-
A Multiparameter and Web Based Modular On-line Monitoring System
for High VoltageMotors and Generators
R. Haslimeier, B. FruthPD Tech Power Engineering AG
Seestrasse 14b, 5432 Neuenhof, SwitzerlandE-Mail:
[email protected]
Abstract: An integrated monitoring system for highvoltage
rotating machines is presented, which measuresparameters relevant
for risk assessment through onesingle platform. These are among
others, partialdischarge, rotor magnet field, endturn vibration
andoperating parameters. The system incorporates a webserver for
remote access.
INTRODUCTION
Unexpected faults in the insulation of rotating high-voltage
machines can incur high costs in materials,repairs, and, in
particular, in machine outage time. Inthe deregulated energy market
the strategy has thereforeevolved from periodic to condition-based
maintenance.On-line condition monitoring of the most
criticalcomponents of an electrical machine has become
anindispensable element of a modern risk managementstrategy. Other
than so-called off-line inspection, data isextracted without
interrupting the operation or beingdestructive. Furthermore, the
machines are tested whilstthey are being stressed by thermal,
electrical, ambient,and mechanical (TEAM) forces. These stresses
maylead to substantially different results and dielectricproperties
[1] as compared to standstill analysis.On-line monitoring is a
continuous quality control of theelectrical machine and the ability
to provide a "riskfactor forecast" allows to:
* Take immediate remedial action* Prevent catastrophic failures*
Plan for outage* Minimize lost production
On-line monitoring data and data trends help to clear upageing
and fault dynamics and thus underpinmaintenance and warranty
contracts and decisionswhether to replace or rehabilitate machines
orcomponents thereof.
MONITORING DEVICE PHILOSOPHY
The web-based modular (multiparameter) on-linemonitoring system
for high voltage motors andgenerators presented here, acquires and
processes datafrom the most critical areas of the machines.The
multi-parameter acquisition function of the deviceallows monitoring
a variety of different physicalquantities through one platform.
This allows analyzing
the interdependence of the parameters (root causeanalysis).
The system is equipped with a LAN interface and anembedded web
server, providing access for differentworkgroups for expert data
analysis and remotemaintenance.Its modular concept allows expanding
or configuringthe system according to the plant operator's needs
bothin terms of budgetary constraints and physical necessity.A
flexible and "configurable" design allows the systemto be connected
to already installed sensors, thusavoiding to re-install
proprietary sensors.
Partial Discharge
Partial discharge (pd) monitoring is the preferredmethod to
evaluate the performance of stator windingand to detect its
defects. High local electrical fields inconjunction with insulation
degradation may lead topartial discharges, which may increase over
time andeventually lead to failure. Frequently, these defects
arecreated by the TEAM-stresses and not by the electricalfield
alone and discharges may be caused by e.g.mechanical abrasion,
thermal deterioration or surfacecontamination.The partial discharge
detection principle is based on theso-called PRPDA or phase
resolved partial dischargeacquisition principle [2]. The partial
discharge pattern(example see Fig. 1) images the stochastic
properties ofthe underlying gas discharge mechanism [3] and
cantherefore be related to certain defect types. The
patternsthemselves are numerically treated in order to
deriveintegral trending quantities as described by certainstandards
[4] the patterns themselves can be analysedby experts , and/or, by
"artificial intelligence" routines[5].
8.4n.-.-
6. n
C4.On-
2.On Mi0.0-1
0 50 100 150 200 250 300
-16384-4096-1024-256-64-16-4-1
1-4244-0189-5/06/$20.00 2006 IEEE.
360 0
718
FLFHighlight
FLFHighlight
-
Fig. 1: partial discharge pattern indicating twoindividual
partial discharge sources; 1: endturn surfacedischarge; 2:
discharge between bar and core
The partial discharge module consists of fourindependent input
channels with variable input filterswhich can be adapted to the
characteristic frequencyband of electrical machines (around 0...
.10MHz, [6]), tothe local electromagnetic disturbance spectrum and
tothe frequency range of the pd sensor. The sensorsthemselves have
specific "antennae gains". In order tocompensate these a
calibration procedure is necessary,which relates a calibrated
charge impulse to the detectorreading (see [4] for details about
the procedure).Therefore, various pd sensors as current
transformers,Rogowski coils, coupling capacitors (80 pF to 9nF)
andmodified surge capacitors are in use with this device.Parallel
acquisition of 4 channels allows applying acoincidence technique
for removal of impulsedisturbance if necessary.
Ozone Concentration
Gas discharges in air lead to formation of nitrogenoxides and
ozone. The latter may be used to augmentand verify electrical
partial discharge readings and todetermine the corrosive risk of
ozone formation. Ozoneis a highly aggressive gas and may
predominantly attackpolymers as e.g. natural rubber, which is used
for seals.Ozone concentration is a qualitative parameter, as
themeasurable concentration strongly depends on sensorlocation and
environmental parameters. Ozone level isfed into the system by an
analogue (4...2OmA) signalfrom a UV absorption type detector
capable ofmeasuring ozone concentrations of some ppb to lppm.
Endturn and Bar Vibration
Endturn vibration may have various origins as endturnresonances
or loose windings, which may be caused byloose wedges, side
packings or spacers..The vibration may depend on temperature
causingsoftening of impregnating resins or on mechanicalforces
under different loads. Winding vibration maycause mechanical
abrasion of the insulation, interturnfaults by continuous friction
and gradual loosening ofspacers between the windings.The occurrence
of resonance is therefore time-dependent and plots of amplitudes
vs. time andfrequency are used for analysis, so-called FFT vs.
timepatterns, (Fig. 2).In order to record the vibration of the
endturns,accelerometers are installed on both sides of themachine.
These are placed on the endcaps (Fig.3). Thislocation is generally
on high electrical potentialaccompanied by strong magnetic fields,
which requirestransducers of special design. Preferably, biaxial
fibreoptical transducers [7] are installed which measure
twodirections of the vibration (tangential and radial), whichallows
to describe the spatial displacement (e.g. Fig. 4).
Standard accelerometers can be connected provided the
interference due to magnetic field is controlled and
theirelectrical insulation is sufficient.The authors have installed
such systems on large turbinegenerators and pump storage
generators, which havetypically a strongly variable load
regime.
Fig. 2: endturn vibration, FFT vs. time pattern duringmachine
start-up (frequency in Hz, Time in s, amplitudecoded in colour)
Fig. 3: biaxial fibreoptical sensor placed on the endcapsin the
endturn region
Fig. 4: time integral of the biaxial displacement of
anendcap
1-4244-0189-5/06/$20.00 2006 IEEE. 719
FLFHighlight
FLFHighlight
-
Overheat Detection and Hydrogen PurityMonitoring
In order to monitor potential hotspots and overheating ina
rotating machine a system called core monitor is used[7]. These
systems are often combined with hydrogenpurity monitors in hydrogen
cooled machines. Eventhough these devices are designed as
stand-alonesystems, their connection to the monitoring unit
wasdesirable in order to facilitate data correlation andprovide LAN
and Internet access.
Rotor Insulation Monitoring
Interturn and ground faults in the rotor insulation lead toa
deformation of the magnetic field in the air-gap and tochanges of
the so-called shaft voltage [7, 8].The magnetic field in the
air-gap is measured by so-called "air gap search coils" or
so-called Hall-effectsensors which were installed by the authors.
Whereasthe airgap installed sensors measure the magnetic
fielddirectly (e.g. every turn of the exciter winding in
2-polerotors) the shaft voltage is an image of the wholemagnetic
loops through the length of the generator.Therefore a variety of
effects are embedded in onesignal which may complicate evaluation.
The permanentmeasurement of the shaft voltage requires a
brush-likepickup on the generator shaft, or, the voltage of
someisolated bearings may be measured. The appearance ofcertain
harmonic frequencies and the symmetry of themagnetic field sensors
is commonly used as an indicatorfor potential problems.
Operating Parameter and Environmental Data
The system records operating parameters as power andtemperature
and allows the connection of furthertemperature and humidity
sensors. These quantities areused for parameterisation of the
monitoring data.
SYSTEM ARCHITECTURE
The monitoring system is designed to be used as aportable
instrument for periodic checks in conjunctionwith a portable
computer or to be temporarily connectedto a LAN. Or, it can be used
as a permanently installeddevice.
There are several software modules:
Embedded Software
These modules are installed in the monitoring deviceitself and
include:
* data acquisition management, including a databasefor parameter
and raw data storage and trending.
* Parameter extraction software, which calculatesscalar trending
values as FFT's (fast Fouriertransforms) and processing of the
partial dischargepatterns.
* Alarm thresholds can be defined for all scalarvalues. A
selection of scalar values can be re-converted in an analogue
4...2OmA signal whichcan be routed to the plants digital control
system. Ifa threshold is exceeded an alarm relay can
beoperated.
* An embedded webserver which allows remoteaccess to the system
with a standard web browser.
* Security and password handling* LINUX operating system
User Interface Software
Thais software is typically installed on a (portable) PCand
includes:* Tools for system set-up and alarm configuration*
Download and display of stored data* Performance of manually
started measurements
with external expert analysis software* Data display and further
evaluations
Hardware Design
The monitoring device is based on an industrialembedded PC
technology, and is free of moving parts.Modules can be added by
plugging in the necessaryinterface boards.
EXAMPLE: INTRA-/INTERNET MONITORINGOF A HYDROELECTRIC
POWERPLANTSCHEME
The plant complex in Switzerland starts in Upper Valaiswhere at
the foot of the Gries glacier Aegina AG built astorage basin with a
storage capacity of 18 million m3,and 400 m further down the
mountain the Altstafelpower plant. At the elevation of Robiei the
tributariesfrom the Gries, the catchment area between
thetributaries and from the Bedretto Valley are collected inthe two
equalizing basins Robiei and Zot. This watercan be pumped into the
Cavagnoli storage basin, or usedto drive the turbines in the Bavona
Hydro Power Plant.As major players in the energy sector the Maggia
andBlenio power plants use the water from the Maggia andBrenno
river catchments areas to produce a total of2,100 GWh hydroelectric
power.
The actual machine park is around 50 years old and willbe
successively modernised. Therefore monitoring isessential for
timing and identification of critical units.Each of the about 21
machines to be monitored isequipped with a monitoring device, which
is assigned afixed IP address.The data supplied by the instruments
are transmitted viathe company network (overhead lines) to the
powercompany's central server and analyzed by the software.The
device is compatible with the customer's existinginfrastructure,
thereby simplifying installation andreducing initial and operating
costs. Since power plantsare frequently interconnected by an
in-house Intranet,
1-4244-0189-5/06/$20.00 2006 IEEE. 720
FLFHighlight
-
central data storage in the parent company is
readilyfeasible.The central server is equipped with a web
interfacewhich records all evaluated data in the form of
trendcurves and partial discharge patterns. External expertscan
view this data online, reconfigure and check theinstruments and do
detailed analysis should the needarise. The connection is made via
a routing table whichassigns the various ports to the respective IP
addressesof the measuring devices .
In the development of the monitoring system the accentwas placed
for the one part on the exploitation offindings from research and
practice, and for the otherpart on the integration of practical
experience andcontemporary technical expertise.The PD pattern
measuring system features automaticPD pattern storage, recording of
the most prominenttrend data, integrated evaluation algorithms and
systemInternet-connectivity
Automatic evaluation of the data proves to be difficultdue to
the fact that several variables are needed toevaluate the condition
of the insulation with any degreeof surety. For this reason manual
evaluation of the PDpatterns by an expert supported by numerical
tools isstill the state of the art. The expert normally requires
notonly the trend data, but also the development of the PDpattern
itself. The system was developed for the event ofexperts not being
available at the site location. It cantherefore be configured to
measure automatically, tomake a provisional evaluation, and if
necessary to signalan alarm.
Every 0.5 h the measuring system fully automaticallyrecords the
PD pattern of all individual phases andstores them in a temporary
internal memory. Once dailythese patterns are weighted on the basis
of thecalculated apparent charge, and the one with highestvalue is
saved. The internal memory capacity isadequate to store the trend
data for the last 90 days.Independent bi-stable contacts are
available for thestatus output and can be programmed by any
giventrend value to a "Warning" and "Alarm" value, so that anon
expert user is informed easily about the systemstatus
Conclusion
We have presented a modular multi-parametermonitoring system. We
expect from a future correlationof an abundance of parameters a
more precisepossibility to reconstruct the occurrence of faults and
animproved tool which helps to avoid unplanned outagesand improves
preventative maintenance actions.In the area of research, in
particular the effects oftemperature and vibrations on the PD
properties areinvestigated. Further characteristic parameters can
beadded or modified at any time. On the basis of these
additional parameters further evaluation algorithms canbe
developed which provide more exact information onthe condition of
the windings and which uncoverproblems at an earlier time and with
great precision.
Internet solutions allow remotely implementing newalgorithms,
interactively optimising procedures andallowing human expert
support without the necessity oftravelling of a specialist to a
site, which has proven tostrongly reduce overhead and support
cost.
REFERENCES
[1] B. Fruth and G. Liptak, "Dielectric Properties ofMica Epoxy
Composites Subjected to Thermal andThermoelectrical Aging", 6h Int.
Symp. High VoltageEngineering, New Orleans, USA, paper 21.02.
1988
[2] B. Fruth, J. Fuhr, "Partial Discharge PatternRecognition - A
Tool for Diagnosis and Monitoring ofAging", CIGRE, paper 15/33-12,
1990
[3] L. Niemeyer, B. Fruth, F. Gutfleisch,"Simulation of Partial
Discharges in InsulationSystems", Proc. 7th Int. Symp. High
VoltageEngineering, ISH, Dresden, paper 71.05, 1991
[4] International Standard, High Voltage Testtechniques -
Partial Discharge Measurements, IEC60270:2000
[5] G. Ujvari, G. Pascoli, C. Kral, C. Kollmitzer,"Partial
Discharge Pattern Recognition Algorithm forTurbo Generator
Monitoring", SDEMPED 2005Symposium on Diagnostics for Electric
Machines,Power Electronics and Drives, Vienna, Austria,
7-9September 2005
[6] H. Zhu, V. Green, M. Sasic, S. Halliburton,"Increased
Sensitivity of Capacitive Couplers for In-Service PD Measurement in
Rotating Machines, IEEETransactions on Energy Conversion, Vol. 14,
No. 4,December 1999
[7] A. J. Gonzalez, M. S. Baldwin, J. Stein, N. E.Nilsson,
"Monitoring and Diagnosis of Turbine-DrivenGenerators, Prentice
Hall, ISBN 0-13-606096-X, 1995
[8] Z. F. Posedel, "Shaft Voltage in ElectricMachines, Shaft
Grounding and Monitoring System",IEEE Shaft Voltage Control and
Monitoring, July 23 -July 17, 1995, Portland, Oregon
1-4244-0189-5/06/$20.00 2006 IEEE. 721