1. D6.3.104 Field test results Pusula field test August 2014 Henry Rimminen VTT Technical Research Centre of Finland

2. 2 11/07/2016 Concept Earth faults can be detected more easily when observing sum of three-phase current in medium voltage lines If the sum deviates from zero, fault is detected Wireless, low-cost, enery-harvesting sensors are used for this

3. 3 11/07/2016 2nd gen. devices installedNormal state signal

4. 4 11/07/2016 1st generation wireless line fault detector  1st generation devices:  Field test results presented in D6.3.5  And also in H. Rimminen, A. Kostiainen and H. Seppä, ”Demonstration of a Low-cost Fault Detector for Sum Current Measurement of Overhead MV Lines, ” in proc. The IEEE 4th European Innovative Smart Grid Technologies (ISGT), Oct. 6 - 9, Lyngby, Denmark, 2013.  Minimum tripping threshold was around 5 A on a 180 A measurement range  Earth faults up to 330 Ω were detectable  10 out 32 faults were detected in Masala  Measured waveforms matched well with substation reference  Self-sustained operation was not achieved with 37 A load current

5. 5 11/07/2016 2nd generation improvements  Analog phase locked loop (PLL) for locking the sample trigger to 50 Hz. The old had a digital one, which introduced quantization noise  PLL lockable also to electric field sensor  New current transformer for energy harvesting  New power management circuit for better efficiency  Self-sustained operation is acheived at 30 A and above (verified at MIKES)  Dual measurement coil for better absolute accuracy  Radio changed to ZigBee for lower power consumption and for compatibility with ABB  External antenna for better reception

6. 6 11/07/2016 Comparison (1st gen. v.s. 2nd gen.) New harvester 1st gen. 2nd gen. External antenna ZigBee radio processor Simple processor with separate bluetooth radio Li-Po battery Ni-MH battery Dual measurement coil Intelligent power management Electric field sensor

7. 7 11/07/2016

8. 8 Base station  Texas Instrumets SmartRF05 EB is used as the ZigBee base station

9. 9 11/07/2016 Packet timing 1666 µs 1200 µs 1666 µs 1200 µs 1666 µs All balls listen for command Sample 5Sample 6 Sample 7 Sample 8 Sample 9 … All balls listen for command Ball1 sends 4 current and voltage samples (no.1-4) Sample clock (locked to 50 Hz) 1/600 S/s = 1666 µs …Sample 4 Ball2 sends 4 current and voltage samples (no.1-4) Ball3 sends 4 current and voltage samples (no.1-4) Timing tolerance = 466 µs Sample moment

10. 10 11/07/2016 Pusula results  21 out of 29 faults were detected (72%). With 1 st gen. device in Masala the detection rate was 31 %.  There were 13 false detections. No comparative value from Masala. Vast majority of the false alarms were caused by synchronization errors. Synchronization errors cause packet collisions (see low margins in packet timing slide). They may be caused by the low load current (2% of measurement scale). We expect that higher load current would lead to less noisy synchronization and to lesser packet collisions.  Threshold for sum current was set to 2.0 A.  Load current was 2.1 A (measurement scale of indicators was 100 A)  Current lock-in was used in sampling.  Electric field lock-in was no possible, since the electric field signals were not in 120 degree phase shifts. This may be because the pole was severely tilted and there was earth conductor 1 m below the middle phase.  The 2 nd gen. fault indicators were not affected by heavy rain. Rain caused severe problems with 1 st gen. devices in Masala field test.

11. 11 11/07/2016 Fault recording example, High-ohmic earth fault (10 kΩ) Fault current ~2.1 A No change in voltage Fault current 2.1 A No dip in voltage

12. 12 11/07/2016 Fault recording example, Low-ohmic earth fault (0 Ω) Fault current 11.0 A Clear voltage dip

13. 13 11/07/2016 Fault recording example, short circuit (0 Ω) Severe voltage dip ~300 A fault current (limited by clipping)

14. 14 11/07/2016 Summary  The detectors worked moderately even in heavy rain  Up to 10 kΩ earth faults were detected  The false detection rate was high, which was most likely due to too large measurement range in reference to the small load current  In general, the 2 nd generation fault detectors performed better than the 1 st generation.

15. 15 11/07/2016 VTT creates business from technology