Incipient Faults in Distribution Systems: Experiences, Use Cases,

and Case Studies

Mirrasoul J. Mousavi, ABB Corporate Research

IEEE PES GM Invited Panel

July 18, 2016 , Boston, MA

1

Agenda

• Introduction

• Incipient fault characteristics with field-recorded illustrations

• Detecting re-occurring incipient faults

• Conclusions

2

© ABB Group

August 5, 2016 | Slide 3

Research Roadmap: Intelligent Solutions

Incipient faults has been an R&D&D topic for many years

• 2013-2016 (condition-based control applications)

• 2011-2013– Time-tagged sensor data and derived intelligence:

integration into control center

– Use of sensor data for real-time feeder performance, asset management, grid analytics, and modeling funded in part by DOE

• 2010-2011– Focus on multi-IED feeder and station intelligence

applications for substation servers

• 2008-2009– Focus on high-end feeder intelligence applications

embeddable in IEDs or substation servers in-line with the advent of Smart Grids.

• 2007&prior– Focused primarily on low-cost modules for analysis of

non-operational data e.g. disturbance data for providing feeder intelligence (e.g. breakers, batteries, lines/cables, IEDs, and feeder components)

Feeder

Intelligence

Substation

Intelligence

Control Center Intelligence

Ris

k/Im

pact

Fie

ld

Contro

llers

IED

s

Substa

tion

Serve

rs

Netw

ork

Manager

Info

rmatio

n/D

ata

Ratio

Field

Devices

Pyramid of Intelligence

• Taps into available data from IEDs (32 spc min. res.)

• No new sensors/CTs/PTs required

• Real-time and automated analysis of disturbance

records from field IEDs

• Detection and Control Center notification within

seconds with the end-to-end process fully

automated• End-User value

• Knowledge: situational awareness (WWWWH)

• Enables incipient fault/abnormality detection

• Faster outage detection and notification e.g. fuse-cleared

faults

• Reduces OK-on-arrival truck rolls

• Confirmation of switching events, power on, and power off

• Field verified and validated algorithms • Available for pilot deployment in IEDs, substation

automation controllers, and integration with control center

Data Acquisition and Processing Flow

05 August 2016

| Slide 4

© ABB Group

August 5, 2016 | Slide 5

• Feeder Fault (Type I)

Case # 955 in MDB DFEVAS OMS

Predicted Actual

Time of Event 12/13/2008, 7:44 AM 12/13/08 8:02 AM

Substation XYZ XYZ

Feeder Number 1234 1234

Phase B B

Event Classification Short-duration

Feeder Fault (High) Cable Fault

Infrastructure UG (80%) UG

Equipment Category N/A Cable

Clearing Device Fuse Fuse

Clearing Device Size [10A,65A]

[40A,0.981] 40A

Cause of Failure N/A Cable Failure

Location PMZ, Segment X Primary Feeder

Time of Restoration N/A 9:45AM

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

-2000

0

2000

Recorded Waveforms in [TOLL1765-LS2] DPU-LS2-081213-074419061.cfg

A

a

b

c

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

-10

0

10

kV

0 0.1 0.2 0.3 0.4 0.50

500

1000

Current Phasor Analysis

A

a

b

c

0 0.1 0.2 0.3 0.4 0.50

5

10

Voltage Phasor AnalysiskV

Filename: DPU-LS2-081213-074419061.cfgTrigger Date: 12/13/08Trigger Time: 07:44:19.610000Faulted Phase: bPeak Phasor Fault Current (A RMS): 1232Peak Inst. Fault Current (A): 2485Fault Duration (cycles): 0.781Load Change (kW): -41

Prefault Loading Phase: A; B; C: Current (A): 366; 296; 328 Power (kW): 2831; 2250; 2539 Power Factor: 0.967; 0.951; 0.97 Voltage (kV RMS): 8Postfault Loading Phase: A; B; C: Current (A): 364; 290; 326 Power (kW): 2822; 2209; 2528 Power Factor: 0.967; 0.95; 0.97 Voltage (kV RMS): 8

Operator Message:

•A Cable Fault event on B phase has just been detected on Primary feeder 1234 out of XYZ substation on Dec 13, at 7:44AM that could have been cleared by a 40A fuse (Rel. probability: High).

IED

Adjacent Zone

Primary ZoneUpstream Zone

Comprehensive Analysis, Detection, and AssessmentWhat/When/Where/Why/How (Available in Substation Controller)

Illustrative Case #1Incipient fault lasting 9+ months

Initial Incipient Fault September 11, 2007

02:42 PM

• Ifault = 422 A RMS

• No outages or customer calls

139 Incipient Faults thereafter

• Ifault = 100’s – 1000’s A RMS

• Multiple faults per day

Permanent Fault June 14, 2008

12:19 AM

• Ifault = 2626 A RMS

• Customer call

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

-1000

0

1000

Recorded Waveforms

A

a

b

c

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

-10

0

10

kV

0 0.1 0.2 0.3 0.4 0.50

500

Current Phasor Analysis

A

a

b

c

0 0.1 0.2 0.3 0.4 0.50

5

10

Voltage Phasor Analysis

kV

Filename: DPU LS1 070911 144232023.cfgTrigger Date: 09/11/07Trigger Time: 14:42:32.230000Faulted Phase: aPeak Phasor Fault Current (A RMS): 422.3Peak Inst. Fault Current (A): 1287.5Fault Duration (cycles): 0.219Load Change (kW): 7

Prefault Loading Phase: A; B; C: Current (A): 124; 119; 128 Power (kW): 962; 938; 1012 Power Factor: 0.986; 0.983; 0.989 Voltage (kV RMS): 7.96Postfault Loading Phase: A; B; C: Current (A): 125; 118; 130 Power (kW): 969; 933; 1025 Power Factor: 0.984; 0.981; 0.989 Voltage (kV RMS): 7.96

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

-5000

0

5000

Recorded Waveforms in [BUCK1270-LS1] DPU-LS1-080614-001906046.cfg

A

a

b

c

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

-10

0

10

kV

0 0.1 0.2 0.3 0.4 0.50

2000

Current Phasor Analysis

A

a

b

c

0 0.1 0.2 0.3 0.4 0.50

5

10

Voltage Phasor Analysis

kV

Filename: DPU-LS1-080614-001906046.cfgTrigger Date: 06/14/08Trigger Time: 00:19:06.460000Faulted Phase: aPeak Phasor Fault Current (A RMS): 2626Peak Inst. Fault Current (A): 4954.7Fault Duration (cycles): 0.875Load Change (kW): 52

Prefault Loading Phase: A; B; C: Current (A): 77; 90; 100 Power (kW): 591; 704; 774 Power Factor: 0.981; 0.981; 0.974 Voltage (kV RMS): 7.94Postfault Loading Phase: A; B; C: Current (A): 92; 89; 102 Power (kW): 643; 695; 788 Power Factor: 0.892; 0.978; 0.973 Voltage (kV RMS): 7.94

Illustrative Case #2Incipient fault lasting 3 hours

•1108A RMS

•No outages or customer calls

Initial “C” Phase Incipient Fault March 8 at 6:05:55 PM

•1600 – 2438A RMS

•Generally less than ½ cycle

6 Single blips thereafter

•2776-4274A RMS

•Over a few non-contiguous cycles

9 Multiple blips thereafter

•4077A RMS followed by a customer call

Permanent fault captured

March 8 at 9:07:53 PM

Illustrative Case #3 Primary zone: Evolving fault

• A phase-A fault evolves into a phase-B fault• No OMS data!

Illustrative Case #4Incipient Fault leads to a Permanent O/H Fault

• 2564A RMS

• No outages reported around that time

• Cause was tree inside maintenance Corridor

• Feeds traffic and street lighting

“A” Phase Fault on Jan 31 @ 12:04:59 AM

Outage registered 7:41AM

Opportunity to fix the problem before an outage call

Current waveforms

Voltage waveforms

© ABB Inc. August 5, 2016 | Slide 10

Avoiding False Positives

• Example of proper cable fault detection

• Examples of ‘false positives’

– Inrush

A B

Incipient Fault Characteristics Solidly-grounded Circuits

Common Characteristics: magnitude, duration, inception angle, harmonics, long and short term trends (not exhaustive)• Do not draw sufficient current for O/C protection• Do not last sufficient enough for coordinated O/C protection• Fault duration on an average less than half a cycle.• Peak fault current could be multiple times the RMS load current Harmonic analysis not necessarily required.

• Increasing trend in the normalized instantaneous peak fault current towards eventual fault time

• Increasing trend in 0th, 2nd, and 3rd harmonic. • Fault inception angle close to 0 degrees w.r.t. voltage peak• Intermittent on/off behavior • Could persist for as little as hours

Recording begins on August 19

• First case of incipient fault recorded on September 11

• Current Peak- 1287A Duration-0.22 cycles

• No Records on Utility Outage System

• A catastrophic failure occurred on June 14

• 65 A Fuse blown

• Peak Current – 5000A

After 139 repetitive incipient faults over 9+ months!

Illustrative Example: Splice failure

View of the catastrophic failure

Close-up view of erosion of

splice bodyThe center splice body is a non-faulted

splice from the same location on different

phase, this splice is in good condition.

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

0

2000

4000

A

a

b

c

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5

-10

0

10

kV

0 0.1 0.2 0.3 0.4 0.50

2000

4000Current Phasor Analysis

A

a

b

c

0 0.1 0.2 0.3 0.4 0.5

6

8

Voltage Phasor Analysis

kV

View of the catastrophic failure

09/11/07 10/31/07 01/31/08 03/22/08 06/13/081000

1500

2000

2500

3000

Peak I

nsta

nta

neous C

urr

ent

Magnitude (

A)

Arrival Time

Global Max=

Global Min=

Global Avg=

2798A

1087A

1720A

Note: This plot represents the absolute value of peak.

09/11/07 10/31/07 01/31/08 03/22/08 06/13/08-3000

-2000

-1000

0

1000

2000

3000

Peak I

nsta

nta

neous C

urr

ent

(A)

Arrival Time

Global Max=2798A

Global Min=-2781A

No. of Positive peaks=63

No. of Negative peaks=83

Time Domain Analysis

Peak FAULT CURRENT by Polarity

Polarity not a significant determining factor for fault initiation.

Peak Abs. FAULT CURRENT

09/11/07 10/31/07 01/31/08 03/22/08 06/13/085

10

15

20

25

30

35

40

Norm

. P

eak I

nst.

Curr

ent

Magnitude

Arrival Time

Note: The absolute value of peak instantaneous fault current has been

normalized with respect to the load rms current.

09/11/07 10/31/07 01/31/08 03/22/08 06/13/0850

100

150

200

250

300

Load R

MS

Curr

ent

(A)

Arrival Time

Global Max=

Global Min=

Global Avg=

250A

55A

124A

Note: First two cycles have been used to calculate the nominal load current.

RMS LOAD CURRENT

• Peak fault current is 5-6times the RMS load current

Time Domain Analysis

• See progressive trend towards failure• repeated faults compromise insulation integrity

• Normalized instantaneous peak fault current has an increasing trend.

09/11/07 10/31/07 01/31/08 03/22/08 06/13/080.25

0.3

0.35

0.4

0.45

0.5

Dura

tion (

cycle

s)

Arrival Time

09/11/07 10/31/07 01/31/08 03/22/08 06/13/08-15

-10

-5

0

5

10

Phase D

iffe

rence (

degre

e)

Arrival Time

Note: Phase refers to the angle of fault initiation w.r.t the nearest voltage peak.

• Duration lies

between 0.25-0.47 cycles- notpicked up by conventional relayalgorithms.

INCEPTION ANGLE w.r.t.voltage peak

DURATION

Phase difference close to 0, averages around -1.16 degrees

Time Domain Analysis

09/11/07 10/31/07 01/31/08 03/22/08 06/13/080

10

20

30

40

50

60

70

Inte

r-arr

ival T

ime

Arrival Time

• Inter-arrival Time – demonstrates intermittent on/off behavior

INTER-ARRIVAL TIME

Time Domain Analysis

09/11/07 10/31/07 01/31/08 03/22/08 06/13/080

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Norm

aliz

ed C

urr

ent

Harm

onic

s

Arrival Time

I0th

I2nd

I4th

I6th

I8th

09/11/07 10/31/07 01/31/08 03/22/08 06/13/080

0.2

0.4

0.6

0.8

1

Norm

aliz

ed C

urr

ent

harm

onic

s

Arrival Time

I1th

I3nd

I5th

I7th

I9th

Note: Harmonic Currents have been

normalized with respect to the

fundamental current over the fault cycle.• Even harmonics generally appear with higher magnitudes than that of odd harmonics –sub-cycle duration.

Frequency Domain Analysis

• The 2nd Harmonic isdominant.

ODD HARMONICS

EVEN HARMONICS

09/11/07 10/31/07 01/31/08 03/22/08 06/13/080

5

10

No

rma

lize

d C

urr

en

t H

arm

on

ics

DC

09/11/07 10/31/07 01/31/08 03/22/08 06/13/080

10

20

Fundamental

09/11/07 10/31/07 01/31/08 03/22/08 06/13/080

5

10

15

2nd

09/11/07 10/31/07 01/31/08 03/22/08 06/13/080

5

10

Arrival Time

3rd

• DC, Fundamental, 2nd and 3rd harmonic normalized currents generally have a positive correlation with the failure.

• They may be used as symptom parameters, although implementation details would vary.

•Fundamental suffices in most cases!

Frequency Domain Analysis

• Available feature in feeder protection IEDs

• Detecting an individual instance of an incipient fault

• Applies to both O/H, U/G, and mixed feeders with incipient, fuse-cleared, or self-cleared faults

• Analog inputs

– Phase currents – Ia, Ib, Ic

• Logical inputs

– Block – on/off control of module

• Internal logic

– Per-phase CFD counters - sc_a, sc_b, sc_c

• Logical outputs

– Cable fault detection – SCFDDetect

– Three-phase cable fault event detected –EventDetect_3ph

– Expanding number of logical outputs

– Map to trigger DFR

SCFDAnalysis (SRC

Core)

SCFDDecision (SRC

Core)

sc_

a

sc_

b

sc_

c

Ia Ib Ic

SC

FD

De

tect

Eve

ntD

ete

ct_

3p

h

block

ph_pu

cycleMultipler

Adaptive_ph_pu

SCFD

Incipient Fault Detection in IEDs

Analysis Block

Decision Block

© ABB Inc. August 5, 2016 | Slide 22

Benefits

• Primary value of incipient fault detection is knowledge, i.e., dispatchers will know what they didn’t previously know when a feeder fault occurs that is either self-clearing or are cleared by non-communicative device, e.g., reclosers or switches, or unintelligent device, e.g., fuse

• Algorithm is able to detect incipient and permanent faults on overhead and underground feeders that are solidly grounded (ANSI type)

• High ease of use factor with feature available in feeder protection and control relay eliminating need for extra or unique instrument transformers (CT, VT) or system sensors

Detecting the Long-term Trend

• Incipient faults do not cause outage until they lead to permanent faults

• A series of repetitive occurrences is a major sign of an impending failure

• Detecting individual incidents of an incipient activity is a prerequisite step for alarming (necessary but not sufficient)

• Just-in-time alarming to avoid or manage an unplanned outage requires active monitoring over the activity period

• Could be hours, days, or weeks depending upon many stress factors including thermal and electrical

24

Detecting The Long-term TrendHuman-in-the-loop Approach

• Keep a daily count of detected incipient activities

• Initiate mitigation plans by intuition after a certain number of occurrences register in a given period

• Pros– Simple approach

– works for one-off situations

– Lower deployment barrier

• Cons– Not scalable

– Not sustainable due to operator fatigue

– Prone to operator shift changes

– Has to be standardized across the system

25

• Effective automated

techniques are required

for just-in-time detection

and alarming.

• Alarming should be

based on risk tolerances

(trade-off between

missed detection and

early warning)

Gained Experience-Other

• Be mindful of benefits misalignment if Operations are siloed from

Engineering/Standards. Significant value for this kind of technology is realized at

the company level.

• Over 90% of incipient and permanent faults occurred on laterals

• Detection and location is harder on laterals

• Do not cause breaker trips and SCADA alarms

• Integration with DMS/Control Center is required to make operational impact.

• Sub-cycle and incipient fault location remain an industry challenge although some

methods recently have been reported!

• Need to deal with feeder modeling inaccuracies for detection and localization

• Bad connectivity data

• Incorrect phasing

• Missing information (conductor length, size, material)

• As-built vs. as-operated models

Acknowledgements and Disclaimer

• US Department of EnergyWe gratefully acknowledge the financial support of the US Department of Energy under Award No. grant DE-OE0000547. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assume any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

• Xcel Energy/Next Generation

• ABB EPMV/USCRC

For more information, contact the author at mirrasoul.j.mousavi@us.abb.com