnection.
Brett Efaw and Ellen McKinney
COEC Reliability and COEC T&D Planning
Idaho Power Company
Boise, Idaho USA
2. Research that addresses PV system interconnection
that mitigates on-going distribution feeder issues.3. Research that determines how solar irradiance and
cloud cover affects the operation of PV systems on
distribution feeders.
Among the research that addresses distribution feeder
simulation are Sandia National Laboratories [1] and EPRI [2]
each of whom developed software for modeling distribution
feeder operation with PV systems interconnected. Additional
research in feeder modeling includes determining optimal PV
locations and sizing [3]-[a] as well as researching the effects
of high-penetration PV onto distribution feeders t5l-t61.
Among the research that addresses PV interconnection as
mitigation to on-going feeder issues are using PV systems as
non-traditional solutions to distribution feeder problems [7],
using PV systems with energy storage for feeder load
smoothing t8l-t91, and using PV systems with CVR to
manage feeder voltages [0].
Among the research that determines how solar irradiance
and cloud cover affects the operation of PV systems on
feeders is research of a cloud shadow model to recreate the
power generated by rooftop PV systems [l] as well as
research ofhow PV system desigrr can change the dispersion
ofPV energy across a feeder because ofpassing clouds [12].
Common to the referenced research is the focus on PV
systems interconnected onto a distribution feeder. The
research reported in this paper is different because it seeks to
find the relationship between solar intensity / cloud cover and
a distribution feeder's actual loading. In other words, this
research is concemed with how sunshine or clouds may
impact the amount of distribution feeder load created
throughout the day.
III. RESEARCHDESIGN
Two questions formed the basis of the research desigrr:
l. Is there a statistical relationship between solar
intensity and load?2. What other relationships might there be between
load and meteorological parameters?
Exhibit No. 14
Case No. IPC-E-17-13
D. Angell, IPC
Page 1 ofS
In April 2013, a project was desigrred to answer the
research questions. The scope of the project included the
following tasks:
. Study a residential feeder to determine weather /
solar monitoring locations.r Install weather / solar stations and gather solar,
weather, and load data.o Analyze the data for possible correlations.
The research project included purchasing and installing
solar intensity monitors, PV panels, and power metering and
recording equipment. Studying the relationships between load
and meteorological parameters other than solar intensity, wind
speed, and ambient temperature were considered to be outside
the scope ofthe project.
The research was limited to data obtained from a single
residential feeder typical for the installation of roof-top PV
systems; feeders that might host larger, utility-scale
installations were considered to be outside the project's scope.
IV. SITE SELECTION CRITERL{
The study feeder is located in a city of approximately
215,000 population. The city's climate is semi-arid, with
average monthly sunshine ranging from a peak of400 hours in
July to 105 hours in December [3]. The distribution feeder
chosen to host the solar iradiance weather stations (SIWSs)
and collect solar irradiance data met the following conditions:
o Covering a geographical distance greater than 3.2
kilometers in an east-to-west direction to be able to
track the effects of cloud cover as the clouds move
from one SIWS site to the next.. Being comprised of mostly residential and light
commercial customers.. Being one where the installation of the SIWSs
would be as accessible as possible, located near the
operating office where the data would be analyzed.
The study feeder is summer-peaking rated 34.5kV and
20MVA. A 34.5kV feeder typically covers a greater
geographical area than a 12.5kV feeder, thus being more
likely to meet the condition set for geographical configuration.
The feeder is configured such that the distance from its
eastern-most location to its western-most location is
approximately 5.6 kilometers.
Three SIWSs were installed on the feeder at locations that
were (l) as geographically equidistant from one another as
possible and (2) as free from impediments to irradiance data-
gathering as possible.
Each station was initially designed to include global
horizontal irradiance (GHI) monitoring, point-of-array (POA)
monitoring in the westerly direction, wind speed monitoring,
wind direction monitoring, ambient temperature monitoring,
and GPS time synchronization. POA monitoring in the
southerly direction and globally was added later. Westerly
POA monitoring was chosen because of the sun's location at
the time of the feeder's 2012 peak. Later, the southerly POA
was added to emulate typical PV system installations.
V. DATACOLLECTION
Solar and weather data was captured onto an SD card - at
a data capacity of 2G - located in the data logger. At a set
time interval, the data was collected by removing the SD card
from the data logger, copylng the data from the SD card to a
laptop's hard drive, deleting the data from the SD card, and
replacing the SD card back into the data logger.
The following data was collected from each of the SIWS
sites:
r UTC time. Wind speed (the average over a ten-second interval)
in meters / second (m/s). Wind gust (the peak over a ten-second interval) in
m/so Wind direction in degrees (with 0" being North)
o Local battery voltage (over a ten-second interval)r Ambient temperature in oC
r Solar irradiance POA South in watts lmeterz
(watts/m2). Solar irradiance POA West in watts/m2
o Global Horizontal irradiance in watts/m2
Initially, data was collected every two weeks because of
the need to ensure that a minimum amount of data would be
lost should an issue with the logger arise. Data was collected
on May 24, Jlulr,e 4, July I, July 15, August l, August 8,
September 12, and October 2. Load data for the feeder was
also collected in ten-second intervals to match the times of the
collected irradiance data.
VI. DATA ANALYSIS
A. Cross-correlation analysis
To find relationships between solar intensity and load /
other meteorological parameters, the data was nominalized so
that each variable ranged between I and 0. That allowed the
two correlated time-series to be more easily relatable when
graphed as well as to provide a corlmon axis from which to
compare results for different days. Figure I shows a typical
nominalized load shape and irradiance shape for the westerly-
configured sensor on SIWS0I on the system peak load day of
July 1,2013.
Sh. 1: W..rrly SoLr .nd Ldd Srnd.idh.d S..i.r ,ot 0iJUL201 3
r.0
0.8
0.6
0.0
0:00 ::00 r0!600 8:00 r0:00 r2:m 14 00 !6:@
Hour
r8:00 20:00 21100 :a00
D{.fom bi 21lo $paffih 21 413
/1t,/t
/
Figure l. Southerly solar and load, 7-1-2013
Exhibit No. 14
Case No. IPC-E-17-13
D. Angell, IPC
Page 2 of 5
Even though data was collected at l0 second intervals,
cross-correlations were calculated at 5-minute intervals to
allow for quicker computation. For each correlation, the load
time-series was held constant while the other time-series -southerly, global, and westerly sensor configurations;
temperature; and wind speed - were shifted across a tl2-
hour range. Cross-correlations were graphed for each ofthe 93
days of the study at each of the three SIWSs. Fifteen cross-
correlation computations were made for each day of the study
at each of the three SIWSs, resulting in the production of 4915
cross-correlation graphs.
For the July l, 2013, SIWS0I, example Figures 2 through
4 show the graphs ofthe various cross-correlation calculations
at +12 hours:
Sir l: Soudr.rly Sohr b! Lo.d Cl64omhon FuGion lor ()lJUL20l3
1
08
0.0
0l
01
-02
-0.4
-0.6
-0.8
n.0
-12 n0 -8,6 .a .2 0 I 4 6 8 10 12
L.s (Hour.)
\
Figure 2. Cross-correlations for Southerly Solar and load at SIWS0 I
on 7-l-2013
St I i Glob.l Solrr by Load CdrcomLtlon Fsdid io. 01 JUL2013
10
08
06
0.4
0.0
-02
.04
-0.0
-08
r0
l:-10 -8 .0 -4 -201
L.f (Hour.)t_.aFl
10 I:
Figure 3. Cross-correlations for Global Solar and load at SIWS0 I
on 7-1-2013
Sib l: W.srly Soa.. b, Lo.d Clw4mhid FuHfth tn olJuulolt
10
08
0_l
0_2
0_0
-0.2
-0.6
-0.8
n0
i:i0 -8 -6 -:0?
bs (Hou6)
I:?cil
a 0 8 l0 12
Figure 4. Cross-conelations for Westerly Solar and load at SIWS0I
on 7-l-2013
For the clear and sunny July l, 2013, the correlations
between the sensor contiguration's time series and the load's
time series are fairly significant. For the worst case westerly
configuration versus load, the maximum correlation of 0.81
occurred when the feeder's load data lagged the solar intensity
data by two hours. On the same day and SIWS, the maximum
correlation was 0.90 when the load data lagged the solar
intensity data by four hours on the southerly-facing sensor,
and the maximum correlation was 0.92 when the load data
lagged the solar intensity data by four hours on the global
sensor.
B. Maximum cross-correlations / kernel densities
While the cross-correlations provided information relating
to the relationships between irradiances and loads for each of
the individual days of the study period, the question remained
regarding if any relationships over the entire 93-day study
period could be identified. To answer that question, maximum
cross-correlations were calculated to create kernel densities.
For each time series pair, the maximum cross-correlation
and the time lag when the maximum cross-correlation
occurred were identified for each day of the study period.
Next, the results from each day were used to estimate a
bivariate kemel density for each pair of time series. This
bivariate density estimate yielded a point that represented the
most common occwrence over the course of the study for the
strength of the linear relationship and the amount of time the
offset occurred. The ranges ofboth the linear relationship and
the offset in time were also observed from the bivariate
density estimates.
The density plots in Figure 5 show, for SIWS0I, the
estimates between the load and the southerly sensor
configuration for solar intensity both 2-dimensionally and 3-
dimensionally.
Exhibit No. 14
Case No. IPC-E-17-13
D. Angell, IPC
Page 3 of 5
Sh 1: krinum Cd}Comhon
Figure 5. Example kemel density for SIWSO1 southerly solar versus load
over the study period
The most common daily occurrence was a strong linear
relationship of 0.94 between the solar intensity and load
lagged by approximately four hours. Over the course of the
entire summer, the range of the strength of the relationship
was generally greater than 0.80 with the lag generally between
three and five hours.
For each of the three SIWSs, twelve kernel densities were
calculated that showed irradiances versus load relationships,
irradiances versus ambient temperature relationships, and
irradiances versus wind speed relationships. It was from
analyzing these thirty-six calculations that the results of the
study, and answers to the research questions, were formed.
VII. RESULTS
A. Relationships to load
Firsl there seems to be a statistical relationship between
solar intensity and load, but with the understanding that
correlation does not imply causation.
o Solar intensity analyzed at the southerly-configured
sensors tends to lead load from 3.96 to 4.13 hours
with a correlation of 0.94 across all three SIWS
locations.r Solar intensity analyzed at the global-configured
sensors tends to lead load from 4.00 to 4.02 hours
with a correlation ranging from 0.95 to 0.96.. Solar intensity analyzed at the westerly-configured
sensors tends to lead load from 1.76 to 1.94 hours
with a correlation ranging from 0.88 to 0.91. The
curves ofthe westerly-configured irradiances tend to
peak closer to the time of feeder peak loads, but with
lower certainty than the southerly- and global-
confi gured irradiances.
Next, there seems to be a strong correlation between
ambient temperature and load data, but a tepid correlation
between wind speed and load data.
r Ambient temperature tends to lead load from 0.70 to
0.86 hours with a correlation ranging from 0.96 to
0.97 across all three SIWS locations. While not a
surprising result, what was a little surprising was the
range in lead time between the three SIWSs.o Wind speed tends to lead load from 1.80 to 2.38
hours with a correlation ranging from 0.68 to 0.70
across all three locations. This tepid statistical
relationship between wind speed and load also was
not surprising considering the low auto-correlation
of wind speeds.
B. Disclaimers
When drawing conclusions based upon the analysis of the
data, a few disclaimers need to be considered.
. The conclusions are limited by the study time
period. The conclusions would likely have been
different ifdata had been analyzed for seasons other
than summer.r The conclusions are limited by the locations of the
study's SIWSs. The conclusions would likely have
been different ifdata had been collected (l) from a
different type of feeder or (2) from a different
geographical area.
These disclaimers, however, can also be considered
opportunities to continue the research to test the statistical
methods and analysis used in the study.
VIU. DECUSSIoN
A. Lessons learnedfrom the research design process
As with most research, the most difficult part of the design
was choosing and applying the most appropriate statistical
tool. After the data was collected, an iterative process ensued,
beginning with making a simple correlation between daily
peak loads and daily peak irradiances, and realizing that using
such a tool with two variables that were clearly unrelated
yielded unsatisfactory results. Time-shifting the data, first
though the auto-correlations and then cross-correlations,
provided a means of accessing the daily relationships between
the variables. Finally, applying maximum cross-correlations
and plotting kernel densities allowed the statistical
relationships - calculated over the entire 93-day study period
- to emerge. Any similar research in the future will also needto implement a cross-correlational / maximum cross-
correlational / kernel density approach to data analysis.
A couple decisions made early in the research desigr
positively affected the ability to answer the research questions:
the choice to collect the data from irradiance sensors rather
than from solar panels, and the choice to add a southerly-
Exhibit No. 14
Case No. IPC-E-17-13
D. Angell, IPC
Page 4 of 5
exposed sensor to the global and westerly sensors at each of
the three SIWSs. The decision to collect data from three
somewhat equidistant locations along the feeder route, while
not necessary for answering the research questions, did
provide the opportunity to validate the data collected and will
be a source of analysis of geographical relationships between
the variables.
B. Possible applications ofthe research results
The analyzed data set has already been implemented. The
data has been used with OpenDSS software to perform
generation interconnection studies of utility-scale PV systems.
Other applications of the data have been in assisting with
resource planning as part of the utility's integrated resource
plan and in collaboration with Sandia National Laboratories
for their variability and GHI to POA conversion studies.
Probably the most interesting possibilities for applying the
research results would be for:
r Recommending preferred PV orientations to
commercial customers for their rooftop applications
that would best support reducing the effects of
feeder summer peak loads.r Designing a demonstration project of a PV system
coupled with energy storage to extend peak load
reduction at the end ofa distribution feeder.
C. Opportunities for further research
Data continues to be collected from the three SIWSs
beyond the initial study period, resulting in more than a year's
worth of data having been collected. Opportunities have been
identified regarding additional studies that could be made with
the study data as well as with the additional data collected
post-study.
l) Using this research study's data.
The SIWSs were geographically configured along the
feeder route, spaced equidistant from one another. While that
fact was not pertinent in answering the research questions, it
could generate a follow-up research project to determine the
geographical relationships between irradiance data collected
from the three SIWS sites. The data collected could be also be
used to calculate the solar energy density (Watt-hours/m2) per
time interval for each site at each solar intensity orientation.
Another use of the research data could be to run irradiance
statistics over a multiple-day time period to see if a
relationship is evident as a predictor ofthe seasonal peak load.
Finally, the data could be used in combination to
determine how PV installation for home- and business-owned
systems could provide optimal benefit the utility.
2) Branching out from this research study data.
Other interesting areas for further research can be sorted
according to (l) using the additional data that has been
collected subsequent to the study period and (2) implementing
additional types ofresearch design protocols. Some ideas for
using the additional data collected include:
r Analyzing data from other seasons ofthe year, such
as the time of the feeder's winter peak load or the
time of the feeder's minimum peak load.
r Analyzing cross-correlations of solar irradiance to
wind to find any complimentary relationships.. Analyzing data collected in summer 2014 compared
to data collected in summer 2013.
Some ideas for implementing additional design protocols
include:
o Determining what effects humidity might have on
solar intensity and load.o Including cross-correlations of the loads from
adjacent feeders to follow the effects ofcloud cover.. Analyzing data from additional feeder types where
solar might be of interest.r Including volatility in irradiance statistical analyses.o Correlating wind speed to solar volatility.o Designing a study similar to this research desigrr for
a winter-peaking area.
IX. REFERENCES
tll M.J. Reno and K. Coogan, "Gnd Integrated Distributed PV (CrridPV)',
Sandia National laboratories, SAND2O I 3-6733, August 20 I 3.
121 M. Rylander and J. Smith, "Stochastic Analysis to Derermine Feeder
Hosting Capacity for Distributed Solar PV', Electric Power Research
Institute (EPRI), Palo Alto, CA: 2012. 1026640.
t31 A. Hoke, R. Butler, J. Hambrick, and B.Kroposki, "Maximum
Photovoltaic Penetration Levels on Typical Distribution Feeders",
National Renewable Energy l-aboratory (NREL), NREUJA-5500-
55094, July 2012.
t4l J. Quiroz and M. Reno, "Detailed Grid Integration Analysis of
Distributed P\l', Sandia National Laboratories, SAND report 2012-
4s33,2012.
t5l H. Thomas, K. Lynn, and A. Razon, "Current Results of the US DOE
Hi gh Penetration Solar Deployment Project", Photovoltaic Spec ial ists
Conference (P\SC), 2012 3* IEEE, pp.73l -736,3 - 8 June 2012.
t6l P. Gotseff, J. Cale, M. Baggu, D. Narang, and K. Caroll. "Accurate
Power Prediction of Spatially Distributed PV Systons using Incalized
Irradiance Measurements", Power and Energt Society General
Meeting, 2014 IEEE, pp. 1 - 5,28 - 3l July 2014.
t7l J.E. Bigger, E.C. Kem, and M.C. Russell, "Cost-effective Photovoltaic
Applications for Electric Utilities", Photovoltaic Specialists
Conference, 1991, Conference Record ofthe Twenty Second IEEE, pp.
486-492, vol. 1,7- ll October 1991.
t8l M.T. Arii A.M.T. Oo, A.S. Ali, and G. Shatiullah, "Impacts of Storage
and Solar Photovoltaic on the Distribution Network", Universities
Power Engineering Conference (AUPEC), 2012 2!d Australasian, pp.
| - 6,26 -29 September 2012.
t91 O. larova, F. Cheng, S. Abdollahy, H. Barsun, A Mammoli, D.
Dreisigmayer, S. Willard. B. Arellano, and C. van Zeyl, "Analysis of
Bettery Storage Utilization for Load Shifting and Peak Smoothing on a
Distribution Feeder in New Mexico", Innovative Sman Grid
Teclmologies (ISGT),2012 IEEE PES, pp. I -6, 16-20 Jarr'nry2012.
[0] R. Singh, F. Tuffner, J. Fuller, and K. Schneider, "Effects of
Distributed Energy Resources on Conservation Voltage Reduction
(CYR)", Power and Energt Society General Meeting, 201 I IEEE, pp. 1
-7,24-29 July2011.
Il] C. Chengui and D.C. Aliprantis, "Cumulus Cloud Shadow Model for
Analysis of Power Systems with Photovoltaics", Power Systems, IEEE
Transactions on, vol. 28, no. 4,pp. 4496 - 4506, Nov. 201 3.
[12] H. Moavani, D.K. Click, R.H. Meeker, R.M. Reedy, and A
Pappalardo, "Quanti$ing Solar Power Variability for a [arge Central
PV Plant and Small Distributed PV Plant", Photovoltaic Specialists
Conference (PVSC), 2013 IEEE 39rh, pp. 0969 - 0972,16 - 2l June
2013.
[3] The Weather Network Web site
http ://www. theweathernetwork.com/forecasts/statisticVc028 93,
Novernber 12,2014
Exhibit No. 14
Case No. IPC-E-I7-13
D. Angell, IPC
Page 5 of 5