NETA Handbook Series I, Online Diagnostics Vol 2-PDF

THE “GO-TO” STANDARDS FOR ELECTRICAL SAFETY AND RELIABILITY

ANSI/NETA MTS-2011 - New Edition

Online Diagnostics Handbook

This standard should always be referenced when writing maintenance specifications or performing routine testing on electrical power systems.

Volume 2

The ANSI/NETA Standards for Acceptance and Maintenance Testing Specifications for Electrical Power Equipment and Systems!

ANSI/NETA ATS-2009 This standard should always be referenced in design specifications or when performing acceptance testing on power system installations.

ANSI/NETA MTS-2011

ANSI/NETA ETT-2010 This standard ensures that your acceptance and maintenance tests are being preformed by qualified technicians who are certified in accordance with ANSI/NETA ETT requirements.

Available in Bound, CD ROM, or PDF

MAINTENANCE TESTING SPECIFICATIONS FOR

ELECTRICAL

POWER

EQUIPMENT AND SYSTEMS

Order Your copy of the ANSI/NETA Standards Today!

Become a NETA Affiliate and Save! To order: visit www.netaworld.org Or call 888-300-NETA (6382)

Published by InterNational Electrical Testing Association

STANDARD FOR

Published by InterNational Electrical Testing Association

On-Line Diagnostics Handbook Volume 2

Published by InterNational Electrical Testing Association

AFFILIATE PROGRAM

How Do You ensure Safety and Reliability?

Join as a NETA Affiliate. NETA and our NETA Accredited Companies invite YOU to become a part of NETA’s technical community and take advantage of the many benefits of becoming a NETA Gold or Standard Affiliate. As a NETA Affiliate you can participate in NETA’s Technical Working Committee, qualify for substantial discounts, and access industry leading experts and technical resources.

Hire a NETA Accredited Company! NETA has been connecting designers, specifiers, architects and users of electrical power equipment and systems with independent, third-party electrical testing companies since 1972. NETA Accredited Companies test the complete system in accordance with industry codes and standards to provide accurate test reports you can count on every time.

For a complete listing of NETA Accredited Companies turn to page 120 or visit www.netaworld.org.

For more information on NETA or our ANSI/NETA Standards give us a call at 888-300-NETA (6382)

Call or go online to sign up for your NETA Affiliate Subscription today! Go to www.netaworld.org or give us a call at 888-300-NETA (6382).

On-Line Diagnostics Handbook Volume 2

Table of Contents Testing and Solutions to Restore Commercial Office Building Space Affected by EMI .....1 Michael L. Hiles

Field Application of Partial Discharge Testing for Medium-Voltage Switchgear ...............4 Cal Patterson

On-Line and Off-Line Testing of Electric Motors .......................................................10 Timothy M. Thomas

Advanced Spectral Analysis ....................................................................................14 Pete Bechard

Experiences in the Monitoring of Partial Discharges .................................................19 Claude Kane

Partial Discharge Testing in Cables .........................................................................23 Craig Goodwin

Advanced Condition-Based Assessment of Medium and High Voltage Electrical Systems Without Requiring an Outage ............................29 R.R. Mackinlay and Don A. Genutis

What is Partial-Discharge Resistance? .....................................................................35 Ralph Patterson

On-Line Shielded Cable Partial Discharge Locating — An Overview ............................37 Don A. Genutis

Energy Spent on Installing Power Quality Monitoring Pays Off in Industrial and Commercial Environments ....................................................................................40 Wieslaw Jerry Olechiw

The Future of Predictive Technologies is Here Now ....................................................43 Claude Kane

Published by

InterNational Electrical Testing Association 3050 Old Centre Avenue, Suite 102, Portage, Michigan 49024

269.488.6382

www.netaworld.org

On-Line Diagnostics Handbook Volume 2

Table of Contents (continued) Case Studies in On-Line and Off-Line Motor Analysis ...............................................46 David L. McKinnon

Battery Testing Techniques .....................................................................................51 James G. Cialdea, P.E. and Mikel Hensley

Corona Imaging — See the Invisible ........................................................................55 Don A. Genutis

Comprehensive Transformer Monitoring to Increase Reliability ................................58 Claude Kane

On-Line Fault Analysis of DC Motors ......................................................................62 David L. McKinnon

Voltage Sag Testing for Commercial and Industrial Equipment ..................................67 Andreas Eberhard

NOTICE AND DISCLAIMER NETA Technical Papers and Articles are published by the InterNational Electrical Testing Association. Opinions, views, and conclusions expressed in articles herein are those of the authors and not necessarily those of NETA. Publication herein does not constitute or imply any endorsement of any opinion, product, or service by NETA, its directors, officers, members, employees, or agents (hereinafter “NETA”). All technical data in this publication reflects the experience of individuals using specific tools, products equipment, and components under specific conditions and circumstances which may or may not be fully reported and over which NETA has neither exercised nor reserved control. Such data has not been independently tested or otherwise verified by NETA. NETA makes no endorsement, representation, or warranty as to any opinion, product, or service referenced in this publication. NETA expressly disclaims any and all liability to any consumer, purchaser, or any other person using any product or service referenced herein for any injuries or damages of any kind whatsoever, including, but not limited to, any consequential, special incidental, direct, or indirect damages. NETA further disclaims any and all warranties, express or implied including, but not limited to, any implied warranty or merchantability or any implied warranty of fitness for a particular purpose. Please Note: All biographies of authors and presenters contained herein are reflective of the professional standing of these individuals at the time the articles were originally published. Titles, companies, and other factors may have changed since the original publication date. Copyright © 2009 by InterNational Electrical Testing Association, all rights reserved. No part of this publication may be reproduced in any form or by any means, electronic or mechanical, without permission in writing from the publisher.

1

On-Line Diagnostics Handbook — Volume 2

Testing and Solutions to Restore Commercial Office Building Space Affected by EMI NETA World, Summer 2003 Issue by Michael L. Hiles Field Management Services, Corp.

A company with sensitive electronic display equipment was in the market for a new location in midtown Manhattan. The performance and stability of their display equipment rested on the electricity of the environment, and a valuable real estate deal turned on the ability to adapt an available 30,000 square foot space to meet the stringent requirements of the high-tech tenant. There were two key components to ensuring that the proposed space would be acceptable: power quality and magnetic field levels. In terms of power quality, the company required not only continual power without spikes or sags but also a power system designed to handle the load composed primarily of computers, which are common nonlinear, harmonic-creating sources. The second requirement, for low enough magnetic fields to allow sensitive monitor operation, proved even more difficult to achieve. Preliminary measurements taken by the local electric utility and confirmed by power quality consultants, revealed that substantial parts of the proposed 30,000 square foot space would be unsuitable for the business requirements of the prospective tenant. The litany of problems included power quality issues, power frequency magnetic fields, and dc fields from a subway line. These concerns needed to be resolved before a lease could be signed. The building management working with the local electrical utility contracted with Los Angeles/New York magnetic field specialist, Field Management Services (FMS) to analyze the extent of the problems and to make recommendations for corrective action. At the same time, the tenant, the building owner, and FMS worked toward an interference specification agreement that could form the basis for an acceptable lease agreement. Although the lease agreement would be a legal document, it would be shaped by highly specialized technical work with the building’s power system. Power quality, building wiring practices, and EMF mitigation would each play a role.

Solutions to the existing problems required coordination between the building’s electrical department, power quality consultants, and the local utility. There were three distinct areas of concern:

• Power frequency (ac) magnetic fields — in excess of 200 milliGauss (mG) • DC magnetic fields — with shifts of more than 800 mG

• Slowly varying (sub-10 Hz) magnetic fields — with shifts of more than 500 mG. Each of these field sources required separate measurement and analysis. Each also required the simultaneous development of a unique threshold interference specification, as there are no performance or safety specifications for any of these sources of interference. Finally, each source required a course of mitigation specifically engineered to meet its threshold specification. For this discussion, we will focus on the power frequency magnetic or electric fields.

What were the interference thresholds? First on the agenda was the establishment of acceptable ambient field levels at 60 Hz and low-order harmonics. The tenant had different requirements for each of three, and possibly more, computer/monitor combinations. There is considerable literature on the interference threshold of computer monitors, but the conclusion is generally that each monitor/computer system is unique in its sensitivity to external fields and that the range of sensitivity is considerable, as much as 200 percent variation. Therefore, the most prudent course would be to test each of the tenant’s computer/display combinations at actual site locations and to correlate the results with simultaneous magnetic fields’ strength readings. Based on the boundary measurements of this work, it would then be possible to construct a system

2 of Helmholtz coils that would duplicate the threshold conditions of each computer system, and, in a laboratory setting, test the limits of these systems and various mitigation strategies. The most demanding systems were large format (21inch), high-definition displays, driven by high-end Macintosh computers. These systems were also the most critical for the tenant. They would be used in a classroom setting, training students in their use. These systems became unstable in anything greater than a six mG field. Next, the computer systems used for administration purposes would be equipped with standard Windows 17-inch displays. These systems exhibited sensitivity at or above 10 mG. Finally, the 17-inch LCD (flat screen) displays, also used by administration personnel, were essentially immune to interference throughout the space.

What levels were in the space and what were their sources? A thorough magnetic field survey revealed four areas of elevated ac fields: two that were along perpendicular walls (most likely caused by high voltage utility feeder cables in the street), one that was adjacent to a partially-abandoned utility transformer vault, and, finally, one area that was adjacent to the main building network protection vault. The maximum field levels ranged from 60 to over 200 mG. With the exception of the area in front of the main vault, the ac magnetic fields were of a type called spatial fields, the result of high-current loads and separated conductors. Measurements in these areas display a rapid rate of decline as distance is increased from the source. In contrast to this, values in certain areas adjacent to the main electrical network protection vault declined at a much slower rate, the signature rate of a net-current source. This distinction is critical to the field analysis and significantly limits the available options for remediation. It warrants discussion. In a typical ac circuit, if the vector sum of the phase currents and the neutral (and/or ground wire if present) add to zero, the magnetic field at a distance from the circuit will be solely due to the differences in distance and direction from the point of measurement to the individual source conductors and the current contained in the circuit. Magnetic fields produced as a result of conductor separation and the consequent decrease of natural flux cancellation are known as spatial magnetic fields. Spatial magnetic fields typically decay with the distance (source to point of measurement) squared, 1/d². Effective mitigation measures may include shielding, reconfiguration of the conductors so as to maximize cancellation, and/or moving all conductors so as to increase distance. In multigrounded, four-wire electric utility distribution and transformer systems and in many industrial and commercial building-wiring systems some portion of the neutral current may return to the source transformer via building grounds. In other cases, neutrals of different circuits may be tied together, thus allowing current from one neutral to

On-Line Diagnostics Handbook — Volume 2 return to the transformer via alternate neutral conductors. In any of these instances, the vector sum of the currents for any given circuit may not add to zero. When the vector sum of the phase, neutral, and parallel ground wire (if present) for a given circuit does not equal zero, a net-current condition is present. This circuit condition creates net-current magnetic fields. Although net-current magnetic fields also decay with distance, the rate of decay is less effective, 1/d versus 1/d². The slower rate of decay produces higher levels of fields across a much larger area than would be expected from the more common spatial fields. Worse, when net-current conditions are present, the fields caused by net currents cannot be shielded by conventional means, and reconfiguration of conductors is usually not effective at reducing net-current magnetic fields. In general, the best mitigation measure for magnetic fields caused by a net-current condition is to correct wiring problems (poor neutral connections, missing neutrals, improper grounding, etc.) so as to minimize the net current. After this work is complete, the residual, spatial fields can be shielded.

What field mitigation measures were available for these sources? Since the fields fall off with distance, the first level of field mitigation to be considered involves space planning. In this strategy, the space planner designs the use of space to accommodate elevated fields. Where possible, troublesome space is assigned to uses that will not involve sensitive technology. Libraries, meeting rooms, conference facilities, hallways, and bathrooms are all productive uses for areas with elevated fields. To the extent that this strategy would not compromise the tenant’s space plan for the area, each of the areas that had elevated fields could be downgraded for uses that would not involve sensitive technology. Careful space planning, in many cases, will obviate the need for more costly measures. For areas not amenable to space planning and in those areas without net-current fields it was possible to design a custom shield and thereby reduce the fields to acceptable levels. Accordingly, custom shields were designed for critical parts of three of the four areas. In the areas with high-current and net-current fields, a combination of engineering measures and shielding is required. In the three areas without net-current fields, the results of the shielding were so successful that there were no restrictions on the use of the space. In all of these areas, conventional shielding reduced the fields to below 2 mG. The field levels had been reduced such that interference from power frequency fields was no longer at issue. In the area adjacent to the network vault, the fields were substantially reduced by the shielding and to within the desired interference specification of less than an average of 6 mG. However, it is unlikely that these levels would have been achieved without the diligent effort by the building management working with the power quality consultant

On-Line Diagnostics Handbook — Volume 2 to reduce the net-current fields by engineering and wiring measures. This is complicated and tedious work, beginning at the top of the building and working down, mapping the flows of current, first within major conductor systems (bus bars, feeder cables) and proceeding to branch circuits and eventually to individual electrical appliances and outlets. At each step in the effort, current readings are taken and, as required, wiring corrections are made in accordance with the National Electrical Code. Slowly, the corrections in the upper floors were reflected in the field strength readings in the area nearest the electrical vaults. Finally, FMS designed a shielding system that reduced the residual fields, including the net-current fields, to below a maximum of 1 mG. Impressive as this result is for this project, equally important for the building is the work to reduce the net currents throughout the building. This work required the building engineering staff to clean the building from top to bottom. The net-current corrections had the ancillary benefit of eliminating the majority of elevated fields throughout the upper floors of the building and reducing the possibility of electrical shock and fires. The requirements of this tenant brought together organizations with quite different skills, contributions, and agendas. In the end, the agreements that concluded with a lease for a large part of a Class A building were the result of this disparate group working productively together. Moreover, the lease document defined the technical limits and economic risks of all of the parties and was based on realistic, real world specifications. For 10 years prior to establishing Field Management Services, Michael Hiles was President and CEO of NoRad Corporation, a manufacturer and marketer of EMF shielding and measurement devices for video display terminals. He is a founder of the American Teleconferencing Association. Mr. Hiles studied electrical engineering at the University of Wisconsin and has completed professional studies in electrical engineering, business administration, and economics at UCLA. He is a member of IEEE, BEMS, and SMPTE.

3

4

On-Line Diagnostics Handbook — Volume 2

Field Application of Partial Discharge Testing for Medium-Voltage Switchgear PowerTest 2004 (NETA Annual Technical Conference) Cal Patterson, Magna Electric Corporation

Partial discharge testing on medium voltage switchgear has, for the most part, been limited to manufacturers. Until recently; the measurement of PD in the field has been complicated by the influence of noise and, to some extent, mystery and limited knowledge. It is well accepted that partial discharge is a leading cause of insulation failure. This is to say that most insulation failures begin with some form of partial discharge. It would then seem obvious that acceptance testing and periodic field measurement of partial discharge should be valued as a predictive tool allowing time for repair of detected PD locations well in advance of failure. Modern technology (electronics, computers, etc.) has recently made partial discharge measurement in the field fairly simple and reliable. The focus of this paper will be to discuss the general theory of phase resolved partial discharge measurements, the types and causes of partial discharges, the unique advantages of using these measurements for acceptance as well as routine insulation evaluation tests. There is good evidence that continuous monitoring of PD activity of switchgear may soon be a standard. The information contained in this paper is not intended to be comprehensive or necessarily complete. It is intended rather to be a general introduction to application of partial discharge testing for medium voltage switchgear. There are several components and materials used in the manufacturing of switchgear. As it is not the purpose of this paper to elaborate on each material or component, I list these for consideration only. Each component and material, although designed to be PD free, can have inherent or developed defects that contribute to partial discharge locations. Other contributing factors can be construction, initial installation and environment.

Materials • Epoxy • Porcelain • Tapes (insulating and semi conductive) • Glastic (fiber-glass) • Varnishes • Air gaps • Fluids - Oil, PCB • SF6 • Vacuum Components • Instrument Potential transformers • Current transformers (bar & window) • Insulating stand-offs (epoxy, porcelain) • Oil filled bushings (designed capacitance) • Bus (fluidized, taped) • Bus windows / barrier boards • Stress cones • Capacitors (surge caps, PD sensors) • Inductors / reactors • Breakers (SF6, vacuum, air, oils) • Dry type transformers • Oil filled transformers • Breakers • Non load break switches

5

On-Line Diagnostics Handbook — Volume 2 What is Partial Discharge? Contrary to popular belief, measurement and evaluation of partial discharge is not black magic or witchcraft. Scientists throughout the world have developed effective instruments and mathematics for measuring and analyzing partial discharge. PD is a localized electrical discharge in an insulation system that does not completely bridge the electrodes. There is a very small amount of energy in a single discharge commonly measured in Pico-coulombs. Often there can be several locations in a switchgear lineup where partial discharges are being produced. Partial Discharges are avalanche occurrences that produce measurable electrical quantities in a very large frequency spectrum. These discharges can also produce visual and audible signs as well as ozone production resulting in the white powder often observed. Although the energy contained in one partial discharge is very small; cumulatively they can have a significant destructive effect. Partial discharge activity over time destroys insulation through chemical and heat related activity. The destructive nature of partial discharge varies with the applied voltage. Below 4.16 kV operating levels partial discharges are far less destructive. As applied voltages are increased PD activity has significantly more destructive result. From experience, we do not expect to measure large quantities of partial discharge in switchgear. Motors and generators are a different story and it is usually expected there will be some established levels of PD. Switchgear usually represents a more black and white, go no go situation. If the switchgear is producing even small discharges, it is usually good practice to remedy the situation very soon or at least understand what is causing the discharges. There are some types of discharge that can exist indefinitely without causing failure. However left unattended these could potentially mask more important Pd events. Partial discharge is certainly not full discharge where the applied potential completely bridges the gap between anode and cathode. It is rather sparking that occurs across locations within the insulating systems that do not meet designed dielectric strength. The discharges are a result of electrical stress points within the insulation. Examples of where these stress points can be expected are as follow. • • • • •

Gas filled voids within insulation Voids between insulation and high voltage conductors Voids between insulation and ground plane Contaminated insulators Conductors under floating potential

A small explosion may best describe these rather complex occurrences. In any case, these rather small events produce currents, voltages and radio waves that can be detected using simple sensors that detect and provide suitable signals for instrumentation Because partial discharges are a result of the applied alternating potential they are closely and specifically related

in time. This is to say, we expect to observe the discharges in the rising portions of the applied sine wave. As the voltage rises, the stress across the voids becomes excessive and a discharge results when the void reaches it’s breakdown potential. A term frequently used is ‘clumping’ where we expect to see a build up of activity (PD Pulses) in the 0 to 90 degree window and the 180 to 270 degree window. Practically we see PD pulses in the 1st and 3r quadrant of applied voltage. This is an important signature of partial discharge activity and shown in the diagram below.

Reference Partial Discharge Quantities Partial discharge standards describe several integral quantities that are commonly used to assess the level of partial discharge that is occurring in insulation. These quantities and their definitions are commonly used standards. Most frequently used integral quantities are as per ASTM D1868 standard, quoted below are: “3.1.10 average discharge (corona) current (It)—the sum of the absolute magnitudes of the individual discharges during a certain time interval divided by that time interval. When the discharges are measured in coulombs and the time interval in seconds, the calculated current will be in amperes.

where:

It = average current, A,

t0 = starting time, s,

t1 = completion time, s, and

Q 1 Q 2 Q n = partial discharge quantity in a corona pulse 1 through n, C.

6 3.1.15 partial discharge (corona) power loss (P)—the summation of the energies drawn from the test voltage source by individual discharges occurring over a period of time, divided by that time period.

On-Line Diagnostics Handbook — Volume 2 The graph shown below is commonly used as a visualization of the integral quantities listed above. The horizontal scale could also be represented in coulombs.

When pulse height analysis is used, the summation over a period of time of pulses above a preset level of corona usually determined by background noise multiplied by the instantaneous test voltage at the time of the pulses in the specimen is approximately equal to:

where:

P = pulse discharge power loss, W,

nj = recurrence rate of the jth discharge pulse in pulses/ second. Qtj = the corresponding value of the partial discharge quantity in coulombs for the particular pulse.

Vj = instantaneous value of the applied voltage in volts at which the jth discharge pulse takes place. 3.1.15 partial discharge apparent power loss (Pa or PDI)— the summation over a period of time of all corona pulse amplitudes multiplied by the rms test voltage.

where:

Pa = apparent power loss in time interval (tl - t0), W, It = average corona current, A, and Vs = applied rms test voltage, V.

3.1.15 partial discharge (corona) pulse rate (n)—the average number of discharge pulses that occur per second or in some other specified time interval. The pulse count may be restricted to pulses above a preset threshold magnitude, or to those between stated lower and upper magnitude limits.”

Partial Discharge Sensors To measure partial discharge sensors are connected in some manner to the switchgear under test. The following is a brief summary of the most common types and how they are applied. Sensors are normally high frequency static components designed to act as low impedance paths for high frequency pulses. Various designs measure PD voltages, PD currents or radio waves. Connections to PD instrumentation are made using coaxial cables. Sensors should always be located as close to the specimen to be measured as possible since partial discharge pulses attenuate rapidly over distance.

Common Types of Sensors 80pf to 500pf capacitors — These sensors are connected directly to the high voltage components of the switchgear. In general and where practical, they are preferred as they effectively capture PD signals as well as reject low-level noises that can interfere with PD measurements. These devices are normally installed permanently.

7

On-Line Diagnostics Handbook — Volume 2 RFCT (radio frequency current transformer) — High frequency current transformers placed around cable shields or surge capacitor grounding conductors. These sensors convert partial discharge currents to measurable PD voltages.

Typical Switchgear Sensor Locations and Connections

RFVS (radio frequency voltage sensors) — are usually connected to low voltage windings of current and potential transformers. The capacitance formed between the high voltage and low voltage windings act as a low impedance path for higher frequency PD pulses. These devices are most commonly used for walk in tests. Where noise is not prevalent these sensors can be installed permanently.

Examples of Partial Discharge Location and Damage The picture below is an example of partial discharges that resulted from improper placement of the bus bar through the current transformer. Detection of this defect was made during a walk in test.

Current Transformer

The picture below is an example of partial discharge caused by moisture and contamination. The treeing can be clearly seen where the electrical stress points will continue to move as the carbonized branches grow and move the stress point towards ground. This defect was found during a walk in PD test.

Schematic Showing Connection of RFVS to Current Transformers

8

On-Line Diagnostics Handbook — Volume 2

Glastic Barrier

In both the above examples the activity and destructive process had been going on for several years.

Measuring Partial Discharge Measurement of partial discharge can be accomplished by simply connecting an oscilloscope to PD sensors. This is only practical for people with extensive experience Through the use of modern electronics and computers the process of collection and analysis of PD activity is now to large extent automated. In general there are only two quantities that can be measured. These are partial discharge pulse amplitude measured in volts and phase relation in degrees. As previously mentioned, partial discharges produce pulses that are measurable in a very wide frequency band. By choosing a suitable frequency bandwidth it is possible to have both adequate sensitivity and acceptable signal to noise ratios. For purposes of this explanation frequency is defined as the average width of a PD pulse. It is important to understand that a suitable sample size is required to accurately capture most of the partial discharge activity occurring in said specimen. A typical sample size is 60 complete power system cycles. The reason for this is that not every defect discharges every cycle The measuring device is usually triggered from the power system voltage. Depending on instrumentation capability several cycles of data are captured during each automated trigger. PD pulses are then extracted from the sample and stored in a spreadsheet like matrix. They are stored in the matrix according to phase relationship and amplitude. There are a pre-defined number of matrix windows. When the sampling and storing process is complete computer analysis can then massage the data into the various graphic displays and integral quantities described above. The diagram below is a picture of a typical oscilloscopic trace of raw partial discharge pulses.

Raw Partial Discharge Oscilloscope Trace

The PD matrix shown below is actually an animated power point that describes the measuring process. Starting from 0 degrees PD pulses begin to appear and continue to occur until the applied voltage reaches 90 degrees. At this point discharges stop. At 180 degrees as the voltage is increasing discharges again begin and continue to accumulate through to 270 degrees. Several cycles of power system voltage are required to adequately measure all the PD activity that may be present. Notice also that this matrix is divided into 15 degree widows by 22 magnitude windows. Pulses are placed in the matrix within these boundaries. Polarity of the pulses change from 1st to 3rd quadrant. Pulses from 0 to 90 degrees have negative polarity and pulses in the 3rd quadrant have positive polarity as the applied potential is very slightly reduced towards ground during a discharge. Distributions of negative and positive pulses are used for more advanced analysis.

9

On-Line Diagnostics Handbook — Volume 2 Partial Discharge Measurements vs. Traditional Insulation Tests Traditional Tests There are several off-line tests that are used for acceptance and periodic insulation evaluation. Although these tests yield an average value of insulation quality they often cannot identify specific defects. The reason for this is due to a masking effect. This not to say that they are not valued as reasonable tests but rather that they have specific limitations. DC hipot tests only measure currents resulting from parallel resistive paths. As well dc hipot tests cannot detect locations where partial discharges occur as d.c. voltages do not produce partial discharge. In an A.C. hipot test both capacitive and resistive currents are combined where the capacitive current is much larger than the resistive current. The resistive current of course tells us about insulation quality but is not available. A better alternative to the A.C. hipot is a power factor or dissipation factor test that at least separates the actual watts loss (resistive current). Isolated but significant insulation defects can still be masked by the overall well being of the insulation using a power factor or dissipation factor tests. The power factor or watts loss yield an average value of insulation integrity. This is the same reason we cannot perform a power factor test of a transformer winding in parallel with long shielded cables. The above tests have other common disadvantages. • Equipment must be taken out of service • Often components and connections are removed or altered to facilitate the tests • Usually in switchgear the main incoming bus is not available for testing

Partial Discharge Tests • Tests are usually performed on-line • All equipment that is normally in service is connected • Actual voltage and operating conditions are used for the test • Detects failing insulation vs. failed insulation • There is no masking effect so even very small defects are detected • Frequently walk in tests can be done using temporary sensors connected to low potential components such as VTs and CTs. • Measures quantities that lead to complete failure of insulation • Measurements are trend able and predictive in nature

Continuous Measurement of Partial Discharge Recently a few manufacturers of partial discharge instrumentation have taken these measurements to the next logical stage. Now available are monitors that can be applied to take continuous readings of partial discharge activity. Because there are conditions and situations that are not easily controlled such as contamination, temperature, humidity and system activities continuous monitoring is preferred over periodic measurements. All of the above variables can affect partial discharge activity and time to failure. Often these variables can cause considerable acceleration of insulation deterioration that could be missed using only periodic PD measurements. Warning and alarm outputs from continuous monitors alert maintenance to rapidly changing or new partial discharge activity.

Conclusion Partial discharge testing of medium voltage switchgear whether periodic or continuous must be used for complete insulation integrity analysis. No longer are experts required to conduct these tests. Rather sophisticated equipment is now available that can be easily used in the field. Partial discharge tests measure quantities on-line relating to failing insulation vs. failed insulation and in most cases repairs or replacement of equipment can be achieved several years in advance of complete insulation break down. Although traditional tests are most common for acceptance and subsequent field tests it is obvious that they only provide part of the picture. When traditional tests (go no go tests) are used in conjunction with PD tests there should be little doubt as to insulation quality. Partial discharge tests provide assurance that insulation remains integral between scheduled field maintenance.

References Eaton Corporation “Universal Partial Discharge Analyzer Technology.”

Eaton Corporation “InsulGard™” — Continuous partial discharge monitor Eaton Corporation “Claude Kane, Alexander Golubev — Cutler-Hammer Predictive Diagnostics

Cal Patterson received his Electrical Engineering Technology diploma in 1973 at the Saskatchewan Technical Institute in Moose Jaw, Saskatchewan. He has been employed with Magna Electric Corporation since January 2003 as a senior project manager. Previously, he worked for Cutler-Hammer Engineering Services as Senior Engineering Technologist and specialized in predictive diagnostic services. Prior to that, he worked with SaskPower from 1973 to 1998 as a system test technician. Professional affiliations include Member of Saskatchewan Applied Science Technologists and Technicians (SASTT), since 1973.

10

On-Line Diagnostics Handbook — Volume 2

On-Line and Off-Line Testing of Electric Motors NETA World, Summer 2004 Issue by Timothy M. Thomas Baker Instrument Company

Abstract Modern processing plants and facilities demand a high degree of motor reliability. This paper seeks to present the most current, effective, and accepted methods of electrically testing and delineating trends for the operational health of electric motors. The benefits and features of modern highvoltage electrical test equipment and testing methodologies are also discussed.

Introduction: The Need for Motor Testing The steady, safe, and efficient operation of electric motors is essential to the productivity of all plants and facilities. Some facilities, including electrical utilities, pulp and paper mills, and innumerable others, have many critical and/or expensive motors. A motor failure could be catastrophic, causing lost production and costly emergency repairs. For

Figure 1 — Burnt One Horsepower Motor

example, a motor failure at a nuclear plant can cost up to one million dollars a day for critical motors and may have a disastrous, long-lasting impact. Even failures at a wastewater treatment facility can have a huge, negative environmental effect and can be very costly. Motors fail due to numerous operational circumstances including power condition, mechanical influences, and environmental hazards. According to recent IEEE and EPRI studies, at least 35 to 45 percent of motor failures are electrically related. Monitoring the motor’s “electrical health” is, unquestionably, an important and vital consideration. Determining a trend for the historical operating condition of a motor makes early detection of any decline in the motor’s health possible. Planning downtime and having only minor reconditioning repairs instead of a major rewind or replacement is far less expensive in both repair costs and lost production. Since electric motors begin deteriorating the instant they are started, it is necessary to monitor their operating condition on a routine, periodic schedule. Periodic monitoring and trend creation of collected and correctly diagnosed data provides the technician with evidence needed to prepare for downtime before a catastrophe occurs. It is no longer practical to test a motor merely with a megohmmeter in order to determine its condition. Plants and facilities depend on a complete predictive maintenance program (PMP) to monitor their operations and plan their repair schedules. A good PMP requires both static (off-line) and dynamic (on-line) testing with educated and trained technicians monitoring data routinely with quality equipment. Besides voltages and currents, on-line test equipment must be able to capture and determine trends for torque ripple and torque signatures as well as rotor-bar side bands. Off-line testing with modern, high-voltage test equipment is essential to getting reliable data. The voltages required to test motor windings correctly cannot be reached with impedance-based or low-voltage test equipment.

11

On-Line Diagnostics Handbook — Volume 2 On-Line Testing Effective dynamic test equipment must be able to collect and determine trends for all essential data affecting the operation of electric motors. Power condition (including voltage level, voltage unbalance, and harmonic distortion), current level and current unbalance, load level, torque signature, rotor-bar signature, effective service factor, and operating efficiency should be tracked and trends determined for them. On-line testing is performed at the motor’s motor control center (MCC), at the load side of a variable frequency drive or at an installed port, which allows for online testing without opening the MCC. Data is collected through a set of current transformers and corresponding voltage probes. The collected, processed, and analyzed data provides the technician with an overall view of the normal operational environment to which the motor is subjected on a daily basis and of how the motor is responding within this environment. Often a motor is subjected to incoming power problems including low or high voltage, voltage unbalance, and harmonic distortion. Lower voltage causes higher current and, therefore, more heat. Higher voltage causes lower power factor and ultimately higher losses. A small amount of voltage unbalance creates an exponential amount of current unbalance which causes temperature increases. Harmonic distortion also causes thermal stress in motors. Any of these voltage problems can cause severe overheating in the motor, even without reaching an overload situation, and excessive heat is the insulation’s major enemy. Some motors are subjected to physical obstacles that cause undue stress. Overgreasing, misalignment, and overtightened belts all cause thermal stress. Many motor failures can be traced to load related situations. Erratic torque signatures can be an indicator of load related problems. Broken or cracked rotor bars can also cause premature motor failures. On-line testing identifies these problems and routine delineation of their trends will reveal the rate of decline. The effective service factor is also an important test of the overall health of a motor. Two elements affect the service factor number: real operating power condition (voltage quality) and steady-state load conditions. The effective service factor number represents the thermal stress caused by these two conditions on the motor. On-line testing can be used to create trends for all these motor conditions. Dynamic testing schedules should be tailored individually according to operating time, criticality, and any other important element of operation. Generally, an on-line test should be performed at least quarterly. Motors beginning to show obvious decline or thermal overstressing should be monitored more closely until the motor can be statically tested or removed from operation and repaired. New and recently repaired motors should be tested as soon as they are returned to service in order to provide a historical record (or baseline) of their performance when the motor is at its “best.”

Off-Line Testing In general, motors are quite reliable and, when correctly maintained, can be expected to provide at least one hundred thousand hours of continual operation. That is to say, a new motor operated within nameplate parameters should give at least eleven years of steady use. Unfortunately, motors are almost always subjected to a variety of damaging elements with the end result being a rise in operating temperature. Thermal aging of the insulation is the major cause of insulation failure. Years of testing and numerous studies have shown that, as a rule-of-thumb, “for every 10 degrees centigrade increase in temperature, the winding life is decreased by half.” (See Crawford.) Besides thermal problems, other causes of insulation failures include incoming line-related problems. Spikes caused by lightning and surges created by switching and contactor closing contribute to insulation breakdown. Motors are also subjected to mechanical influences, including bearing failure, environmental hazards, and magnet wire damage caused during the manufacturing process. Even the physical movements of the windings during startup cause wear to the insulation system — especially the magnet-wire insulation, as D.E. Crawford proved.

Figure 2 — Damaged Coil

Correct testing of all components of a motor requires a series of tests designed to emulate the conditions the motor will see in the field. It has been proven in numerous studies that low-voltage testing, including capacitance, inductance, impedance, etc., is not an effective tool in detecting weakness in the insulation. Quality off-line test equipment can perform winding resistance tests, insulation resistance tests, high potential tests, polarization index, and surge tests at IEEE-, NEMA-, and EASA- accepted standards. Top quality test equipment will automatically run a series of preprogrammed tests and provide a complete final report. This automatic equipment will stop testing before any damage is done to the windings.

12 The resistance test verifies the existence of dead shorts within the turn-to-turn coils, shows any imbalances between phases due to turn count differences, locates poor wire connections or contacts, and finds open parallel coils. DC insulation resistance testing detects faults in groundwall insulation or motors that have already failed to ground. Weak groundwall insulation (prior to copper-to-ground failure) can only be found successfully with high potential tests. The groundwall insulation system consists of the magnet-wire insulation, slot liner insulation, wedges, varnish, and, often, phase paper. A dc high potential test should be performed at twice line voltage plus 1000 volts since motors will see voltage spikes of at least that level during each startup. High potential testing is necessary to verify winding suitability for continued service. Surge testing detects potential faults in both interturn winding and phase-to-phase insulation systems. Turn-toturn faults will not be seen by any other method of testing including megohmmeter and high potential tests. Potential faults can only be seen when the coils see more than 350 volts from turn-to-turn or coil-to-coil, as described by Paschen’s Law. The typical mechanism of fault progression is a turn-to-turn short causing excessive heat and progressing within seconds or minutes to copper-to-ground faults. Faults are much more likely to occur between turn-to-turn winding coils due to the added stress caused by bending and exaggerated during the winding process. The ground wall insulation is generally many times stronger and more capable of withstanding voltage spikes and other stresses.

Case Studies • At a large wastewater treatment facility in Florida, 14 identical motors were scheduled for predictive maintenance. These motors were 40 horsepower aerators for a large treatment tank and operated continuously. Static tests were performed on all 14 with each receiving passing marks on all tests. When dynamic testing was completed, it was noted that 13 of the 14 motors were running within expected parameters at approximately 85 percent load while the remaining motor was running at just over 30 percent load. Further inspection revealed a sheared coupling on the motor running at reduced load. The operators had no way of detecting the problem, and the location of these motors made visual inspection difficult. The dynamic testing found a problem that was costing the customer in both wasted kilowatt usage and production. • Twelve 60 horsepower pump/motors were tested at a large office building. Six were chilled water pump/motors and six were condenser water circulating pump/ motors. All 12 were installed at the same time and ran continuously. Dynamic testing was performed one day on the motors, and all appeared to be operating within expected parameters. The motors were shut down for scheduled, annual, routine building maintenance. Static testing was planned for the following morning. Resistance tests appeared normal on all but two motors.

On-Line Diagnostics Handbook — Volume 2

Figure 3 — Paschen’s Law at Atmospheric Pressure in Air

These two did not pass high potential testing at the preset voltage. Three others failed the surge tests. The five motors were removed from service, disassembled, and inspected. Two were found to be extremely dirty, while three had no visual damage. All five were reconditioned, re-tested, and placed again in service. The off-line testing prevented five potential catastrophic failures and allowed the customer to dictate the downtime.

Conclusions Integrating on-line and off-line testing into a PMP provides the technician with verification of his motor’s condition. Both technologies are necessary in order to have a complete picture of a motor’s health. Collecting both online and off-line data on a routine schedule allows for early warning of impending failures and opens the opportunity window for planned downtime. Performing resistance, high potential, and surge testing along with dynamic testing provides the technician with a total picture of the motor’s condition and allows him to track its rate of decline. Modern test equipment includes enhanced and detailed reporting. Reports are easily generated, providing a written hard copy of test results and making diagnosing and comparing of data clearer and more accurate. Setting up and managing a program to monitor the motors within any facility is essential to insure the safe and continued operation and production of the facility. In most cases, a correctly managed and operated PMP will save a plant or facility much more than it will cost to implement, administer, and manage.

On-Line Diagnostics Handbook — Volume 2 References Crawford, D. E., “A Mechanism of Motor Failures,” General Electric Company 570 E 16th Street, Holland, Michigan 49423, 75CH1014-EI-19.

Paschen, F., Paschen’s Law, 1889.

Motor Reliability Working Group, “Report of Large Motor Reliability Survey of Industrial and Commercial Installations”, Parts I and II, IEEE Transactions on Industry Applications Society, Vol IA-21, No. 4, pp. 863, 1985. “Improved Motors for Utility Applications,” EPRI EL4286 Final Report, Vol. 1, 2, October 1982.

Timothy M. Thomas holds a Bachelor of Science degree in engineering science from Florida State University. He is currently working as an applications engineer at Baker Instrument Company in Fort Collins, Colorado. He has extensive field experience in gathering, compiling, diagnosing, and reporting on both static and dynamic motor circuit data and is proficient with Baker’s test equipment. He holds a certificate as a Level III Vibration Analyst. The previous nine years he worked as General Manager of an electric motor rewind shop in Florida where he set up and managed the predictive maintenance program. His main field of interest is predictive maintenance of electric motors and related equipment.

13

AFFILIATE PROGRAM

How Do You ensure Safety and Reliability?

Join as a NETA Affiliate. NETA and our NETA Accredited Companies invite YOU to become a part of NETA’s technical community and take advantage of the many benefits of becoming a NETA Gold or Standard Affiliate. As a NETA Affiliate you can participate in NETA’s Technical Working Committee, qualify for substantial discounts, and access industry leading experts and technical resources.

Hire a NETA Accredited Company! NETA has been connecting designers, specifiers, architects and users of electrical power equipment and systems with independent, third-party electrical testing companies since 1972. NETA Accredited Companies test the complete system in accordance with industry codes and standards to provide accurate test reports you can count on every time.

For a complete listing of NETA Accredited Companies turn to page 120 or visit www.netaworld.org.

For more information on NETA or our ANSI/NETA Standards give us a call at 888-300-NETA (6382)

Call or go online to sign up for your NETA Affiliate Subscription today! Go to www.netaworld.org or give us a call at 888-300-NETA (6382).

14

On-Line Diagnostics Handbook — Volume 2

Advanced Spectral Analysis NETA World, Summer 2004 Issue by Pete Bechard PdMA Corpration

Current-Signature Analysis As industries continue to look for new methods of identifying and predicting equipment failures, manufacturers of predictive maintenance equipment are developing new tools to add to their arsenal of available technologies. Newly developed methods of extracting information from the line current supplied to a motor have uncovered information on both the electrical and mechanical health of the equipment. Not just the power supply and motor, but now tracking and trending of information deep into the load and shaft line components can be done through the line current as well. This article discusses the fundamentals of these new current-demodulation methods and shows how they are being used to identify both electrical and mechanical anomalies existing in plants today. It also discusses how using this new feature helps bridge the communication barrier between the mechanical and electrical departments relating to vibration and electrical power analysis.

Since 1985, current-signature analysis (CSA) has been growing as a preferred predictive maintenance tool to identify damaged rotors and air-gap eccentricity in induction motors. CSA is based on the observation that variances in the stator-rotor air gap are reflected back into the motor’s current signature through the air-gap flux affecting the counter electromotive force (CEMF). These changes in CEMF then modulate the running current turning an induction motor into an efficient transducer. By performing a fast Fourier transform (FFT) on motor current, the power cables can act as permanently installed test leads for predictive maintenance applications. An FFT is a mathematical operation that extracts the frequency information from a time-domain signal and transforms it to the frequency domain. The frequency domain is a graph of the amplitude of a signal at a given frequency. In the frequency domain, the height of the peak represents the amplitude of the signal. Figure 1 shows the relationship between the time domain (t), the frequency domain (f ), and amplitude (A).

Rotor-Bar Damage

Figure 1 – Relationship between Time, Frequency, and Amplitude

In CSA, the pole-pass frequency (FP) appears as sidebands surrounding the line frequency (FL) after performing an FFT on a captured signal. The synchronous magnetic pattern of the stator rotates faster than the rotor cage. This implies that any given rotor bar is passed by all of the magnetic poles in one rotation of the slip frequency. The rate at which this occurs is termed the FP. Often in vibration analysis the term for this is called the motor pole-passing frequency (PPF). (PPF = motor slip x number of poles.) The difference in amplitude between the FL and the FP is an indication of rotor health. Empirical research has shown that a difference of over 54 decibels indicates a healthy rotor while less than 45 decibels indicates a degraded (i.e., highresistance joints, cracked or broken bars) condition exists

On-Line Diagnostics Handbook — Volume 2

15

in the rotor. An example of CSA showing a damaged rotor is shown in Figure 2.

Figure 3 – Current Spectrum of a Motor with Air-Gap Eccentricity Figure 2 – Spectrum of a Motor with Damaged Rotor

Stator-Rotor Air-Gap Eccentricity Air-gap eccentricity describes the measurable distance between the stator and rotor within the motor. Manufacturers take great care to ensure that air-gap eccentricity is kept to a minimum. Typical maximum levels for large induction motors are between five and 10 percent. There are two types of eccentricity: static and dynamic. Static eccentricity is when the minimum air gap is fixed in space, such as when the rotor is misaligned along the stator bore. Dynamic eccentricity describes the condition when the minimum air gap revolves with the rotor. A bowed rotor results in a dynamic eccentricity. If the distance between the length of the stator bore and rotor is not equal throughout the entire circumference, varying magnetic flux within the air gap creates imbalances in the current flow, which can be identified in the current spectrum. The effect of this condition is seen as multiple sidebands of odd harmonics of the line frequency powering the motor. These sidebands will develop around the eccentricity frequency (FECC). FECC = (# of rotor bars) x (RPM/60) Figure 3 shows CSA results from a motor operating with an excessive air-gap eccentricity. The more severe the eccentricity becomes, the more the amplitude of the peaks will increase.

Mechanical Components Developing the ability to condition and filter the current signal passing through a motor’s windings expands CSA to detect load variations related to mechanical processes. The term used for this process is demodulation. Demodulation is the process by which a signal is recovered from a modulated carrier. Modulation is the process by which some characteristic of a carrier (the 60 hertz applied to the motor) is varied (the rotor flux creating CEMF) in accordance with a modulating wave. Simply put, the load

variations that repeat at a constant frequency are reflected into the stator currents through the motor’s CEMF. Remove the 60-hertz signal and these frequencies become apparent. Removal of the 60-hertz portion of the signal (demodulation of the carrier frequency) reveals repetitive load variations for analysis. PdMA is currently using amplitude demodulation of the current signal to greatly expand the capabilities of the Emax tester. Using a software-driven mode of demodulation to remove the 60-hertz signal, the ability to detect motor speed, pole pass, mechanical pass-through, and reflected frequencies is greatly enhanced. These mechanical and reflected frequencies are related to load variances from items such as belts, gears, pumps, fans, and other mechanical components. To evaluate the magnitude of these frequencies, an FFT is performed on the demodulated signal resulting in a spectrum for analysis. Without the demodulation, many of these load-related frequencies are buried in the signalto-noise ratio of the captured data. The following examples of using motor-current demodulation to evaluate equipment condition are from a large public aquarium. The data was gathered using PdMA Corporation’s MCEmax motor tester with Advanced Spectral Analysis. When using demodulated-current analysis to monitor mechanical components, it is important to establish a baseline when the equipment is known to be in satisfactory condition. After identifying frequencies related to specific components and conditions, any significant increase in amplitude should be investigated. Rotor Unbalance/Misalignment: The number of poles, the power system frequency and, to a lesser extent, the motor load determine the speed of an ac induction motor. A twopole ac induction motor being powered by a 60-hertz line frequency runs at a speed slightly lower than 3,600 revolutions per minute (RPM) or 60 cycles per second (hertz). A four-pole motor runs at a speed less than 1,800 RPM or 30 hertz, and so on. By utilizing current demodulation, the speed of the motor can be identified by a peak in the spectrum and monitored for changes in amplitude. A properly balanced and aligned motor has a frequency peak related

16 to its speed that is barely visible. When the motor is out of balance, or misaligned, this peak’s amplitude will increase. As the condition increases in severity, multiples of the speed frequency develop in the demodulated-current spectrum. Figures 4 and 5 demonstrate the change in amplitude of the running speed and 2X running speed during a precision alignment of a pump and motor.

On-Line Diagnostics Handbook — Volume 2 disappeared and how much lower the amplitude of the belt frequency is after the work has been completed. These frequencies can now be easily monitored detecting possible problems developing in the belt drive of this system.

Figure 6 – Demodulated-Current Spectrum Prior to Belt Alignment Figure 4 – Demodulated-Current Spectrum Prior to Alignment

Figure 7 – Current-Demodulation Spectrum after Belt Alignment Figure 5 – Current-Demodulation Spectrum after Alignment

Belts: When transmitting power to the load via a belt attached to the motor, changes in alignment can be evaluated using the demodulated-current spectrum. Evaluation of the current spectrum is similar to alignment in that increases in the amplitude of the belt frequency and the development of multiples of the belt frequency indicate a problem. To calculate belt frequency requires the operator to know the diameter of the pulley mounted on the motor and the length of the belt. Belt Frequency = 3.142 (D/L) x (RPM/60), where D is diameter of the motor-mounted pulley L is the length of the belt RPM is the motor speed In the following example, the dramatic change in the demodulated-current spectrum can be seen after proper tensioning and alignment was performed on a drive belt. In Figure 6, the belt frequency is 8.188 hertz, and there are elevated peaks at multiples of the belt frequency. Notice in Figure 7 how the multiples of the belt frequency have

Fans/Centrifugal Pumps: Fan blades and centrifugalpump vane frequencies can be monitored in a demodulated-current spectrum at a frequency equal to the number of blades (or vanes) times the FP. Increasing amplitude at this frequency as well as a possible increase at the motorspeed frequency peak is an indication of possible blade or pump-vane damage. After initial installation or verification that the pump or fan is in satisfactory condition, identify the vane frequency and record the amplitude of the peak. With baseline amplitude for the equipment established, the demodulated-current spectrum is used as a simple and efficient method to monitor the equipment. Figures 8 and 9 are a comparison between two identical horizontal pumps. Figure 8 is typical for this application with the pump-vane frequency amplitude of 0.027 decibel. In Figure 9, pump PF-8.6A pump-vane frequency amplitude is 0.046 decibel – nearly double that of all the other identical equipment platforms. Additional testing was performed on pump PF-8.6A, and it is currently scheduled for inspection of the impeller.

17

On-Line Diagnostics Handbook — Volume 2

Figure 10 – Current-Demodulation Spectrum of Motor/Pump with Loose Foundation

Figure 8 - Typical Current-Demodulation Spectrum for Several Identical Pumps

Figure 10 is a demodulated-current spectrum from an induction motor powering a pump with possible loose mounting bolts. When the motor is properly mounted to its foundation, this peak is usually not even visible. Having identified this peak in a demodulated-current spectrum indicates that a thorough inspection of the motor’s foundation is warranted. If the condition worsens the amplitude of the frequency peak will increase. Figure 9 – Current-Demodulation Spectrum of Pump with Suspected Impeller Damage

Loose Motor Foundation: Loose foundation bolts, soft foot, or a distorted bedplate can lead to air-gap eccentricity in a motor. Over time an improperly mounted motor can develop a distorted frame due to thermal expansion and contraction as the motor heats up and cools down. Additionally, a loose motor foundation will make it all but impossible to maintain correct alignment with the powered load. Loose motor foundation can be detected in a demodulated-current spectrum by an elevated peak at half of the motor’s running frequency. If the amplitude of this peak is increasing over time, the condition of the motor’s foundation, mounting bolts, and shims should be investigated.

Motor Bearings: A roller bearing will have a set of unique defect frequencies, and current demodulation utilizes these frequencies to evaluate a bearing. Frequencies are based on the size and design of the bearing. These frequencies are monitored for possible defects in the inner race, outer race, ball (roller), and cage of the bearing. Calculating the inner race, outer race, and ball frequencies uses the formulas listed in Table 1. The formulas for calculating frequencies are a little imprecise because axial loading and slippage affects them in an unpredictable manner. In order to ensure that the dimensions and contact angle are correct, the technical documents from the bearing manufacturer need to be reviewed or, if no documents are available, the manufacturer needs to be contacted. Bearing manufacturers can provide these frequencies for each bearing they manufacture.

Table 1 BPFI = Ball Pass Frequency, Inner Race n BPFI = — 2

Bd 1 + — cos Pd

RPM

FTF = Fundamental Train Frequency 1 FTF = — 2

Bd 1 + — cos Pd

RPM

Bd = Ball diameter n = Number of rolling elements

BPFO = Ball Pass Frequency, Outer Race n BPFO = — 2

Bd 1 – — cos Pd

RPM

BSF = Ball Spin Frequency Pd BSF = —— 2Bd

1–

Bd — Pd

2

(cos

Pd = Pitch diameter of the bearing = Contact Angle

RPM

18 However, good approximations of bearing frequencies for most common ball bearing are as follows: Outer race fault = (# rollers) x (RPM/60) x (0.4) Inner race fault = (# rollers) x (RPM/60) x (0.6) Fundamental train frequency = (0.4) x (RPM/60) Slippage and contact angle variances still need to be accounted for. Because of these variances, actual bearing frequency could be slightly higher or lower. When setting up to monitor these frequencies using a demodulated-current spectrum, ensure the envelope or band is plus and minus 10 percent the estimated bearing frequency. It is important to understand that, although one can monitor these bearing frequencies utilizing a demodulated-current spectrum and that an increase in amplitude is cause for investigation, detailed analysis to determine actual bearing condition should be performed with a vibration analyzer.

Conclusion Analyzing a motor’s current can effectively improve the efficiency and effectiveness of any maintenance organization. As more empirical data from the field is gathered, it is becoming clearer that mechanical components can be monitored through the use of a motor’s power leads. Using current analysis in conjunction with other predictive maintenance equipment can lead to significant savings in cost by reducing the man-hours spent on collecting data. Current analysis can be used to monitor belts, gears, alignment, and other mechanical components. Use these new features in current analysis to bridge the communication barrier between the mechanical and electrical departments relating to vibration and electrical analysis.

References 1. White, G. D., Introduction to Machine Vibration, DLI Engineering Corporation, 1997.

2. Schoen, R. R., Kamram, F., Habetler, T. G., and Habetler, R. G., “Motor Bearing Damage Detection Using Stator Current Monitoring,” IEEE Transactions on Industry Applications, Vol. 31, No. 6, November/December 1995.

3. Azovtsev, A., and Barkov, A., Rolling Element and Fluid Film Bearing Diagnostics Using Enveloping Methods, VibroAcoustical Systems and Technologies, Inc. 4. Jones, R. M., “A Guide to the Interpretation of Frequency and Time Domain Spectrums,” SKF Condition Monitoring, Revision 1, February 19, 1009.

On-Line Diagnostics Handbook — Volume 2 5. Thomson, W. T., “A Review of On-Line Condition Monitoring Techniques for Three-Phase SquirrelCage Induction Motors Past present and Future,” The Robert Gordon University, Schoolhill, Aberdeen, Scotland. Pete Bechard, a native of California, has been living in Tampa, Florida, since retiring from the United States Navy six years ago. He graduated from Columbia College with a degree in Business Administration. During eleven years of his Naval career, Pete was the lead supervisor for several electrical workshops and departments responsible for the maintenance, repair, and operation of both ac and dc rotating equipment. He also spent three years working as a quality assurance officer for nuclear submarine repairs conducted at the Pearl Harbor Submarine Base. During his time with PdMA Pete has completed formal qualification requirements as an instructor/facilitator through Langevin Learning Services. He has also completed formal training in Servo motor operation and repair and harmonic current/power quality in industrial distribution systems. His travels with PdMA have sent him as far as Singapore for motor reliability workshops and as close as The Florida Aquarium in Tampa where PdMA provides MCEmax services as part of their predictive maintenance program.

19

On-Line Diagnostics Handbook — Volume 2

Experiences in the Monitoring of Partial Discharges PowerTest 2005 (NETA Annual Technical Conference) Claude Kane Eaton Electrical Predictive Diagnostics

In today’s competitive environment, increasing demands are being placed on the management of physical assets. It has become imperative to capitalize on advances in technology that allow new approaches to the maintenance of these physical assets. These include reliability-centered maintenance, predictive diagnostics, condition monitoring and expert systems. Concerned customers and suppliers are taking advantage of the convergence of these new technologies to implement proactive maintenance programs to improve the performance and extend the life of their installed base of equipment. The electrical industry has a very limited number of “predictive tools” and those available are generally labor intensive and require some level of expertise. One tool is the continuous monitoring of medium and high voltage equipment for partial discharges. This includes motors, generators, switchgear, bus duct, cables and transformers. For over fifty years companies have been performing partial discharge measurements on electric equipment in order to obtain information as to the quality of insulation on operating equipment. Partial discharge activity is a well-accepted indicator of insulation deterioration. Consequently, trending of partial discharge activity provides equipment owners the opportunity to reduce forced outages and increase productivity of plant equipment. Traditionally, periodic testing is performed, but advances in technology allow the continuous monitoring of partial discharge activity.

nologies to implement proactive maintenance programs to improve their company’s bottom line. One such condition based maintenance technology is the measurement, monitoring and analysis of an electrical phenomenon know as Partial Discharge (PD), a wellknown and accepted precursor or indicator to the failure of insulation systems in higher voltage electrical equipment. Although it has been successfully applied to equipment rated as low as 2,300 volts, customer value is low. Value is created at voltage levels of 4,000 volts and above. Some examples of defects found by the measurement and analysis of partial discharges are shown in Figures 1 through 4. In all cases the equipment were tested while in service while under normal; operating conditions.

Introduction Advances in technology are allowing new approaches to maintenance. These include reliability-centered maintenance, predictive maintenance, condition monitoring and expert systems. Trend setting organizations are increasingly taking advantage of the convergence of these new tech-

Figure 1 — Corona Damage on end turns of 6.9 kV motor caused by contamination and improper spacing between coils

20

On-Line Diagnostics Handbook — Volume 2 Traditional Approach of PD Measurements Traditionally, PD measurements have been performed on equipment on a periodic basis, one to four times per year. The periodicity is based on the analysis of the most current reading, trend analysis, availability of resources, etc. Typical monitored equipment includes motors, generators, switchgear, cables, bus duct and transformers. Some of the deficiencies in performing periodic PD measurements as well as periodic off-line tests such as insulation resistance and over potential testing are:

Figure 2 — Coil insulation deterioration due to loose coils in a stator slot. Significant PD levels were generated due to poor quality rewind of stator

1) Time duration between tests are typically six to twelve months (if on-line) and frequently much longer for offline tests. Many problems will manifest themselves in a much shorter period of time. 2) During the performance of a periodic test a problem is detected, one starts to ask the following questions: • When did the problem start?

• How fast is it degrading? Not only the velocity of change, but also the acceleration of change. • How fast will it continue to degrade?

The true answer is — one does not really know! In a periodic mode one will need, at least, two consecutive tests to establish a rate of defect development, which could mean another year of testing. Therefore, one does not have enough qualitative and quantitative data to make a proper judgment at the time of problem detection, so arbitrary decisions are made. Figure 3 — Tracking on a shutter in switchgear. Close to having a phase to phase fault

Figure 4 — Deterioration of main bus insulation at the bus supports

3) In many cases PD activity is unstable. It may not be active today, but active tomorrow. Wide variations of PD activity are frequently observed based on data from continuous monitoring. Many factors can affect PD activity. A few include: • Voltage • Load • Temperature • Humidity • Vibration • Pressure As more and more data is collected in conjunction with other dynamic factors, much is being learned. Correlation of the certain dynamics with PD activity provides additional insight and can improve diagnostics. Most accepted PD standards focus on off-line partial discharge tests. OEM’s test labs commonly have their own test specifications and acceptance levels vary from OEM to OEM and are specific for specific insulation and equipment. Most PD standards dealing with on-line PD measurements avoid the stating of specific levels or quantities to determine acceptability limits. Instead, trending is stressed.

21

On-Line Diagnostics Handbook — Volume 2 In order to perform proper trend analysis, one should make sure all dynamics are nearly identical at each testing interval. This is very difficult, and sometimes impossible, to accomplish and can be time consuming. At this time, knowledge on correction factors to account for the change in dynamics is more academic than practical. Another reason for the avoidance of establishing levels is that even a very low level of PD activity that is trending upward will be indicating the development of a defect. Conversely, a high level of PD activity that is stable should be of concern, but at least one knows that the problem or defect is not getting worse, which should provide some level of comfort. 4) Time consuming. Taking periodic readings is labor intensive and frequently requires an expert to take and analyze the data to extract valid information. Maintenance organizations that are being driven to do more with less are looking for ways to increase reliability with fewer resources. The only true way to reduce maintenance costs is to take the "work" out of it, which is the essence of condition-based maintenance.

Continuous Monitoring of PD Activity A continuous monitoring system will overcome all of the deficiencies listed above and will be an effective condition based maintenance tool: Advantages include: 1. Finding a problem quickly. The monitor will identify a problem in its earliest stages of defect development. This will provide one with sufficient information as to the growth rate and the severity of the defect. 2. Providing information as to which phase the defect is in and generally what type of defect such as: • Corona

• Surface Discharge

• Void type of defect (Insulation Delamination)

• Conductive tape deterioration (Slot Discharge) • Loose high voltage connection that is arcing

It is even possible to localize even further the location of the defect. 3. Since no labor is required to perform the tests, continuous monitoring allows the use of limited resources to finding solutions to problems instead of finding problems.

6. Requiring no outage to perform the test, therefore there is no loss of asset productivity. 7. No introduction of infant mortality failure patterns via more invasive testing procedures. 8. Reduction of forced outages and increased safety of personnel. One will always be aware of conditions and/or problems. 9. Correlation of other dynamics such as temperature, humidity and load current to PD activity, which provides additional insight for diagnostics. There will be no need to go to several sources and gather the information. 10. Provides the opportunity for remote diagnostics. The expert does not need to come out to the field for basic diagnostics. A site visit by an expert will be the exception and not the rule. This can be done my emailing data to an expert or perhaps have a modem connected to the monitor so the expert can dial-in and upload the information for analysis and even provide a special test if necessary. 11. Evaluation of a piece of equipment is based on its own history and not by comparison to other equipment. This will make the detection of subtle problems easier. 12. Easily monitor worsening conditions so one can defer or accelerate repairs and allow time to plan an outage.

Case Study A Canadian pulp and paper plant has two identical 49MVA, 13.8 kV turbine generators. One machine shows very little PD activity. The second machine has similar low PD activity except that the C-phase sensors showed elevated levels. Both machines are monitored on a continuous basis. Gradual increases in phase to phase (C – A) PD activity were observed over the past year on the C Phase sensor. Starting mid March 2004, increased PD activity was then observed from the C phase sensors Figure 5. The first generator was scheduled for an outage in early May 2004 based on the manufacturer recommendations suspecting loose wedges. Based on the continuous monitoring of partial discharge activity, the first generator showed

4. Reducing unnecessary maintenance because the monitor will be constantly testing and will have accurate data on which to base decisions. 5. Collecting more accurate data as tests are conducted under real operating condition

Figure 5 — Partial Discharge Intensity (PDI) and pulse repetition rate trend for almost one year on a 49 MVA generator.

22

On-Line Diagnostics Handbook — Volume 2

this to be of little concern. But with the increase of PD activity on the second generator, it was decided to inspect both units during this outage. During visual inspection the first generator, no loose wedges or signs of PD activities were observed. On the second generator, initial visual inspection also showed no indication of PD activity. Off line PD tests were then performed on the second generator by applying phase to ground voltage on each phase with the other phases grounded. Little to no PD activity was observed. Based on the fact that the on-line PD activity was observed to be phase to phase type discharges, it was decided to bring in a second HV source. One HV source was connected to Phase A and the other to Phase C in order to simulate phase to phase voltage stress. Upon application of this unique test method, significant PD activity was observed. At the same time a second visual inspection was recommended with specific focus on the A and C phase connection ring bus. Severe insulation degradation was found between the A and C phase connection buses hidden in a very restricted viewing area. As shown in Figure 6, the defect is due to improper spacing between the A and C phase connection ring bus.

Figure 6 — Signs of discharge activity between the A and C phase connection ring bus.

To repair this type of defect, a major repair would be required. Due to time constraints, temporary repairs were performed allowing another one to two years of additional operation before a major repair is required. PDI and pulse repetition rate significantly decreased as shown in Figure 8 while PD magnitude only dropped 20-30%

Summary This article has reviewed the features and advantages of a continuous monitoring system for partial discharge activity on a wide variety of equipment. A continuous monitoring system can save the user a considerable amount of time and money and provides more accurate data allowing for more informed decision. Claude Kane has over 35 years of experience in the installation and maintenance practices for power distribution, transmission, and generation equipment. He started with Westinghouse in 1972 as a field service engineer in Kansas City and has held a number of technical and management positions throughout his career. His last 10 years have been spent developing technological products and emerging markets for predictive diagnostic and prognostic equipment. He is currently one of the principal owners of Electrical Diagnostic Innovations, Inc. based in Minneapolis.

23

On-Line Diagnostics Handbook — Volume 2

Partial Discharge Testing in Cables PowerTest 2005 (NETA Annual Technical Conference) Craig Goodwin HV Diagnostics Inc.

Abstract On-site partial discharge measurements have historically been carried out on various electrical apparatus such as switchgear, motors, generators etc. The application of partial discharge testing for cables has traditionally been left to research laboratories and cable manufacturers in specially designed noise-free screened test bays, often referred to as Faraday cages. Practical, technological and economic limitations prevented its use as a viable diagnostic testing tool in the field. What many people once thought to be a technical impossibility, partial discharge diagnostic testing of cables in the field, is today, a real and effective maintenance tool which when correctly applied, offers tremendous advantages to the electrical industry. This paper provides an overview of partial discharge in medium voltage cables, briefly highlighting some of the current diagnostic measuring methods that are being employed in the field and some of the limitations of partial discharge diagnostic testing.

Introduction The main objective of any diagnostic test is to identify a defect or defects in a non-destructive way, so that correct preventative action can be taken before the defect(s) results in an in-service failure. Partial discharge (PD) diagnostic testing of cable systems in the field is intended to measure and locate those singular or multiple clusters of PD producing defect sites that would or could result in a near term cable failure. Cable systems refer to the cable itself and connected accessories such as joints and terminations. The results of PD diagnostic tests are used to assess the condition of the cable system.

Figure 1 — PD cable test data showing different PD producing defect sites. A PD cluster and a PD singularity are clearly seen

What is Partial Discharge in Cables? The IEC/IEEE defines PD as a “localized electrical breakdown that only partially bridged the insulation between conductors and which may or may not occur adjacent to a conductor.” In effect they are very small “electrical sparks” or discharges that occur in or on the surface of the cable insulation system and do not as individual incidents cause a complete collapse of the applied voltage. Partial Discharges occur as singular events of very short duration (few nano seconds) and they generate light, sound, heat and electromagnetic pulses. The unit of measurement for partial discharge amplitude is the Pico Coulomb (pC). Another “unit” that is often used to measure PD is milli-volts (mV). Sources of PD in extruded cables include breakdowns in voids, cavities, along an interface, between an energized electrode and a floating conductor, in an electrical tree etc. The picture below (Figure 2) shows a pending cable failure that occurred when PD activity between the outer semi

24 conducting shield and a corroded neutral wire resulted in pitting through the shield and then finally through the insulation of the cable. Sources of PD in PILC cables include breakdown from dry, brittle or cracked paper, voids, cavities, moisture ingress, carbonized tracks, waxing of liquid etc. The characteristics of a PD depend on several factors including the location, size, type of material, applied voltage, temperature, and can vary with time.

On-Line Diagnostics Handbook — Volume 2 Figure 3: The sequence of diagrams below illustrates the propagation progression of the traveling waves created by a PD event in cable in the time domain. The acquisition of the PD signature by a PD detector is also shown.

Source of PD

Near End (Test End)

Far End

Cable PD Detector

2

1

Source of PD

Near End (Test End)

Far End

Cable PD Detector

3

2

Figure 2 — An XLPE cable with a corroded neutral showing pitting marks that were caused by Partial Discharge activity

For a partial discharge to occur, the localized field stress at the site of a defect must be sufficient to exceed the inception stress, also known as the PDIV or partial discharge inception voltage. Once this inception voltage has been met, repetitive partial discharge activity will normally continue until the applied voltage is decreased to a level known as the PDEV or partial discharge extinction voltage. This PDEV level is typically 65 to 75% of the PDIV level. Therefore defects that have Partial Discharge Inception Voltages close to the operating voltage of the cable system, are more likely to be initiated by an over voltage transient like those generated by switching or lightning for example. Even after the transient has subsided, the normal system voltage can be sufficient to sustain these Partial Discharges, resulting in a possible cable failure. When a PD occurs in a cable, a traveling wave is created at the source of the PD site by the electromagnetic pulse. One half of the traveling wave sets off towards the one end of the cable – call it the “near end” and one half travels towards the other end – call it the “far end”. If special PD sensors are placed on the cable system, these PD pulses can be detected using special PD detectors. Please refer to Figure 3 below, which shows a graphical representation of the sequence of events that result from a Partial Discharge in a cable insulation and the measurement of these events with a single ended Time Domain Reflectometry (TDR) type detector.

Source of PD

Near End (Test End)

Far End

Cable PD Detector Reflection at far end

4

3

Source of PD

Near End (Test End)

Far End

Cable PD Detector

5

4

25

On-Line Diagnostics Handbook — Volume 2

Source of PD

Near End (Test End)

Far End

Source of PD

Near End (Test End)

Far End

Cable

Cable PD Detector

PD Detector

Triggering of PD Detector and Reflection at Near end

15

6

5

10

Source of PD

Near End (Test End)

Far End

Source of PD

Near End (Test End)

Far End

Cable

Cable

PD Detector

PD Detector

18

7

6

11

Source of PD

Near End (Test End)

Far End

Source of PD

Near End (Test End)

Far End

Cable

Cable

PD Detector

PD Detector

Distance from far end = time • pulse velocity / 2 Distance from near end = cable length – distance from far end

9

7

12

Source of PD

Near End (Test End)

Far End

Cable PD Detector

Reflection on Near end

10

8

Source of PD

Near End (Test End)

Far End

Cable PD Detector Reflection on far end

12

9

19

As can be seen from the sequence of diagrams above, the PD detected by a PD detector connected to one end of the cable, will “see” the apparent, also know as the “virtual” image of the originating PD. As the pulse travels down the cable, attenuation will occur. The original pulse created by the partial discharge will therefore flatten and widen as the signal is attenuated. For a given PD source, the virtual charge seen by the measuring equipment is therefore proportional and not equal to the amount of charge at the actual PD site. The PD sensors used to detect these PD initiated traveling waves are normally capacitors or inductors that are connected either externally or internally on the cable system. Capacitors are generally coupled directly to the cables conductor at one or both ends of the cable. Capacitive coupling can also be achieved by connecting sensors to the semi-conducting shield of cables and accessories – normally done by the manufacturer and mainly found in transmission cables. Inductive sensors can be placed around the cable at various locations along the length of the cable.

26 The repetition rate of PD can vary from very intense, to a seldom and sometimes random occurrence. The repetition rate is defined as the ratio of the total number of PD pulses coming from a specific PD location in a specified time interval and the duration of this time interval. The PD pulse(s) can also have a relationship to the phase angle of the applied voltage. This can cause a pattern of Partial Discharge to occur relative to the phase of the applied voltage. This phase resolved pattern is often used to gain a better understanding as to the severity of the PD.

Figure 4 — Phase Resolved PD Pattern showing phase relationship to applied voltage and the PD intensity

Electrical Noise As mentioned earlier, the signal levels produced by partial discharges that are detected during a PD measurement are very small and become even smaller as they travel and are attenuated by the cable. Unfortunately, electrical noise that is usually endemic to the field environment can also occur in various magnitudes in the same frequency range as these PD signals. Typical sources of this noise include AM radio stations, speed control motor drives, switch mode power supplies etc. Although this electrical noise does not originate in the cable system, it can inadvertently be detected and measured by the PD detection equipment. Therefore noise filtering techniques need to be employed to try and “pluck” out the PD signals from the noise signals. The PD signal typically has to be above the ambient noise values, to successfully trigger the PD detection system. This needs to be done carefully without also blocking out the PD signals. Too much filtering can falsely indicate a cable that does not have any PD producing defects present, while too little noise filtering will result in a measurement where it may be difficult, if not impossible, to detect any PD activity. Good noise filtering is especially important on those cable insulations where even small amplitude PD signals can cause a cable failure to occur. This is especially the case with XLPE insulated cables where some serious PD producing defects may only be a few pC in magnitude.

On-Line Diagnostics Handbook — Volume 2 Brief Summary PD Measurement Methods Available There are two main modes of operation for performing a PD test in the field – either ON-LINE or OFF- LINE.

Off-line PD measurements require that the cable be deenergized (removed from the power system). The Cable is then re-energized with a time varying (ac) external voltage source so that the applied voltage to the cable can be adjusted to below and above the normal operating voltage. These external ac voltage sources are typically 60Hz resonant, Very Low Frequency (VLF) or Oscillating Wave systems. The maximum voltage that is usually applied is between 2 to 3 time’s the normal operating phase to ground voltage. Since the cable is taken off-line, capacitor sensors are often used in this method to detect the PD pulses. The biggest advantage of the off-line test is the ability to elevate the excitation voltage and therefore the electrical stresses applied to the cable system. This is similar to a treadmill stress test that is performed when a heart specialist wants to look for potential heart defects. The elevated stress can “expose” those pending defects that exist in the cable that may be excited into PD activity by a high voltage transient during normal operation. Both the PDIV and the PDEV can be measured and detected using this method due to the variable voltage. In addition, the characteristics of PD with applied voltage can be measured and detected. Another advantage of an off-line test is the ability to perform a PD calibration. A PD calibration test is usually performed by injecting a known voltage pulse into the cable, integrating the measured result based on the characteristics of the cable and measurement equipment, and then adjusting the gain of the system to read the correct PD value based on the known calibration value. The biggest disadvantage of an off-line test is having to switch the cable system out of service. There is also a risk of cable failure associated with any off-line test when the applied voltage is elevated above the normal operating voltage. The more the test voltage exceeds the normal operating voltage, the greater the risk becomes. There is an argument that if a cable cannot sustain 2 times its rated voltage, it is in very poor condition to start with and it is only a matter of time before it would fail in service anyway. In addition if a fault occurs during the test operation, the outage will occur in a controlled environment with no customer interruption. On-Line PD measurements are performed without removing the cable system from service. The excitation voltage comes directly from the power system and cannot be adjusted. No elevation of the applied voltage also means that no elevation in the electrical stresses in the cable can occur. As it is not possible to connect directly to the cable’s conductor, the PD sensors used are usually of the inductive type.

27

On-Line Diagnostics Handbook — Volume 2 Calibration of the on-line measuring system is not possible. The biggest advantage and most attractive feature of the on-line test is that the cable does not need to be switched out of service. PD activity can also be seen under various loading conditions. A test failure is very unlikely during an on-line test as the voltage remains constant. Most people would agree that an off-line PD test is a more thorough assessment of the cable system due to the ability of changing the applied voltage. However the relative ease of not having to switch out a cable is very enticing to many cable owners who are either not able or are not willing to disconnect the cable. Time Domain and Frequency Domain Measurements: There are also two main PD measuring detection methods that can be employed during a PD test. The first is the detection of PD in the Time Domain (as shown in Figure 3). The second detection method is performed in the frequency domain. Generally Off-line tests use the Time Domain approach and On-line methods use the Frequency Domain approach. More information on the above two methods is available, however it is beyond the scope of this paper.

Can Partial Discharge Detect Water Trees? Water trees are the primary aging mechanism for extruded medium voltage cables. They are the source of most insulation defects and failures in extruded cables. There is absolutely no partial discharge associated with water trees and therefore NO PD detection method can measure or locate water trees. Some commercial vendors of PD diagnostic equipment or PD diagnostic services wish this were not the case, but there are too many research papers to support this fact, and even though some of these vendors have promised to produce independent research to support their arguments, to the best of the author’s knowledge, none have ever done so. The best available methods for detecting water trees is the use of global assessment methods like tan delta / dissipation factor. There are many cases from the field where a cable shows very high dissipation factor results, indicating aged insulation, but no PD activity measured in the cable. When the cable insulation is later inspected in the laboratory, mature water trees are abundant and the insulation is often in an advanced stage of aging and deterioration. Although an electrical tree (that does generate PD) may emanate from a Water tree, the likelihood of detecting it during the short duration of a non-continuous on-line test is very small. Even with an off-line PD test, one may pick up the proverbial needle in the haystack when one of the thousands and thousands of Water Trees initiate an Electrical Tree, but odds are not good.

To highlight this point, an independent research facility called SINTEF conducted one of the few round robin “consumer report” type tests to evaluate the most effective diagnostic system for detecting water trees in aged XLPE cables. They never even considered partial discharge as a diagnostic technique.

What Partial Discharge Characteristics Are Typically Analyzed? The measurement approach of partial discharge diagnostics in a cable is often based on statistics and not just on one single event. The Partial Discharge activity builds up a multidimensional matrix of characteristics that can be analyzed. These include: • PDIV and PDEV of PD

• Phase Resolved Pattern Distribution of PD activity • Magnitude of PD

• Repetition rate of PD

• Pulse Shape and frequency of PD • PD Trending with time • PD Location

Is All PD Bad? The biggest challenge in PD diagnostics is no longer the technical ability to detect and measure PD in noisy field environments, but the interpretation of the PD test data. What is the severity of the detected PD in a cable system? The severity of PD is often dependent on the type and location of the PD producing defect, the type of insulating material, operating conditions like temperature, voltage etc. Not all PD will result in insulation failure. Some materials are known to be PD resistant or PD tolerant like PILC cable and accessories – they can tolerate PD for long lengths of time, without damaging the insulation; other materials are PD intolerant such as XLPE. A 20pC PD coming from an electrical tree in a XLPE cable will normally result in imminent cable failure, while a PILC cable would not be affected by such a small, insignificant partial discharge value.

Conclusion There is no doubt that Partial Discharge diagnostics is a tremendous condition assessment tool for the electrical industry for both maintenance and acceptance testing of cables. However the technical wizardry of detecting such small signals in a cable has often been oversold in the market. It must be remembered that not all defects produce PD and not all PD can result in failure. While PD diagnostics can generally detect a cable that is in poor condition, a cable with no PD is not necessarily a cable that is in good condition. A combination of PD with other diagnostics such as tan

28 delta is often the best approach to successfully address the condition of the cable system. As the knowledge base and field experience grows, so will the interpretation of PD test data. A lot of the PD interpretation techniques are therefore becoming proprietary and a closely guarded secret in this fast growing and maturing industry. Craig Goodwin is technical specialist at HV Diagnostics for Baur test equipment. He has spent the last 15 years working in the fields of cable testing, cable fault location, and cable diagnostics in both a field and a research environment. He has worked on cable projects with EPRI and is a working group committee member of the new IEEE400.3 guide for field testing shielded power cables. He is the author of several technical presentations made at PES – ICC and is an active member of the IEEE, ASTM and ICC. He has a BS in Electrical Engineering.

On-Line Diagnostics Handbook — Volume 2

THE “GO-TO” STANDARDS FOR ELECTRICAL SAFETY AND RELIABILITY

ANSI/NETA MTS-2011 - New Edition

Online Diagnostics Handbook

This standard should always be referenced when writing maintenance specifications or performing routine testing on electrical power systems.

Volume 2

The ANSI/NETA Standards for Acceptance and Maintenance Testing Specifications for Electrical Power Equipment and Systems!

ANSI/NETA ATS-2009 This standard should always be referenced in design specifications or when performing acceptance testing on power system installations.

ANSI/NETA MTS-2011

ANSI/NETA ETT-2010 This standard ensures that your acceptance and maintenance tests are being preformed by qualified technicians who are certified in accordance with ANSI/NETA ETT requirements.

Available in Bound, CD ROM, or PDF

MAINTENANCE TESTING SPECIFICATIONS FOR

ELECTRICAL

POWER

EQUIPMENT AND SYSTEMS

Order Your copy of the ANSI/NETA Standards Today!

Become a NETA Affiliate and Save! To order: visit www.netaworld.org Or call 888-300-NETA (6382)

Published by InterNational Electrical Testing Association

STANDARD FOR

Published by InterNational Electrical Testing Association

29

On-Line Diagnostics Handbook — Volume 2

Advanced Condition-Based Assessment of Medium and High Voltage Electrical Systems Without Requiring an Outage PowerTest 2005 (NETA Annual Technical Conference) Presenter R.R. Mackinlay, High Voltage Solutions Ltd Co-Author Don Genutis, Hampton Tedder Technical Services

Introduction Partial discharge (PD) is becoming increasingly viewed as the best diagnostic for insulation, particularly for on-line measurements. Clearly this applies to insulation, which exhibits and is degraded by PD activity. However, even for insulation, which is designed to be PD free, the knowledge that the system is actually PD free is still a vital part of the diagnostic process. Hence PD measurements, which are accurate and reliable, always contain important information about the high voltage (HV ) equipment to which they apply. With the development of on-line PD methods, several problems still exist to be solved to yield the optimum diagnostic for insulation. These are: • Noise interference from RF radio broadcasts

• Noise interference from pulses other than PD pulses

• Location of PD pulses on cables and static equipment (e.g. switchgear, transformers etc) • Estimation of remaining service life for the PD activity measured in the HV equipment under test

The remaining service life of equipment exhibiting PD activity is briefly dealt with for restricted types of equipment, otherwise this is left for a future paper. This paper addresses the main issue of making reliable PD measurements, namely the identification and removal of noise sources, and the location of PD events. To be able to do this allows the identification of discharging equipment, and hence the assessment of the risk to reliable service performance.

This paper describes the development of methods to distinguish between noise and PD activity, and to distinguish between several different types of PD. The recognition is done using simple algorithms, which use the waveform shape of the PD (and noise) pulses, to discriminate between the different types. Partial discharges (PD) in voids and cavities in general produce very similar pulse shapes with very fast pulse widths. Values of a few tens or hundreds of picoseconds are typical. The pulse times are closely related to the breakdown time across the cavity. However, at the terminals of the high voltage equipment (or using an electromagnetic pickup if these are used) the pulse shape is dependant on the parameters of the detection circuit. Often these are determined by the equivalent circuit of the power equipment at the detection point. The observer may not have much control over the detection circuit in these cases. Hence it might appear that using the waveshapes of the pulses to distinguish between noise and PD (and between different types of PD) does not have a bright future. However, there are two classes of PD pulses, which do retain a reasonable consistency of behavior, for both on-line and off-line testing. These are cable PD’s and PD’s from local equipment (switchgear etc). In this case, local equipment means equipment within a few meters of the measurement point. The algorithm used in the development described in this paper, essentially makes three categories of pulsed signals. These are cable PD’s, local equipment PD’s, and all others are regarded as noise. This simple algorithm gives remarkably good results in a wide variety of circumstances, and has now been successfully incorporated into several commercially available products.

30

On-Line Diagnostics Handbook — Volume 2

PD Events from High Voltage Cables In the special case of partial discharge in cables, the cavity responsible for the PD discharges into a real impedance. This is the surge impedance of the cable and is purely resistive at the point of launch. The resulting pulse is virtually monopolar (see Ref 1) with fast risetimes and very short pulse width. This pulse travels outward from the originating site, and arrives at the detection point wider and smaller due to attenuation and dispersion on its travel down the cable. The pulse is in general well defined by the time it is detected, and in practice retains much of the same characteristics as a consequence of originating in the cable. Figure 1 shows a typical cable pulse, with computer-generated cursors to measure the risetimes and pulse properties. Figure 2 — High frequency current transformers (CT’s) around cores of 33kV cable for PD measurement (CT’s are arrowed)

Figure 1 — Pulse from a partial discharge in a cable (showing computer generated cursors)

Providing the risetimes and widths fit into sensible values for a PD pulse, then the event is categorized as a cable event. The risetimes are typically between 50nSec and 1mSec, with pulse widths in general being less than 2mSec. Actually for XLPE cables these values are generally shorter than this, as the loss and dispersion are considerably less for XLPE cables. The pulse risetime and width are dependent on both the pulse shape at the cable end, and the detection circuit. Hence the neat method of using the risetime (or pulse width) as a measure of the location of the pulse cannot be rigorously used, as the detection circuit is not known, and for example, may contain a large inductance and always then give slow risetimes and longer pulses. However, the risetime is nevertheless a valuable tool in the initial localization of the pulses, as for on-line PD detection using high frequency current transformers (CT’s) the detection circuits generally have large bandwidth (>20MHz) and simple localization works reasonably well.

Figure 2 shows the current transformer arrangements for a 33kV XLPE cable, where the CT’s can be placed around each of the single cores, above the ground strap take off point. Normally the current transformers are placed around the ground strap. The PD pulses traveling down the cable to the termination, have an equal and opposite polarity on the conductor and screen respectively. Hence it does not matter whether the CT’s are placed in the earth strap, or the conductor. The important criterion is that only one of the ground or conductor currents is intercepted. In practice the two signals are very similar, but often the noise content can vary between the two detection points.

Figure 3 — Distribution of pulse risetimes from a cable

Figure 3 shows a cable with a spread of risetime events, where there are clear bands to the risetimes recorded. In principle, a graph could be constructed from figure 3, with the ordinate calibrated in meters. This would assume a relationship between the risetime and the distance traveled for PD pulses in cables. This is actually reasonably well established (depending on the cable type), but the main problem is the unknown detection impedance at the termination. For example, if the detecting impedance contained a large inductance, then the risetimes of the pulses would be dominated by the detection impedance, and not the distance traveled by the PD pulses. In cases like this, a relationship between risetime and distance cannot be made. The response

31

On-Line Diagnostics Handbook — Volume 2 of the detection circuit could be measured, and if possible, the relationship could be correctly established. This will be the subject of a future paper. The results in figure 3 were taken from a 33kV paper insulated cable tested on-line. Notice that the risetime of the PD events for this cable are around 200nSec. The cable was around 2km in length, and the locations were all from a single splice located at 1600m. The huge advantage of carrying out PD detection using the waveforms of the PD pulses is that it allows for correction of the magnitude of the discharge almost irrespective of the distance down the cable the pulse has traveled. Particularly for paper-insulated cables where the attenuation is large, making magnitude only measurements of PD pulses leads to poor conclusions about the peak magnitudes. A PD of the same size can easily look a factor of 10 or 20 smaller when seen at some distance down the cable. Hence a very severe event remote to the measurement site can be observed as quite a modest PD if the attenuation has reduced the peak magnitude by a factor of say 20 or so. Using the waveform to measure the area under the curve of the PD current, a measure of magnitude can be made which is much less sensitive to attenuation. The measurement of charge is made by simply integrating the charge under the current curve: Charge =

end_of_ pulse

I *dt =

start_of_ pulse

Precise Localization of PD Events on Cables Using a Transponder In the special case of testing high voltage cables, the usefulness of measuring PD activity is vastly increased if a localization of the PD origin can be made. For short cables, a double-ended method of timing the arrival of the PD pulses is the most effective. This requires a sensor at both ends of the cable, and a calibration of the cable travel time for both legs. For longer cables, getting a radio frequency cable to the far end sensor is not usually possible. Several methods have been attempted to overcome this problem. GPS sensors can be used to tie two PD recorders to the same absolute time. This method takes a lot of care is setting up, and requires a great deal of marrying up signals after the data is recorded.

end_of_ pulse

const *V * dt

start_of_ pulse

where ‘const’ is the multiplier which converts the current to voltage. This will include the transfer impedance of the current transformer and such factors as the cable impedance and amplifier gain. Hence using this method for calculating charge, the values of PD magnitudes can be measured in Pico Coulombs, provided that the detection impedance may be assumed to be the cable surge impedance, or a correction factor used. In practice the variation in the change of surge impedance at the ends tends to be less than 20% or so, when compared to measurements made in the ground straps of splices in the middle of the cable, where the surge impedance will be close to the surge impedance. As an example of measuring the charge in this way, at 3km on an 11kV paper insulated mass impregnated nondraining (MIND) cable, the area measure had only reduced by a factor 3. However, the amplitude measurement had reduced by factor of 15. This means that for on-line PD tests, without the need for any calibration, the cable events can all be expressed in pico coulombs.

Figure 4 — The trigger unit and large pulse generator of the transponder unit

Alternatively, the timing information from the remote end of the cable can be directed back down the cable cores themselves. A simple method has been adopted to achieve this. A transponder consists of a detector and trigger level for PD pulses, linked to a large voltage pulse generator, which launches pulses back onto the cable under test. Figure 4 show a photograph of the two battery operated devices. The trigger levels for the input device spans the range 1mV – 1V. The linked pulse generator is capable of generating 1mSec pulses at 200V into an open circuit. If high frequency current transformers are used as the detection and launching sensors, the system can be used successfully with cables of 2 or 3 km or more. The accuracy of the device is dependant on the risetime of the PD pulses, as the trigger must occur somewhere on the rising edge.

32

On-Line Diagnostics Handbook — Volume 2 These so-called transient earth voltage (TEV) pickups detect the electromagnetic pulses from inside the local equipment. They are essentially a capacitive coupling with the metal surface. This type of sensor is only suitable for local PD pickup, as the sensors do not pass low frequencies. Typical cutoff frequencies are in the region of 2MHz -5MHz, and signals must be above this frequency to be detected. Hence for local PD, if the frequency content is typically above this cutoff value, the chances are extremely high that the pulses are from PD’s originating in local equipment. The argument also applies to testing equipment with a high frequency current transformer as the pickup; as long as the upper frequency cutoff is above say 20MHz, so that the larger frequency content of the local PD pulses is observable. Figure 7 shows a typical resonant response from a local PD event. Notice the fast oscillations denoting the large frequency content. Figure 5 — Location of PD pulses with and without transponder location pulses

Figure 5 shows the results of carrying out the locations of a cable PD with and without the help of a transponder. In this case the location of the PD event is very close to the measurement substation, and the very large transponded pulse is clear. The total cable length was around 750m. The transponder developed for localization of cable PD’s is battery powered, allowing remote substation location measurements to be made, even if no local test power is available (a circumstance which is actually quite common). This method for localization has proved very successful largely as a result of its simple and direct application. If the PD pulses are clear, then the transponder method is very robust, and locations are always clear and unambiguous.

PD Events from Local Equipment For on-line testing local equipment (switchgear, transformers, bushings etc) for PD activity, the most effective sensor is an electromagnetic pickup, placed on the outside of the grounded metalclad surface. Figure 6 shows a typical TEV probe.

Figure 6 — Transient earth voltage probes (TEV’s or capacitive pickups)

Figure 7 — Waveform of local equipment event using a capacitive sensor

The algorithm for recognizing a local PD event is therefore very simple. If the frequency content of the event is above a critical cutoff, (somewhere in the region of 2MHz to 5MHz) then the event can be counted as a local event. This method cannot apply to ultrasonic signals and sensors, all of which have much smaller frequency content. Other methods apply for the ultrasonic case. It is reasonably surprising that such a simple algorithm for recognizing local PD events is effective. However, pulses with larger frequency content which are not PD related, are in practice quite rare. The authors have seen some local processing equipment (intelligent switchgear monitors) which produced similar high frequency pulses, but with this exception, the vast majority of pulses with larger frequency content tend to originate from PD’s within switchgear or similar equipment.

On-Line Diagnostics Handbook — Volume 2

33

Noise Reduction from Continuous Wave Interference

In practice for noise reduction of a waveform spanning the whole power cycle, the length of the segments which are noise reduced are typically 100 Sec - 200 Sec. Figure 9 shows waveforms before and after the noise reduction process. In figure 9 the reduction in the standard deviation is around 7.6 between the two cases. The waveform interference is reasonably sinusoidal, and the results would be expected to be good in this case, but similar practical cases also yield similar reduction ratios. The double pulse in figure 9 is now clear and can be used for location purposes, whereas before the noise reduction, the second pulse is not evident.

A large problem in PD detection comes in the form of radio broadcast interference. This is particularly the case for on-line measurements made on cables attached to switchgear which is not metalclad, although it can sometimes be a problem for metalclad switchgear. To overcome this major problem, a spectral subtraction method was developed. Again the ideas for this were simple, and are shown schematically in figure 8. Divide the longshot waveform up into suitable segments (typically 100-200uSec long) Calculate frequency and phase of largest amplitude RF noise for each segment Optimize the amplitude of the single frequency

Operation of Event Recognizer Method The implementation of the event recognizer algorithms starts with recording the complete waveform of a single power cycle of data. This is recorded at full sample rate (normally at least 100MSamples/sec). This ‘longshot’ recording is now RF noise reduced if required.

Subtract A*Sin( ) from the original signal for each segment

Figure 8 — Method for reduction of RF interference from waveforms

Essentially the largest frequency in the waveform is calculated, and this frequency is subtracted from the waveform, leaving all the pulse data unchanged, but without the interfering RF noise. As the RF radio broadcast noise is in general in the medium or short waveband, the carrier is in general amplitude modulated. Hence the best results are achieved if the length of the waveform is made over a time in which the modulation does not change much.

Figure 10 —PD distributions across the power cycle in separate categories of cable PD and noise

Figure 9 — Waveforms from before and after the application of the noise reduction process

The resulting waveform is then analyzed for all the pulse data, and separated into populations of cable, local equipment and noise events. These categories can now be treated separately if required. The pulsed data is now really a list of PD events with properties such as position in the cycle, channel number (if this applies), category of pulse, and all the pulse data from the curve fitting and frequency analysis. This per-cycle analysis provides for very powerful techniques for continuous monitoring and spot testing. The traditional data result-

34

On-Line Diagnostics Handbook — Volume 2

ing from PD measurements was a record of the peak and count values. In the new method, the peak and count data can be carried out for each category of pulses, and hence in essence for each type of high voltage equipment. Figure 10 shows the results of 2 minutes of data recording, which recorded around 6 power cycles of complete waveform data. In this case, the PD pulses were mixed with a large number of different noise sources. The event recognizer has separated out the good data from the PD recordings, so that all the cable PD events are correctly identified. This was verified by manual inspection of the individual waveforms. Figure 11 shows two typical waveforms representing noise and cable PD respectively. If only peak and count methods had been used, all of these pulses would count as valid data. Using the event recognizer technique, noise can now be distinguished from the cable PD events. The recordings were made on an oscilloscope with an open PC, so the event recognizer software can be run internally on the device which records the data.

Conclusions The paper has described several new tools which can help to make PD (partial discharge) detection for high voltage equipment much more robust and reliable, by removing two of the largest obstacles to making good PD measurements on high voltage equipment, namely: • Noise reduction of radio Interference from broadcasts and power line carrier signals • Separation of noise pulses from cable PD pulses and local equipment PD pulses The methods described follow simple algorithms, and produce reliable results in practice. All the older traditional methods of measuring peak, counts and distributions for PD activity have all been retained. The new event recognizer tools essentially sit as an add-on to the original data. However, the clarity and discrimination given by the new event recognizer methods will allow huge improvements in the quality of the PD data recorded. This especially applies to on-line PD measurements. For the case of high voltage power cables, a location method based on a transponder placed at the remote end of the cable is described, which allows localization of PD events on cables to be made. This is particularly useful for on-line PD measurements where it is notoriously difficult to interpret waveforms to give good location data. The tools have been developed into several commercial products, where the results from the users (not the authors) have been very encouraging.

References 1) Kreuger, F.H. “Partial discharge detection in high voltage equipment”, Butterworths, 1989.

Figure 11 —Waveforms of noise and cable PD respectively from the data in figure 10

This improves data transfer rate between waveform recorder and the PC by a factor of between 2 and 5 times. Hence by using this method, a large part of the classic difficulties of making PD measurements can be transformed by classifying the pulses into useable categories. This means that less experienced operators can have much more confidence in PD measurements, and that more experienced operators can carry out diagnostics very much more reliably from one set of data.

Dr. Ross Mackinlay is recognized as a worldwide expert in the field of partial discharge testing. Ross resides, and received his education in England, obtaining his Bachelor of Arts degree at Lancaster University, and his PhD at Oxford University. After working at Bangor University investigating laser plasma interactions, Mackinlay was employed at the Electricity Council Research Centre, performing electrical distribution research. In 1989, he became the head of Cables and Dielectrics for EA Technologies, and in 1993 served as Manager of Cables and Power Technology. Shortly thereafter, Mackinlay formed High Voltage Solutions to perform on line partial discharge testing, and to develop partial discharge testing equipment. Mackinlay has over 30 years of experience applying partial discharge measurements to energized cables and switchgear. He has made several significant contributions to on line diagnostic methods for cables, switchgear, and other electrical equipment and has published several papers on these subjects. Additionally, he teaches the partial discharge graduate course at the University of Manchester in England.

35

On-Line Diagnostics Handbook — Volume 2

What is Partial-Discharge Resistance? NETA World, Spring 2006 Issue by Ralph Patterson Power Products & Solutions, Inc.

Particle discharge (PD) is defined by IEEE as an electrical discharge that only partially bridges the insulation between conductors and that may or may not occur adjacent to a conductor. Partial discharges occur when the local electrical field intensity exceeds the dielectric strength of the dielectric involved, resulting in localized ionization and breakdown. Depending on the intensity, PDs are often accompanied by emissions of light, heat, sound, and radio influence voltage (with a wide frequency range). When voids are present in solid dielectrics and the electrical field is sufficiently high, air (or other gas) inside the voids ionizes and creates breakdown pulses across the voids. These pulses are referred to as discharges. Most soliddielectric insulations degrade under the presence of partial discharge and lead to premature failure of the insulation. However, some dielectric materials are relatively insensitive to PD activities. The ability of solid dielectrics to withstand voltage under the presence of partial discharge is called partial-discharge resistance.

Why is Partial-Discharge Resistance Important? In today’s triple-extrusion/dry-curing technology or with the PD factory test one may tend to believe that discharge resistance is not necessary for new cable. While it is true that today’s cables contain fewer apparent PD activities than their predecessors, there is no assurance that they are completely free of partial discharge. As stated in the Association of Edison Illuminating Companies (AEIC) specifications CS5-97, CS6-87, and CS7-93, the pass/fail criterion for the factory PD test is 5-pC (picocoulomb). A single 5-pC signal requires at least three 10-mil voids, nine 5-mil voids, 100 1-mil voids, or other equivalent void combinations to simultaneously discharge.

Higher test voltages enable smaller voids to discharge but do not necessarily increase the magnitude of PD. Regardless of how high the test voltage is, there is always probability that some voids in cable insulation will escape the detection of factory PD testing. These voids can discharge under normal lightning/switching surges.

Factory Partial-Discharge Test Procedure Since the gaseous by-product of curing takes time to diffuse out of the cable, the voids are pressurized for a period of time after cable extrusion. The detectability of partial discharge is decreased due to the gas pressure. In accordance with Insulated Cable Engineers Association (ICEA) publication T-24-380, the manufacturer shall perform a partial discharge test prior to shipment. The manufacturer shall wait a minimum of 20 days after the insulation extrusion process before the test is performed. The 20 day waiting period may be reduced by mutual agreement between the purchaser and manufacturer when effective degassing procedures are used. Each shipping length of completed cable shall be subjected to a partial discharge test. The partial discharge shall not exceed the values in Table 1. The test voltages for partial discharge measurements are listed in Table 2.

Table 1 Partial-Discharge Requirements Vt/Vg ratio

1.0

1.5

2.0

2.5

pC – Limit

5

5

5

10

ICEA S 108-720-2002 Vt = Test Voltage, Vg= Cable Line-to-Ground RatedVoltage

36

On-Line Diagnostics Handbook — Volume 2 Table 2

Test Voltages for Partial-Discharge Measurements Cable Voltage Rating (kV)

Test Voltages (Vt) in KV Corresponding to Vt/Vg Ratio 1.0

1.5

2.0

2.5

69

40

60

80

100

115

65

100

135

160

120

70

105

140

175

138

80

120

160

200

161

95

140

185

230

230*

135

200

265

N/A

* 230 kV class cables are partial discharge tested up to 2.0 Vg only. ICEA S 108-720-2002

Theoretically, spaces originally filled by water and/or water soluble materials also may become sites for PD if water diffuses away, leaving a gas-filled pocket, and the void shape is maintained. These spaces are not detectable in factory PD tests since there is no air or gas at the time of measurement.

Field Testing Discharge sites also may be introduced at splices or terminations during installation. The discharge may affect the surface of cable insulation in contact with the devices and cause failure in the termination or splice. A shielded power cable is composed of a conductor, conductor screen, insulation, insulation screen, a coaxial insulation shield, and a jacket. The coaxial shield causes the electrical field within the cable to be equally distributed between the conductor and the shield so that the insulation is equally stressed, thus allowing the minimum insulation thickness. When a cable ends, the shield must be separated from the core by a sufficient distance to reduce the electrical interaction between the two. Unfortunately, when the shield is ended abruptly, a destructively high electrical field is generated at the terminus. This stress must be relieved to insure the life of the installation. The surface of the insulation that lies between the conductor and the shield terminus is under an electrical stress and is susceptible to electrical discharge that causes erosion and chemical decomposition. This vulnerable area can be fortified by the choice of materials and/or the geometry. Cable preparation is extremely important to the reliable performance of any termination. When preparing the cable for the termination as per the manufacturer’s instructions, extreme care must be taken when removing the insulation semiconducting shield. The underlying insulation must not be nicked or cut and the semiconductive layer that remains on the cable must not be lifted up on the end. These occur-

rences introduce air voids into the system, which will cause premature failure. If the cable preparation is done correctly and the termination is installed as per the manufacturer’s instructions, the termination will perform. If testing in-service cables installed before the mid 1980s a PD occurrence may not necessarily indicate that the PD is from service aging. During the last 15 years cable has been manufactured with less than 5 pC discharge magnitudes. This was not always the case, as we evolved over the last 30 years to this plateau. In 1976 the AEIC cable committees introduced this standard. This prompted the triple extrusion process which was needed to meet this stringent requirement. Of additional importance is the newly-discovered relationship between partial-discharge activity and the watertreeing phenomenon, a degradation mechanism recognized as the main reason for many premature cable failures in extruded dielectric cables. Discharge activity may develop during the service life of cables regardless of the dischargefree design of cables.

In Conclusion Since discharge activities may exist in today’s cable as it is produced, installed, terminated, and aged in service, it is beneficial to consider partial-discharge resistance as a very important property for overall cable performance.

References AEIC CS5-94, Specifications for Cross-Linked Polyethylene insulated Shielded Power Cables Rated Through 46 kV AEIC CS6-87, Specifications for Ethylene Propylene Rubber Insulated Shielded Power Cables Rated Through 69 kV AEIC CS7-93, Specifications for Cross-Linked Polyethylene Insulated Shielded Power Cables Rated Through 138 kV. Insulated Conductors Committee, 117th Meeting Spring 2005 ICEA S-108-720, Extruded Insulation Power Cable Rated Above 46 Through 230 kV ICEA Publication T-24-380-1994, Guide for Partial-Discharge test procedure Ralph Patterson is President of Power Products and Solutions, a NETA Accredited Company located in Charlotte NC. His professional background includes working as a design engineer of transformers and as a specifying engineer of insulated conductors. He has more than 25 years in power engineering particularly in insulation diagnosis and evaluation of electrical distribution equipment. He serves on the NETA Standards Review Council and Board or Directors, is the NETA liaison for the IEEE Insulated Conductor Committees working groups and received NETA’s 2001 Outstanding Achievement Award.

37

On-Line Diagnostics Handbook — Volume 2

On-Line Shielded Cable Partial Discharge Locating — An Overview NETA World, Fall 2006 Issue by Don A. Genutis Group CBS

Introduction & Background

Partial Discharge in Cable Systems

First and foremost, all shielded cables and their terminations or splices should be partial discharge free, just as the cable was when it was manufactured and tested at the cable manufacturer’s plant. This is especially true for acceptance testing and as a basis for trending those existing service-aged facilities, which may have some partial discharge activity starting. Partial discharge is the partial failure of medium- and high-voltage insulation. As the insulation fails, it produces signals that can be detected using advanced technologies. These signals are a symptom, or Side Effect, that is produced by the partial failure of the insulation. Therefore, the detection of partial discharge in cable and other insulation provides an early warning of eventual or impending failure. Partial discharge testing can be performed on-line without requiring an outage and can be economically applied on all facility equipment in a survey fashion. By doing so, failures will be reduced, and reliability will be greatly enhanced. In addition to the obvious economic and convenience advantages of on-line partial discharge testing, research has indicated that on-line partial discharge measurements better represent the true condition of the insulation, since measurements are taken under actual operating conditions. This is due to a composite effect of several complex physical interactions within the operating cable system that are not present in its off-line state. The prime effects include service factors such as operating temperature and its effect on partial discharge via thermal expansion and contraction characteristics, material properties, mechanical loading influences, current-based power factor, and harmonic frequency effects. Additionally, long-term energization affects insulation characteristics, as the insulation may respond differently because it has been energized for a long contiguous time, as compared to initial short-term energization during an off-line test.

When evaluating the integrity of a typical cable installation, it is important to consider all elements of the cable system – terminations, splices, and the cable itself. If any of these components fail, power is lost and service is interrupted until corrective actions can be carried out. Reliability statistics indicate that approximately 90 percent of cable system failures occur at splices and terminations. This is likely due to workmanship defects developed from the inability to create flawless insulation specimens in the field under varying conditions by craftsmen of varying skill levels as compared to the repeatability of producing cable insulation in a controlled manufacturing environment that utilizes laboratory partial discharge testing for quality assurance purposes. Imperfections in cable system insulation can be caused by defects, contamination, poor workmanship, faulty installation, and many other problems. These imperfections create localized stresses in the electrical field which in turn create localized breakdowns that lead to eventual complete failure. These localized breakdowns create high-frequency pulses that can be detected and evaluated with specialized sensors and instruments. This technology continues to advance, and presently a great deal of information can be gathered by processing these pulses including the discharge magnitude, discharge power, number of pulses per cycle, pulse phase angle, and specific information related to each individual pulse including pulse width, risetime, frequency and other detailed information. This information can then be filtered, sorted, and classified to determine the existence of partial discharge activity, the level of danger that the discharge represents, the type of discharge that is occurring, and the location of the discharge.

38

On-Line Diagnostics Handbook — Volume 2

Locating Partial Discharge in Cable Systems It is very valuable to the facility manager to know if partial discharge activity exists on a specific feeder cable and the threat level that this activity represents. Additionally, it is also important to know where the activity exists so that corrective actions can be planned (prior to a complete failure) and taken in a nonpanic emergency mode. In order to effectively assess cable system integrity, partial discharge measurements are typically taken at locations where the cable shields are grounded. These locations usually exist at each termination and splice, but certain installation practices may not include shield grounding at the splice points. When shield grounding does occur at the splices, relative partial discharge location can be determined by the partial discharge pulse magnitude and frequency since pulse magnitude and frequency decrease as the distance from the partial discharge source increases. Therefore, locations along the cable system where the highest magnitude and highest frequency pulses are found indicate the most likely location of the partial discharge activity. Different approaches must be taken to locate partial discharge in cable systems that do not have grounded splice shield points, and these approaches can also be applied to confirm partial discharge location in cable systems that do have grounded splice shield points. Depending on noise levels, partial discharge pulses can be detected a thousand feet or more away from the partial discharge source. By applying special noise reduction filters to the partial discharge measurement system, it is usually possible to detect partial discharge in the longest runs of cable just from the terminations. Even though the length of the cable run and the level of background noise may adversely influence the accuracy of the test results, it is still worth attempting the test since, at a minimum, termination flaws can be identified. When partial discharge occurs in a cable system, two pulses of similar size and characteristics propagate away from the partial discharge site towards the terminations. Depending on the cable insulation type, shield construction, and other factors, the speed in which the pulses travel is relatively consistent. For instance, partial discharge pulses travel at a speed of approximately 468 feet per microsecond (142.65 meters / u-sec.) in XLPE insulation. As can be seen in Figure 1, the time difference between the first two pulses (the initial pulse and the reflected pulse) is noted as delta T1-T2. The two pulses will continue to travel up and down the cable. These two pulses are reflected at exactly the cable return time (delta T1-T3) from the original pulse set, creating two sets of pulses, each spaced at the cable return time, delta T1-T3. Then the partial discharge location can be calculated from the following formula:

Figure 1 — Partial discharge pulse train seen from the measurement end of the cable system.

Location from Measurement End (in % Cable Length) = 100*(1- ∆T1-T2/∆T1-T3) Under certain circumstances, such as an extremely long cable or noisy background conditions, the reflected pulse may fall below the noise threshold as it attenuates along the cable and may not be distinguishable at the measurement end. In those cases, it is possible to apply a remote pulse transponder to the far cable end. The transponder detects the reflected pulse as it reaches the far end and then injects a large pulse back on the cable shield that will easily be distinguished at the measurement end.

Figure 2 – On-line cable partial discharge mapping showing discharges from a splice located approximately 45% of the cable length from the measurement end and a secondary discharge occurring from the termination at the measurement end

In addition to performing hand calculations to determine partial discharge location, special software is available that calculates and plots partial discharge location as seen in Figure 2. This cable partial discharge mapping software can provide partial discharge location with more accuracy and more efficiently than hand calculations.

On-Line Diagnostics Handbook — Volume 2 Conclusion Partial discharge testing is a vital tool for determining the health of cable systems. This technology can be applied economically in an on-line survey fashion to provide excellent cable system condition assessments. This information can then be used to channel maintenance resources to the areas that require the most attention. Several advanced methods can be used to identify, quantify, and locate partial discharge, depending on the cable system construction and the partial discharge location accuracy requirements of the facility. Mr. Genutis received his BSEE from Carnegie Mellon University. He was a NETA Certified Technician for 15 years and is a Certified Corona Technician. Don’s technical training and education is complemented by nearly twenty-five years of practical field and laboratory electrical testing experience. He is presently serving as Vice President of the Group CBS Eastern U.S. Operations and acts as Technical Manager for their subsidiary, Circuit Breaker Sales & Service located in Central Florida.

Other Acknowledgements: The author would like to thank Harold Orum of Halco Service for his valuable “Devils Advocate” and “Sounding Board” assistance with this article. Harold is a NETA Affiliate and has accumulated more decades of electrical experience than he will admit to.

39

40

On-Line Diagnostics Handbook — Volume 2

Energy Spent on Installing Power Quality Monitoring Pays Off in Industrial and Commercial Environments NETA World, Winter 2006-2007 Issue by Wieslaw Jerry Olechiw Dranetz

the same boat, given how many machines are now governed by programmable logic controllers (PLC’s). Auto assembly lines are cases in point, with PLC-driven robots at virtually every phase of production. Power quality fluctuation can affect the ability of all these devices to work correctly; it can even trip them off-line and shut them down. Interestingly, computers are susceptible to power fluctuations more than any other device. Because of the ongoing struggle to produce computers at the lowest cost possible — which can subsequently result in highly competitive pricing at the commercial and retail level — manufacturers, over the years, have removed a capacitor here and there to save a few pennies. These small savings per unit, multiplied by the staggering number of units that are annually sold, add up very quickly. However, the quest for savings also results in a less robust device, one that is far more susceptible to power quality fluctuations. As a result, new SEMI-47 and ITIC sensitivity curves have been developed by industry to address these issues. (See Figure 1.) 3 ms

20 ms

50 ms

100 ms

200 ms

500 ms

1000 ms

10 seconds Steady State

100 87

90

Original

80 70

30 20 10

SEMI Original CBEMA

New CBEMA

40

al CB EMA C

50

urve

60

Origin

% of Equipment Nominal Voltage

You have just experienced a power outage in your house. It lasted just one minute, so it was over before you even noticed it. You see that the clock on your VCR is flashing, so you reset it. You see that the clocks on your microwave and oven are also flashing, so you take care of those as well. Other than the few minutes it took to reset these devices, you have come out of this power event relatively unscathed. Consider the same scenario at a chemical processing plant. You are processing a plastic formulation, and it is in the molten or liquid state. Then the power outage hits, and the entire line shuts down. The plastic stops flowing. Although the power comes back quickly, you have a mess on your hands. You have lost a large amount of costly raw material. Worse, you have to perform a full cleaning of the machine and the pipes to extract the suddenly rock-hard plastic. It’s a time-consuming, costly chore, and you can not start the line back up until the cleanup is finished, which means a serious loss of productivity. It is clear from these diverse examples that while there is seldom a reason for monitoring power quality in a residential setting, the need to do so in industrial and commercial settings is critical. If continuous voltage and current are not available, equipment might stop working, affecting the entire production process, The consequences, financially and otherwise, may be far-reaching. As machinery has become more sophisticated in recent years — specifically due to the widespread use of microprocessors — the ramifications of poor power quality have grown substantially. Even a minor drop in voltage can render the microprocessor ineffective. A vast number of industries depend on microprocessor-driven equipment. Data centers, with their high volume of computers, are certainly primary targets. Healthcare facilities, which rely on sophisticated diagnostic and treatment apparatus such as MRIs, X-rays, and dialysis machines, are another. Manufacturers are in

New CBEMA SEMI One cycle = 16.67 milliseconds @ 60 Hertz

0 1/2 cycle

3 cycles

12 cycles

30 cycles

60 cycles

Figure 1 — Voltage Tolerance Curves

On-Line Diagnostics Handbook — Volume 2 There are two types of power interruption. The worst, of course, is a blackout (like the 2003 power outage that encompassed a large portion of the Northeastern United States). At the other end of the spectrum is voltage sag which is defined by the IEEE 1159 standard as reduction in voltage between 10 percent and 90 percent below normal voltage levels for a duration between 0.5 cycle and one minute. In terms of consumer products, a 10 percent drop in voltage might cause the clock on your VCR to go out and then reset. Again, the consequences in this instance are relatively minor for a residential setting; however, they are far more devastating in an industrial setting. The need to monitor power quality in the industrial and commercial areas from the point of generation through the point of use is unquestionable. But the value of monitoring goes beyond reacting to a specific event. There is a proactive, preventive element as well. With proper monitoring, the user can examine what happened in the power circuit, determine where the problem occurred, and, with some analysis, pinpoint the source of the problem. Perhaps a large motor starts up and draws as much as six times full load during acceleration which reduces the overall voltage available and causes the equipment on the same circuit to succumb to the voltage sag. If that situation can be recognized through observation of past event information, the cause can be detected, and preventive measures can be taken. Possibly a reduced voltage starting method can be used, or critical devices can be removed from the motor circuit so they will not be affected by the sag. Voltage fluctuations can be mitigated by the use of an uninterruptible power supply. Surge protectors can be used to mitigate the effects of lightning strikes or surges. If it is determined that the problem is being caused by the electricity supplier, the supplier should be contacted immediately. Having evidence of a problem really helps in getting attention from the utility, which typically has ways of improving voltage quality through the use of voltage regulators and capacitor banks. Once the decision is made to engage in power quality monitoring, another issue must be addressed: whether to implement a portable monitoring or permanent monitoring system. While portable monitoring can be effective, it carries certain drawbacks: • If a power problem is suspected and a portable monitor is applied to the circuit to track the problem down, the procedure will be ineffective unless the problem occurs again while the monitor is in operation. It is not unlike bringing your car to a mechanic and praying he will hear the same noise you heard the day before. • The database of information captured will be relatively small; thus, it may not be of much help in terms of predicting power events over time.

41 To begin with, permanent monitoring ensures that power quality is being recorded 24/7. If a problem occurs, it will be recorded for later analysis. Secondly, permanent monitoring enables users to build an extensive database of event information, allowing trending over a long period of time. This factor is critical in facilitating the creation of a sound preventive maintenance program. Clearly, there are a number of benefits that accompany choosing a permanent monitoring device over a portable one. There are a number of permanent monitors available on the market today, so it is important to take several factors into account when specifying a permanent monitoring solution. There are devices that will capture events triggered by both voltage and current. Some devices will trigger on voltage only Some monitors employ a web-based browser environment, allowing multiple users to view and share the same information simultaneously. The plant engineer, plant manager, and utility operations person can all be analyzing a power event together, increasing the chances of a quicker resolution. The value of a user-friendly interface should not be underestimated. By utilizing sophisticated software, a quality instrument can present information in a way that allows nontechnical personnel to understand it without extensive interpretation. The software should be able to analyze the wave patterns and display and relay the exact nature of the difficulty to the user in simple, readable fashion. Artificial intelligence is another element of a quality system. In most utility distribution systems, capacitors are used to help regulate voltage. These are typically switched in at some predesignated hour of the morning. Superior instrumentation can recognize these capacitor-switching events and can identify the direction from which they came. This capability is also helpful in pinpointing the source of fast voltage and current transients. When voltage sag occurs, it is not uncommon for the utility to believe it happened on the customer side and vice versa. By positioning a reliable permanent monitor on the point of common coupling – the spot where the utility connects to the facility – the monitor can, in an independent and unbiased manner, determine the event source. Because each user has a different power-monitoring requirement, modularity is a vital feature. Users are looking for systems that they can configure based on their individual needs. Some instruments are equipped with four voltage and four current channels. The highest level of configurability is available in instruments that offer the choice of voltage, current, and data acquisition modules to build from one to four (or even more) virtual instruments in a single, compact format. By combining four modules in one instrument for applications that previously required two or more instruments, users will save money, prevent integration aggravation, and gain physical space. It goes without saying that compliance with the latest standards is an essential ingredient for any power quality instrument. The key standards include the European standard IEC 6100 4-30, and the American standard IEEE 1159.

42 Predictive maintenance based on trended values and events can be employed to anticipate or prevent problems. Likewise, continuous monitoring allows users to monitor the performance of the power system and the frequency of power quality events and to develop a database from which long term system health can be predicted and preventive maintenance steps initiated. At that point, the device becomes invaluable. Wieslaw Jerry Olechiw holds a BS in Mechanical Engineering from Michigan State University and an MBA from Northeatern University. In his 30 years of engineering experience, he has worked in utility, commercial and industrial facilities in power, pollution control, and process engineering. Currently, he is the Vice President of Business Development and Utility Sales for Dranetz-BMI and Electrotek Concepts.

On-Line Diagnostics Handbook — Volume 2

43

On-Line Diagnostics Handbook — Volume 2

The Future of Predictive Technologies is Here Now NETA World, Spring 2007 Issue by Claude Kane Electrical Diagnostic Innovations, LLC

Introduction In today’s competitive environment, it is imperative to capitalize on advances in technology that allow new approaches to the maintenance of a facility’s physical assets. These include reliability-centered maintenance, predictive diagnostics, condition monitoring, and expert systems. These approaches have had a significant factor in reducing unscheduled downtime and realizing gains in lost profit opportunities. Studies have shown that 80 percent of equipment failures occur on a random basis and only 20 percent are age related. This indicates that whatever our current maintenance practices are, they are not overly effective. Reasons for this phenomenon include: • Many defects develop quickly (weeks, days, or hours) that are not caught during normal periodic mainte-

nance. Time intervals between outages have increased over time; therefore, less maintenance is being performed.

• Some components of equipment are rarely, if ever, inspected or tested such as the main bus structure of medium-voltage switchgear and bus duct.

• Outages are shorter and more compact; therefore, less effective maintenance is provided due to budget and time constraints.

• Equipment has been “valued engineered” which means all of the tolerances in clearances, etc., are at the absolute minimum to reduce cost and yet pass standards. The old adage of a built in “safety margin” is now a misnomer.

• Many new insulating systems are being utilized and are not time tested and are being thermally, mechanically, and electrically Table 1 stressed to their limits. Failure Rates and Average Downtime Per Year for Common • Many past periodic on-line mainMedium-Voltage Industrial Electrical Equipment tenance activities have been severeIEEE Standard 493-1997 Failure Avg. Avg. Downtime ly limited due to new arc flash and All Industry Equipment Type Rate/Year Hours/Outage Hours/Year safety requirements. Switchgear – Bus Only 7 Sections

0.0119

261

21.7413

Large Power Transformers

0.013

1076

13.988

MV Synchronous Motors

0.0318

175

5.565

MV Induction Motors

0.0404

76

3.0704

Underground Cables 3000 feet

0.00613

53

0.97476

Cable Terminations

0.000814

284

0.693528

Small Power Transformers

0.0025

217

0.5425

MV Circuit Breakers

0.0064f

89.3

0.57152

Bus Duct – 30 feet

0.0038

128

0.4864

Note: Average downtime hours per year is calculated by multiplying the failure rate per year times the average hours per outage.

Many times money is spent when not necessary. Many repair/replace decisions are made due to limited qualitative and quantitative information. Also, many times ultrasafe decisions are made due to not having enough valid and reliable information or made due to internal corporate politics (what the boss wants to hear).

44

On-Line Diagnostics Handbook — Volume 2

Reliability and Failure Statistics The IEEE Standard 493-1997, IEEE Recommended Practice for Design of Reliable Industrial and Commercial Power Systems — Gold Book, completed end-user surveys over two time periods. This data is relatively old, but at the same time it does represent the majority of existing medium-voltage electrical equipment in service today. Table 1 and Figure 1 show the comparable hours of downtime per year of a variety of medium- and high-voltage industrial equipment. The cost of an unplanned outage can be calculated by multiplying the average downtime hours per year by your hourly plant cost of an outage. Even with a low hourly outage cost, a rapid ROI can be realized. It must be noted that switchgear and large power transformers are identified as two of the most critical assets, mainly due to the high number of hours per outage. We know from experience that we already perform extensive maintenance on transformers such as visual inspections, power factor testing, and dissolved gas analysis, yet the failure rate is in the top four. We also know we perform limited maintenance on the main bus compartments of switchgear and bus ducts, mainly due to outage constraints. Small Power Transformers 1% MVCircuit Breakers 1% Underground Cables 3000 feet Bus Duct - 30 feet 2% 1% Cable Terminations 1%

MV Induction Motors 6%

MV Synchronous Motors 12% Large Power Transformers 29%

Switchgear - Bus Only 7 Sections 47%

Figure 1 — Percentage of Downtime Hours per Year by Equipment Type

On-Line Monitoring Use of on-line monitoring and predictive technologies can alleviate the shortcomings inherent in many of our current maintenance practices. If you think these capabilities are beyond your budget constraints, it is time to rethink. The same advances that allow us to put a powerful computer on our desk for a relatively low price has put these new monitors and technologies in the reach of most budgets. Many of these technologies have become more intelligent and thus require less expertise in interpreting the results. This is not always the case, and it is difficult for companies to employ enough qualified personnel at each site to understand the large variety in data from various types of equipment. Many monitors provide more information than the typical end user

can understand but are available to the expert for additional diagnostics and prognostics. A new acronym has emerged for these intelligent electronic devices (IED). Unfortunately there is no one technology that is the Holy Grail for electrical equipment condition assessment. In many cases, multiple technologies need to be employed in order to perform a complete diagnosis. Most monitoring systems are designed to alert the user of a problem or unusual condition and provide additional diagnostic data. Most of these technologies have built-in communication capabilities that allow for remote predictive monitoring via Ethernet, serial communications such as ModBus, DNP 3.0 or customized data streams and wireless and analog modems. Remote predictive monitoring allows companies to streamline expertise at centralized locations or outsource to the appropriate experts. This can be done on a continuous or periodic basis or be event driven. Event driven systems send out an alert via a phone, pager, e-mail or fax to the appropriate person, who then communicates back to the monitor for further analysis and recommendations.

What New On-Line Predictive Technologies Are Available Today? Infrared Monitoring Performance of infrared surveys is probably the most common predictive technology used today. It is utilized on all types of equipment found in the electrical power distribution systems. It is somewhat limited on medium-voltage equipment such as switchgear since it is difficult to scan certain areas due to limited access. Also, more problems are found in low-voltage equipment due to the higher current levels. Today, there are wireless infrared sensors available that can be placed at connections such as bolted connections and finger clusters that will communicate back to a central source. The US Navy has also utilized specialized infrared sensors in low-voltage switchboards for years that connect back to a central location. Dissolved Gas Analysis (DGA) DGA on transformer oil has been utilized longer than any other predictive technology for electrical equipment. Improvements over time on the analysis of the results have led to better diagnostics. DGA used to be limited to just the oil in the main tank but has now been validated for use in load tap-changers and oil circuit breakers. There are several on-line continuous gas monitors available in the market today. These include Key Gas and Seven Gas monitors. Also, portable instruments are available to provide on-site DGA rather than sending the samples to a lab and having to wait for results.

45

On-Line Diagnostics Handbook — Volume 2 Motor Current Signature Analysis By analyzing the signatures of currents on three-phase motors, one can diagnosis many problems such as broken rotor bars, air gap problems, misalignment, improper mounting, eccentricity, and problems with gear boxes and loads. Most of this technology is based on making periodic on-line measurements, but strides are currently being made to build this intelligence into the motor starter protective devices. Electrical Partial Discharge A partial discharge (PD) is a small spark that occurs in an insulation system that does not go completely from phaseto-phase or phase-to-ground. Therefore, measurement and analysis of this discharge can provide an early warning of insulation failures in equipment rated 4,000 volts and higher. Measurement of PD can be done on a periodic or a continuous basis. In either case, specialized sensors must be installed to detect these pulses created by the small spark. Proven applications of this technology include cables, bus duct, switchgear, motors, generators, and transformers. PD monitoring on medium-voltage switchgear is an excellent use of this technology and can compensate for the lack of traditional maintenance on main bus structures. Large Power Transformer Monitoring Systems Most industrial facilities do very little monitoring of their large power transformers. Modern day systems now control and monitor all aspects of a transformer including temperatures, loads, cooling systems, pressures, bushings, and windings. Four great examples include: • The monitoring of the loads on the cooling fans and pump circuits to indicate abnormal conditions such as locked rotor or loss of cooling capacity. • Monitoring the temperature differential across the connection board between the main tank and the load tapchanger compartment • Continuous monitoring of the power factor and capacitance of high voltage bushing

• Central data concentrator and communication RTU for all third party monitors such as PD and DGA.

Summary Experience has taught us that half the battle in ensuring electrical equipment reliability is to keep it clean, dry, cool and exercised. If these four items are accomplished, then currently available on-line predictive technologies can take care of the rest. With depleting technical expertise in electrical equipment and the pressures of running facilities 24 x 7 x 365, these new technologies offer the ability to fill the maintenance gap while reducing costs and increasing power system reliability. Due to the decreasing cost of these IEDs and the high cost of unplanned outages, a rapid ROI is achieved. Claude Kane has over 30 years of experience in the installation and preventive and predictive maintenance practices on a large variety of power distribution and generation equipment. He graduated from the Milwaukee School of Engineering in February 1973. He started with Westinghouse as a field service engineer and has held a number of technical and management positions. He has presented numerous technical papers on the subject of partial discharge at many IEEE, NETA, and other technical conferences. He also was on the committee for the development of the IEEE Guidelines for the Measurement and Analysis of Partial Discharge on Rotating Equipment (P-1434). Claude is the President of Electrical Diagnostic Innovations, LLC located in Minneapolis, MN.

46

On-Line Diagnostics Handbook — Volume 2

Case Studies in On-Line and Off-Line Motor Analysis NETA World, Spring 2007 Issue by David L. McKinnon PdMA Corporation

I. Introduction

Problem

Current signature analysis (CSA) has become the standard for detecting broken rotor bars by analyzing the sidebands around line frequency. Another useful tool is demodulated current spectrum analysis (DCSA) that enhances the ability to detect broken rotor bars especially on two pole motors. Our first case study will present a situation where CSA and DCSA were used to detect broken rotor bars on a two pole motor. The polarization index (PI) is a standard measurement in the IEEE 43-2000 standard for insulation testing. PI is a ratio of the measured resistance at ten minutes divided by the resistance measurement at one minute. This ratio can be used for analyzing the general health of the insulation system of the motor. In our second case study, we present a modification to the standard polarization index test. By plotting the resistance measurement every five seconds, a graph called a polarization index profile (PIP) is obtained. The resulting PIP profile can then be used for additional analysis of the insulation system that cannot be obtained from the standard polarization index. Power quality is a measure of the quality of the voltage and current supplying a motor or other load. By analyzing the harmonics, voltage and current unbalance, overvoltage or undervoltage, and overcurrent conditions, a technician can determine what may be causing nuisance trips, voltage swells or sags, and other power system related situations as we will examine in our third case study.

During routine EMAX testing in June 2003, it was observed that the peak dB level of the pole pass sideband was 0.7419 dB, which exceeded the alarm set point of 0.3 dB (Figure 1). Rotor bar problems were suspected; however, when vibration analysis was performed the results indicated a healthy motor. Due to the conflicting indications between the EMAX and vibration tests, the decision was made to monitor the motor and trend the test results.

II. Case Study 1 — Broken Rotor Bars

Action Taken

Motor Specifications: 3500 HP, 2 Pole, ac Induction, 4160 Volts, 3590 RPM.

Figure 1 — Demodulated Current Spectrum Showing the Pole Pass Sideband at 0.7419 dB.

The motor was retested periodically until 5/12/2004. At that time it was removed from service. The trended data indicated a 1420 percent increase in the peak level of the pole pass sideband from 0.1814 dB at a running speed of

47

On-Line Diagnostics Handbook — Volume 2 3591 RPM on 8/15/2001 to 2.5851 dB at a running speed of 3592 RPM on 5/12/2004 (Figure 2). Trending the data from tests taken between 2001 and 2004 showed an exponential increase in the pole pass peak levels, which is typically indicative of at least one or more broken rotor bars. Additionally, there was a 275 percent increase in the load variation, from 0.855 percent in 2001 to 2.345 percent on 5/12/2004 (Figure 2). The load variation should be constant from test to test under normal operating and motor conditions. The current spectrum (Figure 3) of test data taken on 5/12/2004 showed an increase in sideband activity around the fundamental frequency, which also indicated broken rotor bars.

was sent to the motor shop where it was disassembled. A visual inspection found 22 of 51 rotor bars broken or cracked (See Figure 5).

Figure 4 — Rotor Influence Check (RIC) Test of the Motor Before Disassembly. Notice the Repeated Variations in the Inductance Waveforms, Which are Indicative of Broken Rotor Bars.

Figure 2 — Pole Pass Sideband of 2.5851 dB at a Running Speed of 3592 RPM.

Figure 5 — Some of the Broken Rotor Bars Found Using On-Line and Off-Line Tests.

Root Cause It was determined that bad braze joints between the bars and end rings from a rotor repair performed in 2001 caused the broken and cracked rotor bars. Figure 3 — Hi Resolution Current Signature Spectrum Centered Around Line Frequency.

All EMAX test results indicated broken rotor bars in the motor. The motor was pulled and a RIC test was performed, which also indicated a rotor anomaly (Figure 4). The motor

Savings Cost to repair the motor was $90,000 (repair $60,000 plus $30,000 planned plant downtime). Had the motor run to failure the cost would have been $370,000 ($170,000 cost of new motor plus $200,000 unplanned downtime). Total savings were $280,000.

48

On-Line Diagnostics Handbook — Volume 2

III. Case Study 2 — Polarization Index Profile In March 2005, a PIP test was performed on an induced draft (ID) fan motor resulting in the PIP shown in Figure 6. Notice the low PI value and fast initial rise-time of the resistance to a relatively low value overall. These are indicative of an insulation system containing a significant amount of moisture. IEEE 43-2000 recommends an insulation resistance of no less than 100 megohms and a PI value of >2.0 for this motor.

Figure 8 — Polarization Index Profile of a Healthy Insulation System

When insulation systems become contaminated with debris such as dirt, carbon dust, etc., the PIP will have a significant amount of spiking in the profile throughout the test as shown in Figure 9.

Figure 6 — Polarization Index Profile (PIP) of an Insulation System Containing a Significant Amount of Moisture

Figure 7 shows the moisture around the power cable entrance that caused the low overall PIP and low PI value.

Figure 9 — PIP of a Motor That Has a Contaminated Insulation System

Figure 7 — Condition That Caused the Unacceptable PIP.

The moisture was dried from all cables, components, etc., and another PIP performed a few days later. Figure 8 shows the resulting PIP after the dry-out. This PIP is indicative of a motor with a healthy insulation system.

On-Line Diagnostics Handbook — Volume 2 Figure 10 shows the contamination on the stator windings of the motor.

49 There are eight motor line-ups and five dc drives on the system. This is unique to this power distribution system. All of the other dc drive power transformers supply power only to the dc drives and are not used in combination with ac motors. This suggested the power quality issues may be generated by the drives themselves. Further testing was performed to compare the test results when the winder was not running and when it was running. As shown in Figure 12, when the winder was not running (1/14/2004), current harmonics (THD) were less than 2 percent, voltage harmonics were less than 1 percent, current was 105 percent of rated current, and the system voltage was 475 volts. With the winder running (1/15/2004), current harmonics (THD) were greater than 5 percent, voltage harmonics were greater than 7 percent, current was 108 percent of rated current, and system voltage was 453 volts, a drop of 22 volts.

Figure 10 — Stator of the Motor Containing the Contaminated Insulation System.

IV. Case Study 3 — Power Quality Problem Initial test data on a winder motor indicated the voltage harmonics are greater than 5 percent and the running current at 107 percent of full load amperes (FLA) as seen in Figure 11.

Figure 12 — Test Results with the Winder Down (1/14/2004) and with it Running (1/15/2004)

Figure 11 — Test Results Page with the Winder Running

Action Taken The power distribution system to MCC M4-50 was researched to identify what might be causing the harmonics. Transformer #31, a 13.8 kV to 480 volt, 2500 kVA transformer, is the feed for the MCC M4-50’s 460 volt service.

These power quality issues were reported to the appropriate department managers and then filed away. Earlier this year, the E/I supervisor came to the E/I reliability office to ask about the power quality report on MCC M4-50. A new microprocessor-based dock lock system for their eight-bay tractor trailer loading dock had been purchased and a new power feed for the controllers had been provided by the E/I supervisor’s crew. The system had been installed by the vendor and was under warranty. The system worked fine

50

On-Line Diagnostics Handbook — Volume 2

most of the time, but at other times the locks would open and close on their own. The controller manufacturer’s field service group including their design engineer had made several trips to the mill. They replaced several electronic cards and two complete controllers. The E/I supervisor’s crew got the power for the new feed from MCC M4-50. The E/I supervisor now remembered the report and asked if the power quality issues might still exist on the supply from M4-50. After a review of the report, line filters were installed on all of the new controllers as shown in Figure 13. The problem for the dock lock controller disappeared.

Figure 13 — Line filters Installed on a New Controller.

IV. Summary Three case studies were presented: broken rotor bars, polarization index profile, and power quality. Broken rotor bars can be detected using a battery of on-line and off-line testing. On-line testing includes demodulated current and current signature analysis. Off-line testing includes a rotor influence check, which graphically depicts broken rotor bars. In our second case study, a modification to the standard polarization index test provided a very good benefit in analyzing the health of insulation systems. In our last case study, power quality was used to analyze a power system and correct harmonic issues caused by the installation of motor drives. David L. McKinnon received his BS in Electrical Engineering from New Mexico State University in 1991 and an MBA from the University Of Phoenix in 2002. He has worked in the field of magnetics for over 14 years. During the past four years, he has worked for PdMA Corporation as a Project Manager for hardware and product development of motor test equipment.

51

On-Line Diagnostics Handbook — Volume 2

Battery Testing Techniques PowerTest 2007 (NETA Annual Technical Conference) Presenter James G. Cialdea, P.E., Three-C Electrical Co., Inc. Co-Author Mikel Hensley, ProgUSA

Abstract This paper presents a method for testing of substation battery banks ‘on-line’ without the need to disconnect the load. The final terminal battery voltage is increased so all of the capacity is not used during the test. A review of battery testing is presented including: recommended routine maintenance, preparation for load testing, and load testing.

Introduction Even though batteries are critical to the safe and reliable operation of many types of equipment they are still not a priority on the list of items to be maintained. The major reasons for this are: 1. Lack of understanding of maintenance requirements for batteries 2. Cost to properly maintain batteries

3. Downtime necessary for proper testing of capacity Proper understanding of a battery system design is crucial for designing a proper testing and maintenance program for the system. A good question to ask is “Is this battery system properly sized for the application?” Another question, “Is the system properly monitored?” Many systems have no annunciation, or annunciation is not connected so if a problem does occur it can go unnoticed for quite some time. A failure of a battery system can create a catastrophic situation.

20MW Generator Damage after dc System Failure – Machine lost dc Oil Pumps and Breaker Failed to trip. Unit motorized for 45 minutes. Shaft sheared in 3 places. Repairs exceeded $3M and 6 months downtime.

In this paper we are going to look at some specific battery types and applications – flooded cell (lead-acid) and sealed valve regulated for use in Substations and Power Plants. NiCads and UPS applications are not included here.

Bank Design Basically, a battery system is designed to provide power to a load if the ac is interrupted, for whatever reason. While the ac is present, the typical design is for the battery charger to provide the constant power draw of the load and have enough capacity to charge the batteries within a certain number of hours. Since the battery charger has current limiting and ramping functions on it’s output, instantaneous loads (like a breaker tripping) is handled by the battery bank. A note to remember: battery chargers are not designed to

52 operate without a battery connected. Some unfiltered units will put a high level of ac on the dc system if disconnected from the battery and powered solely by the charger. Also, even if the charger has a filtered output, the current limiting circuits may not allow for enough current flow for breakers to operate. The first step is to properly size the battery, or verify it is properly sized. To do this determine the load characteristics – constant draw, power requirements without ac, and the duration needed to provide power to load without ac. A detailed methodology for these calculations can be found in IEEE 485-R2003 and IEEE 1187-2002. From these calculations charger output ampacity, battery amphour rating, bank voltage, and terminal voltage can be determined.

Installation A key issue is proper installation. Manufacturers’ instruction manuals give a thorough procedure for installation of the battery system. Besides proper connection torque, a critical issue typically missed is charger calibration. The charger must be calibrated for the particular bank. This will be dependant on the number of cells and the environment that the bank will operate in. There are two output calibrations required: float voltage and equalize voltage. The battery instruction manual will provide ranges for these voltages on a per cell basis. Higher per cell voltages charge the cells faster, but also add heat which will shorten the life of the cells. Once the voltage per cell is selected it is multiplied by the number of cells to determine the bank voltage. The charger is then calibrated to those values. Alarm setpoints also need to be set.

Maintenance Recommended routine maintenance can be found in IEEE 450-2002 and IEEE 1188-2005. Basically, maintenance can be broken down into two types: on-line and off-line. On-line maintenance consists of visual inspections, voltage measurements, specific gravity, and impedance measurements. Off-line is load testing. On-lie maintenance is performed on a monthly, quarterly, and annually basis. Monthly: Visual Inspection, bank voltage, electrolyte levels, charger output, room temperature, pilot cell voltage and electrolyte temperature. Quarterly: Monthly plus: individual cell voltages, specific gravity and electrolyte temperature of 10% of cells. Annually: Quarterly plus: specific gravity and electrolyte temperature of all cells, connection resistance, and individual cell impedance.

On-Line Diagnostics Handbook — Volume 2 Battery Load Testing Load Testing is typically recommended annually for sealed batteries and every five years for flooded batteries. This involves taking the battery off-line which will require either a temporary bank or a shutdown of the facility or equipment that the battery provides power for. This is typically expensive and inconvenient for the Customer. To perform a meaningful load test, the following steps are required: 1. The battery must be in good condition before the test is performed and preparation is required: a. Equalize charge – per manufacturer’s recommendations – usually 48 to 72 hours b. Float charge – minimum of 48 hours after equalize c. Measure charger ripple – important – can ruin batteries d. Check connection resistance – most common problem for battery bank e. Check specific gravity f. Clean bank 2. The test parameters need to be defined: a. What type of test will be performed? In most applications, a straight discharge test will be performed. This test puts a constant current draw on the bank until the bank reaches a pre-determined final terminal voltage. The other option is to design a test that actually simulates the loading of the battery. This is a custom test and will not be discussed in this paper. b. Determine final terminal voltage. The final terminal voltage is the design voltage at which the bank has delivered all of its capacity. For a substation application, this is typically 1.75 volts per cell. c. Determine the test current and duration. This requires information from the battery manufacturer. The manufacturer provides tables for different terminal voltages that show the cell output current and the length of time the cell will deliver that current. The tables are not linear, so a battery that is 100 amphours will not necessarily deliver 400 amps for 15 minutes or 50 amps for two hours.

53

On-Line Diagnostics Handbook — Volume 2 d. Determine the minimum voltage for a cell to stop the test. 3. Perform the test a. Once the test parameters have been defined the equipment is setup and programmed for the test. Individual cell monitoring during the test provides a better profile of the individual cells although this is not required by the standard. Also the connections to the cells should be arranged to monitor cell to cell connection voltage drop.

4. Calculate the bank capacity a. The bank capacity is calculated based on the time it takes to reach terminal voltage compared to the manufacturer’s specification. The IEEE standard now applies a temperature correction factor to the calculated capacity. The prior standard applied a temperature correction factor to the test current for the test. Either method is acceptable for determining bank capacity. % Capacity = (Ta/(Tm x K)) x 100 Ta = Actual time of test to terminal voltage Tm = Manufacturer’s specified time to terminal voltage K = Temperature correction factor from IEEE 450-2002 Table 1 b. Determine suitability of bank and new frequency of testing based on capacity.

Typical Load Test Setup

b. On application of the test current, the bank voltage will drop then recover slightly. The test is continued to the determined terminal value. 125

120

115

110

105

100

95

Bank Voltage vs. Time for Constant Current Test

c. The test is then performed until the bank terminal voltage is obtained.

Some other tests that can be performed are a charger load test – does the charger provide rated current output and what is the ripple at full load? An infrared scan can be performed during the load test for additional information on connections and cell condition. Following the test, the bank needs to be charged per the manufacturer’s instructions. The bank will not be ready for full service until recharged. This can take from 4 to 24 hours depending on the system design.

Battery Load Testing On-Line With modern equipment it is possible to perform battery load testing on-line. There are some important considerations before performing the test. A higher end test voltage needs to be chosen to make sure that the battery has some capacity left after the test should it be called upon to supply power to the load either during or immediately after the test. Since the load test will drain the batteries, if they are called upon immediately after the test they will not provide full design capacity as they are starting at a reduced capacity. If this condition is not acceptable, then on-line testing can not be performed. Usually, batteries for substations are designed for ‘worst case’ and will have more than sufficient capacity to trip breakers if required. Another important item is dc low voltage alarms or trips. These must be identified and possibly defeated.

54

On-Line Diagnostics Handbook — Volume 2

The higher terminal voltage per cell from the manufacturers spec sheet will have a lower test current than the off-line test. This reduced current does decrease the stress on the battery. Since the test is not bringing the bank to the design terminal voltage, it does not meet the IEEE specification for load testing. However, the test will give a capacity of the bank related to the manufacturer’s specifications for the battery. In cases where an off-line test is impractical, the on-line test is a good option to get the bank capacity. The test is performed in the same manner as the off-line test except the load is monitored and subtracted from the calculated test current. The load bank makes up the difference between the test current and the load current so the bank sees a constant current equal to the test current.

Disadvantages: 1. Test terminal voltage is increased and test current is reduced and from levels used for an off-line test. This reduces the ‘stress’ on bank during the test and does not drain the bank to its full design capacity. 2. Possibility of failure of bank or problem under test could compromise load.

Conclusion Regular battery maintenance is essential for safe and reliable operation of the dc system. Maintenance needs to include load testing. For some cases, on-line load testing may be a cost effective method to perform regular load tests to insure battery system integrity.

References 1. IEEE Std. 450-2002 Recommended Practice for Maintenance, Testing, and Replacement of Vented Lead Acid Batteries for Stationary Applications 2. IEEE 1188-2005 Recommended Practice for Maintenance, Testing, and Replacement of Valve-Regulated Lead-Acid (VRLA) Batteries for Stationary Applications 3. Enersys PowerSafe CC Performance Specifications Publication No. US-CC-PS-001 April 2006 Test Connection for On-Line Test

Advantages of this method: 1. Test can be performed without interruption to the load. 2. Back up battery bank is not needed. 3. Test can be performed at any time, does not need a scheduled outage.

James Cialdea is Vice President and Director of Engineering for Three-C Electrical Co., Inc., Ashland, MA where he has performed installation, testing, maintenance, troubleshooting, design services, and studies on electrical distribution systems for industrials, municipals, institutions, utilities, and power plants throughout New England for over 25 years. He holds a Bachelor of Science in Electrical Engineering, Power Option, from Worcester Polytechnic Institute and has graduate courses from Rensselear Polytechnic Institute’s Center for Electric Power Engineering. He is a Registered Professional Engineer, Master Electrician, and Licensed Construction Supervisor. He is a member of the Commonwealth of Massachusetts Electrical Code Advisory Committee (6 years), Member IEEE (25 years), Associate Member NETA (10 years), Member Massachusetts Electrical Contractors Association (18 years, 12 years on State Board of Directors, and 2 years as President)

55

On-Line Diagnostics Handbook — Volume 2

Corona Imaging — See the Invisible NETA World, Summer 2007 Issue by Don A. Genutis Group CBS

It felt like a fine spring morning in eastern Oregon as my colleague and I rolled up the power plant driveway, although the date on my Blackberry indicated mid-February. After some friendly greetings and hand-shaking with maintenance personnel, we signed in at the control room and viewed the not too lengthy or boring safety video. Then it was time to don our hard hats and head toward the substation – all half-million volts of it.

It was dry and sunny as our work boots crunched against the coarse substation gravel rocks. The massive bushings and insulators did not appear to have much, if any, condensation from the night before as we gazed across an acre of GSU transformers, SF6 dead-tank breakers and other equipment. Ideal conditions for a corona image survey. After powering up our new corona camera and making a few minor adjustments, we began doing what no other person had ever done before in this substation – we were viewing an invisible phenomenon known as corona. What we and plant personnel saw during the next few hours that morning would better our understanding of high voltage engineering in a manner that no text book possibly could. We saw corona from insulators that required cleaning, sharp edges from hardware connections, poor bus work corona suppression, and dangerous insulator corona caused by cooling tower residue deposits. As the prevailing wind puffed another cloud

Figure 1 — Dangerous corona occurring from middle of the insulator. The corona from the top corona ring is harmless.

Figure 2 — GSU transformer bushing corona is seen flirting with flashover.

This column focuses on electrical inspection methods and technologies that are performed while the electrical system remains energized. Although no-outage inspections can be very valuable tools, always remember to comply with proper safely guidelines when conducting energized, on-line inspections.

56 from the cooling tower onto the westernmost insulators, we could hear the corona crackling intensify as the camera revealed near flashover conditions. “When it sounds like this, I tell my guys to get out of the substation, ” our customer stated. Not a bad idea, I thought. Next, we followed our customer to a corner of the substation were we examined a flash-damaged insulator laying in the gravel, then observed the additional weather sheds added recently to the in-service insulators nearest the cooling tower. For the most part, the increased surface creepage of the modified insulators were doing their job by eliminating most of the corona. However, as can be seen in Figure 1, one insulator was clearly showing dangerous corona that consistently occurred from a defect located half way down the insulator. Additionally, we found severe corona activity occurring from the center phase GSU transformer bushing as shown in Figure 2. We left the power plant feeling that the customer greatly appreciated the value of our new technology and would use the knowledge gained that day to reduce future flashover likelihood. One of the greatest difficulties when working with electricity is that it cannot be seen. The technician must generally determine its presence based upon meter deflection or simple visual or audible warning signals. Perhaps even more difficult to appreciate is the concept of electrical fields surrounding conductors or insulators in air. A visual representation of the electrical field is shown in Figure 3, which illustrates the voltage gradient surrounding an insulator. Medium- and high-voltage equipment is carefully designed so that electrical fields are contained. If the electrical field is compromised by foreign materials, poor design, or other factors, corona will occur. Corona is responsible for the generation of ultraviolet radiation, ozone, acids, heat, mechanical-erosion through ion bombardment, electric power loss, and electromagnetic interference of radio communications. Corona by-products destroy insulation and lead to catastrophic failure while remaining invisible to the eye. Corona can occur due to: • poor component or system design

• improper installation or workmanship

• temporary contamination by icing, salt sprays near highways or oceans, fog, agricultural chemicals, and other chemicals

On-Line Diagnostics Handbook — Volume 2

Figure 3 — Electrical field surrounding an insulator

Today, a relatively new technology allows invisible corona to be visible through a special imaging device. The handheld camera seen in Figure 4 clearly displays the ionized air created by harmful corona activity by indirectly detecting the associated ultraviolet radiation. The camera user can change the color of the corona activity to provide better contrast against varying backgrounds, and the camera records the corona image to allow subsequent transfer of the image to a PC for report generation. This highly effective technology is quickly becoming an indispensable tool for the technician. In addition to the obvious maintenance applications for all types of medium- and high-voltage equipment, the corona camera can also be used for:

• damage due to storm wind or lightening • damage due to vandalism

• cumulative damage due to corona activity caused by weather conditions • cumulative damage due to contamination from nearby sources

Until recent technology breakthroughs, ultrasonic detection was one of the only methods capable for detecting corona. This technology is limited, especially when scanning high voltage equipment from a distance since only the relative direction of the problem source can be determined.

Figure 4 — Hand held corona camera in use

57

On-Line Diagnostics Handbook — Volume 2

Figure 5 — No corona from breaker during shop high potential test

Figure 6 — Unwanted corona from poor test lead clearance

• supplementing partial discharge and ultrasonic surveys

The future use of corona imaging promises to provide many new applications and many associated new surprises, but one thing is certain — we will never think of corona in the same manner, for now we can see it.

• acceptance testing of new or modified installations • product quality assurance • design verification

• field and shop high potential testing • general research

We can now even have a much greater understanding of equipment insulator and conductor behavior when conducting routine high potential testing. Figure 5 displays the absence of corona while performing an ac high potential test of a medium-voltage circuit breaker, indicating good insulator and conductor condition, while Figure 6 shows unwanted corona generating from inadequate clearance of the energized test conductor during the same test. It is apparent that the breaker was good but our temporary shop test setup was not.

Mr. Genutis received his BSEE from Carnegie Mellon University, has been a NETA Certified Technician for 15 years, and is a Certified Corona Technician. Don’s technical training and education is complemented by nearly twenty-five years of practical field and laboratory electrical testing experience. He is presently serving as Vice President of the Group CBS Eastern U.S. Operations and acts as Technical Manager for their subsidiary, Circuit Breaker Sales & Service located in Central Florida.

58

On-Line Diagnostics Handbook — Volume 2

Comprehensive Transformer Monitoring to Increase Reliability NETA World, Winter 2007-2008 Issue by Claude Kane Electrical Diagnostic Innovations, LLC

The time to failure on transformers is relatively unknown. The use of on-line comprehensive transformer monitoring has become a very useful and cost effective way to help identify problems early. By identifying problems early, they can be fixed before they escalate into a complete failure. Within the industry, there is a common debate over whether monitoring should be installed on new transformers or whether it should be added as a retrofit to older transformers. Each side of this debate is easily defended.

Monitoring on New Transformers The most compelling case for including comprehensive transformer monitoring on new transformers is the cost. Transformer prices have risen dramatically in the past few years. Some figures indicate a price increase of 300 to 400 percent over the past 10 years. During this same time, the technology available for transformer monitoring has improved dramatically in features, reliability, and cost. When specifying a million dollar asset, the addition of a comprehensive transformer monitoring system is often considered a prudent investment.

Figure 1 — Utility Failure Data

Most companies are feeling severe operations and maintenance (O&M) budget constraints. By utilizing the condition data from on-line transformer monitoring systems, companies can automate some of the routine testing and data collection tasks providing savings on the O&M side of the business. Hence, the addition of a monitoring system on a new transformer will result in a small percent increase in the transformer capital budget while helping to conserve precious O&M dollars in the future. In addition to O&M savings, transformer monitoring systems will provide the ability to improve transformer reliability. Transformers don’t have to be old before they fail. Transformer design or production problems can cause a failure within the first few months of service. The following data was published in T&D World magazine and illustrates the infant mortality failures and the subsequent random nature of failures experienced by one medium-sized midwestern utility. While failures can be unpredictable, one thing is for certain; all transformers left in service will eventually fail. In general, those customers who are utilizing comprehensive transformer monitoring systems typically have an end objective of eventually covering the majority of their transformer fleet. Certainly, the retrofit approach is necessary if any utility plans to cover all transformers. However, if new transformers are being purchased without comprehensive monitoring equipment, it can be very difficult to gain on the problem. As a result, we see most companies include comprehensive monitoring systems in all new transformer purchases and then subsequently attack the existing fleet of transformers employing additional transformer monitoring systems as their budgets and resources will allow.

59

On-Line Diagnostics Handbook — Volume 2 Monitoring as a Retrofit to Existing Transformers Certainly, there are plenty of justifications for monitoring of new transformers. However, the most seductive objective of comprehensive transformer monitoring is to predict a transformer’s end of life. We often get the question posed to us, “Can you tell us when the transformer will fail with enough advanced warning to respond?” H. Lee Willis , one of the authors of Aging Power Delivery Infrastructures, explains the challenges in predicting a transformer’s end of life. This theory also applies to any equipment. With present [periodic testing] technologies, it does not seem possible to predict time-to-failure exactly. In fact, capability in failure prediction for equipment is about the same as it is for human beings. 1. Time-to-failure can be predicted accurately only over a large population (set of unit). Children born in 2000 in the United States have an expected lifetime of 76 years, with a standard deviation of 11 years. Service transformers put into service in 2000 have an average expected lifetime of 43 years with a standard deviation of 7 years 2. Assessment based on time-in-service can be done, but still leads to information which is accurate only when applied to a large population. Thus, analysts can determine that people who have reached age 50 in year 2000 have an expected 31 years of life remaining. Service transformers that have survived 30 years in service have an average 16 years of service remaining.

3. Condition assessment can identify different expectations based on past or existing service conditions, but again this is only accurate for a large population. Smokers who have reached age 50 have only a remaining 22 years of expected lifetime, not 31. Service transformers that have seen 30 years service in high-lightning areas have an average of only 11 years service life remaining, not 16.

healthy, neither his relatives nor his doctors knew whether it would be another two years or two decades before he died and his will was probated. Now that he lies on his deathbed with a detectable bad heart, failure within a matter of days is nearly certain. The relatives gather. Similarly, in the week or two leading up to failure, a power transformer generally will give detectable signs of impending failure: an identifiable acoustic signature will develop, there will be internal gassing, and perhaps detectable changes in leakage current, etc.

The presence of on-line transformer monitoring can greatly improve the ability to capture impending failures that would otherwise be missed using time based condition monitoring techniques. The next sections of this paper will outline the latest advancements in on-line transformer monitoring being used by companies to identify transformer problems early and thereby prevent major failures.

Comprehensive Transformer Monitoring (Transformer Management Systems – TMS) Transformer monitoring in its most basic form has been around for many years. Historically, companies used alarm contacts from the various gauges on the transformer, and this was considered to be the monitoring system. As technology has advanced, there is now a vast array of monitoring tools and technologies available to provide better information which will give earlier indication of problems. To determine what should be included in an on-line monitoring system, transformer customers often use the failure history of their transformer fleet as a guideline. The data, shown in figure 2, was obtained from Doble Engineering Company’s review of transformer failures from 1993-1998.

4. [Periodic] Tests can narrow but not eliminate the uncertainty in failure prediction of individual units. All medical testing in the world cannot predict with certainty the time of death of an apparently healthy human being, although it can identify flaws that might indicate likelihood for failure. Similarly, testing of a power transformer will identify if it has a “fatal” flaw in it. But if a human being or a power system unit gets a “good bill of health,” it really means that there is no clue to when the unit will fail, except that failure does not appear to be imminent. 5. Time to failure of an individual unit is only easy to predict when failure is imminent. In cases where failure is due to “natural causes” (i.e., not due to abnormal events such as being in an auto accident or being hit by lightning), failure can be predicted only a short time prior to failure. At this point, failure is almost certain to advanced stages of detectable deterioration in some key component.

Thus, when rich Uncle Jacob was in his 60s and apparently

Figure 2 — Transformer Failure Data

Transformer monitoring systems can be designed to economically monitor most of the transformer failure modes shown above. With a properly designed system, detection rates of at least 60 - 70 percent are easily achievable.

60

On-Line Diagnostics Handbook — Volume 2

In addition to deciding what to monitor, companies also need to decide how to monitor their equipment. The traditional approach was to install function specific products to perform a wide variety of discrete monitoring functions:

logic appropriate for its own system’s needs with alarm setpoints and algorithms which may be unique for the given transformer.

• Temperature Monitoring

• Cooling System Monitoring • LTC Monitoring

• DGA Monitoring

• Moisture-in-Oil Monitoring • Bushing Monitoring

• Partial-Discharge Monitoring This approach is similar to the first electronics used in automobiles. The electronics for each function were separate. The alternator/voltage regulator, electronic ignition, fuel injection, ignition timing (spark advance), climate control, emission control, and sound system were all separate. Now many of these functions are integrated into one system. Not only does this reduce the number of discrete components and thus improve the reliability, it also provides many additional functions such as improved engine performance and lower emissions. It also provides the capability to perform more luxurious options such as automatic volume adjustment on the sound system to account for road noise and other similar functions. TMS systems typically have the temperature, cooling system, and other essential functions built in. The more advanced functions of DGA monitoring, bushing monitoring, and partial-discharge monitoring may be performed by stand alone devices. However, the data would still be consolidated into the TMS providing the customer with a single point of connection to the transformer which will provide all the necessary information. In addition to simplifying the customer’s installation, it also makes the SCADA connection easier as there would only be one point list and one communication link that need to be tested and supported. The two fundamental types of architecture for a TMS system are centralized data processing and distributed processing. Systems are available on the market for both, but the trend appears to favor the distributed processing. Just as with early computer systems where the competing architecture was main frame vs. PC, the advantages of distributed processing eventually won over most applications. The distributed processing systems used for transformer monitoring have the following advantages: • Reliability: If communication connection is lost temporarily, the systems continue to operate. • Customization: The sensors and control/monitoring hardware can be optimized for each individual installation without increasing the complexity of the central system. Each system is also programmed with the

• Commonality: Even if each installation has different monitoring needs, there is no need to make changes to the central logic. A common point list can be utilized with grouped logic.

• Bandwidth: As monitoring systems become more complicated, more and more information is being collected and analyzed. Most of the raw data is meaningless, and it is only the analyzed data that is of any concern. Distributed systems will process the data and send the actionable information upstream and store the raw data locally in the rare case that it is needed later. This significantly reduces the amount of data which is being added to the communication link. • Scalability: As each new site is added, it will bring with it additional processing capability.

• Protection from Obsolescence: As technology changes, the new distributed intelligent TMS systems can be purchased without concern over compatibility with previous hardware. Individual installations can be revised or upgraded independent from one another. • Mobility: Distributed processing would be accomplished using hardware installed on the transformer. If the transformer is ever moved, the history, alarm settings, and monitoring logic will automatically move with it.

As with computer systems, the distributed processing still benefits from centralized data storage and reporting. One of the most important objectives of a successful monitoring system is to get the right information to the right people. Hence, the information sent back through the SCADA or dcS system must be customized for the intended audience and must be in an actionable format.

Conclusions The time to failure for transformers can be unpredictable. Fortunately, most transformers will provide an identifiable sign of an impending failure in the days or weeks prior to the failure. On-line monitoring offers the best opportunity to prevent these failures by identifying transformer problems early. While there is a wide variety of monitoring devices available in the marketplace today, the data provided by these discrete sensors and IED’s can be greatly enhanced when correlated with loading, temperatures, and other critical information. Comprehensive transformer management systems offer the best opportunity to capture and consolidate these signals to provide asset managers the timely, accurate, and meaningful information they require to prevent failures by identifying transformer problems early.

On-Line Diagnostics Handbook — Volume 2 Only by innovating, by taking a new approach and changing the way they plan, engineer, operate, and in particular, manage their systems, can companies get both the system and financial performance they need. Claude Kane has over 35 years of experience in the installation and maintenance practices for power distribution, transmission, and generation equipment. He started with Westinghouse in 1972 as a field service engineer in Kansas City and has held a number of technical and management positions throughout his career. His last 10 years have been spent developing technological products and emerging markets for predictive diagnostic and prognostic equipment. He is currently one of the principal owners of Electrical Diagnostic Innovations, Inc. based in Minneapolis.

61

62

On-Line Diagnostics Handbook — Volume 2

On-Line Fault Analysis of DC Motors NETA World, Spring 2008 Issue by David L. McKinnon PdMA Corporation

Abstract Over the last 20 years, current signature analysis (CSA) has become an established tool for on-line fault analysis of ac induction motors. Presently, very little research has been performed using current signature analysis on dc motors. This paper is a brief introduction to on-line fault diagnosis of dc motors using current signature analysis.

Introduction This research initiative was undertaken to further develop on-line fault detection of dc motors using current signature analysis in both the time and frequency domains. These faults include differential current, shorted armature windings, shorted field windings, and off magnetic neutral plane brush positions. To detect the various faults in dc motors, we must develop a methodology to correctly differentiate normal operating conditions from those of fault operating conditions. The first step is to establish a baseline of normal

Figure 1 — No fault - full load, full speed, and brushes at zero magnetic neutral axis

operating conditions. Once a baseline of normal operating conditions is established, a method of differentiating fault operating characteristics from baseline characteristics must be developed. The primary differentiating methodology used in this study was a visual comparison of fault operating conditions to the baseline condition. For this study, a deterministic fault condition is considered one in which there is an obvious visual or numerical change in either or both the time or frequency domain. Visual changes may include variations in the waveforms in the time domain or the number of peaks, their amplitude, or their location in the frequency domain. For our purposes, numerically deterministic changes are those that exceed the measurement error sensitivity of the equipment in use by more than a specified amount beyond the maximum measurement error. For example, if the error sensitivity of the equipment is one percent of reading and the specified change is one percent, the minimum fault differential required would be two percent.

Figure 2 — Turn-to-turn short – full load, full speed, and brushes at zero

63

On-Line Diagnostics Handbook — Volume 2 Discussion Turn-to-Turn Short Many turn-to-turn or commutator bar-to-bar faults occur from carbon dust buildup. Carbon dust from the brushes builds up on the commutator creating a short between commutator bars. To simulate the worst case of this fault condition, two wires that terminated on adjacent commutator bars were shorted together on the armature of the dc motor. The motor was then run and a current signature analysis in both the time and frequency domains was performed. Figures 1 and 2 show a comparison of the current signatures in the time domain of a no-fault condition to a turn-to-turn short (faulted condition). In the no-fault condition, there is no modulation of the 120 Hz carrier frequency. In the fault condition shown in Figure 2, the waveforms have a modulation of the 120 Hz carrier of approximately 17 Hz. This fault condition is further noticed in the frequency spectrums shown in Figures 3 and 4. The frequency spectrum shown in Figure 3 is the no-fault condition. Figure 4 shows the frequency spectrum of the turn-to-turn fault condition. Notice the dramatic increase in the harmonics throughout the spectrum.

Figure 4 — Turn-to-turn short - full load, full speed, and brushes at zero

Figure 5 — Coil group short - full load, full speed, and brushes at zero

Figure 3 — No fault - full load, full speed, and brushes at zero

Coil Group Short Figure 5 shows a fault operating condition in which an entire coil group is shorted. Notice this increase in modulation as compared to the turn-to-turn short shown in Figure 2. Figure 6 shows the frequency domain of the coil group fault operating condition. There is a significant increase in the harmonics throughout the spectrum.

Coil-to-Coil Short Figure 7 shows the time domain of a fault operating condition in which two coil groups are shorted together. Notice the increase in modulation as compared to the coil group short shown in Figure 5. Figure 8 is the frequency spectrum produced by this fault.

Figure 6 — Coil group short – full load, full speed, and brushes at zero

64

On-Line Diagnostics Handbook — Volume 2 Brush Position Detecting when the brushes are off the magnetic neutral axis can be difficult, especially if the motor is inaccessible during operation. Using voltage analysis in the time domain makes the job of properly setting the brushes for the desired load much easier. The voltage waveforms in Figure 9 appear to be clean (i.e., without noise). When the brushes are off the magnetic neutral axis, the voltage waveforms in the time domain have a lot of hash as shown in Figure 10.

Figure 7 — Coil-to-coil short - full load, full speed, and brushes at zero

Figure 10 — Brushes off magnetic neutral axis

Figure 8 — Coil-to-coil short – full load, full speed, and brushes at zero

Figure 11 — No field ground, tester grounded at drive

Field Ground

Figure 9 — Brushes precisely on magnetic neutral axis

To detect field winding grounds, the time domain waveform of the line-to-neutral voltage should be analyzed for anomalies. Under no fault conditions, the field voltage will be of significant amplitude as shown in Figure 11. When there is a field ground, the voltage will be very low as shown in Figure 12.

65

On-Line Diagnostics Handbook — Volume 2

Figure 12 — Field grounded to tester ground

Figure 14 — No brush ground, tester grounded at the drive

Brush Ground Brush grounds are very similar to field grounds. To detect a grounded brush, analyze the time domain of the line-toneutral voltage waveform. Under no-fault conditions, the field voltage will be of significant amplitude as shown in Figure 13. When there is a grounded brush, the voltage will be very low as shown in Figure 14.

Figure 15 — No fault condition – 0.09% differential current

Figure 13 — Brush grounded to tester ground

Differential Current Differential current may be analyzed by comparing two current waveforms in the time domain. There are two predominant situations where differential current analysis may provide insight to fault conditions that may otherwise be overlooked. One of these is when two or more cables feed a single terminal such as those found in semihigh current situations. Another situation for which differential current analysis may be used is in comparing the A1 to A2 currents. Situations may occur in which one of the main power cables may have bypass current. Bypass current may occur from the

Figure 16 — High resistance fault condition – 2.44% differential current

66 high frequency switching found in dc drives. Differential current may also occur when alternate return paths offer a lower impedance than the primary feed cables. Numerical analysis is the primary methodology used when analyzing differential current. The deterministic differential will vary according to the application. For our study, we used a two percent differential (one percent equipment + one percent minimum differential) to compare two cables feeding a single terminal. One of these cables had a resistance inserted into the line to represent a high resistance connection of one of the cables in a multicable situation. Figure 15 shows a situation of balanced currents between two cables connected to A1 terminal. Figure 16 shows a situation in which one of the cables connected to A2 terminal has a high resistance connected in series to simulate a high resistance connection. Note the 2.44 percent differential current met our established criteria of at least two percent for a fault condition to be deterministic.

Summary Our research has shown the use of current and voltage signature analysis in both the time and frequency domains may provide useful insights to on-line fault analysis of dc motors. Many common faults such as shorted turns or commutator bars, grounded windings, and off magnetic neutral axis faults may be detected using on-line current and voltage signature analysis. Trending these over time may provide an indication of a developing fault in the motor. As with all tests, cross correlation between technologies is imperative in the decision making process. David L. McKinnon received his BS in Electrical Engineering from New Mexico State University in 1991 and MBA from the University Of Phoenix in 2002. He has worked in the field of magnetics for over 14 years. During the past four years, he has worked for PdMA Corporation as a project manager for hardware and product development of motor test equipment.

On-Line Diagnostics Handbook — Volume 2

67

On-Line Diagnostics Handbook — Volume 2

Voltage Sag Testing for Commercial and Industrial Equipment NETA World, Spring 2008 Issue by Andreas Eberhard Power Standards Lab

Modern equipment can be sensitive to brief disturbances on the ac power mains. Electrical systems are subject to a wide variety of power quality problems which can interrupt production processes, affect sensitive equipment, and cause downtime, scrap, and capacity losses. The most common disturbance, by far, is a brief reduction in voltage, lasting for a few hundred milliseconds. These voltage sags, known as voltage dips in the IEC literature, are the most common power problem encountered. Besides fuse or breaker operation, motor starting, or capacitor switching that trigger voltage sags, they are also caused by short circuits on the power distribution system. These short circuits can be caused by snakes getting across insulators, trenching machines hitting underground cables, lightning ionizing the air around high-voltage lines, etc.

Figure 1 — Voltage sag immunity testing has been common in the semiconductor industry for years, where it has proved its economic value. New IEC standards for voltage dip immunity will expand this kind of testing and certification to any other industry.

A decade ago, the solution to voltage sags was to try to fix them – store up enough energy somehow, and release it onto the ac mains when the voltage dropped. Some of the old solutions included UPS, flywheels, and ferroresonant transformers. More recently, engineers have realized that this is really a compatibility problem and it has at least two classes of solutions. The power quality can be improved or the load equipment can be made more tolerant of the voltage sags. The latter approach is called voltage sag immunity, and it is the basis of several compliance standards. Voltage sag immunity testing has been common in the semiconductor industry for years, where it has proved its economic value. New IEC standards for voltage dip immunity will expand this kind of testing and certification to any other industry.

Standards for Voltage Sag Immunity We will discuss the three main voltage sag immunity standards in this article: IEC 61000-4-11, IEC 61000-4-34, and SEMI F47. However, there are also many other voltage sag immunity standards including IEEE 1100, CBEMA, ITIC, Samsung Power Vaccine, international standards, and military standards from the Department of Defense. IEC 61000-4-11 and IEC 61000-4-34 are closely related standards. They both cover voltage dip immunity. IEC 61000-4-11 Ed. 2 covers equipment rated at 16 amperes per phase or less. IEC 61000-4-34 Ed. 1 covers equipment rated at more than 16 amperes per phase and was written after IEC 61000-4-11, so it has better explanations. SEMI F47 is the voltage sag immunity standard used in the semiconductor manufacturing industry. It is used both for semiconductor equipment and for components and subsystems in that equipment. Enforcement is entirely customer-driven. The purchasers of semiconductor equipment know the economic consequences of sag-induced failures and generally refuse to pay for new equipment that fails the SEMI F47 immunity requirement. SEMI F47 is going through its five-year revision and update cycle.

68 All three standards specify voltage sags with certain depths and durations, for example, 70 percent of nominal for 500 milliseconds. The percentage is the amount of voltage remaining, not the amount that is missing. These sags are applied to the equipment under test (EUT). Each standard specifies pass-fail criteria for EUT when a voltage sag is applied. The IEC standards have a range of pass-fail criteria, but the SEMI F47 standard is more explicit.

On-Line Diagnostics Handbook — Volume 2 shifts during your phase-to-phase sags – something that sag generators designed for these standards do automatically. Typical suppliers of compliant sag generators include Keytek (www.keytek.com), Power Standards Lab (www.PowerStandards.com), and Schaffner (www.schaffner.com) . The IEC standards require phase shifting during sags on 3-phase systems, but sags on all three phases simultaneously are not required.

Test Equipment Required

Figure 2 — A typical example for a voltage sag ride-through curve that is used in the process industry.

Three-Phase Testing For three-phase EUT’s, the sags are applied between each pair of power conductors, one pair at a time. If there is a neutral conductor, this implies that there are six different sags at each depth-duration pair; three different phase-tophase sags, and three different phase-to-neutral sags. If there is no neutral conductor, there are just three different sags at each depth-duration pair in the standard; just three different phase-to-phase sags. In all of the standards, all three phases are never sagged at the same time. Note that IEC 61000-4-11 and 61000-4-34 specifically forbid creating phase-to-phase sags by sagging two phase-to-neutral voltages simultaneously, an approach that is permitted in SEMI F47. Instead, you must create phase

A voltage sag generator is a piece of test equipment that is inserted between the ac mains and the EUT. It generates voltage sags of any required depth and duration. Some include preprogrammed sags for all of the IEC, SEMI or MIL standards. Because a common EUT failure mechanism is a blown fuse or circuit breaker during the current inrush after a voltage sag, the sag generator must be specified for delivering large peak currents — typically in the hundreds of amperes. This peak current requirement in the IEC standards means that electronic amplifier ac sources generally can only be used for precompliance testing, not certification. Voltage sag generators must handle hundreds of amperes at three-phase voltages, while still staying portable. Built-in standards help speed up testing; built-in digital oscilloscopes help the test engineer diagnose EUT problems. Portability of sag generators is a key consideration. It is often impossible to bring larger room-sized industrial equipment to a test lab. Instead, the test lab must come to the equipment, bringing a sag generator. In general, the largest portable sag generators can handle no more than 200 amperes per phase at 480 volts. Some of the standards, such as SEMI F47, offer specific advice about how to test EUT’s that require more than 200 amperes by breaking them down into subsystems. Many conformance certification labs subcontract voltage sag testing to labs that have engineers who have the training and experience both to perform the sag testing, and to help diagnose EUT failures. This is an especially attractive approach when certifying large, industrial loads. For smaller commercial and industrial loads, many labs choose to rent a voltage sag generator. Such a rental often comes complete with over-the-phone engineering support from an experienced sag testing engineer. This can be the best way to get started on voltage sag immunity testing

What Makes Voltage Sag Testing Different?

Figure 3 — The IEC standards require phase shifting during sags on 3-phase systems, but sags on all three phases simultaneously are not required.

Unlike most other emissions and immunity testing, the test engineer must control and manipulate all of the power flowing into the EUT. For smaller devices such as personal computers, this is not a great challenge. But for larger industrial equipment, perhaps rated at 480 volts, 3-phase, at 200 amperes per phase, with an expected inrush current of 600 amperes or more, the test engineer must be prepared for serious performance and safety challenges.

69

On-Line Diagnostics Handbook — Volume 2

This kind of testing requires a fully-functional EUT and someone who knows how to operate it. The only way to determine if an EUT is immune to the required voltage sags is to have it fully operating during the voltage sags. In many cases, the sags will need to be applied during different stages of EUT operation. Often, the EUT is not ready on time for voltage sag testing. Development work may need to be completed, or no one is available to operate the EUT, or the supplies to operate the EUT (raw materials, cooling water, compressed air, etc.) are not available, or the EUT software is not functioning correctly. Test engineers should plan for these kinds of problems. EUT failure mechanisms can be complicated, too, and the test engineer will be expected to help diagnose them. The built-in digital oscilloscopes in most sag generators will help, but the test engineer must figure out where to connect the channels to circuits inside the EUT.

Common EUT Failure Mechanisms During Voltage Sags Figure 4 — Voltage sag generators like this Industrial Power Corruptor from Power Standards Lab handle hundreds of amps at three-phase voltages, while still staying portable. Built-in standards help speed up testing; built-in digital oscilloscopes help the test engineer diagnose EUT problems.

Figure 5 — The voltage sag test engineer will insert a sag generator between the ac source and the Equipment Under Test. Often, high currents (200 amps) and high voltages (480 volts 3-phase) must be handled.

The voltage sag test engineer will insert a sag generator between the ac source and the equipment under test. Often, high currents and high voltages must be handled. Some software comes with extensive safety checklists. Some of the checklist items are obvious such as who on the test team is trained in CPR and the location of the closest fire extinguisher. Some are less obvious such as how to get access to at least two upstream circuit breakers.

The most common failure mechanism is the obvious one: lack of energy. This can manifest itself in something as simple as insufficient voltage to keep a critical relay or contactor energized to something as complex as an electronic sensor with a failing power supply giving an incorrect reading, causing EUT software to react inappropriately. The second most common failure mechanism, surprisingly, occurs just after the sag has finished. All of the bulk capacitors inside the EUT decide to recharge at once, causing a large increase in ac mains current. This increase can trip circuit breakers, open fuses, and even destroy solid-state rectifiers. Most design engineers correctly protect against this inrush current during power cycling, but many do not consider the similar effects of voltage sags. EUT’s that are tested with sag generators that lack sufficient current capability will incorrectly pass the test if there is insufficient current available to blow a fuse or trip a circuit breaker in a half-cycle. Another common EUT failure mechanism is a sensor detecting the voltage sag and causing the EUT to shut down. In a straightforward example, a three-phase EUT might have a phase-rotation relay that incorrectly interprets an unbalanced voltage sag as a phase reversal and consequently shuts down the EUT. A more obscure example is an airflow sensor mounted near a fan that might detect that the fan has slowed down momentarily. The EUT software might misinterpret the message from this sensor as indicating that the EUT cooling system has failed. In this case, a software delay is the obvious solution to improve sag immunity. Yet another common EUT failure mechanism involves some obscure sequence of events. For example, a voltage sag is applied to the EUT, and its main contactor opens with a bang. But further investigation reveals that a small relay wired in series with the main contactor coil actually opened, because it received an open relay contact from a stray water sensor. That sensor, in turn, opened because its small 24 volt

70

On-Line Diagnostics Handbook — Volume 2

Figure 6 — Top graph shows a typical voltage sag. The next graph shows the current waveform, which was about 40 amperes peak before the sag, but increases to 450 amperes peak after. The next graph shows the same current, this time as an rms value. Before the sag, it is about 23 amperes rms (this equipment was rated at 30 amperes), but after the sag the current increases to 175 amperes rms – typical behavior of an EUT. The final graph shows the output of a dc supply during this sag.

dc supply output dropped to 18 volt during the sag. In this case, the solution is an inexpensive bulk capacitor across the 24 volt dc supply. Many other failure mechanisms can take place during voltage sags. The question to the test engineer will always be: how do we fix this problem? Usually, there is a simple, low-cost fix once the problem is identified. Only in extreme cases should devices that eliminate voltage sags on the ac circuit be considered as this is the most expensive possible solution. With the ever-increasing use of sophisticated controls and equipment in industrial, commercial, institutional, and governmental facilities, the continuity, reliability, and quality of electrical service has become extremely crucial to many power users. The power quality is unlikely to get better in the future, so the ultimate goal for any product manufacturer is to make its product immune to voltage sags. Just as we want modern cars to be able to drive through the frequent bumps and potholes that exist on the road today, we should expect every electrical product to be able to ride through the voltage sags that we have come to expect with today’s power quality.

Figure 7 — Picture of latest low-cost Power Quality Sensor PQube® for permanent installation.

Andreas Eberhard is member of various power quality and safety standard committees around the world. He has over 10 years experience in product compliance based on international standards. He is Vice President of Technical Services at Power Standard Labs.

NETA Accredited Companies The following is a listing of all NETA Accredited Companies as of August 2011. Please visit the NETA website at www.netaworld.org for the most current list. A&F Electrical Testing., Inc...................................................................................Kevin Chilton Advanced Testing Systems ............................................................................Patrick MacCarthy American Electrical Testing Co., Inc. ......................................................................Scott Blizard Apparatus Testing and Engineering ....................................................................... James Lawler Applied Engineering Concepts .................................................................... Michel Castonguay Burlington Electrical Testing Company, Inc. ........................................................... Walter Cleary C.E. Testing, Inc. ........................................................................................... Mark Chapman CE Power Solutions of Wisconsin, LLC............................................................. James VanHandel DYMAX Holdings, Inc. ....................................................................................... Gene Philipp Eastern High Voltage ....................................................................................... Joseph Wilson ELECT, P.C. .................................................................................................Barry W. Tyndall Electric Power Systems, Inc. .................................................................................. Steve Reed Electrical and Electronic Controls ..................................................................... Michael Hughes Electrical Energy Experts, Inc............................................................................... William Styer Electrical Equipment Upgrading, Inc. .......................................................................Kevin Miller Electrical Maintenance & Testing, Inc........................................................................ Brian Borst Electrical Reliability Services ..................................................................................Lee Bigham Electrical Testing, Inc. ................................................................................. Steve C. Dodd Sr. Elemco Services, Inc. ...................................................................................... Robert J. White Hampton Tedder Technical Services ....................................................................... Matt Tedder Harford Electrical Testing Co., Inc. ................................................................... Vincent Biondino High Energy Electrical Testing, Inc..................................................................... James P. Ratshin High Voltage Maintenance Corp. ........................................................................... Eric Nation HMT, Inc. .........................................................................................................John Pertgen Industrial Electric Testing, Inc. ........................................................................ Gary Benzenberg Industrial Electronics Group ................................................................................. Butch E. Teal Industrial Tests, Inc. .............................................................................................. Greg Poole Infra-Red Building and Power Service ............................................................ Thomas McDonald M&L Power Systems, Inc. .................................................................................. Darshan Arora Magna Electric Corporation ................................................................................... Kerry Heid Magna IV Engineering – Edmonton ...................................................................Jereme Wentzell Magna IV Engineering (BC), Ltd. ........................................................................ Cameron Hite

Setting the Standard

MET Electrical Testing, LLC .......................................................................... William McKenzie National Field Services...................................................................................... Eric Beckman Nationwide Electrical Testing, Inc. ...............................................................Shashikant B. Bagle North Central Electric, Inc. ...............................................................................Robert Messina Northern Electrical Testing, Inc. .......................................................................... Lyle Detterman Orbis Engineering Field Service, Ltd. ....................................................................... Lorne Gara Pacific Power Testing, Inc. ...................................................................................Steve Emmert Phasor Engineering ........................................................................................... Rafael Castro Potomac Testing, Inc. ........................................................................................... Ken Bassett Power & Generation Testing, Inc.......................................................................... Mose Ramieh Power Engineering Services, Inc. ..................................................................... Miles R. Engelke POWER PLUS Engineering, Inc. ...................................................................Salvatore Mancuso Power Products & Solutions, Inc. ........................................................................ Ralph Patterson Power Services, LLC ........................................................................................ Gerald Bydash Power Solutions Group, Ltd ...........................................................................Barry Willoughby Power Systems Testing Co. ............................................................................... David Huffman Power Test, Inc. ..............................................................................................Richard Walker POWER Testing and Energization, Inc. ............................................................... Chris Zavadlov Powertech Services, Inc. ................................................................................... Jean A. Brown Precision Testing Group .................................................................................... Glenn Stuckey PRIT Service, Inc. ........................................................................................ Roderic Hageman Reuter & Hanney, Inc....................................................................................... Michael Reuter REV Engineering, LTD ................................................................................ Roland Davidson IV Scott Testing, Inc................................................................................................Russ Sorbello Shermco Industries ............................................................................................... Ron Widup Sigma Six Solutions, Inc. ....................................................................................... John White Southern New England Electrical Testing, LLC ................................................. David Asplund, Sr. Southwest Energy Systems, LLC .......................................................................Robert Sheppard Taurus Power & Controls, Inc. ............................................................................... Rob Bulfinch Three-C Electrical Co., Inc.................................................................................James Cialdea Tidal Power Services, LLC ....................................................................................Monty Janak Tony Demaria Electric, Inc. ............................................................................ Anthony Demaria Trace Electrical Services & Testing, LLC ...................................................................Joseph Vasta Utilities Instrumentation Service, Inc. ........................................................................Gary Walls Utility Service Corporation.................................................................................. Alan Peterson Western Electrical Services ......................................................................................Dan Hook

Setting the Standard

About NETA NETA (InterNational Electrical Testing Association) is an association of leading electrical testing companies; visionaries, committed to advancing the industry’s standards for power system installation and maintenance to ensure the highest level of reliability and safety. NETA is an accredited standards developer for the American National Standards Institute (ANSI) and defines the standards by which electrical equipment is deemed safe and reliable. NETA is also the leading source of specifications, procedures, testing, and requirements, not only for commissioning new equipment but for testing the reliability and performance of existing equipment.

QUALIFICATIONS OF THE TESTING ORGANIZATION An independent overview is the only method of determining the long-term usage of electrical apparatus and its suitability for the intended purpose. NETA Accredited Companies best support the interest of the owner, as the objectivity and competency of the testing firm is as important as the competency of the individual technician. NETA Accredited Companies are part of an independent, third-party electrical testing association dedicated to setting world standards in electrical maintenance and acceptance testing.

Hiring a NETA Accredited Company assures the customer that: • The NETA Technician has broad-based knowledge — this person is trained to inspect, test, maintain, and calibrate all types of electrical equipment in all types of industries. • NETA Technicians meet stringent educational and experience requirements in accordance with ANSI/NETA Standard for Certification of Electrical Testing Technicians, (ANSI/NETA ETT). • A registered Professional Engineer will review all engineering reports. • All tests will be performed objectively, according to NETA specifications, using calibrated instruments traceable to the National Institute of Science and Technology (NIST). • The firm is a well-established, full-service electrical testing business.

CERTIFICATION NETA Certified Technicians conduct the tests that ensure that electrical power equipment meets the ANSI/NETA standards’ stringent specifications. Certification of competency is particularly important in the electrical testing industry. Inherent in the determination of the equipment’s serviceability is the prerequisite that individuals performing the tests be capable of conducting the tests in a safe manner and with complete knowledge of the hazards involved. They must also evaluate the test data and make an informed judgment on the continued serviceability, deterioration, or nonserviceability of the specific equipment. NETA, a nationally-recognized certification agency, provides recognition of four levels of competency within the electrical testing industry in accordance with ANSI/NETA Standard for Certification of Electrical Testing Technicians, (ANSI/NETA ETT).

Setting the Standard