Sei sulla pagina 1di 8

1

Quasi-Static Time-Series Test Feeder for PV


Integration Analysis on Distribution Systems
Barry A. Mather, Member IEEE

quasi-static time-series analysis could be used to investigate


Abstract-- This paper presents a quasi-static time-series PV integration impacts such as:
test feeder based on the well-known 34 node test feeder [1].
This test feeder is useful for the evaluation and • Voltage excursions beyond the desired voltage range
quantification the distribution system impacts of • Voltage regulator device operation frequency
distributed PV integrations. The load profiles and solar • System power loss characteristics
resource profiles used are both publically available. The
quasi-static time-series test feeder has been developed In order to build industry confidence in the results provided
using data for the year 2010. The fixed time period by these new tools this paper proposes a testing platform in
between power flow solutions is 1 min making the the form of a quasi-static time-series test feeder based on the
presented quasi-static time-series test feeder appropriate well-known IEEE 34 node test feeder [2] for use in general
for use in investigating voltage regulation device PV integration analysis. Previously published works on using
operations. A load allocation algorithm is also proposed the IEEE 34 node test feeder as a platform for distributed
for assigning spot and distributed loads in the 34 node test generation (DG) modeling focus on static single solution
feeder to time varying load profiles based on rate class power flow analysis [3,4], focus on optimal DG placement [5]
load profiles. A simple model for modeling the real AC or focus on circuit protection [6,7]. This paper specifically
power output of a PV system based on discrete global outlines the methods and data used to simulate the operation
horizontal irradiance (GHI) is also presented. The results of the IEEE 34 node test feeder, including voltage regulation
of analysis completed using the quasi-static time-series device operation, over the course of 1 year (2010) with a
test feeder are shown for various single-site 1MW PV simulation time step of 1 min. All the data used is publically
deployment scenarios and voltage regulator compensation available making the proposed quasi-static time-series test
settings. feeder appropriate for use by industry and academia alike.

Index Terms—34 node test feeder, quasi-static time-series Section II describes the development and implementation of
analysis, power flow, voltage regulator operation simulation, PV the proposed quasi-static time-series test feeder. Example
integration, distribution system impacts. analyses for 1MW single-site utility scale PV system
integration, based on the simulation of the quasi-static time-
I. INTRODUCTION series test feeder, are presented in Sec. III. Section IV
The increasing number and size of photovoltaic (PV) systems concludes the paper.
either being interconnected or requesting interconnection at
the distribution system level has resulted in an increased II. THE QUASI-STATIC TIME-SERIES TEST FEEDER BASED ON
interest in distribution system modeling for use in determining THE IEEE 34 NODE TEST FEEDER
the impacts of such PV integrations. Accordingly, many
distribution system modeling professionals have begun to A. General requirements for quasi-static time series power
develop the appropriate modeling tools to evaluate and flow simulations
quantify these impacts. One of the tools of most interest Quasi-static time-series analysis is essentially a sequential
currently is the ability to investigate the operational series of discrete power flow solutions. “Quasi-static” refers
characteristics of a modeled distribution system and an to the concept that a static steady-state power flow solution is
interconnected PV system over a set time period (typically a being used to evaluate an inherently dynamic, i.e. non-steady-
day or a year in simulation length). This type of analysis, state, system. Typically, if the system is changing slowly
called quasi-static time-series analysis as defined in [1], is compared to the discrete time interval between quasi-static
typically used to evaluate the low frequency dynamics of a power flow solutions the low frequency dynamics of the
distribution system. In terms of PV interconnection studies a system are well represented in the quasi-static time-series
analysis. For distribution system modeling this means that the
accuracy of a quasi-static time-series analysis will depend on
Barry A. Mather is with the Distributed Energy Systems Integration the time period between steady-state power flow solutions and
(DESI) Group at the National Renewable Energy Laboratory (NREL), the settling time of the system after a dynamic event that is
Golden, CO 80401 USA (e-mail: barry.mather@nrel.gov).
desired to be evaluated in the analysis. “Time-series” refers to

978-1-4673-2729-9/12/$31.00 ©2012 IEEE


2

the causal time dependence of the next steady-state power The voltage regulators specified in the 34 node test feeder
flow solution to the system state determined by the last have well defined regulator control set points provided. These
steady-state power flow solution. include the voltage regulation bandwidth, regulating voltage
level, and line compensation values R and X. The PT ratio and
For the development of a quasi-static time-series PV primary CT rating are also given for completeness. It was not
integration test feeder the operational characteristics that were indicated whether the voltage regulators were reversible. For
of interest were the operation of the voltage regulation PV integration analysis the reversibility of the voltage
devices. In distribution systems such devices are typically regulators will be one of the many important factors to
controlled using low bandwidth controllers that often have consider when using the quasi-static time-series feeder.
time delay lockouts to disable device operation for a fixed
C. Load profiles and feeder load allocation
period of time following a change in device state. These time
delays are often in the range of 30s to 2 min. A simulation Since the goal of a quasi-static time-series analysis is to
time period of 1 min was chosen for the development of this investigate or evaluate the operation of the distribution system
quasi-static time-series test feeder based on the compromise over a period of time it is important that the distribution
between the time delay operational characteristics of typical system loading be as realistic as possible. Unfortunately, the
voltage regulation devices and the available load and solar 34 node test feeder has only static loading specified. This
resource data discussed in Secs. II.C and II.D. Both datasets loading data was assumed to be the peak load for each load
are developed for the year 2010. point on the system as would typically be used in basic
B. IEEE 34 Node Test Feeder Characteristics planning power flow studies.
Figure 1 shows a single line diagram of the 34 node test
Four hourly load profiles based on load research statistics
feeder as described in [2]. Individual nodes are denoted by
containing 8,760 data points per year were utilized for the
number starting with 800 at the substation bus bar connection.
development of the quasi-static time-series test feeder. They
This 24kV test feeder is very long, with a maximum line
length to the substation from node 848 of nearly 36 miles, and are available online from Southern California Edison [8]. The
is lightly-loaded. The circuit includes two voltage line four load profiles used include domestic (DOM), commercial
regulators connected between nodes 814-850 and 852-832 (COM), industrial (IND) and street lighting (SL). Table I
respectively. To static capacitors are also connected to the shows the 2010 static load profiles used for the four load
feeder at nodes 844 and 848 with ratings of 100kVAr/phase classes integrated into the quasi-static time-series test feeder.
and 150kVAr/phase respectively. As the targeted quasi-static time-series simulation step time
822 was 1 min (525,600 data points/solutions per year) the load
profiles were resampled at 1 min intervals with simple linear
820
interpolation applied between the hourly data points. The use
818 826 of load profiles based on hourly load research statistics is not
800 Reg #1
802 806 808 812 814 850
optimal as the resulting circuit is not as variable as would be
816
824 expected from actual circuit measurements. Additional
810 828
random variation on top of the resulting load profiles was
considered but as the load profiles solely represent the
864
830 averaged load profile of a statistically significant number of
Reg #2
840 836 860 834 832 852 customers from the same rate class (as they are derived from
854
858 load research statistic data) it is difficult to justify such
862 842 888 856 variation without assuming some number of customers behind
each load point. Alternatively, the approach was taken that
838 844 890 assumed each load point in the 34 node test feeder served a
statistically significant number of customers and that these
846
customers were all of the same rate class. Such assumptions
848 were required since no additional information on the actual
loading characteristics of the 34 node test feeder is publically
Fig. 1. Single line diagram of the IEEE 34 node test feeder showing the available. As better load models become publically available
location of the two voltage regulators and two shunt capacitors. they should be incorporated into the proposed quasi-static
time-series test feeder for a more realistic simulation of the
The bus voltage at node 800 is 1.05pu for all three phases modeled distribution circuit. As currently implemented and
possibly suggesting that the substation contains a load tap proposed test feeder does allow the relative evaluation of the
changer (LTC) for regulating substation bus voltage. Such an
primary impacts of PV integration but does under represent
LTC is not defined in the original 34 node test feeder and has
the effects of higher frequency load variations.
not been added to the proposed quasi-static time-series test
feeder.
3

TABLE I secondary circuit are not considered in the quasi-static time-


LOAD PROFILE DATA SOURCES AND DESCRIPTIONS
series test feeder as it has been assumed that the PV systems
Load Profile Rate Class Description Data File of interest in this analysis (1MW or greater) will be connected
DOM Domestic Single/Multiple DOMSM10.DLP using their own independent distribution transformer.
General Service, Non-demand Metered,
COM Small Commercial GS110.DLP
General Service, Demand Metered, Medium D. Solar Resource Data and PV system model
IND Commercial/Industrial GS210.DLP
In order to realistically model the variable nature of actually
SL Street and Area Lighting stlt.txt
PV system due to the diurnal characteristic of the solar
resource and the effects of clouds, actual 1 min global
Each of the spot and distributed loads (modeled as a spot load horizontal irradiance (GHI) data was used to drive the quasi-
in the middle of a line section between nodes [9]) were static time-series test feeder. There are a limited number of
allocated to a rate class load profile using the simple algorithm publically available sources for 1 min GHI data. The source
shown in Figure 2. If the load was a balanced 3-phase load it proposed for use in the quasi-static time-series test feeder is
was assigned the industrial (IND) load profile. If the load was from the National Renewable Energy Laboratory’s Solar
a single-phase load and was relatively small, ≤10kVA, it was Resource Research Laboratory Baseline Measurement System
allocated to the street lighting (SL) profile. If not already [10]. Specifically, the 1 min averaged GHI data from the LI-
allocated into the IND or SL profiles the load was allocated 200 [11] was utilized. These irradiance values are collected in
into either the domestic (DOM) or commercial (COM) load Golden, CO and while they reflect the somewhat unique
profiles for loads with a total kVA load on all phases of variability characteristics of the region they are available for
≤125kVA or >125kVA respectively. This load allocation public use and are of high quality.
algorithm resulted in a good mix of load profile assignments
to each load profile identified by rate class. In order to allow the investigation of various distributed PV
deployment scenarios the 1 min GHI data is used to derive the
modeled output of a PV system using the simple model shown
in Fig. 3.

TABLE II
LOAD PROFILE DATA SOURCES AND DESCRIPTIONS

Spot Loads
Node Balanced kVA Load Allocation
860 Y 108 IND
840 Y 48 IND
844 Y 720 IND
848 Y 108 IND
890 Y 675 IND
830 N 65 DOM
Distributed Loads
Nodes Balanced kVA Load Allocation
802-806 N 84 DOM
808-810 N 24 DOM
818-820 N 51 DOM
820-822 N 205 COM
Fig. 2. Flow chart showing the load allocation algorithm used for each spot
816-824 N 7 SL
and distributed load in the 34 node test feeder in order to associate loads with 824-826 N 60 DOM
load profiles. 824-828 N 6 SL
828-830 N 10 SL
The resulting load allocation for the quasi-static time-series 854-856 N 6 SL
test feeder is shown in Table II. It is important to note that the 832-858 N 22 DOM
858-864 N 3 SL
load model (Δ or Y and Z, I, or PQ) associated with each spot
858-834 N 49 DOM
and distributed load was not modified from the 34 node test 834-860 N 219 COM
feeder and was not considered as part of the load allocation 860-836 N 125 DOM
algorithm. It should be noted that the IEEE 34 node test 836-840 N 60 DOM
feeder does not model distribution transformers or the 862-838 N 42 DOM
secondary distribution circuit connecting customers to the test 842-844 N 14 DOM
feeder. All loads are lumped by phase and referred to the 844-846 N 68 DOM
primary circuit voltage. The potential impacts on the 846-848 N 34 DOM
secondary circuit due to the integration of PV onto the
4

test feeder and the quasi-static time-series analysis shown in


Sec. III, OpenDSS [12] was used via the COM interface [13]
as the main power flow solution solver. MATLAB was used
as a scripting and control tool to manage load and PV system
profiles. MATLAB also was used to collect pertinent
distribution system data at each simulated time step. The
modeled controls for the voltage regulators were implemented
solely in OpenDSS excepting regulator time delay. Since a
MATLAB script was used to advance the quasi-static time-
1 series solution through time the time delay of the voltage
1000
regulators was realizable only in discrete quantities of 1 min
time periods (1 min, 2 min, etc.) For simplicity the effective
voltage regulator delay implemented was 1 min or both
voltage regulators. This was accomplished simply though the
Fig. 3. Simple PV system model using discrete global horizontal irradiance
(GHI) values (Ee[n]), and system settings, Ee,sys,sat and Po,sys,STC, to generate the
discrete sampling nature of the quasi-static time-series
real AC power output of a PV system (Po,sys[n]). analysis. Longer delays are possible, again in an integer
multiple of 1 min intervals, and would be implemented within
This highly simplified model relating discrete samples of the the MATLAB script.
GHI (Ee[n]) and the discrete PV system power output
(Po,sys[n]) uses two static inputs. One input is the power output Although some detail is given above regarding the software
of the PV system if the PV array associated with the PV used to develop the quasi-static time-series test feeder it
system had an incident GHI of 1000W/m2. This is equivalent should be noted that this test feeder was developed with the
to the irradiance condition specified by the standard test aim of providing a basic model which could be solved by
condition (STC) commonly used to test PV cell performance. many distribution system analysis software packages. The
The other model input is related to the relative sizing of the results from these solutions should be compared leading
PV inverter and the PV array. Ee,sys,sat denotes the GHI level eventually to a consensus as to the validity of such quasi-static
which just begins to saturates the PV inverter. At irradiance time-series analyses for determining the distribution system
levels above Ee,sys,sat the inverter will be forced to produce less impacts of PV integration.
power than available from the PV array. The inclusion of
these two inputs into the model allows for the modeling of any III. BENCHMARK QUASI-STATIC TIME-SERIES ANALYSIS
size of PV system and allows for the investigation of how The quasi-static time-series test feeder was used to analyze
different relative ratio between PV inverter size and PV array various PV deployment scenarios in addition to the base case
size affect the interconnected distribution system. for the feeder when no PV was deployed. Only single station
1MW PV systems with PV inverters that saturate at a GHI of
One drawback from using this simplified model include the 1200W/m2 were investigated. The analyses presented in the
limitation that the PV systems are inherently being modeled as following sections are provided as a starting point for
flat plate collectors and no provisions for investigating the comparing test cases for the quasi-static time-series test feeder
effects of fixed tilt or tracking arrays is provided. and are not presented for the specific inspection of the impacts
Additionally, the expected variability of a PV system of PV integration.
intuitively decreases as the name plate power rating and thus
the PV array’s geometric size increases. Using the averaged 1 A. Analysis of feeder voltage profile
min data likely properly approximates some size of PV Figure 4 shows the maximum, average and minimum per unit
system. Unfortunately, the exact size of PV system that is phase A voltage for Wednesday, June 16th, 2010 present at
approximated is not yet known. Future studies in the area of each node on the 34 node test feeder starting at the substation
PV variability will improve the ability to properly model the (800) and following the main feeder trunk to node 840 for
time dependence (variability) of various PV system sizes. case when no PV systems are interconnected. The location of
the voltage regulator #1 is easily identified approximately 20
Additionally, the model assumes that the PV inverter is miles from the substation feed with a vertical dotted line.
operating with a power factor of unity. Modeling PV inverters Voltage regulator #2 is also visible at a distance of about 33
with volt/VAr support functions requires a modification to the miles from the start of circuit and is also denoted by a vertical
model and the calculation of the equivalent kVA for which a dash line. For this single typical summer day the voltage
PV inverter will saturate (typically due to current rating profile graph shows that the minimum voltage on the circuit is
limits). at node 852 located just before voltage regulator #2. The
minimum voltage is 0.9836 pu. The maximum voltage during
E. Quasi-static time series power flow analysis
the day, ignoring voltage levels near the substation is located
There are a number of options for simulating the operation of immediately after voltage regulator #2 at node 832. The
the quasi-static time-series test feeder over the period of 1 maximum voltage for the day is 1.0467 pu. Inspecting the
year and at a discrete time step of 1 min for a total of 525,600 entire simulation dataset the yearly minimum circuit voltage
individual power flow solutions. For the development of the along the main feeder trunk from node 800 to 840 is 0.9519
5

and this voltage is present at node 814 otherwise identified as base case. The maximum voltage, again discounting the
the primary side of voltage regulator #1. The maximum yearly voltages seen near the substation, for the day is 1.0489 pu and
voltage, again ignoring nodes near the substation, is 1.0496 is located at the point of interconnection of the modeled 1MW
pu. This maximum voltage is seen at the output of voltage PV system. Inspecting the yearly dataset the minimum
regulator #2. voltages seen on the feeder are 0.9728 pu and 0.9764 pu for
the primary input nodes for voltage regulators #1 and #2
respectively. The maximum voltages seen are 1.0748 pu and
1.15 1.0517 pu again for the primary input to voltage regulators #1
Vpu,max
and #2 respectively.
Vpu,avg
1.1 Vpu,min The results of this example feeder voltage profile analysis
mirror the common expectations and concerns of PV
integrations at a considerable distance from a substation.
Voltage

1.05 Specifically, the analysis indicates that the expected minimum


and maximum voltages present on the feeder increase
compared to the base case voltage profiles for the circuit
1 without any PV interconnections.

Table III presents the three phase node voltages present at


every node along the main feeder trunk for the quasi-static
0.95
0 10 20 30 time-series solution corresponding to Wednesday, June 16th,
Distance from Substation (mi) 2010 at 12:15PM. Node voltages are given for both the base
Fig. 4. Maximum, average and minimum per unit voltage on phase A along case where no PV system(s) are interconnected to the test
the 34 node test feeder from node 800 to 840 with no PV system(s) feeder and for a 1MW PV system (Ee,sys,sat = 1200W/m2)
interconnected (base case) for Wednesday, June 16th, 2010.
connected at the end of the circuit (node 840). These voltages
are given to facilitate the comparison of various distribution
1.15 system analysis software packages and the simulation results
Vpu,max they produce when implementing the quasi-static time-
Vpu,avg series
1.1 Vpu,min TABLE III
THREE PHASE NODE VOLTAGES ALONG THE MAIN TRUNK OF THE QUASI-STATIC
TIME-SERIES TEST FEEDER ON JUNE, 16 , 2010 AT 12:15PM FOR THE BASE
TH
Voltage

CASE AND THE CASE WHERE A 1MW PV PLANT IN INTERCONNECTED AT NODE


1.05 840.

Voltage
1
No PV 1MW PV at 840
Node φA φB φC φA φB φC

0.95 800 1.0500 1.0500 1.0500 1.0500 1.0500 1.0500


0 10 20 30 802 1.0488 1.0494 1.0493 1.0493 1.0499 1.0498
Distance from Substation (mi) 806 1.0479 1.0490 1.0488 1.0489 1.0499 1.0497
Fig. 5. Maximum, average and minimum per unit voltage on phase A along
the 34 node test feeder from node 800 to 840 when a 1MW PV system is 808 1.0324 1.0420 1.0403 1.0402 1.0493 1.0474
interconnected at node 840 (end of the line) and the PV inverter(s) saturates at 812 1.0142 1.0342 1.0301 1.0299 1.0488 1.0444
an array irradiance of 1200W/m2 for Wednesday, June 16th, 2010. 814 0.9997 1.0279 1.0219 1.0216 1.0483 1.0418
850 1.0184 1.0151 1.0156 1.0152 1.0156 1.0158
Figure 5 shows the maximum, average and minimum per unit 816 1.0182 1.0150 1.0154 1.0151 1.0155 1.0157
phase A voltage for each node voltage along the test feeder 824 1.0134 1.0103 1.0110 1.0133 1.0139 1.0141
line as in Fig. 4 except that a 1MW PV system has been
828 1.0130 1.0099 1.0106 1.0131 1.0138 1.0140
modeled as interconnected at node 840 (end of the line). For
830 1.0035 1.0011 1.0016 1.0095 1.0111 1.0106
this deployment scenario the voltage profile impacts of a
854 1.0032 1.0009 1.0014 1.0094 1.0110 1.0105
1MW PV system integrated at the end of the main line are not
852 0.9864 0.9851 0.9855 1.0031 1.0061 1.0046
overtly apparent. Closer inspection of the date shows that the
addition of the 1MW PV system results in lower minimum 832 1.0357 1.0344 1.0348 1.0282 1.0312 1.0297
voltages and higher maximum voltages seen along the 858 1.0346 1.0333 1.0338 1.0284 1.0315 1.0300
distribution feeder. For Wednesday, June 16th, 2010 the 834 1.0333 1.0321 1.0326 1.0287 1.0319 1.0303
minimum voltage recorded is at node 852 at a value of 0.9919 860 1.0331 1.0319 1.0324 1.0290 1.0322 1.0307
pu for the 1MW PV system interconnected at node 840 case. 836 1.0330 1.0318 1.0324 1.0296 1.0327 1.0313
This is a minimum voltage increase of about 0.008 pu over the 840 1.0330 1.0318 1.0323 1.0298 1.0329 1.0315
6

test feeder. The simulated voltage regulator tap positions may line compensation disabled. A small improvement in the
vary depending on initial circuit conditions and the number of tap operations was observed but the feeder voltage
distribution system analysis software used for the regulation may have been adversely affected. The number of
implementation of the quasi-static time-series test feeder. In voltage regulator tap operations shown in Table V can be used
order to better compare circuit simulation solutions the tap to compare alternate implementations of the proposed quasi-
positions that result in the voltage profiles presented in Table static time-series test feeder using various distribution system
III are given in Table IV. analysis software packages.

TABLE IV
VOLTAGE REGULATOR TAP POSITIONS ON JUNE, 16TH, 2010 AT 12:15PM FOR

Power Output (kW)


THE BASE CASE AND THE CASE WHERE A 1MW PV PLANT IN INTERCONNECTED 1MWSTC sys. output
AT NODE 840. 1000

500

0
0 6 12 18 24
Time of Day (hour)

Regulator #1 Tap Pos. (N)


16
φ
A
8 φ
B. Analysis of voltage regulator operations B
φ
0 C
The voltage regulator tap position and tap operation count
were collected during simulation runs of the quasi-static time- -8
series test feeder. This example analysis was completed to
-16
show how voltage regulator operations would be impacted by 0 6 12 18 24
PV system deployments. Table V shows the tabulated voltage Time of Day (hour)
Regulator #2 Tap Pos. (N)

regulator tap operations for three test cases modeled in the 16


quasi-static time-series test feeder. The test feeder was solved φ
A

for the base case to develop a baseline measurement for the 8 φ


B

“normal” number of tap operations. The numbers shown in 0 φ


C

Table V are for a complete year of operation. For the base


case the average number of tap operations per regulator -8
(single phase regulator controls) was about 7 tap operation per -16
day. These results were obtained using the voltage regulator 0 6 12 18 24
Time of Day (hour)
settings and compensation defined in the 34 node test feeder.
Two cases are shown for the cases where a 1MW PV system Fig. 6. Top: Modeled PV system power output for Wednesday, June 16th,
was deployed at node 840. The first test case utilized the same 2010 for a 1MW system whose PV inverter(s) saturate at an array irradiance
regulator line compensation settings as defined in the 34 node of 1200W/m2. Middle: Voltage regulator # 1 tap position and movement for
the case when a 1MW PV system modeled in the top figure is interconnected
test feeder. A significant increase in the number of operations at node 840 (end of the line). Bottom: Voltage regulator #2 tap position and
was observed. The final case shown in the tabulated data is for movement for the case described above.
the same PV deployment scenario but with voltage regulator

TABLE V Figure 6 shows the daily PV system power output and the
TOTAL NUMBER OF VOLTAGE REGULATOR TAP CHANGE OPERATIONS FOR 2010 voltage regulator #1 and #2 tap positions for Wednesday, June
FOR VARIOUS PV DEPLOYMENT AND VOLTAGE REGULATION COMPENSATION
CASES
16th, 2010. The position of the voltage regulator taps are
shown for all three phases. The PV system power output
profile shows a day that is mostly clear during the morning
hours with thick intermittent clouds appearing in the
afternoon. The voltage regulator tap positions and movements
show the efforts expended by the voltage regulators to
accommodate for the variability of the interconnected 1MW
PV system that is interconnected at node 840.

Figure 7 shows the same data as Figure 6 but for only for the
hours of 10:00 a.m.—1:00 p.m. for a closer look at the voltage
regulator tap positions and movements during periods of high
solar resource variability. Plots like those shown in Figures 7
and 8 can be used to determine voltage regulator operations
7

and can possibly be used to develop modified voltage


regulator control set points and settings.

500 N
op,A
1500 N

Percent Reg. Op. Increase


Power Output (kW)

1MW sys. output op,B


STC
400 N
op,C
1000
300
500
200
0
10 11 12 13 100
Time of Day (hour)
Regulator #1 Tap Pos. (N)

16 0
φ
A
8 φ 0 10 20 30
B
φ
Distance from Substation (mi)
0 C

Fig. 8. Percent tap operation increase for regulator #1 for the interconnection
-8 of a 1MW PV system along the length of the main 3-phase trunk of the 34
node test feeder (from 800 to 840) with standard regulator R+jX
-16 compensation.
10 11 12 13
Time of Day (hour)
Regulator #2 Tap Pos. (N)

16
φ
A
500 N
8 φ op,A
B
N
Percent Reg. Op. Increase

φ op,B
0 C
400 N
op,C
-8
300
-16
10 11 12 13
Time of Day (hour) 200
Fig. 7. Top: Modeled PV system power output for 10:00 a.m.—1:00 p.m.,
Wednesday, June 16th, 2010 for a 1MW system whose PV inverter(s) saturate 100
at an array irradiance of 1200W/m2. Middle: Voltage regulator # 1 tap position
and movement for the case when a 1MW PV system modeled in the top figure
is interconnected at node 840 (end of the line). Bottom: Voltage regulator #2
0
tap position and movement for the case described above.
0 10 20 30
Distance from Substation (mi)
The results of additional PV integration analysis using the Fig. 9. Percent tap operation increase for regulator #2 for the interconnection
quasi-static time-series test feeder are presented in Figs. 8 and of a 1MW PV system along the length of the main 3-phase trunk of the 34
9. These figures show the percent increase of the voltage node test feeder (from 800 to 840) with standard regulator R+jX
compensation.
regulator tap operations for the possible integration of a 1MW
PV system at any node along the main 3-phase feeder line cost of reduced voltage regulation capability along the feeder.
from node 800 to node 840. The modeled 1MW PV system’s Still, the example analysis shows the kinds of analyses that
inverter(s) saturate at a GHI of 1200W/m2. Figure 8 shows the can be investigated using the proposed quasi-static time-series
effect of integrating the 1MW PV system along the line for test feeder.
voltage regulator #1 and Fig. 9 shows the same for voltage
regulator #2. IV. CONCLUSIONS
This example analysis shows some interesting results as to the This paper presents a quasi-static time-series test feeder based
effect of PV system placement on the test feeder. Not only is it on the well-known 34 node test feeder. This test feeder has
apparent that locating the PV system near the substation specifically been developed to evaluate and quantify the
results in fewer voltage regulator operations but also that distribution system impacts of distributed PV integrations.
locating the PV system on the primary side of the voltage The load profiles and solar resource profiles used are both
regulator, as opposed to the secondary, if connection near the publically available. The quasi-static time-series test feeder
voltage regulator is required may also reduce voltage regulator has been developed using 2010 data and has a discrete solving
operations. Further analysis of this finding is required as such time period of 1 min between power flow solutions. This
a reduction in voltage regulator tap operations my come at the results in a total of 525,600 power flow solutions per year. A
8

load allocation algorithm is also proposed for assigning spot


[6] Silva, J.A., Funmilayo, H.A., and Butler-Purry, K.L. “Impact of
and distributed loads in the 34 node test feeder to time varying
distributed generation on the IEEE 34 node radial test feeder with
load profiles based on rate class load profiles. A simple model overcurrent protection,” in Proc. of IEEE North American Power
for modeling the real AC power output of a PV system based Symposium, pp. 49-57, 2007.
on discrete global horizontal irradiance (GHI) is also [7] Bulter-Purry, K.L. and Marotti, M. “Impact of distributed generators on
presented. The results of analyzing the quasi-static time-series protective devices in radial distribution systems,” in Proc. of IEEE
Power and Energy Society General Meeting, pp. 1-6, 2011.
test feeder are shown for various PV deployment scenarios [8] Southern California Edison, 2010 Static Load Profiles, online resource:
and voltage regulator compensation settings. http://www.sce.com/AboutSCE/Regulatory/loadprofiles/2010loadprofile
s.htm, accessed Aug., 2011.
V. ACKNOWLEDGMENT [9] Kersting, W. H., “Radial distribution test feeders,” in Proc. of IEEE
Power Engineering Society Winter Meeting, vol. 2, pp. 908-912, 2001.
[10] NREL Measurement and Data Center (MIDC) Solar Radiation Research
This work was supported by the U.S. Department of Energy Laboratory (SRRL) Baseline Measurement System (BMS), online
resource: http://www.nrel.gov/midc/srrl_bms/, accessed Aug., 2011.
under Contract No. DOE-AC36-08-GO28308 with the [11] Li-Cor LI-200 Pyranometer datasheet, online resource:
National Renewable Energy Laboratory. http://www.licor.com/env/products/light/pyranometer.jsp, accessed
Aug., 2011.
[12] OpenDSS electric power distribution system simulator, online resource:
http://sourceforge.net/projects/electricdss/, accessed Aug., 2011.
VI. REFERENCES [13] R. Dugan, OpenDSS Level 2 Training, online resource:
[1] IEEE P1547.7 D5.1 Draft Guide to Conducting Distribution Impact http://www.smartgrid.epri.com/doc/OpenDSS%20Level%202%20Traini
Studies for Distributed Resource Interconnection, IEEE Standard, Aug. ng.pdf, accessed Aug., 2011.
2011.
[2] Distribution System Analysis Subcommittee of the IEEE Power
VII. BIOGRAPHY
Engineering Society, IEEE 34 Node Test Feeder, online resource:
http://www.ewh.ieee.org/soc/pes/dsacom/testfeeders/index.html, Barry A. Mather (S’2003, M’2010) received the B.S.
accessed Aug., 2011. and M.S. degrees in electrical engineering from the
[3] Dugan, R.C. and Kersting, W.H., “Induction machine test case for the University of Wyoming, Laramie, WY, in 2001 and
34-bus test feeder-description,” in Proc. of IEEE Power Engineering 2004 respectively and the Ph.D. degree also in electrical
Society General Meeting, Oct. 2006. engineering from the University of Colorado, Boulder,
[4] Owuor, J.O., Munda, J.L. and Jimoh, A.A., “The IEEE 34 node radial CO, in 2010. In March 2010 he joined the Distributed
test feeder as a simulation testbench for distributed generation,” in Proc. Energy Systems Integration (DESI) Group at the
of IEEE AFRICON, pp. 1-6, 2011. National Renewable Energy Laboratory (NREL),
[5] Gandomkar M., Vakilian, M. and Ehsan, M., “Optimal distributed Golden, CO, where he is currently researching
generation allocation in distributed network using Hereford Ranch distributed photovoltaic energy system integration issues. His research
algorithm,” in Proc. of IEEE International Conference on Electric interests include the analysis and mitigation of high-penetration PV
Machines and Systems, vol. 2, pp. 916-918, 2005. integration into the distribution system as well as advanced power electronic
applications in renewable energy systems.
.

Potrebbero piacerti anche