This article examines four alternatives used for core and winding repair when localized damage occurs due to a failure. It shows examples where some of these techniques were used, as well as the technical aspects of these repairs.
By Mauro Uemori, Edson Alves and Michel Spiridon
The most likely component of a hydroelectric plant to fail is the generator, according to a study by SINTEF.1 A second study by Enel on 250 machines shows that stator faults, although less frequent than other faults, are responsible for longer outages.2 Other studies on this topic have drawn similar conclusions.3
The conventional solution for these faults involves replacing the stator core and winding. This solution has many benefits, but it is costly and requires a long outage.
In some cases, the conventional solution is not preferred due to the cost and/or outage limitations. A solution must be found to allow for at least temporary operation of the unit until a permanent repair is performed. Developing an efficient repair method for these cases by reducing the cost and outage time of the machine is essential. This can make sense if there is localized damage that affected the core and stator winding, for example due to a short circuit and/or damage caused by a part that broke in the rotor.
This article compares the various repair methods used when one of these faults occurs, evaluating the benefits and impacts of each solution:
- The conventional solution is to completely replace the core and winding. This solution is used when the damage affected a significant area, making the repair difficult, or when possible, fixes are only temporary, limiting the use or reliability of the generator. It is also used when components are at the end of their useful life and/or design improvements are necessary.
- Another solution is partial replacement of the core and winding, meaning the core is partially disassembled and the damaged laminations are replaced. The winding in the affected area is also replaced. Other parts that have not been affected significantly are only repaired. This solution is typically used when the damage has affected a relatively small area, although sometimes such repair can make sense for relatively large damaged areas. For this case, this article presents an example where this solution allowed the replacement of damaged stator laminations and reinstallation of the stator winding in a short time.
- A third solution is core repairs and winding (without disassembly). The core is repaired without disassembly, while the winding can be partially disassembled and/or repaired. This solution is used when the damage affected a relatively small area, or in the case of a relatively large area, when economic factors such as operation requirements and repair reliability are important. This article presents an example where significant damage was fixed without core disassembly.
- Finally, cosmetic repairs are used typically when the damage is localized and minor. In some cases, they are also used to fix serious damage. The repair is necessary to prevent additional heating that will gradually affect the winding and core, leading to a significant failure over time. The term “cosmetic” is used because the damaged area is replaced by one insulating material, acting as a cosmetic component. The affected core area is removed by cutting, sanding and straightening the laminations. Usually such a repair is followed by an EL CID test to determine if a short between laminations remains. If the stator laminations are short-circuited, acid etching can be used to remove the shorts. A high capillary resin is then applied, and an insert is added to act as a cosmetic component.
The photos above provide an example of a cosmetic repair to a damaged winding and core.
Table 1 provides a comparison summary of these methods, including the outage duration, advantages and disadvantages of each.
Table 1 – Repair Methods
|Total Replacement||1 year or more; Rarely done in less than a year.||New components (improves the lifespan) Uprating (when possible)||High cost|
Long outage time
|Partial Replacement||1 month or more||Short outage period Elimination of damage due to fault||Requires spare parts to be available|
|Repair without replacement||1 month or more||Short outage period Damage repair due to fault||Requires spare parts to be available.|
|Cosmetic repair||1 Week or more||Short outage period||Unless root cause is addressed, new failures may occur|
Inspections and tests executed during the repair
A visual inspection is the first logical step of the damage investigation, to localize the damage and assess its extent and severity. Effectiveness of the visual inspection depends on physical access to the core and winding, so some disassembly may be required.
Most of the time, removal of generator cover plates will be necessary to access the winding end-turns. The removal of a cooler or the opening of cover plates on the generator frame can give access to the back of the core to assess possible deep-seated damage. It might be possible to look at the stator winding wedging system with the rotor in place, from the space available between two rotor poles. When this is not possible, removal of two to four, depending on the size of the unit, can facilitate access to the stator bore surface.
Depending on the type of damage (deep-seated damage after a short circuit or significant damage on the core air gap surface), it might also be required to remove the rotor before an inspection. Tools such as a set of mirrors or a borescope can ease the inspection. A borescope can be used to assess the condition of the stator winding radial wedging system in the air gap or to detect deep-seated damage from ventilation ducts.
During an inspection after a forced outage, attention should be paid to any unusual signs on and around the winding, such as foreign materials and debris. In one example described in this article, aluminum debris was found in the end-turns and ventilation ducts. Particular attention should be paid to unit cleanliness (contamination). Black dust may indicate overheating and help to localize a ground fault. Evidence of looseness, overheating, electrical deterioration, mechanical damage, dusting and corrosion could be signs that can help locate the fault and identify its root cause. The inspection should not be limited to the winding end-turns but also include the stator connections (group, serial and circuit ring connections), winding circular, phase and neutral busses, bracing elements (ties, bracing ring, etc.) and wedging system. The inspection should also include the ventilation ducts, bore and back surface of the core to help identify in-slot damages.
During the inspection, aspects such as generator history, previous maintenance and outages, and the unit’s condition before failure should be taken into account. Useful guidelines of aspects to be considered during an inspection, inspection forms and recommended tools are available.4 Additionally, IEEE Std 492-1999 provides more information on generator inspections.5
Description of tests
The main tests involved in this type of repair are basic and similar to tests done to evaluate the condition of the unit.6 A brief description of each test is below:
- EL CID: This test was developed to overcome the need for a high power supply required for rated flux testing. Usually, a variable transformer supplied from an electrical outlet is all that is required to excite the stator. Even if a low flux density excitation does not induce heat generation in EL CID, the faulty current created near laminations with short circuits is directly detected. The basis for the detection is that fault currents will have a mainly resistive component in respect to the supply voltage and will be shifted by 90 degrees in respect to the supply current. The currents are detected using a sensor (called Chattock coil) and phase separator. An advantage of EL CID is that it can be carried out with the rotor in place, by removing two or three poles. The results are in the form of traces that can be interpreted by skilled personnel. In general, a 100 mA reading requires close examination and possible rated flux testing to verify damage. The shape of the measured current also helps to locate a fault, whether it is on a stator tooth or deep-seated inside the lamination stack. Theory and more information about the test setup and its interpretation is available.7
- Core loop: The core loop test, also called core magnetization test, consists in applying 1 Tesla or nominal flux to the stator core using a toroidal winding. Heat develops at any point where the insulation is damaged. The temperature of the stator core is monitored using infrared camera and/or thermocouples. As a generally accepted criterion, a core is considered damaged and needing repair if a hot spot is detected, with temperature being 5 to 10 degrees Celsius higher than the surrounding area. Theory and more information about the test setup and its interpretation is available.8
- Insulation resistance: The insulation resistance test measures the ohmic value between a conductor (part of the winding) and the ground (stator core). The insulation resistance of a winding in good condition usually reads in Megaohm as insulation restrains the current flow from the conductor to ground. The lower the resistance, the higher the chance of insulation defect. This is probably one of the most widely used tests in diagnostics and maintenance. This test is usually done at a voltage that is lower than the generator rated voltage (usually 5 or 10 kV DC). The polarization index, which is the variation of resistance over time, also helps to identify contamination problems. Theory and more information about the test setup and its interpretation is available.9
- D.C. ramp: This test is a variation of the DC high potential test. The main difference is that the voltage is increased progressively, usually at a constant rate of 1 or 2 kV/minute. The evolution of the leakage current over the voltage/time is recorded and displayed continuously for direct interpretation. Analysis of the curve helps identify existing or imminent winding faults as well as characterize it, as shown in section 7.8.2 of IEEE Std 95.10. The main advantage of the DC ramp test over the traditional DC high potential test is that it allows detection of an imminent failure, while giving the opportunity to stop the test before the failure, to avoid a puncture or damaging a winding even more. In the context of winding troubleshooting, this test can be used in conjunction with the insulation resistance test to detect hidden faults while avoiding additional winding damage. By comparing phases, circuits or equal portions of a winding, winding defaults can also be located. However, as this is a high-voltage test, the number of repeat tests should be limited to reduce stress on the insulation.
- D.C. resistance (or conductivity): The purpose of this test is to find damaged copper conductors or joints by resistance or conductivity measurements. A low-resistance ohmmeter or Kelvin bridge can be used. By comparing the resistance of phases, circuits or portions of a winding, differences observed will help locate a fault. If it is possible to easily “open and divide” a stator winding, this test can be used iteratively to locate a fault in a short amount of time (one or two days of site work) Temperature should be taken into account as the DC resistance is strongly affected by temperature.
- Continuity: This test consists of supplying the damaged winding with an AC current while measuring the current though a portion of the winding where a fault is suspected. A significant current drop should be measured on a bar/coil where the insulation is damaged. This test is particularly useful when it is not possible to easily “open and divide” a winding. The use of a flexible current probe enables measuring the current through a bar, coil, group or serial jumper without opening the electrical circuit. An AC source or even a welding machine can be used as only a small current is required. An insulation resistance test might be done before performing this test on separate phases/circuit to identify the faulty phase/circuit first. The tester should have a good knowledge of the winding diagram before performing the test. Care should be taken during this test, as additional damage could be caused to the winding. If the insulation is significantly damaged, a direct path to ground for the current could burn the insulation even more. For this purpose, several people should be distributed around the winding and specifically at the suspicious portion of winding to look for signs of smoke or a burning smell immediately after the winding has been energized.
Core evaluation and/or repair
An EL CID test is highly recommended, and it helps during the repair work but is often not seen as sufficient. As a final test (after the repair has been carried out), a loop test or core magnetization test may be performed to detect hot spots in the core. In cases where the winding is installed and a magnetization test is performed, it is advisable to reduce the flux density in the yoke. Normally a flux density of 1.0 T or less is recommended. This level of flux density can help mitigate or avoid vibration risks in the machine, reducing the risk of damaging the winding while still producing reasonable results.
Stator winding using coils
The most common tests involve DC resistance, continuity, insulation resistance between strands (usually at low voltage), short-circuit between turns, DC Hi pot test (coils replaced), insulation resistance in the winding and DC ramp test (same voltage level as the one used during maintenance).
Stator winding using bars
The most common tests involve DC resistance, continuity, DC Hi pot test (bars replaced), insulation resistance in the winding and a final DC ramp test (same voltage level as the one used during maintenance).
Core and winding repairs
To validate the repair methods proposed in the article, two examples are shown:
- Example 1 – The unit is a generator/pump with a water-cooled winding (390 MVA, 18 kV, 20 poles) and the failure consisted of a rotor-stator contact. During a routine unit shutdown, two aluminum rotor inter-pole supports dislodged and scraped both stator winding and core, causing severe damage to the stator laminations and winding end-turns.
- Example 2 – The unit is a generator (133 MVA, 16.5 kV, 40 poles) that suffered a phase to ground short circuit (fault near the end of the stator core). This short quickly became a two-phase short circuit due to a failure in the cable between the ground cubicle and the neutral of the generator.
Example 1 – Repair without disassembly
Two interpolar aluminum supports broke and caused rotor/stator contact. The stator core was heavily contaminated by aluminum debris, mainly in the first and last packages (upper and lower). About 25% of the core was damaged (~130 m2). In addition, 67 upper bars were damaged, to varying degrees. Damage also occurred in the rotor and air deflectors.
The first step performed was a detailed visual inspection of the damage. Thereafter, the core and winding were cleaned, and preliminary tests were performed to verify the extent of the damage. These tests included insulation resistance and Hi pot (ramp DC), Ohmic winding resistance and EL CID (full core). In the EL CID test, 25% of the core surface had a current measured above 100 mA. The test performed showed that despite the significant area, the damage was superficial.
The photos above are from the visual inspection of the stator bars with conductor damage
One of the damaged areas was chosen to determine and validate the best repair process. Brushes, magnifiers, and belt and rotating sanders were used for cleaning of each spot damaged.
A damaged stator core is visually inspected and sanded
After sanding, an EL CID test was done in the selected area and acid etching was used to remove the short between laminations. Another EL CID test was performed, and the process was repeated until the shorts between laminations were eliminated.
Figure 1: EL CID initial results (at top) and stator core repair (at bottom)
Once the method was validated, a complete core repair was started. After the rework, a core magnetization test was performed, and hot spots were observed (where the temperature was more than 10 K above the surrounding areas). Localized repairs and EL CID were performed again at these areas and magnetization tests were repeated. This process was repeated until all spots were treated.
Bar damages were also repaired, with seven bars replaced. All winding wedges were replaced. The total unit outage period for repairs was five months. The unit has been in operation since 2012.
Example 2 – Repair with partial disassembly
The machine suffered a phase to ground short circuit (fault near the end of the stator core). This short quickly became a two-phase short circuit due to a failure in the cable between the ground cubicle and the neutral of the generator. The failure resulted in two separate and prolonged high-intensity discharges with significant damage to the core and winding.
In addition to the core, four stator bars were damaged beyond repair and several bars suffered damages in the insulation. Group connections, circuit rings, neutral terminals and current transformers were also affected.
Visual inspection of a stator core with damage to the first packet
The first step was to perform a detailed visual inspection of the damaged area. Soon after, a core and winding cleaning was performed, and preliminary tests were carried out to verify the extent of the damage. These tests included insulation resistance and Hi-pot (DC ramp test) and ohmic resistance of the winding. Based on this assessment, a repair with partial disassembly was the best option.
The repair consisted of:
- Replacement of damaged laminations, pressure fingers and end plates;
- Replacement of four fully damaged bars;
- Removal of 37 top bars and 32 bottom bars to allow the lamination replacement;
- Replacement of group connections (pole jumpers) in the damaged area;
- Replacement of damaged circuit rings;
- Replacement of the neutral terminals damaged and current transformers in the affected phase.
The first step in the repair was the removal of the bars, end plates and pressure fingers between slots 34 and 74 (damaged area, see photos below). A support was installed and the laminations on the end were hoisted, allowing removal of the overlapped laminations. This process continued until all damaged laminations were replaced with new laminations.
Figure 2: Diagram (at top) and photo (at bottom) of the area where the laminations were removed
To verify and validate the repair, a core magnetization test was performed. No hot spots were detected (see Figure 3). The test could also have been done by means of EL CID, but a magnetization test was the best option due to the location of the fault (extremity of the core).
Figure 3: Results of the stator core magnetization test
After installation of the circuit rings, group connections and repairs in the neutrals leads, the final tests were performed (insulation resistance, Hi pot and ohmic resistance) and the machine was returned to operation.
The total outage period for repairs was a month. The unit has been in operation since 2015.
While the conventional solution for replacing the core and stator winding has many benefits, sometimes it is not the most effective one when cost and/or outage limitations are taken into account. In the two examples, the repair outages were five months and one month. A typical winding and core replacement outage would have taken 18 to 24 months from notice to proceed.
Alternative solutions such as localized core repair with partial lamination replacement or repair without lamination replacement should be reviewed because they reduce the machine’s outage time and cost. This is especially true if there was localized damage that affected the core and winding of the stator, for example after a short circuit and/or damage caused by a part that has broken in the rotor. It is important to emphasize that such fixes are challenging and need an in-depth understanding of the machine and a skilled repair crew.
1 Bakken, B.H., et al, Start and stop costs for hydropower plants [in Norwegian], Report #TR A5351, 2001.
2 Galasso, G., and M. Marke, “New developments in diagnostic and monitoring techniques for hydro-generators, Waterpower and Dams conference, Barcelona, Spain, 1995.
3 Dehlinger, N., J. Figueroa and D. De-la-Garza, “Stator Core and Winding Repairs after Extensive Damage,” Proceedings of HydroVision International 2013, Clarion Energy, Tulsa, Okla.
4 Kerszenbaum, Isidor, “Inspection of Large Synchronous Machines: Checklists, Failure Identification, and Troubleshooting,” Wiley-IEEE Press, 1996.
5 IEEE Std 492-1999 – IEEE Guide for Operation and Maintenance of Hydro-Generators
6 Dehlinger, N., E. Alves and C. Messier, “Generator condition assessment, an efficient tool to prevent unscheduled outages and minimize their duration,” Proceedings of HydroVision International 2015, Clarion Energy, Tulsa, Okla.
7 Rickson, C., “Electrical Machine Core Imperfection Detection,” IEE Proceedings, Vol 133, Pt B, N 3, 1986.
8 IEEE Guide for Diagnostic Field Testing of Electric Power Apparatus – Electrical Machinery, IEEE 62.2-2004.
9 Recommended Practice for Testing Insulation Resistance of Rotating Machinery, IEEE Std 43 – 2013.
10 IEEE Recommended Practice for Insulation Testing of AC Electric Machinery (2300 V and Above) With High Direct Voltage, IEEE Std 95 –
Mauro Uemori is consultant engineer for GE. Edson Leite Alves is senior engineer for GE. Michel Spiridon is chief engineer at the GE Technology Center in Switzerland.