Deliverable D6.3.5, 2nd Proof of Concept Demonstrator, Line fault detector field test at Masala, with ABB and Fortum 16-19.4.2012 Henry Rimminen, Anu Kärkkäinen VTT Technical Research Centre of Finland 27/06/2012 Deliverable D6.3.5 The field test results from Masala is analyzed in the following This document is the deliverable D6.3.5 (2nd Proof of Concept Demonstrator of the VTT line fault detector) 2 27/06/2012 3 Background Fault detectors were designed to measure magnitude of current and phase of voltage Combination of these reveal phase change caused by faults Both measurements are non-contact Time synchronization enables sum current calculation of three phases with three detectors Bluetooth is used for synchronization and data transmission They harvest operating power from the line current 27/06/2012 Operation 1. 2. Clocks are synchronized with Bluetooth Currently all data is transmitted for analysis. Later only alarms are enough. Sample rate is 1200 Hz (= 24 samples per 50-Hz cycle). 4 27/06/2012 Photos 5 27/06/2012 6 Normal state In normal state (Wed. morning 9:05), the phase current was measured to be 34 A (all currents are RMS). At the power station ABB measured ≈ 33 A. The sum current was measured to be 1.1 A. The three-ball-synchronization worked as planned and required no post synchronization. 27/06/2012 7 Interpretation of fault recordings This helps to interpret the following fault recording figures Topmost figure shows all three line currents (L1, L2, L3) starting 150 ms before the fault. Middle figure shows sum current (L1 + L2 + L3) Bottom figure shows synchronization flags for three fault detectors. After Bluetooth synchronization, the fault detectors maintain their relative synchronization using the line current waveform. They slowly adjust their internal clock so that every 24th sample matches the zero-crossing of the current waveform. Synchronization flag is up when the internal clock and current match. When mismatch occurs, flag goes down. This happens always during faults, so the synchronization flags were used to trigger the recordings. Sum current is at its minimum when all three flags are up. 27/06/2012 8 L3, short to ground L3 shows a current of 38 A before the fault. During the fault it shows 117 A for ~20 cycles and then returns to 38 A. Sum current is below 2 A before and after the fault and 86 A during the fault. Phase curents, recording no. 615, time ke10.03 200 L1 L2 L3 Current (A) 100 0 -100 -200 3512.5 3513 3513.5 3514 3514.5 3515 3514 3514.5 3515 Time (s) Sum current 150 Current (A) 100 50 0 -50 -100 -150 3512.5 3513 3513.5 Time (s) Ball sync flags (used as recording trigger) Clock syncd to waveform (1/0) 1.5 L1 L2 L3 1 0.5 0 -0.5 3512.5 3513 3513.5 3514 Time (s) 3514.5 3515 27/06/2012 9 L2, short to ground L2 shows a current of 36 A before the fault. During the fault it shows 120 A for ~20 cycles and then returns to 36 A. Sum current is below 1 A before and after the fault and 89 A during the fault. Phase curents, recording no. 672, time ke10.10 200 L1 L2 L3 Current (A) 100 0 -100 -200 3950 3950.5 3951 3951.5 3952 3952.5 3951.5 3952 3952.5 Time (s) Sum current 200 Current (A) 100 0 -100 -200 3950 3950.5 3951 Time (s) Ball sync flags (used as recording trigger) Clock syncd to waveform (1/0) 1.5 L1 L2 L3 1 0.5 0 -0.5 3950 3950.5 3951 3951.5 Time (s) 3952 3952.5 27/06/2012 10 L1, 330 Ω to ground L1 shows a current of 35 A before the fault. During the fault it shows 62 A for ~20 cycles and then returns to 35 A. Sum current is below 1.5 A before and after the fault and 26 A during the fault. With 330 Ω resistor, the sum current is more than three times lower than with a 0-Ω fault (26 A v.s. 86-89 A) Phase curents, recording no. 1128, time ke11.42 A). 100 L1 L2 L3 Current (A) 50 0 -50 -100 9472 9472.5 9473 9473.5 9474 9474.5 9473.5 9474 9474.5 Time (s) Sum current 40 Current (A) 20 0 -20 -40 9472 9472.5 9473 Time (s) Ball sync flags (used as recording trigger) Clock syncd to waveform (1/0) 1.5 L1 L2 L3 1 0.5 0 -0.5 9472 9472.5 9473 9473.5 Time (s) 9474 9474.5 27/06/2012 11 L1, bare wire on terrain 1 L1 shows a current of 34 A before the fault. During the fault it shows 128 A for ~20 cycles and then returns to 34 A. Sum current is below 1 A before and after the fault and 94 A during the fault. With bare wire to terrain, the sum current is similar than with the shorts to ground (94 A v.s. 86-89 A). Phase curents, recording no. 1, time ke13.47 200 L1 L2 L3 Current (A) 100 0 -100 -200 72 72.5 73 73.5 74 74.5 73.5 74 74.5 Time (s) Sum current 150 100 Current (A) 50 0 -50 -100 -150 -200 72 72.5 73 Time (s) Ball sync flags (used as recording trigger) Clock syncd to waveform (1/0) 1.5 L1 L2 L3 1 0.5 0 -0.5 72 72.5 73 73.5 Time (s) 74 74.5 27/06/2012 12 L1, bare wire on terrain 2 L1 shows a current of 34 A before the fault. During the fault it shows 129 A for ~20 cycles and then returns to 34 A. Sum current is below 1 A before the fault, but L3 fails to resync afterwards leaving a sum current of 9 A. Sum is 90 A during the fault. This is Identical with the first bare wire on terrain test. Phase curents, recording no. 2, time ke13.54 200 L1 L2 L3 Current (A) 100 0 -100 -200 486.5 487 487.5 488 488.5 489 488 488.5 489 Time (s) Sum current 150 Current (A) 100 50 0 -50 -100 -150 486.5 487 487.5 Time (s) Ball sync flags (used as recording trigger) Clock syncd to waveform (1/0) 1.5 L1 L2 L3 1 0.5 0 -0.5 486.5 487 487.5 488 Time (s) 488.5 489 27/06/2012 13 L3, bare wire on a small tree 1 L3 shows a current of 33 A before the fault. During the fault it shows 65 A for ~20 cycles and then returns to 33 A. Sum current is below 1 A before and after the fault. Sum is 30 A during the fault. This is almost identical to the 330-Ω to ground test (sum is 30 A v.s. 26 A). Phase curents, recording no. 22, time ke14.27 100 L1 L2 L3 Current (A) 50 0 -50 -100 2470.5 2471 2471.5 2472 2472.5 2473 2472 2472.5 2473 Time (s) Sum current Current (A) 50 0 -50 2470.5 2471 2471.5 Time (s) Ball sync flags (used as recording trigger) Clock syncd to waveform (1/0) 1.5 L1 L2 L3 1 0.5 0 -0.5 2470.5 2471 2471.5 2472 Time (s) 2472.5 2473 27/06/2012 14 L3, bare wire on a small tree 2 L3 shows a current of 32 A before the fault. During the fault it shows 64 A for ~20 cycles and then returns to 32 A. Sum current is below 1 A before and after the fault. Sum is 32 A during the fault. This is identical with the first small tree test. Phase curents, recording no. 23, time ke14.32 100 L1 L2 L3 Current (A) 50 0 -50 -100 2808 2808.5 2809 2809.5 2810 2810.5 2809.5 2810 2810.5 Time (s) Sum current Current (A) 50 0 -50 2808 2808.5 2809 Time (s) Ball sync flags (used as recording trigger) Clock syncd to waveform (1/0) 1.5 L1 L2 L3 1 0.5 0 -0.5 2808 2808.5 2809 2809.5 Time (s) 2810 2810.5 27/06/2012 15 L1, L2, L3 shorted together L1, L2, L3 show 34 A before the fault. During the fault they clip at 200 A (peak not rms) for ~10 cycles and then return to 34 A. The recovery to normal current seems to take 100 ms. Sum current is below 2 A before the fault and remains at 4 A for 500 ms after the fault. Then it goes below 1 A. Sum is 80 A during the fault but it is not reliable due to clipping. Phase curents, recording no. 81, time ke15.12 200 L1 L2 L3 150 100 Current (A ) 50 0 -50 -100 -150 -200 5190.5 5191 5191.5 5192 5192.5 5193 5192.5 5193 Time (s) Sum current 150 Current (A) 100 50 0 X: 5192 Y: 0.327 -50 -100 -150 5190.5 5191 5191.5 5192 Time (s) Ball sync flags (used as recording trigger) Cloc k s y nc d to wav eform (1/0) 1.5 1 0.5 0 -0.5 L1 L2 L3 27/06/2012 16 Results Fault phase Fault resistance Fault type (Ω) L3 0 Ground L2 0 Ground L1 330 Ground L1 0 Terrain L1 0 Terrain L3 0 Tree L3 0 Tree L1,L2,L3 0 Short together Phase current (Arms) Fault Normal 117 38 120 36 62 35 128 34 129 34 65 33 64 32 200 34 Fault - Normal 79 84 27 94 95 32 32 166 Sum current (Arms) Fault Normal 86 2 89 1 26 1.5 94 1 90 1 30 1 32 1 80 2 Clipping? no no no no no no no yes The difference of phase current at normal and at fault state (column “Fault – Normal”) follows very closely the sum current. Differences may be caused by magnetic leakage between detectors or the network actually can not maintain currents of the two normal phases. 27/06/2012 17 Problems Tuesday tests did not include one fault detector device (L2) because it broke down due to rain. L2-device was repaired on Wednesday morning Voltage channels of all three detectors recorded blank data during all tests. This was caused by a code bug. The voltage channels were tested with 20 kV in laboratory, but the bug was made after that. Half of the recordings were missed due to intermittent failures of the Bluetooth connection. Cause was most likely moist surface of the device enclosures (the balls), which hindered the quality of the connection. These were not plotted in the results. 18 27/06/2012 Energy harvesting Battery Voltage 3 6 5.8 2.5 Battery voltage (V) 5.6 5.4 2 5.2 Device A 5 1.5 Device B 4.8 1 4.6 4.4 0.5 Device C Measuring? 4.2 4 0 0 5 10 15 20 25 30 Time (h) • • As seen from the figure, the energy harvesting works as planned. When idle (Measuring? = 0) the battery voltages rise. The mean harvested current was approximately 6 mA. The devices consume 3 mA at idle. When measuring, the battery voltage drops, since the devices consume 25 mA. The devices managed to stay operational while measuring 8 hours during days and charging at night. 27/06/2012 19 Discussion As mentioned before, voltage phase was missing due to a code bug However, there was indication that the phase of the current changed during faults, since the detectors fell out of sync during every fault. They have stable quartz oscillators. The detectors indicate successful synchronization if zero crossing of the measured current is within ± 4.4 A (± 5.2º at 34 A). If the criterion is not fulfilled, the sample clock is incremented or decremented at constant steps of 35 µs on 20-ms intervals. In both of the 0-Ω ground faults and in the three-line short fault, the outof-sync flag was on during the whole 400-ms fault duration. The sample clock can adjust itself 700 µs during that time, which equals 12.1º. This means that the phase of the current must have changed at least 17.3º (12.1º + 5.2º) during the aforementioned faults Direction of the phase change cannot be calculated 27/06/2012 20 Summary Voltage phase measurement failed due to a code bug and thus the fault phase calculations were not possible. Problems with Bluetooth caused large amounts of corrupted data Current measurement was successful and the measured value matched the value at the power station. Error was ≈ 1 A, which equals ≈ 0.5% full scale accuracy. Sum current calculation of three phases succeeded and no post synchronization was needed Ground fault resistance at least up to 330 Ω is detectable. Data from higher resistance faults were corrupted. Line-to-line faults were detectable The two fault types can be distinguished (with three detectors) Energy harvesting was successful Many lessons were learned, which creates good basis for next demonstrator 27/06/2012 21 Future work It is necessary to use flash memory in the detectors to avoid radio problems. Final product will not need this, since all data is not stored then. Currently we let the voltage signal clip and measure only voltage phase. According to feedback we should keep it sinusoidal and measure also the voltage waveform. This would reveal amplitude of the voltage, even tough it might be inaccurate. Overall reliability must be improved Possibility of using “lintupallo” as enclosure will be studied The voltage bug was an easy fix 27/06/2012 VTT creates business from technology 22
© Copyright 2026 Paperzz