global monitoring: the paradigm for asset management in the smart

Report
GLOBAL MONITORING: THE
PARADIGM FOR ASSET
MANAGEMENT IN THE SMART GRID
FRAMEWORK
G. C. Montanari, A. Cavallini
Dip. Ingegneria Elettrica, University of Bologna
viale Risorgimento 2, 40136 Bologna, Italy
[email protected]
[email protected]
M. Tozzi
TechImp HQ Srl
via Toscana 11/c, 40069 Zola Predosa, Italy
[email protected]
GLOBAL MONITORING IN SMART GRID
ENVIROMENTS
• Making MV-HV grids ready to be smart
• Diagnostic quantities
• Global Monitoring approach
• A case study
2
Smart Grids: what do they generally mean?
• Smart grid = smart metering + renewables ( distributed power
generation)
• Low, sometimes medium, voltage
• Only this, indeed?
3
Smart Grid the E.U. vision
• E.U. vision: by far the most complete vision
of SG
4
Are Electrical apparatus/assets Smart?
Smart Grids (EU)
• Reliability and quality
• Innovation and
competitiveness
• Nature and wildlife
preservation
• Low prices and
efficiency
Electrical
Apparatus/Assets
• Monitoring Diagnostics
• Remote Control of
Assets and CBM
• Reliable (smart)
Insulation Systems
• Recyclable and New
Materials (nano)
5
Smart Grid the E.U. vision
• Let us review some basic concepts of power
system asset management to see how they
are related intimately to electrical
appartus/insulation system monitoring
• The goal is to demonstrate that advanced
monitoring tools can become the key for
smart grid management
6
Asset management time frames
• Long-term AM (LTAM, yearly and beyond): Strategic
planning, decide which assets need to be replaced (using
which technologies) and grid expansion.
– Improved Materials & Manufacturing Techs
– Permanent diagnostics, operating conditions
• Mid-term AM (MTAM, from monthly to yearly):
Maintenance management: optimal maintenance strategy
and optimal outage plan.
– Bulk diagnostic techniques (tand, pol/depol, dielectric
spectroscopy, DGA) and local diag. techniques (PD) (on line)
7
Asset management time frames
• Short-term AM (STAM, from real-time to
months/year): Operational management: secure
and reliable operation of the system, system
monitoring and control, fault restoration.
– Local degradation diagnostic techniques
(PD, hot spots; on line/off line), recording
operational conditions.
8
The vision set forth: our contribution
to smarter grids
• With improved monitoring tools, the apparatus will be a
source of information as, for instance:
– Apparatus condition
– Apparatus optimum operation
– Synchro-phasors
– Power Quality at the apparatus node
• Advanced communication tools will enable this info to be
available to a broad number of operators for Operational
management, Reliability evaluation, Security assessment,
Stability analysis, Load flow calculations, Power quality
assessment
• This is smartening the grid
9
GLOBAL MONITORING IN SMART GRID
ENVIROMENTS
• Making MV-HV grids ready to be smart
• Diagnostic quantities
• Global Monitoring approach
• A case study
10
Diagnostic properties (Insulation Systems)
Two families of diagnostic quantities:
-Local quantities, related to defects (e.g. Partial Discharges, Hot spots)
-Global quantities, related to bulk degradation
EFFECTIVE DIAGNOSTIC TOOL (for localized defects)
Which properties to choose?
1) Those bringing the fastest ageing rate
2) Those providing info on the global state of an electrical apparatus
decision on the most appropriate maintenance action
11
PD as an effective diagnostic tool
Partial discharges are at the same time CAUSE and CONSEQUENCE of
insulation system degradation and provide often the fastest ageing
mechanism (on organic materials)
EFFECTIVE DIAGNOSTIC TOOL (for localized defects)
Partial discharges occur
when there are defects
within the insulation system
System degradation
increases under Partial
Discharges action
PD = CONSEQUENCE
PD = CAUSE
12
PD definition (IEC 60270)
• Partial discharge (PD): localized electrical discharge that
only partially bridges the insulation between conductors
and which can or can not occur adjacent to a conductor
• PD normally develop in air gaps or on insulation surfaces
13
Insulation Degradation
Partial Discharge
activity
Insulation material
erosion
Partial Discharge
Complete Discharge:
Breakdown
HV electrode
Electrical tree in HV
cable joint
insulation
Epoxy
slab
LV electrode
Formation of
treeing
channels in a
point-to-plane
specimen
14
Diagnostic quantities and CBM
Effective maintenance:
only at the right moment
DGA
Global
Diagnostics
Tandelta
Insulation condition
Vibrations
Partial Discharge Analysis
Insulation ageing
15
Sensors
• For on-line monitoring, sensors are a key issue
for: -) reliability -) sensitivity -) effectiveness of
measurements and cost
• it is possible to design appropriate sensors for
each apparatus and diagnostic quantity
• Regarding PD, it is possible to design the
detector in order to use just one detector for
all sensors (bandwidth).
16
• HF sensors
•Capacitive sensors
•Inductive sensors
•VHF sensors
•UHF sensors
•Acoustic sensors
Best Technical:
ONE detector FOR ALL
Assets.
The detector should be
able to acquire PD data
from all different
sensors
17
Innovative approach to PD diagnosis:
Separation, Identification and Diagnosis (SID)
PD inference is the prerequisite for correct
diagnosis
Separation
Identification
Diagnosis
S
I
D
• Noise rejection
• Potential defect
..harmfulness
• Source separation
..(one source at a ..
.. time)
• Risk assessment
• Maintenance program
• Life extension (trend of the
weakest spots,.time to end point)
18
Pulses coming form different points have
different T/F characteristics
Two PD pulses from sources at different distances from detection point
(broadband detection chain)
Pulse Fre que ncy Spe ct rum
A
0.020
5.5E-4
Pulses coming from
close to the
detection point:
5.0E-4
0.015
4.5E-4
0.010
4.0E-4
0.005
3.5E-4
3.0E-4
0.000
2.5E-4
2.0E-4
-0.005
1.5E-4
-0.010
1.0E-4
-0.015
5.0E-5
0.0E+0
-0.020
0.0 100.0n
300.0n
500.0n
700.0n
0.0
900.0n 1.0u
5.0
10.0
15.0
20.0 25.0 30.0
Fre que ncy [M Hz]
35.0
40.0
45.0 49.0
Higher frequency
content
Pulse Fre que ncy Spe ct rum
B
0.010
5.5E-4
0.008
5.0E-4
0.006
4.5E-4
Pulses coming far
from the detection
point:
4.0E-4
0.004
3.5E-4
0.002
3.0E-4
0.000
2.5E-4
-0.002
2.0E-4
-0.004
1.5E-4
-0.006
1.0E-4
5.0E-5
-0.008
0.0E+0
-0.010
0.0 100.0n
300.0n
500.0n
700.0n
900.0n 1.0u
0.0
5.0
10.0
15.0
20.0 25.0 30.0
Fre que ncy [M Hz]
35.0
40.0
45.0 49.0
Lower frequency
content (due to
attenuation)
19
Categorization induced by TF mapping
20
The concept of PD pattern
0.015
0.010
30
20
240
220
200
180
160
140
120
100
80
60
40
20
10
200180
160
M ag
-0.010
ha
2 4 02 2 0
-0.005
140
120
n it u d
P
0.000
se
ch
an
ne
0.005
l
0.020
40
f o ccu rre n ce
F re qu e n cy o
During AC PD
activity hundreds
of pulses per
second occur
having different
amplitude and
phase !
One PD event is a
pulse having a
large frequency
content (from the
MHz to the GHz
range)
100
80
e ch
ann
60
40
20
el
-0.015
0.0 100.0n
300.0n
500.0n
700.0n
900.0n 1.0u
Amplitude
Phase
•The PD pattern represents the density of
discharges in the phase/magnitude plane.
•It is a 3-D histogram represented through
color maps
Magnitude
-0.020
Phase
21
SID Separation, Identification and Diagnosis
Separation
of pulse
features
10
Feature #3
5
0
-5
-10
10
10
5
5
0
0
-5
ENTIRE ACQUISITION
NOISE
Feature #2
-5
-10
-10
Feature #1
INTERNAL PD
SEPARATION
MAP
22
Identification is the key
• Different defects lead to different degradation
rate in the insulation system
• To assess insulation condition it is necessary to
investigate each PD source separately
• Basing diagnostics on a general level of PD,
without any identification, may be misleading,
since just the predominant phenomena will be
taken into consideration…and the biggest one
may not be the most dangerous
23
Example in rotating machine:
Bar to Bar (B2B) PD can be significantly larger than Slot
PD, but degradation rate associated to the slots may be
faster
24
How to identify different PD types?
PD sources of the same nature give rise to similar PRPD Patterns.
-Internal PD
HV
- Surface PD
HV
- Corona PD
HV
25
Automatic identification (1): Fuzzy logic at work
Statistical marker
evaluation
Fuzzy inference
engine
Mixed stress-grading PD and
microvoid activity in mediumvoltage motor.
Fuzzy ident: 87% surf, 13% internal
Next comes the same phenomenon
recorded in a much more degraded
machine
26
Automatic identification (2)
(V) 3.00
Statistical marker
evaluation
2.00
1.00
0.00
-1.00
Fuzzy inference
engine
-2.00
-3.00
0
45
90
135
180
225
Phase (deg)
270
315
360
Stress-grading PD clearly
predominant in medium-voltage
motor.
This time no fuzzyness in
identification (100% surface, 0%
internal)
27
Automatic identification: 3rd ID level based
on fuzzy logic
28
GLOBAL MONITORING IN SMART GRID
ENVIROMENTS
• Making MV-HV grids ready to be smart
• Diagnostic quantities
• Global Monitoring approach
• A case study
29
Smart Grid Global Monitoring System
Structure
• A system that can correlate
several diagnostic and
operational quantities
to achieve better
condition evaluation
• Endowed with advanced connectivity and data
processing (noise rejection, data compression,
innovative detectors and sensors) tools
• Providing, in real time to SCADA centers, a
valuable estimate of apparatus failure likelihood
30
Smart Grid Global Monitoring System
• Asset Condition Estimator (ACE).
– STAM, MTAM: quantities associated with localized
defects where stress concentration often takes place:
fastest mechanism for insulation failure
• PD
• Hot spots
– MTAM, LTAM: bulk aging, i.e., a generalized loss of
electrical, mechanical and thermal properties of the
system, besides PD and hot spots
•
•
•
•
•
gas levels in oil
Tand
Conduction current
PD, hot spot (e.g. Real Time Thermal Rating)
Vibration signals.
31
Smart Grid Global Monitoring System
• Operating Point Recorder (OPR): log data regarding
– Bus voltages and load currents (Synchro-phasors)
– Readings from temperature probes and/or fiber optic
monitoring systems
– Environmental quantities.
• Note 1: random power flow fluctuations due to
renewable sources: impact on insulation systems???
• Lots of operational data needed to correlate these
fluctuations with failures.
• Note 2: what about power electronics repetitive
pulses? And sporadic voltage transients? See next
32
Smart Grid Global Monitoring System
• Power Quality Monitor (PQM).
– Harmonics: promote hot spot overheating,
mechanical stress and enhance peak voltage
levels.
• Intolerable when series or parallel resonances take
place.
– Surge voltages and voltage dips can threat interturn insulation of transformers and motors
– External short circuits could affect the mechanical
stability of transformer windings.
33
Smart Grid Global Monitoring System
• Communication module
(COM).
– Software, database
management and
communication tools that
allow enhanced data
exchange between SGGMS
and supervisory control
and data acquisition
(SCADA) centers
Remote User
Internet
Web
Server
Database
application
Central Unit
Web
service
Diagnostic
application
Thirdpart
systems
Local
SCADA
Network
Data
Downloader
Acquisition
Unit #1
Acquisition
Unit #2
34
Communication Module
•
•
•
When a permanent diagnostic monitoring system is installed, it could be
useful to save all the data in a central server.
Especially if more than one EUT are monitored, all the data can be saved in
the server and collected in the DATABASE
The DATABASE represents the history of the monitored system
EUT 1: generator
EUT 2: transformer
EUT 3: HV cable
PD/DP sensor
PD/DP sensor
PD/DP sensor
ACQUISITION BOX
ACQUISITION BOX
ACQUISITION BOX
SERVER - DATABASE
35
The on line monitoring system is made up essentially of the following
components:
Sensors (one for each joint/terminal);
Diagnostic units: (PD, Tandelta, DGA) one or more for each asset;
Supervision & Control System (one for each complete circuit);
Ethernet links between the detection units and the Supervision & Control
System.
36
• One can open synoptic views to immediately and
easily understand if, where and when any problem
occurred during the monitoring session
37
• Simple synoptic visualization modes are available for
any electrical apparatus, e.g. generators,
transformers, cables and GIS
38
• One can see the trending associated to each sensor in each
phase of each equipment.
• One can see the recorded data and patterns… one can play
e.g. with the T-F map and set up PD alarms properly.
39
• Folders containing the stored data can be opened
when alarms are raised and patterns associated with
the PD activities (or what else among Diagnostic
Properties DP) can be seen immediately
40
• Advantages:
•
•
•
•
•
Centralized data storage for resource optimization;
Data comparison among electrical apparatuses of the same
family or insulation technologies or within a single electrical
apparatus (e.g. among different phases) or under different
conditions (load, time, maintenance interventions);
Data trending allows the harmfulness level to be evaluated and
threshold criteria to be fixed/modified;
Combined analysis of quantities other than PD (e.g. humidity,
temperature, load, voltage transients, DGA);
Capability of customizing the alarms/warning decision trees
depending on asset manager evaluation.
41
GLOBAL MONITORING IN SMART GRID
ENVIROMENTS
• Making MV-HV grids ready to be smart
• Diagnostic quantities
• Global Monitoring approach
• A case study
42
Background
•
A 250 MVA autotransformer experienced immediately after installation
a significant increase of Hydrogen
•
According to the IEC and IEEE specs, the level and the trend of H2 were
critical after only few months. After one year the H2 level exceeded
1000 ppm
–
Possible PD according to IEC60599 based on Duval Triangle
–
Condition 2 according to IEEE C57.104: Exercise caution- Analyze
for individual gases-Determine load dependence
•
BUT:
–
Is this gas increase actually due to thermal or electrical problem?
–
Is the PD activity, if present, harmful or not?
–
Which type of PD and where is this located?
–
Which is the degradation rate?
–
Which is the best action to be taken reducing costs and increasing
reliability?
43
Actions
1. OIL TREATMENT
2. MONITOR PD+GAS+Bushing Tandelta before oil
treatment and during Spring/Summer (most critical
period)
SCOPE OF THE MONITORING SYSTEM INSTALLATION:
- MONITOR THE TRANSFORMER DURING A CRITICAL
PERIOD TO AVOID UNEXPECTED FAILURES
- ASSESS THE PD HARMFULNESS
- GIVE A PROBABILITY OF FAILURE WITHIN THE
GUARANTEE TIME
44
SGGMS main characteristics
• PD
– UWB detector (16kHz-35 MHz)
– 6 sensors (Tap Adapters)
– Time -Frequency Map Separation algorithm
• DGA
– 2 Gas (H2,CO) + Moisture + Temperature
– Membrane technology/electrochemical
sensors
• Bushing Tandelta/Capacitance
– Leakage Current
– Dissipation factor
– Insulation Resistance
45
GLOBAL MONITORING LAYOUT
1: Acquisition
Box
2: Tap Adapter
for both PD
and TanD
acquisition
2
3: TD Sensor
3
1
4
4: DGA
46
Results before oil treatment
• TWO PD phenomena were detected
on-line:
– A sporadic activity due to small gas
bubbles in the oil. This activity was
intermittent.
– A smaller, but persistent, activity
detected in all the HV phases,
identified as mixed internal/surface
PD.
• H2 level increase about 5 ppm/day
Bubble PD
Surface/Internal PD
47
Main results after oil treatment
• The first activity, due to the bubbles, disappeared after the oil
treatment.
• Second activity was still there, in all three phases at HV side
(230 kV)
Phase 4
Phase 8
Phase 12
48
Nw > 80
PD Trend
Phase 4
Qmax>500 mV
Qmax>250 mV
PD Trend
Phase 8
Nw > 100
Nw > 80
PD Trend
Phase 12
Qmax>300 mV
49
Evaluation of trending
• Necessary to give COMBINED alarms and assess insulation
condition
DGA
DGA
PD
PD
• Meaningful trending! Not influenced by external disturbances
or other PD!
50
Qmax Trend of Phase 4 without separation of Corona and Bubble
from PD at interfaces
Corona
Bubbles
PD
Bubbles
Possible False Alarm
PD
PD + Corona
Just PD
51
TF FILTERING: Smart Alarm setting
TF
FILTERING:
Trending
evaluated
only in this
region of the
map!!
CORONA + PD
CORONA
PD
52
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/14/2010
4/14/2010
4/14/2010
Qmax 95%
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/15/2010
4/14/2010
4/14/2010
4/14/2010
4/14/2010
0.6
4/14/2010
1.2
4/14/2010
4/14/2010
4/14/2010
4/14/2010
4/14/2010
0.8
4/14/2010
0.8
4/14/2010
4/14/2010
4/14/2010
4/14/2010
Qmax 95%
Possible False Alarm due to External Corona
1
Qmax 95%
Threshold Alarm
Level
0.4
Trending
without TF
filtering
0.2
0
1.2
1
Qmax 95%
0.6
Threshold Alarm Level
0.4
0.2
Trending
after having
TF filtered
external
corona
0
53
FACTS after 6 months monitoring
• No significant changes in bushing tandelta values were noted over the
monitoring period (6 months)
• Polarity of detected PD indicated that PD source was not located inside
the bushings.
• The H2 gas levels increased during the monitoring period with average
rate of 5 ppm/day. No significant changes in the rate was noted.
CONSTANT RATE.
• PD activities were detected continuously for 6 months , demonstrating
that gas increase was due to PD
• PRPD pattern investigations suggested that
– There are three defects: one each phase
– PD activity was generated by a constructional defect within the connection
between the bushing and the winding leads. Location of the source was
confirmed also by additional acoustic measurements
54
Monitoring results
EXPECTED RESULTS
OBTAINED RESULTS
MONITOR THE TRANSFORMER DURING A
CRITICAL PERIOD TO AVOID UNEXPECTED
FAILURES
• TRANSFORMER WAS MONITORED
CONTINUOUSLY AND NO CRITICAL CHANGES
IN PD TREND WAS NOTED.
• UNEXPECTED FAILURES DID NOT OCCUR.
• NO FALSE ALARMS WERE GENERATED
ASSESS PD HARMFULNESS
PD ARE NOT YET HARMFUL SINCE PD
TRENDING IS CONSTANT (amplitude alone is
not the only parameter)
GIVE A PROBABILITY OF FAILURE WITHIN THE LOW PROBABILITY OF FAILURE IN ONE YEAR
GUARANTEE TIME (1 YEAR)
IF TRANSFORMER OPERATED AT THE SAME
STRESS/CONDITIONS, CONSIDERING BOTH
THE MONITORING RESULTS AND
TRANSOFORMER HISTORY. BTW, PD INVOLVE
PAPER LAYERS AND CAN BECOME CRITICAL!
SUGGESTED ACTIONS: PLAN VISUAL INSPECTION AND MAINTENANCE ACTION IN THE
MOST CONVENIENT PERIOD (AUTUMN). MEANWHILE, MONITOR THE TRANSFOMER
UNTIL MAINTENANCE IS TAKEN
55
CONCLUSIONS
56
• Advantages of SGGMS:
– Trend evaluation – failure risk assessment
– Action (maintenance) planning
– Problems identified
– Proper and clever alerts activated
– Maintenance planning feasibility -> increase reliability with
cost reduction
– Advanced connectivity
– Proper management of operation (load, availability)
• Smart Grid operations will profit of knowledge of the
availability, reliability and operation capability of each
electrical component of the grid.
57
The vision set forth
1. Condition monitoring tools massively
integrated in electrical apparatuses
2. Information extracted from the monitoring
data stream in a smart way using, for
instance, artificial intelligence techniques.
3. Bidirection information flow.
58

similar documents