Overview of openFLE open source fault location engine

1. Open Fault Location Engine - BETA
July 19, 2012
1 © 2012 Grid Protection Alliance. www.GridProtectionAlliance.org

2. Development Overview
For an input file “1234.pqd”
2 © 2012 Grid Protection Alliance.

3. openFLE Computational Approach
• As an extensible platform, openFLE has
two major layers:
• Get event data:
– openFLE frees developers from the chore of
reliably parsing and positioning event data for
analysis
• Perform the calculations:
– openFLE can be extended with new
algorithms
3 © 2012 Grid Protection Alliance.

4. openFLE Computational Approach
4 © 2012 Grid Protection Alliance.

5. Completed Items for Fully Automated System
• Data Layer
– Discover when files are dropped in a folder
– Parse COMTRADE or PQDIF files (including COMTRADE ’96)
– Reliably and predictably convert time domain data to frequency
domain for each cycle of data available
• Calculate true RMS voltage and current magnitudes
• Curve fit data to determine phase angles
• Computational Layer – A working Example Implementation
– Identify the cycles where a fault is present
– Determine fault type
– Calculate fault impedance based on the fault type
– Calculate fault distance
– Output calculations and logs
– Move input files from drop folder to results folder.
5 © 2012 Grid Protection Alliance.

6. RMS Calculation and Curve Fitting Required
since Fault Data Isn’t Purely Sinusoidal
6 © 2012 Grid Protection Alliance.

7. 30 Values Computed for Each Cycle
• Phase VoltageN – Magnitude and Angle
• Peak Phase VoltageN
• Sequence Voltages – Pos, Neg, Zero, Magnitude
and Angle
• Phase Currents – Magnitude and Angle
• Peak Phase Current
• Sequence Currents – Pos, Neg, Zero, Magnitude
and Angle
Magnitude based on RMS calculation from individual measurements.
Angle based on phase difference to sinusoid fitted to data.
7 © 2012 Grid Protection Alliance.

8. Detecting a Fault
• Assume the first cycle represents pre-fault
condition
• A fault is detected in any cycle where the RMS
current (cycle value) exceeds pre-fault current
by a factor of 5 AND exceeds 500 amps
Improvement: Use line ratings as part of the
basis for fault detection.
• Other simple methods to implement ?
8 © 2012 Grid Protection Alliance.

9. Determining the Fault Type
• For each phase, the fault test was applied
• The representative best fault cycle is
selected as the one with the largest sum of
all RMS currents
• The number of currents which pass the
fault test determine the fault type.
• Other simple methods to implement?
9 © 2012 Grid Protection Alliance.

10. Determining the Fault Location
• Calculated Fault Impedance using the Absolute Value
Method (Reactance method was tested and found to be less accurate.)
• 3 Phase
 Select phase with “purest” sine wave
 ZF = RMS (cycle data) |VN| / |I|
 Distance is ratio ZF/Z1
• Line to Ground
– ZF = RMS (cycle data) |VN| / |I|
– Distance is ratio ZF/ZS (where ZS is loop impedance)
• Line to Line
– ZF = RMS (cycle data) |VL-L| / (|IP1 - IP2| )
– Distance is ratio ZF/Z1
10 © 2012 Grid Protection Alliance.

11. The Test Event Files
• Event 1 – Line-to-Ground (A), DFR, 14.9 miles
• Event 2 – Line-to-Ground (A), DFR, 30.9 miles
• Event 3 – Line-to-Line (AB), DFR, 18.6 miles
• Event 4 – 3 Phase, DFR, 2.3 miles
• Event 5 – 3 Phase, DFR, 5.9 miles
• Event 6 – Line-to-Line (AB), PQM, 29.5 miles
• Event 7 – Line-to-Line (BC), DFR, 7.5 miles
• Event 8 – 3 Phase (C), DRF & PQM 36.4 miles
11 © 2012 Grid Protection Alliance.

12. Data Anomalies
• Bad GPA developed configuration file –
successfully resolved through COMTRADE
configuration information
• One file with bad channel index – openFLE
gracefully ignores file.
• Sign seemed inverted on one set of data.
12 © 2012 Grid Protection Alliance.

13. Seemingly out-of-phase
Measurements
A-phase Voltage
A-phase Current
13 © 2012 Grid Protection Alliance.

14. RESULTS
Two Fault Location Methods Tested
TVA Provided Data Yielded 20 Test Cases
(All values in miles)
Absolute calculation method selected based on
better alignment with actual fault distances.
14 © 2012 Grid Protection Alliance.

15. RESULTS
TVA Calculation vs. Actual Location
TVA Calculated Distance (Miles)
Actual Distance (Miles)
15 © 2012 Grid Protection Alliance.

16. RESULTS
openFLE Calculation vs. Actual Location
openFLE Calculated Distance (Miles)
Actual Distance (Miles)
16 © 2012 Grid Protection Alliance.

17. RESULTS
Calculated Location –openFLE vs. TVA
openFLE Calculated Distance (Miles)
TVA Calculated Distance (Miles)
17 © 2012 Grid Protection Alliance.

18. CodePlex Template Established
http://openFLE.codeplex.com
18 © 2012 Grid Protection Alliance.

19. openFLE Manager
19 © 2012 Grid Protection Alliance.