Predicting the Remaining Life of Generator Components

The winding of hydro generators operating for decades can suffer from damage, such as rust and erosion of the slot corona protection.

Faults on the stator windings account for about 40% of the outage time of hydro generators and result in time-consuming repair outages and serious loss of production. The authors present a method to objectively assess the aging process of high-voltage stator windings where only limited visual inspection is possible. This method can be used to assess the remaining life of the entire generator as well as its individual core components.

By Christoph Wendel, Carl-Ernst Stephan, and Thomas Kunz

Regular maintenance enhances the operational reliability and service life of a piece of equipment, such as a hydro generator. Thus, from the economic point of view the following question arises: Up to what point is it economical to continue with maintenance and repair, or from when would it make more sense to replace complete components or subassemblies, such as the stator winding?

Signs of wear, tear, and aging start to appear with increasing operating time of the equipment. Aging can be defined as the diminishing of the required technical qualities over the course of time. When the risk of an unplanned outage increases significantly or a component experiences a major failure, the end of the lifetime is reached.

The terms ‘lifetime’ and ‘aging’ are not absolute or bound to a fixed rule of expectation. Not least of all, the availability and mode of operation play important roles. When a generator in a run-of-river station reaches a lifetime of 30 years, this appears rather short when considering the mode of operation. On the other hand, 30 years appears long relative to the expectations for a motor-generator in a pumped-storage power station. This variance occurs regardless of the fact that there is no great difference in the construction of the two machines.

With regard to accelerated aging, it is assumed that technical qualities are diminishing faster as compared to the expected performance. These quantities can only be acquired statistically.

A change in operation influences the aging process, making it even more important to regularly check and analyse the generator condition. When the end of the expected lifetime is reached, timely countermeasures can be taken to prevent unexpected failure. These could be in the form of selected repair work or replacement of complete components.

Lifetime prediction is not aimed at pinpointing the actual end of the lifetime to be achieved, that is, it is not the time of total failure. It is much more a measure of the failure probability of the components within a certain time frame. Based on the operating mode up to the time in question and the diagnostic findings, it can be used to more accurately estimate future failure risk.

Reference to current events

Compared to previous years, more recent diagnostic inspections have shown an increased number of machines with poor insulation quality that actually suffered an earth fault. One reason for this is liberalization of the electricity market, which, for instance, came into force in Switzerland on January 1, 2009. This affected the producer/customer relationship. Major customers can now freely choose their suppliers, meaning production need no longer be as close as possible to the user.

In particular, the supply of ancillary services – which include primary, secondary, and tertiary reserves and production of reactive power for voltage stabilization – directly influences the load and thus the aging process of the machines. Before deregulation, over-production and under-production within a control area were compensated for by reducing or increasing production in another area, whereby the compensated power was later returned in MWh. The initial power compensation could be made during a low-tariff period and returned during a high-tariff period, or vice versa. Such compensations were extensively planned, meaning resulting loads on machines and other plant equipment were kept to a moderate level.

Mutual compensation of exchanged power is now made on a financial basis. The production is billed to real market prices, so that the production and supply of control power is attractive to the producer at all times.

Because of their processes, pumped-storage stations are predestined to produce control power, and the responsive supply of this power forces the producer to change the operational mode of the machine. Greater and more frequent load changes increase the thermal and mechanical loads in the generator components.

Specific changes to operating modes: temperature

As a rule, new bars and also complete machines leave the factory with insulation that is not completely cured. Simultaneously with the curing process, the relative permittivity gradually decreases, and chemical degradation of the polymeric compounds of the impregnating resin begins. Thus, the curing process continues during machine operation. Thermal stresses are minimised because solidification takes place in a situation where the machine should be operating for a number of decades. Ideally, the machine insulation should harden at constant temperature, which means approximately constant power, whereby thermal stresses are kept to a minimum and are frozen.

Should the operating mode be changed, this static condition will be lost. When this happens, it can adversely affect the lifetime of the equipment.

The most important aging factors are the so-called TEAM factors (thermal, electrical, ambient, and mechanical).

As mentioned, insulation degradation by depolymerisation progresses continuously, and this process takes place more rapidly with increased temperatures. However, thermal aging also includes the thermally induced mechanical stresses arising in the components as a result of different coefficients of thermal expansion.

Among other things, shear stresses can lead to cracks and delamination of the insulation (see Figures 1 and 2), which result in lower heat conduction and increased partial discharge (PD) activity.


Essentially, electrical aging is limited to degradation of the polymeric compounds in the impregnating resin by PD.

Some tens of thousands of these discharges occur every second inside the stator bars or coils, causing the insulation to age slowly but continuously. PDs with high pulse energy, which generate so-called fast electrons, especially lead to damage. However, the damaging effect may be considered slight, as modern insulation systems possess electrical lifetime expectancies far beyond the attainable machine lifetime.

This means that PD activity is not the main cause of insulation aging; rather it is much more a good indicator of other damaging influences.

In addition, the acquisition of PD activity is of particular importance in the diagnosis because it represents the only practicable online method of assessing the condition of the insulation.

Discharge activity results in continuous ozone production, which subsequently attacks the polymeric structure of the impregnating resin. In addition, the machine becomes dirty during operation, resulting in poorer cooling and changes in the electrical behaviour, especially on the winding surfaces. Furthermore, liquids such as oil or humidity can penetrate the insulation and adversely affect the insulating characteristics. Insulation is subjected to mechanical stressing by vibration, electromagnetic forces acting on the bars, and thermal stresses caused by different thermal coefficients of expansion. This can result in shearing or hairline cracks, which could eventually lead to an electrical breakdown and machine failure. In addition, loose bars in combination with mechanical vibration may damage slot corona protection.

Remaining life prediction for machine insulation

Based on diagnostic experience and results of investigations, the Alstom approach to lifetime prediction is as follows: The lifetime behaviour relative to certain aging factors (for example, electrical aging) is known.

An expected lifetime is attained under certain basic conditions (diagnostic characteristics, operating conditions) and is based on the lifetime experience for the insulation system. This also incorporates the design of the anti-corona system, number of phase separation points, rated voltage, and more. Should the machine be operated more severely or the dielectrical characteristics deviate adversely from a predefined base level, machine aging will accelerate.

Changes in the operating mode or small repairs would result in a slowing down of the aging process. However, the achieved flattening of the curve can never be flatter than that for moderate operation. Only the replacement of complete components, such as the stator winding, could increase the lifetime expectancy again. This would mean beginning with a new component whose lifetime cycle would thus begin at 100% lifetime.

With the exception of very few points, assessment of the condition/remaining life is based on diagnostic and therefore reproducible measurements, not on visual and therefore subjective findings. This decision was made because visually detectable deficiencies must generally be reflected in the test results.

In any event, lifetime consumption is always equal to, or greater than, the respective real operating period of the machine.

The methodology and tools for condition assessment on rotating electrical machines allows the evaluation of the remaining life of active components, particularly the stator winding. Compared to other test programs, the underlying data that are used for calculation of the remaining life are essentially based on the results of reproducible dielectric measurements and not only on subjective visual findings. The standard measurements to collect the necessary data can be performed with 1-2 days.

The diagnostic results can be used for early planning of maintenance and service work to prevent unplanned outages of the power generating units. In addition, the remaining life calculation provides a valuable basis for planning future investments.

Table 1 shows a selection of characteristic values that are investigated systematically and that are considered within the lifetime prediction process. When all of the set parameters for the insulation system are met or are within the limit values, it may be assumed that the machine will completely fulfill the expected lifetime. Curve A in Figures 3 and 4 shows the relationship between lifetime and machine age.


During the first diagnosis, dielectric tests will be carried out on the machine insulation and the results evaluated. In this example, the result of this diagnosis is that the machine is aging considerably faster than initially assumed, or, faster than expected for moderate operation (see curves B and C in Figure 4). Should the machine be operated further at these more severe conditions, the lifetime will be substantially shortened (see curve C in Figure 4).

In this case, the lifetime would be reduced to 50%. In cases where accelerated aging is recognized by the diagnosis and eliminated by repair measures or changing operating conditions, subsequent aging will revert to moderate consumption (see curve B in Figure 4). However, the initial accelerated aging results in an adverse effect on machine age, reducing the lifetime, in this example, to 80% of the initially expected maximum.


Wendel, C., C. Stephan, and T. Kunz, “Lifetime Prediction on Active Components of Hydro Generators – A Way to Optimise Condition-Based Maintenance,” 2009. 

Application to operating machines 

Two 53 year-old hydro generators of identical design, both installed in the same power station and both being operated in the same manner, have for the last 38 years been subjected to regular winding diagnosis using Alstom’s WIDIPRO® process. From the first diagnosis, the two asphalt mica-folium insulated machines were shown to have different dielectric characteristics, and over the years ever greater differences occurred in the visual observations.

The last diagnosis on one of the machines revealed poor dielectric characteristics and poor condition of insulation. Numerous signs of discharges were evident in the winding overhangs on the line-end bars. In addition, prominent signs of vibration were found at many places in the area around the slot ends, coupled with the beginning of erosion of the slot corona protection. After the last diagnosis in 2009, a rewinding of the machine was urgently recommended.

Based on the many dielectric measurements taken over the years and the complete history of all inspections on these machines, these were used among others to validate the ‘Lifetime Prediction’ programme. As the electrical lifetime behaviour is best known and also confirmed by innumerable measurements, only the electrical remaining life was determined. Consequently, only the dielectric findings were considered in the calculations (see Figure 5).

The first insulation breakdown occurred after 36 years of operation, during a voltage test after service work (see point A in Figures 5 and 6). Presumably, the repair of this fault location had no effect on the remaining life. However, a slight acceleration in the ageing process within the winding is noticeable.


In 2004, after 48 years of operation, the first insulation breakdown during operation occurred (see point B in Figures 5 and 6) at a calculated remaining life of 15 years. Assuming the same aging conditions, after the first repair (after 36 years in operation) the remaining winding should have withstood the operational loading until about 2014 (see point D in Figure 6). As the last winding diagnosis in 2009 showed the calculated remaining life to dip into the negative zone (see point C in Figures 5 and 6), it was recommended to fit a new winding within the next three years.

During all this time, the diagnostic measurements on Generator 1 showed hardly any acceleration in the aging process, which is reflected by the almost horizontal shape of the curve showing the predicted end of life (see Figure 6). Most probably, the machine can be operated for another 30 years before a replacement or a new winding will be necessary.

Christoph Wendel is a diagnosis engineer with Alstom Hydro. Dr. Carl-Ernst Stephan is technical director for service activities for Alstom Hydro Europe. Thomas Kunz, R&D/technology director, has overall responsibility for electrical R&D and leads product development programs for Alstom Hydro.

More HRW Current Issue Articles
More HRW Archives Issue Articles

Previous articleFiscal Austerity Should Not Exclude Hydro
Next articleABB generator circuit breakers to protect hydropower plants in China

No posts to display