Lessons from a Major Motor Refurbishment Project

Insulation system refurbishment and upgrade project for 50 motors ranging from 1¼ to 6000 hp incorporates a variety of test methods to uncover defects and provide predictive maintenance baseline for motors returned to service.

In the fall of 1999, JEA (formerly Jacksonville Electrical Authority) initiated a major project to upgrade the insulation and mechanical systems of 50 electric utility power plant motors ranging from 1¼ hp to 6000 hp. The project was part of a broader effort by the utility to sustain a high level of availability and assure reliability in anticipation of the more competitive deregulated power generation market.

A variety of test methods were used to evaluate the motors—hipot, surge, Motor Circuit Evaluation (MCE), (from PdMA Corp., Tampa, FL) vibration, growler, core loss, polarization index profile (PIP), and bearing insulation tests. PIP, a relatively new method of testing insulation systems, was used during refurbishment to evaluate progress in achieving a result consistent with the utility's goal of increased motor insulation reliability.

JEA has been engaged in a substantial improvement of its power generation facilities for the past few years. A major initiative to improve capacity of the power generation facilities includes addition of new and repowered units at its Northside Power Generating Station and other sites.

In addition, the maintenance program for the older units has been changed to improve their reliability through preventive and predictive maintenance enhancements, rotating spares and inventory improvements, improved shops and tools, training, better planning, and implementation of a new computerized maintenance management system.

The results have been impressive, illustrated by the fossil unit system availability rising from a range of 51-74 percent in the 1980s to values consistently in the range of 90-94 percent in 1991 through 1999. This places JEA among leading utilities in North America in terms of its ability to meet customer needs.

In addition, through these and other initiatives, such as diversifying the fuel mix useable by its plants, stockpiling oil when prices are low, and using less of it for power generation whenever feasible, the utility has reduced the average cost of power. Residential customers paid an average of $83.60 in 1981. By May 2000 this had been reduced to an average $68.15, a reduction of 18.5 percent.

JEA's Northside station is the wholly owned centerpiece of the utility. Two of three original units built in the 1970s are currently capable of producing 793 MW using oil and natural gas as fuel. A third unit is being repowered to deliver 295 MW. Four combustion turbine generators of 52 MW each provide peaking power at this site.

Unit 3 has been in operation for more than 20 years. Motors of this unit, especially those that are outside in a high humidity, sea-air, semitropical environment, were beginning to show signs of physical deterioration. Other than vibration and lubricant analysis, there had been little effort in the past to assess condition of the motor internals from either electrical or mechanical standpoints.

Once the JEA Predictive Maintenance Department began to acquire knowledge of motor failure modes and apply new technologies that could collect data to begin the process of condition assessment, many more questions were raised concerning reliability and condition than could be answered easily. Seizing upon the opportunity presented by an extended outage for a major controls and instrumentation upgrade in the fall of 1999, JEA management approved a project to refurbish and upgrade, where possible, all the motors essential to reliability of Unit 3.

Project launch
The motors to be included in the project were identified and the list included in a request for price under the competitive bid process of JEA. Early in the solicitation period, all prospective bidders were advised that the JEA Predictive Maintenance Department would take a highly active role in motor condition assessment and the decision-making process for work to be accomplished over and above that which was to be included in their bid price.

The lead engineer of the department was designated as project leader, empowered to make all of the decisions necessary to meet JEA goals. A project fiscal target was established by the project leader and agreed to by JEA management to cover both the initial bid price and any changes and additions deemed necessary based on condition assessments on the motors included in the package.

The JEA specifications for motor refurbishment, which were quite detailed and proven reasonably effective over many years, provided the essential information for the competitors to provide reasonable price quotes. The quotes ranged up to 300 percent of the winning bid submitted by Eastern Electric Apparatus Repair Co. Work was carried out at the contractor's Baldwin and Orlando, FL, facilities.

Refurbishment steps
The sequence of refurbishment for most of the project motors was as follows:

  • Remove from plant
  • Transport to designated refurbishment facility
  • Record nameplate and other characterization data observable without disassembly
  • Test resistance-to-ground to assure safe value for energizing motor in the shop
  • Run motor on test stand for warm-up and vibration analysis
  • Conduct off-line, low voltage Motor Circuit Evaluation (MCE) testing
  • Disassemble, visually inspect, and record conditions and certain characteristics observed (e.g., number of rotor bars, type of bearings installed)
  • Clean and dry components and conduct post cleaning search for mechanical defects
  • Conduct hipot, surge, core loss, growler, and bearing insulation tests as appropriate to the type and size of motor to be refurbished
  • Project leader and shop supervisor confer on out-of-scope repairs and any modifications to be made
  • Make electrical and mechanical authorized modifications, repairs, and restorations (including dip or over-spray and cure winding insulation)
  • Dynamically balance rotating elements and reassemble motors
  • Install new or additional nameplate with latest motor characteristic information
  • Perform post refurbishment shop tests (motor circuit analysis and vibration analysis)
  • Transport to plant and reinstall
  • Run post installation (baseline) tests (MCE test after electrical connections made and before energizing for the first time, and vibration analysis and on-line electrical analysis at earliest possible time after startup)
  • Make follow-up adjustments as required (e.g., to correct any post installation circuit resistance unbalance and to adjust mechanical alignmentthe most common post installation problem).

There were some exceptions to the sequence, made for economic reasons. For example, after four motors were delivered to the refurbishment activity for confirmation of nameplate data and before they were disassembled, a "replace vs refurbish" decision was made. The estimated cost for basic refurbishment (before determination of need for any additional work) was quoted be about 50 percent of the cost of new motors (new ones averaging $550) with equal or better characteristics (such as higher level insulation class, efficiency, etc.). In addition, the new motors carried a 2-year warranty, twice the length of those that would be refurbished. Since the new motors were immediately available, the decision was made to replace with new rather than to refurbish the old ones.

Some other small motors were not run on a test stand nor tested prior to disassembly, other than for resistance-to-ground. The reasons were conflicting work priorities, refurbishment schedule, and non-availability of JEA personnel at the refurbishment activity at the time scheduled for testing. A concerted effort was made to minimize the impact on the refurbishment progress of having nonshop personnel present or absent because of conflicting JEA work priorities.

JEA Predictive Maintenance Department personnel conducted vibration, MCE, hipot, and surge testing. Refurbishment shop personnel conducted resistance-to-ground, core loss, bearing insulation, growler, and dynamic balance tests.

Delivery condition

Testing and inspection upon delivery to the repair facilities revealed the following:

  • All motors tested for vibration showed readings that were quite low, reflecting the success of in-service efforts over many years
  • Three motors arrived with grounds that could not be cleared. These were scheduled for rewind with upgraded (Class F vs Class B) insulation systems with spike resistant magnet wire
  • The interiors of the stators (cores in the air gap and end turn areas) of seven motors were severely contaminated by bearing lubricant (oil)
  • Seven motors had substantial corrosion on internal and/or external surfaces
  • One large motor had cracked rotor bars, repaired by brazing
  • One large motor winding failed while running in the shop for vibration testing
  • The thermal insulation systems of five out of six large motors had substantial deterioration, requiring replacement. All of these motors had bent or missing (alignment) dowel pins
  • The bearing insulation on five of six large (greater than 2500 hp) motors had low readings of resistance
  • High potential and surge testing revealed no winding turn-to-turn or phase-to-phase defects or instability during testing
  • Using the MCE tester, windings of seven motors were found to be unstable, exhibiting rapidly changing insulation-to-ground readings, some to very low values
  • Core testing showed all motors to have relatively low values, well within the acceptable range of values indicating low (eddy current) core losses.

Other defects included damaged cooling fans (three motors); two worn couplings and two that were too tight; soft foot condition on three motors; defective heaters or cabling on two motors; one motor with a cracked end bell (repaired by welding); three motors with deteriorated phase leads requiring replacement; and four motors with no provision for safely greasing bearings while running, if needed.

Details of defect diagnosis and remedies for several motor groups follow.

Circulating water pump motors
A group of Allis Chalmers 600 hp, vertically mounted, guarded-drip proof, squirrel cage rotor motors is located in open areas closest to the source of cooling water for the plant, the St. Johns River. Humid air contains salt from the nearby Gulf Stream in the Atlantic Ocean.

The salt-laden air attacks the inside of the motor, exacerbating corrosion of any unprotected surfaces. In particular, surfaces in the passages of the rotor cores had corroded, resulting in reduction of their cross sections and cooling airflow.

Corrosion removal required sand blasting and use of rods inside the cooling passages to loosen it from the surfaces, restoring cross section areas. After cleaning, the rotors were dipped in a thinned insulating varnish and cured, a technique found to be a more effective anti-corrosion measure for this area than painting.

Corrosion on the stator cores was also visible, mitigated slightly by the presence of bearing lubricating oil that had leaked past the upper shaft seals. The oil coated air gap surfaces and winding end turns, capturing and retaining dirt carried inside the motor by the cooling air.

During vibration testing one of the motors in the shop, a winding failure occurred. As a result it was scheduled for rewind with an upgraded (Class H vs Class F) insulation system with spike resistant magnet wire, copolymer insulation, and vacuum pressure impregnation process.

Large fan motors
A common defect was found in six horizontally mounted, guarded-drip proof squirrel cage motors. The 6000 hp induced draft, 3000 hp forced draft, and 2500 hp recalculating gas fan motors were manufactured by Allis Chalmers using the same basic design. The common defect found in testing upon delivery was low resistance across the bearing insulation.

This was corrected by disassembling the frame for cleaning and replacement of insulating gasket material. Inspection during disassembly revealed many bent or missing bearing alignment (dowel) pins, which were replaced.

Five of six machines had blanket style fiberglass thermal insulation inside the airflow hoods that had begun to come apart or was hanging loosely from retainers into the airflow passages. This was replaced with fitted sheets of aluminum foil faced compressed fiberglass having the same insulating value and retained close to the interior surface of the hoods.

In addition, after evaluation of the original (all blue) paint scheme for the upper exterior surfaces of the airflow hoods, it was decided to paint them white, to reduce heat absorption and aid the thermal insulation to perform its function.

One of the induced draft fan motors had a history of one bearing repeatedly running hot during the last operating cycle. Supplemental cooling was often needed to keep the temperature below alarm level. Close examination during disassembly revealed that bearing caps had been switched, possibly the last time this motor had been in a repair shop. Returning bearing caps to their original locations eliminated this problem.

The same motor had an unstable resistance-to-ground reading, which was improved somewhat by cleaning, drying, insulating material overspray, and curing. However, the minimum value was never less than about 200 megohms during the restoration process, twice the minimum value of the latest IEEE standard 43-2000. Therefore, the motor was returned to service with this condition, because replacement of the winding was not otherwise justified economically at this time.

The rotor of one of the forced draft fan motors was found to have cracked rotor bar to shorting ring joints. These were repaired by rebrazing them. The same motor was found to have feet that were out of level. The feet were milled to restore the unit to a level condition.

Booster and fuel oil service pump motors
Shaft seals on seven horizontally mounted, squirrel cage induction motors (600 hp Westinghouse condensate booster, 500 hp Allis Chalmers feed water booster, and 350 hp Westinghouse fuel oil service pump motors) with oil lubricated sleeve bearing motors were leaking, allowing oil to infiltrate past them into the air gap and end turn areas of the motors. Since enclosures for these motors are open-drip proof designs, significant amounts of dirt brought in with cooling air had mixed with the oil, clogging air passages.

On these same motors, one sleeve bearing was severely damaged; five other sleeve bearings (on three motors) had excessive clearances; and four bearings (all those on two motors) were threaded. One condensate booster, one fuel oil service, and both feedwater booster pump motors had unstable resistance-to-ground readings.

Cleaning and drying cycle(s) (some required several) eliminated the oil and dirt from the windings. The stators were then dipped and cured to seal them. Final resistance-to-ground readings were stable at a minimum of 2000 megohms for one and greater than 3500 megohms for all the others.

Comprehensive refurbishing process
JEA project specifications covered preservation, preassembly balancing, reassembly, shop testing, and post installation testing.

Preservation—The specification for all motors in this project stressed restoration and upgrading of preservation systems of both internal and external surfaces. It also required specific primers, paints, coats, and minimum dry film thickness(es) on parts exposed to the open air. The smallest of the motors in this project (two-speed 1¼ hp totally enclosed, fan cooled, squirrel cage, induction motors with fans and externally finned casings) used to drive traveling screens at circulating pump intakes were a case in point.

Two of the four had been replaced within the past year. Three of four came into the shop with severe corrosion on external surfaces. Screens on the shrouds of older motors had been weakened by corrosion and required reinforcement. In the case of these motors, white paint was applied after thorough preparation of external surfaces, because they were directly exposed to sunlight, in addition to the presence of (brackish) moisture from the traveling screens on a continuous basis. The same attention to details of preservation was required for all motors in the project.

Preassembly balancing—All rotating elements were dynamically balanced prior to reassembly of the motors. Over 50 percent of the rotating elements had dynamic unbalance reduced by the addition or removal of weight, improving the chances for longer operating life before the next repair.

Reassembly—Sleeve bearings that exhibited any degraded condition and all rolling element bearings were replaced. All shaft seals were replaced and clearances checked to assure minimum lubricant infiltration into motor interiors. Common fasteners were replaced with new ones having the same specifications as the original equipment manufacturers. Custom fasteners were examined for defects and replaced in kind if not repairable.

A new, engraved nameplate was attached with added data beyond that of the old, including identification of bearings installed and number of rotor bars, as well as post refurbishment characteristics called for by the latest National Electrical Manufacturers Association Standards Publication MG-1.

Shop testing—Motors were run as close to normal operating voltage as possible and tested for vibration. All readings taken showed that the low levels that had existed before refurbishment had been sustained. Motor Circuit Evaluation test baseline readings were taken. Careful attention was paid to resistive and inductive unbalance, capacitance-to-ground, and resistance-to-ground values.

A rotor influence check (RIC) graph was performed as a quality check on the reassembly and for reference as a baseline for the future. The RIC graphically illustrates the mechanical-electrical-magnetic field relationship between the rotor and the stator. It is performed by taking a series of inductance readings on all three phases using identical test signals with the rotor at specified intervals of sequential rotation.

Pattern recognition techniques are used to analyze the graphs for such defects as static and dynamic eccentricity and rotor cage defects (such as cracked bars, shorting rings, and connecting joints). The test is done easily in a shop environment even on large motors. Often, performance of a RIC test is difficult if not impossible after a motor is installed in a plant because the motor is coupled to a load that can't be rotated easily when shut down.

The primary means for early detection of rotor defects in most cases is Motor Current Signature Analysis, data from which is sometimes hard to interpret. Once degradation is indicated with this on-line test, the motor can be shut down and uncoupled from its load for performance of an off-line RIC test to confirm the problem.

Further disassembly can be avoided if a comparison of the latest information with baseline test data done in the shop test shows no change in pattern away from what was known to be a rotor in sound condition. Conversely, the latest RIC test may confirm a degraded condition, justifying motor removal and further disassembly for inspection and repair.

Post installation testing—After each motor electrical hookup to its circuit, but before it is energized for the first time, another MCE test is performed to assure the entire circuit is balanced in resistance and inductance and has acceptable capacitance-to-ground and resistance-to-ground values.

As much of the circuit is included in this test as possible and the data used for baseline purposes, since all such routine tests in service are done from the same point (ideally at the motor control center) without disconnecting any part of the motor circuit, whenever possible.

After bringing the motor on line and loading it to the conditions specified for taking data, vibration and on-line motor power analysis testing is performed to establish the baseline for the refurbished motor in the new operating cycle. The most common defect found at this time is usually mechanical misalignment, leading to a work request for correction, primarily to protect motor bearings.

Lessons learned
The final phase of the project involved enumeration and analysis of lessons learned. Action taken on each of the resulting conclusions is indicated in the following discussion.

Specification for refurbishment of motors—The specification used going into the project was found during the actual work to be inadequate in many ways, even though it covered many items in great detail. It did not reflect some of the lessons learned as described below, nor did it reflect some of the best practices of leading electrical apparatus repair shops.

The specification covered both ac and dc motors, creating some confusion and unnecessary complexity. New specifications documents have been under development since then to correct the deficiencies revealed in the project. Refurbishment shop personnel helped this process with ideas on how to improve the specifications.

Types of insulating materials specified—The operating histories and conditions found in JEA motors with polyester insulating systems were consistently worse than those found in systems using epoxy materials. For that reason, the decision was made to prohibit the use of polyester based insulation materials in JEA motors requiring full winding replacement.

A decision was made to accept and specify copolymer based insulation systems (along with epoxy based systems) in JEA motors requiring rewind. This was based on study of the manufacturer's data and in-depth discussions with the refurbishment activity managers who had made the decision to shift their vacuum pressure impregnation (VPI) system to the new material. They are so confident with the decision that they offered a two-part warranty for motors rewound with the new product and their VPI process.

Once completed, they warrant the system for 3 years in storage, as long as certain (reasonable) conditions were met. Further, once placed in service anytime in the 3-year period, the system was warranted against failure for another 2 years.

None of the motors in the Northside station's Unit 3 has a variable speed drive. However, rewound machines are required to have spike resistant magnet wire installed.

The immediate reason for spike resistant magnet wire is that this plant location is subjected to numerous lightning strikes. Several motor windings have failed during the accompanying storms. The site has a single lightning surge capacitor set for protection of the entire facility. Individual busses and motor circuits are not individually protected.

Having spike-resistant magnet wire installed also opens the possibility for lower risk introduction of variable speed drives at some later date, should unit operating profile change and the option becomes economically justified. Some refurbishment facilities no longer use anything but magnet wire with spike resistant coating, so this requirement may eventually have little effect. Including it in refurbishment specifications reduces the risk of future problems, a key goal of this project.

Record of tests performed during the refurbishment—A long-range view was taken concerning future testing and condition assessment of motors at JEA. A succession of analysts over many years in the future in the Predictive Maintenance Department will need organized, easily recallable data in order to make good calls and informed decisions concerning motor conditions and repairs.

In addition, an easy method for comparing pre- and post-refurbishment data (such as vibration readings) was found to be needed in order to evaluate effects of actions taken during the refurbishment. Data recording sheets provided to the apparatus repair activity as attachments to the refurbishment specification were revised to reflect what was needed for anyone wanting to assess and compare past and future conditions. The written data for each motor are organized the same way, also.

Automated units such as MCE testers have their own data organization schemes. Results of some tests, such as core loss analysis, typically provided on a paper printout from the core test computer, cannot be placed into complete perspective unless the information on acceptance and rejection values for the given motor also are provided in writing.

Whole databases can be contaminated when data are included that are taken under different operating conditions or at different places.

For example, vibration data for a motor in the shop (typically with no load) should not be compared to vibration data from the same motor installed in the plant and operating under load. Electrical data taken at the motor connection box cannot be compared to data taken on the same motor from the motor control center.

Accordingly, separate databases were created to segregate compatible information for proper analysis.

Tagging of motor components—As noted previously, switched bearing caps on one induced draft fan motor caused extensive trouble for plant personnel during the previous operating cycle. Evaluation of the cause of this error resulted in the conclusion that improper tagging of components removed during disassembly in the last repair at a refurbishment activity might have been the cause.

Refurbishment activity personnel involved with this project were very diligent in their tagging process and had been provided by management with a tagging system that withstood the rugged environment of a repair facility.

JEA project personnel had observed that tagging systems in other facilities with which they were familiar were not as rugged or reliable as the one in use during this project. As a result it was concluded that any shop used to repair or refurbish JEA motors in the future would be required to demonstrate its system for tagging and prove its ability to stand up to the rigors of its process. This is now a prequalification, preaward requirement of JEA.

Motor lubrication—It was found during this project that at least two and possibly four motors were not accounted for in the Northside station's bearing greasing plan. The motors were modified so they could be greased safely, even when operating, if needed. In addition, it became obvious from poor conditions of sleeve bearings found during disassembly of at least seven of the motors in this project (almost 50 percent of those so equipped) that oil lubrication practices needed attention.

Related to the lubrication of motors is the finding that open, drip proof design motors accumulate dirt entrained in the cooling air, especially when the internals are contaminated by oil infiltrating through poorly fitted shaft seals. Some time in past years, the decision was made to install filters over the cooling air inlets of induced draft, forced draft, and recirculating gas fan motors.

The internals of these motors were found to be notably cleaner than those without such added protection. The conclusion reached is that filtering inlet air might help smaller motors having open, drip proof designs.

These findings along with appropriate recommendations were passed to responsible parties for action. To help make the cases for action, photographs taken during the refurbishment process are available for anyone doubting the findings and needing a little more convincing. The photos are included in a comprehensive report of the refurbishment project, organized for ready reference in the future.

Predictive maintenance ownership

The JEA's Predictive Maintenance Department has acquired "ownership" of motor electrical insulation systems, primarily because it owns the equipment and data used to monitor and analyze condition. To some degree, also, the department has become the "conscience" of the organization when it comes to best practices on both mechanical and electrical sides of motors.

In many cases personnel from this group are the only ones from the utility having contact with contract motor refurbishment activities. However, by making recommendations such as the addition of motors to the grease plan or filter media to the inlets of open design motors there is the potential for adding to the preventive maintenance burden of personnel at the plant. So, the personnel responsible for monitoring condition become the messengers bearing bad news of the need for corrective action and a commitment of a higher work burden for someone else.

In this case the actions taken to improve the motors during this refurbishment project help position JEA Northside Generating Station to sustain its already superb record of fossil unit system availability and customer service at lowest possible cost in the fast approaching deregulated market.

The authors gratefully acknowledge the invaluable support of the managers and staff at Eastern Electric Apparatus Repair Co. Baldwin and Orlando Facilities in granting unlimited access at all hours of the day and night and answering countless questions in the course of performing their work on the JEA motors that were the focus of this project. MT

Steve Tanner is lead engineer and Ray Russell is predictive maintenance specialist in the Predictive Maintenance Department at JEA, the municipal utility for Jacksonville, FL. Jack R. Nicholas, Jr., P.E., is CEO of Maintenance Quality Systems LLC