Improving Fault Response with Real-time
Fault Location Analysis through
Combined Outage and Advanced
Distribution Management
David Lyons – Georgia Power
Brad Williams – Oracle
Kevin Costin - Oracle
Southern Company
Southern Company serves a 120,000
square-mile territory in Alabama, Florida,
Georgia, and Mississippi
Southern Company Overview
• Vertically Integrated – Generation / Transmission /
Distribution
– 4 Operating Companies, SoCo Services (IT, HR),
Southern Linc (Telecom), Southern Nuclear, Southern
Power (wholesale generation)
• Generation – 43,000 MW
‒ Coal – 68%
‒ Natural Gas / Oil – 16%
‒ Nuclear – 15%
‒ Hydro, Renewable – 1%
Georgia Power Company
• 2 .3 Million Customers
• 153 Counties
• 57,000 square miles
• Metro 1.1 Million
Customers
• Outside Metro 1.2
Million Customers
Georgia Power Statistics (Distribution
and Customer Service)
• Georgia Power
• Design, Construct, Operate, and Maintain:
– 1.8 million poles
– 800,000 street lights
– 640,000 transformers
– 68,000 miles of line
– 57,000 sq. miles of service territory
• Serve 2.3 Million Customers(customer = meter)
• Customer Service and Satisfaction – Core Value
Background
• Presentations by other utilities utilizing waveform
data to locate faults.
• GPC’s deployment of SEL 351 Relays
– DEP for watts and Var data
– Replacement for failing Reclosing Relays
The Question
• Can we replicate the success of using fault values and
waveforms to identify fault locations using the SEL
351S’ fault data?
Current System: DCC Operators
• Static fault values based upon CYME studies are
placed in GIS at strategic locations and flow to TCMS
model (500 of 2000 circuits completed)
• DSCADA system pulls fault data directly from SEL
feeder relays
• Fault magnitude and type is displayed on SCADA
screens
• Data is used by DCC Operators to locate probable
fault location (manual process)
Current System: Engineers
• DSCADA system pulls fault data directly from SEL
feeder relays
• Fault magnitude and type is displayed on SCADA
screens and stored on a webpage
• Data is used by engineers to find temporary faults
(using Socket or CYME)
• Static fault values placed in GIS are easily available in
DistView and on TCMS maps at strategic locations
DistView Fault Location “Wagon
Wheels”
Validate the concept
• Collect data about substation breaker operations /
lockouts
– Targets and fault current magnitude (DSCADA
webpage)
– Fault location and cause (TCMS reports)
• Calculate fault current at the fault location
• Calculate difference between the calculated value
and actual value, measured in spans of wire
Validation summary
• 51 events were studied in Duluth, Lawrenceville,
Atlanta, and Cartersville from March 2009 till
February 2010
• Average difference between actual fault current and
predicted location was 6 spans
• 61% of the values were within 5 spans
• 92% of the values were within 15 spans
Desired System
• The desire is a system that will automatically:
– Retrieve fault data
– Identify fault location
– Present graphical information to operator
– Allow first responders to go directly to the fault
(including temporary faults)
Visualization Tool
• Operators wanted visualization integrated into
TCMS.
– Utilize Oracle fault calculation engine, work
agenda item, and graphical displays
– Obtain fault data from SEL relay via existing
SCADA-TCMS ICCP interface (enhanced to include
analog values… functionality exists, but not
currently implemented)
Fault Location: Additional Enhancements,
Data Requirements vs. Existing TCMS
• Enhancements to existing, GIS-derived TCMS
electrical model
– Add source bus impedance (per feeder), wire
impedance data (table)
• For accurate prediction, requires greater level of
wire-size accuracy… cleanup being done in
conjunction with other stimulus project needs
GPC - Visualization Pilot
• 2 Substations electrical models provided to OMS
Vendor (Oracle)
– Peachtree Corners – 7 Circuits
– Tilly Mill – 4 Circuits
• Oracle fault calculator took SEL fault values,
converted them to an impedance, and matched the
impedances to locations in TCMS
• Compared TCMS locations to known faults locations
Visualization Results
• The Fault Calculator, Visualization tool and ICCP data
stream from DSCADA worked as designed
• Time requirements were met.
– Fault locations were displayed in TCMS within 45
seconds of the outage
• Fault locations were often, but not always, predicted
closer to the substation than actual. This indicated
that the fault values were too high.
Next Steps
• Finalized statement of work with Oracle to move the pilot to
production in late 2011
• Continue SEL penetration in substation breakers as part of
SGIG. Currently 1600 out of 2018
• Create a statement of work with Schweitzer to modify the
fault magnitude value for increased accuracy
• Create a statement of work with Schweitzer to modify how
the fault magnitude data is presented in the DNP map of the
SEL351S to make it compatible with the new CRG and non IP
SEL devices connected to RTUs.
SEL modification of Fault
Magnitude data
• Schweitzer
Waveform
data from two
phase fault
• Report steady
state value
instead of
maximum
value
Oak Street 1162 - 11/10/2009 14:13:28.3000
Va
Vb
Vc
Ia
Ib
Ic
Voltage (V)
10000
5000
0
-5000
-10000
6000
Current (A)
4000
2000
0
-2000
-4000
2AB
0.00
Electrotek/EPRI
0.05
0.10
0.15
Time (s)
0.20
0.25
PQView®
GPC Fault data flow
TCMS
SCADA Gateway(s)
SCADA
IP communication based Breaker
Relays, Reclosers, and Fault indicators
Future Fault Data Flow
TCMS
SCADA Gateway(s)
SCADA
IP communication based Breaker
Relays, Reclosers, and Fault
indicators
Non IP
SCADA
Devices
Expected Benefits
• Addresses Arc Flash issues by changing protection
with Hot Line Tags
• Reduce SAIDI by knowing approximate fault locations
• Reduce MAIFE & SAIDI by identifying unknown
operations before they repeat or become permanent
• Reduce labor associated with line patrolling
• Provide diagnostic data for breaker malfunctions &
Customer complaints