EFFECTS OF PHOTOVOLTAICS ON DISTRIBUTION SYSTEM VOLTAGE REGULATION

Peter McNute, Josh Hambrick, and Mike Keesee

National Renewable Energy Laboratory, Golden, Colorado
Sacramento Municipal Utility District, Sacramento, California

ABSTRACT

As grid-integrated photovoltaic (PV) systems become more prevalent,
utilities want to determine if, and at what point, PV systems might
begin to negatively impact the voltage regulation of their distribution
systems. This paper will briefly describe voltage regulation methods in
a utility distribution system. It will also take a preliminary look at
the distribution impacts of high penetrations of grid-integrated PV
systems being installed and operated in a Sacramento Municipal Utility
District community. In particular, the issue of excessive service
voltage and excessive substation voltage due to the reverse power flow
from exporting PV systems will be examined. We will also compare
measured data against modeled data.

Index Terms - Photovoltaic (PV), Distribution System, Voltage Regulation

INTRODUCTION

A three-year, joint project between the National Renewable Energy
Laboratory (NREL) and Sacramento Municipal Utility District (SMUD) began
in March 2008 to analyze the distribution impacts of high penetrations
of grid-integrated renewable energy systems, specifically PV-equipped
SolarSmartSM Homes found in the Anatolia III Residential Community
(hereafter referred to as Anatolia) in Rancho Cordova, California.
SolarSmart Homes combine high-efficiency features along with
rooftop-integrated 2.0-kWac PV systems with no energy storage. When
completely built out, Anatolia will have 795 homes, 600 of which will be
SolarSmart, eventually amounting to 1.2 MWac potential generation. (So
far, only 115 homes have been built for 238 kWac potential generation.)
In particular we are investigating if there will be excessive service
voltage or substation voltage due to reverse power flow from exporting
PV systems.

VOLTAGE REGULATION METHODS

The primary distribution voltage needs to remain within ANSI C84.1
limits: 114 V to 126 V or 11.4 kV to 12.6 kV, over the length of the
feeder. The service voltage is provided to the customer meter and
includes any voltage drop from the primary service through the
distribution transformer and service conductors. The electric supply
system is designed so that the voltage is within the 110 V to 126 V
range (Range A) most of the time and infrequently within the 106 V to
127 V range (Range B). The utilization voltage is the operating voltage
system and loads are designed to operate within. During heavy loading,
SMUD may adjust the substation voltage as high as 12.75 kV to account
for the voltage drop at the end of the feeder.

Load Tap Changer

A load tap changer (LTC) effectively varies the transformer turns ratio
to maintain the transformer secondary voltage at the substation as
primary-voltage changes occur due to changes in loading of the
transmission system, or as the load on the transformer itself varies.
The Anatolia substation voltage is automatically controlled by a LTC on
the secondary of the substation transformer, managed by a voltage
regulating relay. The band center is 123 V (12.3 kV lineto- line at the
substation). The bandwidth is +/- 1.5 V (or 3.0 V) total. The time delay
for adjustments is 60 seconds. The LTC compensates for added load by
increasing the band center linearly from 123 volts at no load to 126
volts at full load (about 20 MVA or 1,200 A).

Capacitors

Capacitors are reactive power sources that affect the voltage by
supplying leading reactive current that compensates for the lagging
reactive current of the load. The Anatolia substation has six
three-phase, 1,800kVAR capacitor banks (hereafter referred to as
capacitors). The capacitors are computer controlled using an algorithm
developed by SMUD called Capcon. During normal operations, one capacitor
is turned on each time the VAR flow exceeds 900 kVAR (from the
substation bank), and a capacitor is switched off each time the VAR flow
exceeds -1,200 kVAR. These set points are modified on days hotter than
100oF, but this setting can be manually overridden depending upon
weather forecasts. These settings automatically adjust for abnormally
high or low bulk system voltages.

SMUD ANATOLIA III OVERVIEW

Distribution System Description

Anatolia is a residential community served by individual single-phase
lateral circuits from a three-phase primary feeder which connects to a
20 MVA, 69 kV/12.47 kV delta-wye transformer at the Anatolia-Chrysanthy
Substation (hereafter referred to as Anatolia-Chrysanthy or substation).
The feeder has several points along the line that allow for switching.
The length of the feeder to the switching cubicle furthest from the
substation is about 18,895 feet and the furthest distribution
transformer, 5K7, is about 22,360 feet from the substation. In Anatolia,
there are a total of 85 singlephase pad-mounted distribution
transformers: 23 are 75 kVA and 62 are 50 kVA. Also connected to the
same substation feeder is a rendering plant fed by two 1,500kVA
transformers, consuming between 20 and 200 kW, a water storage plant,
and a residential community with an estimated 1,000 homes between the
substation and Anatolia. There are two additional feeders that also
connect to the 20-MVA transformer that have an additional estimated
1,000 customer loads each.

SolarSmart Homes Description

SolarSmart Homes are advertised as able to reduce annual residential
electric bills by 60%. They combine cost-effective, energy-efficient
features and a rooftop PV system. Typical SolarSmart Home features
include: radiant barriers to reflect summer heat that would otherwise
enter the attic and cause greater need for air conditioning;
90%-efficient furnaces; 14 SEER / 12 EER HVAC systems;
compact-fluorescent lighting; ENERGY STAR-qualified windows; and
independent third-party verification, required to confirm all
energy-efficiency measures are installed and operate correctly.

A 2.0-kWac PV system can generate a major portion of the electrical
energy consumed in the home. The majority of the Anatolia PV systems
comprise 36 SunPower 63-watt SunTile roof-integrated modules feeding
into a SunPower SPR-2800x positive-ground, grid-connected inverter. The
inverter can be remotely accessed by SunPower if the homeowner elects to
connect it via the internet. The orientation of the PV systems ranges
from southeast to southwest.

MONITORING

The distribution system is monitored at the substation feeder using a
DAQ Electronics ART-073-79-0 Supervisory Control and Data Acquisition
(SCADA). The SCADA is ANSI-class high-accuracy, better than 0.5%. The
sampling rate is every two seconds with an average recorded every five
minutes. Four distribution transformers are monitored using PMI Eagle
440 monitors having an accuracy better than 1.0%. The sampling rate is
256 samples per cycle with an average recorded every five minutes. Four
homes are monitored at the service panel using PMI iVS-2SX+ power
monitors having an accuracy better than 1.0%. The sampling rate is 128
samples per cycle with an average recorded every five minutes.

Solar and metrological data are also collected from a
solar/meteorological station installed at the substation.
One-minute-interval data from the solar/meteorological station are
collected automatically daily.

ANATOLIA RESULTS

SMUD wanted to investigate what effect reverse power flow from exporting
PV systems would have on service and substation voltage regulation.
Figure 1 shows the Home-to-Substation voltage difference (top) and solar
irradiance (bottom) on a clear, cool day, Saturday, March 7, 2009,
representative of a day with relatively low load and high local PV
penetration. Penetration is defined as the amount of PV output divided
by the load at a particular point in time.

At night the substation voltage ranged between 0.4 V to 0.7 V higher
than the home voltage. This is representative of a typical circuit with
voltage drops through the line and transformer impedances. During
daylight hours, this reversed and the home voltage rose as high as 0.7
V, or 0.6%, greater than the substation voltage. The voltage amplitude
was 124.5 V at its peak, so it remained well within ANSI C84.1 limits.
Due to heavy home loads in the morning, the PV system did not begin to
export to the distribution system until almost noon. In March 2009,
there was a 2.0-kWac PV system on the home, 30.1 kWac of PV on the
transformer (twelve SolarSmart Homes), and 238 kWac of PV on the
distribution system (115 SolarSmart Homes). Preliminary estimates of PV
penetration levels on the feeder were 11% to 13% under lightly-loaded
conditions (2.0 MW) and around 4.0% of the total substation transformer
load. Since the PV penetration levels are still relatively low, there
were no adverse effects on voltage regulation.

Modelling Anatolia

To measure the effects of PV on voltage regulation, a Distributed
Engineering Workstation (DEW) model was created that included the
distribution transformers, secondary, and service connections. Once
validated using measured field data, the model will be used to determine
acceptable levels of PV penetration.

The feeder contains significant loads which are not part of the system
under evaluation including a rendering plant, water storage facility and
approximately 1,000 residential customers. Additionally, the substation
transformer and LTC are shared by two unmonitored feeders.

For initial testing and model verification, the unknown residential
loads were represented as a lumped spot load. Eventually, the
residential loads will either be calculated from load research
statistics or distributed based on distribution transformer size. The
rendering plant load was represented as a spot load based on recorded
data.

The load on the other feeders was modeled using a spot load placed
immediately after the LTC. Since the additional feeders share the LTC,
the load on those feeders will affect the position of the tap as well as
the regulation set-point. While the load on the other feeders will not
greatly affect the voltage profile of the system under study, these
loads may affect the voltage regulation.

To verify the topology of the model and to ensure the switching devices
are represented in their correct states, the electrical distances from
the substation of the measurement points were compared against the
values estimated from a GIS map. Table 1 describes this comparison. All
modeled electrical distances are within 3.5% of the estimated distances
as determined from GIS drawings.

Next, the behavior of the secondary-side of the distribution transformer
was validated against real data. The voltage rise measured between the
secondary of the distribution transformers and the service entry to the
homes will be affected by the loads and generation of the other,
unmonitored homes that share the secondary connection.

Figure 2 shows the simulated and measured voltage rise from Home 3 to
the secondary of Transformer 3 (8K6). The data reflects roughly a month
of data where Home 3 is net exporting real power with an inductive load
between 0.18 kVAR and 0.22 kVAR. The secondary system was simulated
assuming uniform generation from all customers connected to the
secondary of the transformer. The generation from the homes was
increased while maintaining a constant inductive load of 0.2 kVAR. The
homes were modeled as constant current loads.

The variation in measured voltage rise data is largely due to
uncertainties with the other customers sharing the secondary connection
as well as the resolution of the voltage measurement device at the home.
Figure 2 indicates that the model reasonably reflects the behavior of
the actual system. With improved monitoring and meter accuracy, the
model could be better validated and, if necessary, adjusted to more
accurately reflect the system.

Currently, there are too many uncertainties on the circuit to perform
any meaningful whole system validation. Efforts are under way to
increase the monitoring on the system so that the overall model may be
better validated. This includes adding meters at critical points to
eliminate many of the unknown loads described above.

CONCLUSIONS

After one year of monitoring the Anatolia SolarSmart Homes Community,
there was no excessive service or substation voltage due to reverse
power flow from exporting PV systems. Preliminary estimates of PV
penetration levels on the feeder were 11% to 13% under lightly-loaded
conditions (2.0 MW) and around 4.0% of the total substation transformer
load. Since the PV penetration levels were relatively low, there were no
adverse effects on voltage regulation. It was possible to see the
effects of the PV systems on the voltage at the individual homes and the
distribution transformers. This slight voltage rise was approximately
0.6% on clear days in comparison to the normal drop of -0.6% without the
PV exporting. The DEW model that has been developed reasonably reflects
the behavior of the actual system. The model will be used to determine
acceptable levels of PV penetration.

================

Photovoltaic Specialists Conference (PVSC)
2009 34th IEEE
Issue Date: 7-12 June 2009
On page(s): 001914 - 001917
Date of Current Version: 17 February 2010