On-line PD testing  & Diagnosis of MV & HV Cables

Contents

Introduction to PD damage to cables

On-line PD Measurement

• Motivations

• Sensors and Measurement

• Location and Field Applications

• Monitoring

Case Studies

Conclusions

Cables

• Paper (PILC, MIND etc.)• XLPE• EPR• PVC• 3 Core and Single Core

• Cables• Mixed cables

with transition joints

Right: 33kV PILC 3-core cable

Left: 11kV PILC ‘Belted; cable

33kV XLPE 3-core cable

33kV XLPE single-core armoured cable

PD Detection – Energies for Different Points in Cable System

Corona at metal contactsElectrical charge RF Electromagnetic radiationAcoustic UltravioletOzone

Discharges on insulator surfaceElectrical charge RF Electromagnetic radiationAcoustic UltravioletOzone

Partial discharge in termination insulation systemElectrical charge RF Electromagnetic radiation (local)Acoustic (local)

Partial discharge in cable insulation or jointElectrical charge RF Electromagnetic radiation (local)Acoustic (local)

PD Damage – Incorrect Fitting of Stress Cone

Tracking on 115kV Termination (before failure)

Failed 115kV Termination (same

type)

PD Damage to Cables – Water Tree Conversion to Electrical Tree in XLPE Cable

•Water trees turn into electrical trees immediately before failure

•PD testing will not detect early stages of water tree growth

Bow-tie water trees

Electrical trees

On-line PD Spot-Testing

• Quick to perform

• Overview of PD activity

• MV and HV Cables, all insulation types

• Advanced diagnostic techniques for noise-reduction

Continuous Monitoring

• Detect load varying and intermittent PD (mostly on PILC)

• Detect incipient faults through changes in pattern

• Temporary and permanent applications

PD Detection Theory

HV Capacitor

• Placed in parallel with cable at termination

• C = 500pF – 1nF (typically)

• C is high impedance to low freq (i.e. 0.1 – 400 Hz) high voltage applied to cable

• C is short circuit to high frequency PD signals (i.e. kHz/MHz range)

• Off-line only

Cable under test

Coupling Capacitor

Measurement Impedance

Sensors

Sensor Attachment Point PD Detection Method

HFCT Power cable earth strap/drain wire or power cable with earth strap/drain wire brought back through sensor.

Current impulses from PD in cables, cable terminations and plant/switchgear cables are terminated into.

TEVMetal-clad plant housing close to vents/seams/gaskets.

Electromagnetic radiation from PD sites in plant that is induced onto the plant metal housing.

Airborne

Acoustic

Over vents in plant housing with line of site to PD source.

Airborne acoustic (ultrasonic) radiation from corona and surface discharges in the plant.

HFCT Sensor Attachment Temporary Outside Cable Box

2

1

HFCT Attachment at Cross-bond Points

Circuit Name

Cable Type

Voltage (kV)

Length (m)

Return Time (μs)

Return Propagation

Speed (m/μs)

Test Date Test method

Hotel Maluri Inn - Taman Maluri - Y Phase

3c XLPE

11 232 2.9 80 27/07/08 OWTS,PD at 1.3U0

All

PD Map of Circuit Hotel Maluri Inn - Taman Maluri - Y Phase

Location (meters)240220200180160140120100806040200

All

Pha

ses

PD

500

450

400

350

300

250

200

150

100

50

0

HFCT

OWTS Supply with inbuilt HV Capacitor Cable under test

OWTS HV Capacitor

HFCT

On-line PD Detection Equipment

PD Detection Equipment and Methodology

Detect cable and local PD

Cable PD• PD signals from power cable components

• HFCT detection

Local PD• PD signals from nearby sources: cable terminations,

switchgear, transformer etc

• HFCT/TEV/Acoustic detection

Wideband Capture of PD signals, BW, 50-400MHz

• Capture all of the PD energy detected by sensor• Capture simultaneously with power cycle and extract

impulsive signals• Separation of PD and noise signals based on wave

shape• Continuous (RF, etc) noises removed with hardware

notch/high pass filtersHardware Filtering/Amplification

Digitiser/DSOPD Data Analysis Software

PD Sensors

Trigger signal

HVPD On-line PD Technology

PD Detection• HVPD-Longshot™ with PDGold© Software

• Wideband PD test system

PD Location• HVPD Longshot™ with PDMap© Software

• Transponder Installed at far cable end

PD Monitoring• HVPD Longshot™ (up to 48 hours)

• HVPD Mini™

On-line PD Signal Measurement

Available Waveform Display

Chan 1 Curs 1 Curs 2

Time (mSec)20191817161514131211109876543210

Cha

n 1

0.015

0.01

0.005

0

-0.005

-0.01

-0.015

Ch 1

Segment Waveform

Time uSec14121086420

Vol

ts (

mV

)

15

10

5

0

-5

-10

-15

Ch 1

Segment Waveform

Time uSec14121086420

Vol

ts (

mV

)

15

10

5

0

-5

-10

-15

Ch 1

Segment Waveform

Time uSec14121086420

Vol

ts (

mV

)

4

2

0

-2

-4

Cable PD Pulse: 16mV, 1160pC Local PD Pulse: 15mV, 24dB Noise Pulse: 6mV

20 ms

15 us

One power cycle of raw data from

HFCT sensor

Normal Distribution of All Events in a Single Power Cycle

4 6 8 10 12 14 16 18 20

1

10

100

Co

un

ts

Peak (mV)

Noise Peak Cable PD Peak

Time (ms)20191817161514131211109876543210

Ch

an

1 (

V)

0.015

0.01

0.005

0

-0.005

-0.01

-0.015

Segment Waveform

Time uSec1086420

Vol

ts (

mV

)

15

10

5

0

-5

-10

-15

Segment Waveform

Time uSec302520151050

Vo

lts (

mV

)

15

10

5

0

-5

-10

-15

1 Power Cycle

Noise and Interferences in On-line PD Measurements

Noise can be problem on-line

• Cables/plant not isolated from network

• Sensors often attached to ground

Noise sources

• Continuous: Radio/TV stations, communications

• Impulsive - Periodic

• Power electronics, e.g. inverter power supplies, UPS• Rotating machine excitation

• Impulsive – Random

• Switching Events• Corona• PD on other cables/plant

Noise Source Summary

switchgear bus

RF broadcast signals

outdoor terminationpower

electronicdevices

PD

from adjacent

feeders

substation earth

overhead lines

switchgear earthswitching

noise/interference

corona/surface

discharge

corona/surface discharge/RF signals/PD in

outdoor equipment

PD

from

transformers

G

PD in switchgear

noise from far end of cable

excitation signals/PD from motor

excitation signals/PD from generator

Noise Discrimination – Hardware Filters

• Preferred over software filtering• Remove interferences in HFCT pass-band• Use in difficult measurement situations

Noise Discrimination – Hardware Filters

13.0m 13.0m 13.0m 13.1m

-0.08

-0.06

-0.04

-0.02

0.00

0.02

0.04

0.06

0.08

CH

3 (V

)

Time (s)

Noisy Data Denoised Data

• Single Frequency RF Noise Reduction

Power Cycle

PD and noise detail

Noise Discrimination – Event Recognition

Segment Waveforms

Time (uSec)140120100806040200

Mai

n se

gmen

t

0.15

0.1

0.05

0

-0.05

-0.1

-0.15

Time (ms)20181614121086420

Vol

tage

(V

)

0.015

0.01

0.005

0

-0.005

-0.01

-0.015

• Power Electronic Switching

Continuous amplitude across power cycle

No phase pattern

Pulses appear as chain 16kHz repetition frequency

PD Test and Location Applications

PD location at RMU

Ch 3Ch 4

Segment Waveform

Time uSec109876543210

Vol

ts (

mV

)

200

150

100

50

0

-50

-100

-150

-200

Ch 3Ch 4

Segment Waveform

Time uSec321

Vol

ts (

mV

)

150

100

50

0

-50

-100

-150

• Signals propagate between cables with little attenuation on two feeder RMUs

• Reduced detection points on network – no need to test every RMU

• Synchronous capture necessary to determine source feeder

Measured PD pulse

Zoom of pulse start

Example Test Configuration for PD Location on Circuit with RMUs

11kV

RMU(Normally closed)

RMU(Normally closed)

Transponder

Section of Cable Undergoing Mapping Test

RMU(Normally open)

HVPD-Longshot

Cable Return Time, L for On-line Measurements

Use from Off-line Tests

Estimate from cable length and average propagation speed for cable type

Measure with on-line TDR (requires impedance change at far end of cable)

PD Site Location,

1001%

L

TPD

Cables with Cross-Bonding

PD Propagation in Cross-bonded Cable Systems

Cross-bond point is seen as change in impedance in cable system

Some of pulse propagates into cross-bond point and onto all cables

Some of pulse is reflected back down source cable

Long-cross bond leads/coaxial cross-bond cables are high impedance to HF PD signals

ZA ZD

ZE

ZFZC

ZB

Cross-bond Propagation Example

Available Waveform Display

Chan 1 Chan 2 Curs 1 Curs 2

Time (uSec)210

Cha

n 1

0.07

0.060.05

0.040.030.02

0.010

-0.01

-0.02-0.03

-0.04-0.05-0.06

-0.07

Cha

n 2

0.035

0.030.025

0.020.0150.01

0.0050

-0.005

-0.01-0.015

-0.02-0.025-0.03

-0.035

50Ω

50Ω

50Ω

50Ω

50Ω

L1

L2

L3

Current Transformers

Cross Bond Point

Inje

ctio

n S

ide

Outgoing S

ide

Pulse Injection Test Set-up

Input signal, L1

Reflection from CB

Pulse on outgoing side of CB, L1

Outgoing scale 50% of Injection scale

Procedure for Testing Cross-bonded Cable System Detection

Sequentially test at terminations and all CB-points

N = 2 + number of joints

Test 1 Test 2 Test 3 Test N

Continuous PD Monitoring

Continuous PD Monitoring Aspects

Detect cyclic changes in activity

• Load varying activity on PILC cables

• Humidity related activity from surface discharges

Detect changes that relate to incipient faults

• Gradual rise

• Sudden rise

• Sudden drop

Carried out on: key circuits, circuits with suspected cyclic PD changes, circuits with high spot-test results

PD and Load Relations

Although PD incepted by voltage, load can have effect

Mostly on PILC cables

Load variations

• Movement of oil/impregnant

• Expansion of conductors

S S M T W T F

Cumlative PD Activity (CT)

Time

30/1

2/20

10

00:0

0

31/1

2/20

10

00:0

0

01/0

1/20

11

00:0

0

02/0

1/20

11

00:0

0

03/0

1/20

11

00:0

0

04/0

1/20

11

00:0

0

05/0

1/20

11

00:0

0

06/0

1/20

11

00:0

0CT

PD

Act

ivity

(P

c/C

ycle

)

10,000

5,000

0S S M T W T F

Examples of Monitoring

• Simultaneous change in PD with monitors on 11kV networkCumlative PD Activity (CT)

Time06

/03/

2010

00

:00

07/0

3/20

10

00:0

0

09/0

3/20

10

00:0

0

11/0

3/20

10

00:0

0

13/0

3/20

10

00:0

0

15/0

3/20

10

00:0

0

17/0

3/20

10

00:0

0

19/0

3/20

10

00:0

0

21/0

3/20

10

00:0

0

23/0

3/20

10

00:0

0CT

PD A

ctiv

ity (P

c/C

ycle

)

10,000

5,000

0

Bemish Road

Cumlative PD Activity (CT)

Time

06/0

3/20

10

00:0

007

/03/

2010

00

:00

09/0

3/20

10

00:0

0

11/0

3/20

10

00:0

0

13/0

3/20

10

00:0

0

15/0

3/20

10

00:0

0

17/0

3/20

10

00:0

0

19/0

3/20

10

00:0

0

21/0

3/20

10

00:0

0

23/0

3/20

10

00:0

0CT

PD A

ctiv

ity (P

c/C

ycle

)

10,000

5,000

0

Lower Richmond

On-line PD Severity Evaluation

PD Level Guidelines

To determine the true severity the PD is posing to the cable, the following should be taken into account

• Cable accessories (joints and terminations) have a higher tolerance to PD than the cable insulation

• Mixed circuits: Paper/PILC has a much higher tolerance to PD than XLPE

PD Level Guidelines

Example of Normalised Distribution of Off-line PD Levels Measured in 33kV Paper and XLPE Cables

Mackinlay, R. & Walton, C.Some advances in PD monitoring for high voltage cablesMV Paper Cables: Asset or Liability?(Digest No. 1998/290), IEE Colloquium on, 1998

Case Study: On-line PDMappingTNB-D PPU – Central Spectrum

May 2006

PDGold Test: Central Spectrum

Cable PD

Phase of Pow er Cycle (deg)360270180900

PD

Mag

nitu

de (

pC) 2,000

1,000

0

-1,000

-2,000

Cable PD Segment Waveform

Time us14121086420

Vol

ts (

mV

)

40

20

0

-20

-40

PDGold Test: Central Spectrum

• Install transponder at Substation B due to lower noise level

Cable PD

Phase of Pow er Cycle (deg)360270180900

PD

Mag

nitu

de (

pC)

2,000

1,000

0

-1,000

-2,000

Cable PD Segment Waveform

Time us14121086420

Vol

ts (

mV

)

20

10

0

-10

-20

Transponder at Central Spectrum

Injection HFCTDetection

HFCT

Trigger Unit

Pulse Generator

Transponder Waveform Measured at PPU

Chan 1 (mV)Cur 1Cur 2Cur 3Cur 5

Waveform data in Time

Time (uSec)9080706050403020100

Vol

tage

(m

V)

15

10

5

0

-5

-10

-15

-20

Direct Pulse

Reflected Pulse

Transponder Injected Pulse

Transponder time delay

PDMap Waveform

• Result: 3x PD Sites found on Cable

All

PD Map of Circuit main substation to central spectrum

Location (meters)4,0003,5003,0002,5002,0001,5001,0005000

All

Pha

ses

PD

2,000

1,800

1,600

1,400

1,200

1,000

800

600

400

200

0

Conclusions

Conclusions

• HFCT sensors allow sensitive measurements to be made at multiple points in the cable system

• On-line PD location can relatively quickly identify weak points for timely remedial action

• Solutions available for complex circuits such as with cross-bonding