optimal settings for multiple groups of smart inverters on ... · optimal settings for multiple...
TRANSCRIPT
Optimal Settings for MultipleGroups of Smart Inverters on
Secondary Systems UsingAutonomous Control
Mobolaji Bello,Electric Power Research Institute (EPRI)
April 25, 2017
Outline
• Use case feeder model• Smart inverter functionalities• Autonomous vs integrated control• Load and solar variability index• Methodology to derive settings• Illustrate feeder response to settings• Show feeder impact from various
inverter settings
• Method to select appropriate settings in a very simple way• Focus is on volt-var settings!
Secondary (LV) Overvoltage Due to Residential Rooftop PV
• Impact at the Customer Level– Customers with PV fed from
same service transformer– Light-load conditions– Each PV system tends to increase
secondary voltage– Results in overvoltage on
secondary
Number of utilities have seen secondaryover voltages where PV customers are fedfrom same transformer
Number of utilities have seen secondaryover voltages where PV customers are fedfrom same transformer
Feeder XYZ OverviewSubstation Transformer
• 69kV – 12.47kV (Delta-Wye)• 25MVA• X1 = 8.98%
Voltage Setpoints
• 1.03pu @ 69kV bus• 1.01pu @ unregulated 12.47kV bus
Voltage Control
• 5 capacitor banks• No feeder regulator
Load Modeling• kW based on reported AMI data• Residential loads assumed 0.9 PF• Commercial loads assumed 0.85 PF
XYZ Loading and Fault Levels
Fault levels• 3 ph. = 11953A• 1 ph. = 12635A
Fault levels• 3 ph. = 11953A• 1 ph. = 12635A
Peak PV• on 2016-04-25 @ 12:00:00• total PV is 2737.310kW and the net load is -307.537kW• total load is 2429.773kW and the net load is -307.537kW
Most back fed• on 2016-04-28 @ 12:00:00• total PV is 2690.920kW and the net load is -538.490kW• total load is 2152.430kW and the net load is -538.490kW
Heavy load• on 2016-06-20 @ 17:00:00• total PV is 775.200kW and the net load is 9102.193kW• total load is 9877.393kW and the net load is 9102.193kW
Fdrname
Loadingcondition Abbr kW
XYZ
Peak (PV-PV1,Net load-NL1,Total load-TL1)
PV1 2700NL1 -300TL1 2400
Most back fed(PV-PV2,Net load-NL2,Total load-TL2)
PV2 2650NL2 -500TL2 2100
Heavy load(PV-PV3, Netload-NL3,Total load-TL3)
PV3 770NL3 9100TL3 9900
Smart inverters
allows the powerfactor to be set at afixed value
Volt-Watt: Limits of Operation
allows the DER to manage poweroutput based on voltage
′ ′′1.0 pu
Power( )
Voltage( )
% A
vaila
ble
VARS
(Q)
Indu
ctiv
eCa
paci
tive
100%(nominal voltage)
Q = 0System Voltage
Dead Band
allows the DER to manage its own reactivepower output in response to local service voltage
Volt-VAR Example Characteristic
Autonomous vs integrated control
Distributionoperator
Substations
Feeders
Integrated control: MV node• Typical DERMS approach• Based on electrical proximity to group,
monitor and measure• Operation and coordination all the SI is key.
Autonomous: inverter node• Inverter measurement is based on
inverter node voltage!
Autonomous InfrequentCommunication
FrequentCommunication
Remote ON/OFF
Power Factor Control (can be schedulebased)
Volt – var (less effective) (more effective)
Volt – Watt
Communication Needs for Grid Support Functions
Project goal was to useAutonomous system !
Categories for Daily Variability Conditions: Sandia’s VI and CI applied
Clear Sky POA IrradianceMeasured POA Irradiance
Clear Sky POA IrradianceMeasured POA Irradiance
HighHigh
ModerateModerate
MildMildOvercastOvercast
ClearClear
VI < 2CI ≥ 0.5
VI < 2CI ≤ 0.5
2 ≤ VI < 5
5 ≤ VI < 10
VI > 10
Variability Conditions: AZVariability Conditions: AZ
Variability Conditions: NMVariability Conditions: NMVariability Conditions: NJVariability Conditions: NJ
Q2 2012 Q3 2012 Q4 2012 Q1 20130
20
40
60
80
100
Perc
enta
ge o
f Day
s (%
)
Season
Variability Conditions: TNVariability Conditions: TN
Q2 2012 Q3 2012 Q4 2012 Q1 20130
20
40
60
80
100
Perc
enta
ge o
f Day
s (%
)
Season
Q2 2012 Q3 2012 Q4 2012 Q1 20130
20
40
60
80
100
Perc
enta
ge o
f Day
s (%
)
Season
Q2 2012 Q3 2012 Q4 2012 Q1 20130
20
40
60
80
100
Perc
enta
ge o
f Day
s (%
)
Season
“PV Measures Upfor Fleet Duty” –IEEE Power/EnergyMarch/April 2013Credit: SANDIA
Best Settings Analysis Framework
0 5 10 15 20 25
1.024
1.026
1.028
1.03
1.032
1.034
1.036
1.038
1.04
1.042
1.044
Hour
Vol
tage
(pu)
Voltages with different voltvar settings
---- Voltvar
---- No PV---- PV base
What is the best voltage response?
Depends on what objective you want to achieve.
Discrete voltagechanges are due tocapacitor switching orinverter status change.
Deta
iled
feed
er m
odel
Blue lines indicatevoltage response usingdifferent volt-varsettings.
1a. Performance Objective
Benefits
Site
Efficiency Power Quality Asset Life Deference ofCapital Spending Reliability Enabling
Redu
ced
dist
ribu
tion
line
loss
es
Impr
ove
cust
omer
effi
cien
cy C
VR
Flat
ter v
olta
ge p
rofil
e
Impr
oved
har
mon
ics
Volt
age
flick
er
Ove
rvol
tage
Redu
ce LT
C ta
p ch
ange
s
Redu
ce li
ne r
egul
ator
tap
chan
ges
Redu
ce s
witc
h ca
p ch
ange
s
Def
er c
apac
itor a
ddit
ions
Def
er li
ne r
egul
ator
s
Def
er re
cond
ucto
ring
Def
er s
ubst
atio
n up
grad
es
Supp
ort d
urin
g m
omen
tary
Supp
ort d
urin
g au
tom
atio
n
Hig
her
Pene
trat
ion
of P
V
Network Deta
iled
feed
er m
odel
1a. Key performance objective for XYZ
• MeanPCCv – average PV interconnectvoltage
• Voltage Variability Index – Voltagevariability index at the inverter terminal
• Consumption – End-use consumption inkWh
• Losses – Total feeder losses in kWh
• Time Above – Seconds that any point onthe feeder is above 105% nominal
• Time Below – Seconds that any point onthe feeder is below 95% nominal
• Max V – Maximum voltage at any point onthe feeder
• Min V – Minimum voltage at any point onthe feeder
• Difference btw Feeder-wide Max and Mini.e. flattened voltage profile (Vdiff.)
Metric based on primary voltage Deta
iled
feed
er m
odel
1b. Detailed feeder model
• Model prep and data cleaning up is huge!• Get a snap shot for the real peak load• Load shape must reflect values for at least one year
Deta
iled
feed
er m
odel
1c. Detailed feeder model – Smart Inverter location
Autonomous system analyzed!
Feeder XYZ
S.Inv Units 48Size(kVA) 6PVs (all units) 512
Deta
iled
feed
er m
odel
Deta
iled
feed
er m
odel
Powerfactor
Voltwatt
Vot-var
• OpenDSS used to take advantage of smart inverter models!• Simulation is conducted at the minute resolution for a 24 hr
period in each combination of scenario.• With the inverter settings in, then, crunching can start ..…
• OpenDSS used to take advantage of smart inverter models!• Simulation is conducted at the minute resolution for a 24 hr
period in each combination of scenario.• With the inverter settings in, then, crunching can start ..…
1d. Simulated Inverter Settings
2. Smart inverter modeling
An automated routine for thousands of analyses. Takes a while becauseof the total number of runs and combinations considered………
Run
quas
i sta
tic ti
me
serie
s mod
el
3. Data processing for performance metrics (load/solar probability)
Determine peak tooff-peak ratioBased on one yr SCADA data
Feeder head identified that midday feeder load level is 30% ofthe time.
• Weighted average of days considered over one year• To examine multiple feeder impacts simultaneously• Weighted average of days considered over one year• To examine multiple feeder impacts simultaneously
Proc
ess r
esul
ts
Process Network Response (illustration)Optimized PerformanceMetric• Black dashed line
indicates rank based onperformance metric
• Trends because controlsettings utilized differentxtics involving set points,bandwidths and slopes
Voltage Constraints• Blue circles indicate no
voltage violations• Green circles indicate
improvement in voltagebut still has someviolations
• Red circles indicateadditional voltageviolationsDifferent Control Settings
(110 unique settings analyzed)
Best
Worst
Proc
ess r
esul
ts
XYZ- Control settings
187 unique settings analyzed187 unique settings analyzed
Sele
ct b
est s
ettin
g
VdiffTime belowANSI
Losses VoltageVariabilityIndex
Recommended settings (Voltvar) –XYZ
Sele
ct b
est s
ettin
g
How much improvements did these provide? XYZ
depends on feeder characteristicsdepends on feeder characteristics
Sele
ct b
est s
ettin
g
Variable day_peak
Variable day_offpeak
Overcast day_peak
Overcast day_offpeak
Clear day_peak
Clear day_offpeak
Recommended Control Settings…XYZ
depends on solar profile
Sele
ct b
est s
ettin
g
Recommended Control Settings…
Performance objective
Recommended Control Settings…
Seasonal Variations
Summary
• Methodology can be applied to determine best recommended settings• There is a unique feeder impact for each of the different control types and
settings.• Periodical updating of settings can be beneficial• Best inverter settings are dependent on
– Performance metric (losses, voltage difference, time below ANSI, VVI)– Feeder characteristics– Load and solar condition (peak load, variable solar; off peak load, clear solar
etc.) Solar and load measurements could be used to automatically update the inverter
settings to best correlate with the current field conditions Communication with the inverters can be used to update settings based on
operator command or automatically based on SCADA measurements.– PV size Total smart inverters on the circuits are needed to see more feeder impact and
effect!
Jeff Smith
865.218.8069
Mobolaji Bello
865.218.8005
Ben York
865.218.8187
Davis Montenegro
865.218.8091