A Comprehensive Review of Shaft Voltages and Bearing Currents, Measurements and Monitoring Systems in Large Turbogenerators

Abstract

Turbine generators are essential for power generation, but the presence of shaft voltages and currents poses significant challenges to their reliability, efficiency, and operational lifespan. These phenomena, arising from electromagnetic induction, poor shaft grounding, rotor excitation systems, and varying operational conditions, can lead to severe damage to bearings and rotors, resulting in costly downtime and maintenance. This study reviews the mechanisms behind shaft voltage and current generation, their impact on turbine generators, and the effectiveness of various mitigation strategies, including shaft earthing brushes, bearing insulation, and advanced health monitoring systems. Furthermore, it explores emerging techniques for measuring and diagnosing shaft voltage and current, as well as advancements in predictive maintenance and condition monitoring. This study further explores the integration of artificial intelligence and machine learning in predictive maintenance, leveraging real-time condition monitoring and fault diagnostics. By analyzing existing and emerging mitigation strategies, this study provides a comprehensive evaluation of solutions aimed at minimizing these electrical effects. The findings underscore the importance of proactive management strategies to enhance generator reliability, optimize maintenance practices, and improve overall power system stability. This research serves as a foundation for future advancements in shaft voltage mitigation, contributing to the long-term sustainability of power generation infrastructure.

1. Introduction

Shaft voltages and bearing currents in large turbogenerators have emerged as critical issues affecting machine reliability, safety, and operational efficiency [1,2]. These unwanted electrical phenomena result from complex interactions between electromagnetic fields, parasitic capacitances, and high-frequency excitations introduced by modern power electronic systems. Left unmitigated, they can lead to accelerated bearing wear, electrical discharge machining (EDM), insulation breakdown, and even catastrophic machine failure [3]. Notable documented examples in reference [4] include multiple generator failures and extensive bearing damage caused by ineffective shaft grounding practices, as reported in industrial case studies. One such case involved complete bearing failure and stator core damage resulting from a high shaft potential due to an open-circuited grounding brush. Another reported incident involved persistent sparking and shaft pitting, despite grounding systems being installed, and ultimately traced back to poor contact quality and maintenance practices. In hydro-generator fleets, statistical studies have linked inadequate insulation and grounding to a high incidence of bearing-related failures [5]. Additionally, reference [6] documented comprehensive assessments of shaft voltages across generating units ranging from 66 to 800 MW, revealing that static excitation systems and design asymmetries contribute significantly to shaft voltage magnitudes, sometimes exceeding 130 V. These voltages resulted in recurrent failures of hydrogen seals, journal bearings, and Permanent Magnet Generator (PMG) units due to arcing and lubrication breakdown. A documented incident involved bearing failure in a speed control unit due to current flow across inadequately insulated components, highlighting the need for rigorous grounding and insulation strategies. These cases highlight the urgency of addressing shaft voltage phenomena through reliable diagnostic and mitigation strategies. These real-world cases highlight the critical role of proper grounding in ensuring generator safety and reliability. Numerous diagnostic techniques, signal processing methods, and mitigation strategies have been proposed across the literature [1,2,3,4,5,6,7,8,9]. However, the diversity of sources, waveform signatures, and grounding system implementations often creates challenges in achieving standardization and effective field application. This article aims to provide a comprehensive review of the mechanisms, diagnostic techniques, and mitigation strategies associated with shaft voltages and bearing currents in large turbogenerators, highlighting both theoretical foundations and practical implementation challenges.

When a turbine generator is powered by a thyristor-controlled static excitation system, the high dv/dt characteristics of common-mode voltage significantly intensify shaft voltage and bearing current issues [6,7,8,9,10]. Additionally, external influences, such as electrostatic interactions with connected loads and capacitive coupling within the static excitation system, can further contribute to fluctuations in shaft voltage and bearing current, exacerbating potential reliability and performance challenges [10,11].

The authors of [12] provide a detailed examination of bearing current damages in turbine generators, emphasizing practical and straightforward methods for mitigating this kind of damage. The authors identify four primary types of bearing damage: pitting, frosting, welding and fluting, discussing their causes and effects extensively. Additionally, the authors of [11,12,13] offer an in-depth overview of bearing currents and categorized shaft current types, which contributes to understanding the sources of shaft voltages and failures related to bearing currents. Their studies focus on recognizing bearing current damage, investigating their origins, detecting these currents, and implementing effective control measures. Understanding and mitigating these phenomena are imperative, as their unaddressed presence can lead to costly downtime, reduced operational efficiency, and increased maintenance burdens. Such failures highlight the importance of proactive measures in safeguarding these essential systems, which directly influence societal energy security and economic stability. Effective shaft earthing or grounding mechanisms and bearing insulation are crucial, particularly in large turbine generators operating at higher power ratings (i.e., ≥75 MW), to counteract the adverse effects of high voltages and circulating currents [1,4]. Inadequate shaft earthing systems can exacerbate problems by potentially increasing circulating currents [2,4].

Although the phenomena of shaft voltage and bearing current in turbine generators have been recognized for over a century, ongoing research remains crucial due to continuous advancements in power electronics and electric machines [14]. Additionally, the integration of deep learning-driven predictive maintenance systems for real-time shaft voltage monitoring has shown promise in mitigating failures before they escalate [15]. Emerging approaches, such as suppression systems and optimized Pulse Width Modulation (PWM), offer alternative solutions, yet their long-term feasibility and effectiveness require further exploration [13,16]. Most research on shaft voltage and bearing current issues primarily focuses on three-phase or multiphase motors, with relatively few studies addressing the challenges and mitigation strategies for large-scale turbine generators. However, in turbine generators, particularly those using thyristor-controlled static excitation systems, high $dv/dt$ common-mode voltages significantly intensify shaft voltage and bearing current problems, leading to electrical discharge machining (EDM) effects, insulation breakdown, and accelerated wear of bearings and other critical components [10,13]. The presence of electromagnetic induction, electrostatic charging, shaft magnetization, and capacitive coupling further complicates these issues, requiring more advanced diagnostic and suppression techniques [10]. While shaft grounding brushes, insulated bearings, and optimized PWM control techniques have been widely explored for three-phase motors, their effectiveness in turbine generators remains under-researched [12].

Recent research studies have emphasized the critical role of high-frequency common-mode voltages in exacerbating shaft voltage and bearing current issues, particularly in inverter-driven and static excitation turbine generators [17,18]. The introduction of electronic switches in excitation systems has intensified the impact of these electrical phenomena, making mitigation strategies more urgent [17]. Studies have shown that the $dv/dt$ characteristics of common-mode voltage contribute significantly to the breakdown of lubrication films, leading to electrical discharge machining (EDM) currents, which accelerate bearing degradation [3,17]. Furthermore, this research has highlighted the need for bearing failure mechanisms, mitigation strategies, and measurement techniques, as most studies focus on three-phase motor configurations, overlooking the increased complexity and unique challenges of large-scale wind, hydro, and steam turbine generators [10,19]. The authors of [20] focus on the critical role of shaft voltages and bearing currents in the reliability of electric vehicle (EV) and hybrid-electric vehicle (HEV) motors, with an emphasis on failure mechanisms, propagation, and mitigation techniques. While the focus of the review is on EV and HEV motors, their findings on PWM-induced common-mode voltages and bearing failure mechanisms are highly relevant to large turbine generators with static excitation systems.

Moreover, the lack of predictive systems and condition monitoring systems for imminent failures limits direct performance comparisons across different power plant environments. By integrating recent findings, this study expands the understanding of shaft voltage and current mechanisms, remedial methods, health monitoring systems, and diagnostic approaches. It provides a critical analysis of existing and emerging mitigation strategies while highlighting research gaps that must be addressed to improve turbine generator reliability and efficiency in modern power systems. In recent years, there has been growing interest in advanced condition monitoring and predictive maintenance strategies to detect and mitigate these failures before they escalate [7,21]. The emergence of high-precision sensing technologies, real-time condition monitoring systems, and artificial intelligence (AI)-driven diagnostics has significantly improved the ability to track early-stage bearing damage, lubrication degradation, and electrical discharge patterns [15,22]. AI-powered predictive analytics can process vast amounts of data, identifying failure trends and potential risks with greater accuracy than traditional monitoring methods [7,21]. To address this gap, recent studies have investigated advanced mitigation techniques, including suppression systems and intelligent condition monitoring solutions that use AI-based predictive maintenance for real-time fault detection [15,23]. Despite these developments, comparative assessments of these mitigation strategies in large-scale turbine generator systems remain scarce, hindering the development of standardized guidelines for advanced condition monitoring systems. This research addresses this issue by reviewing measurement techniques and predictive methods for bearing current analysis and health diagnostics, providing insights into effective monitoring and mitigation approaches for turbine generators. By summarizing recent research on common-mode voltage suppression and identifying current limitations, this paper serves as a valuable reference for improving generator reliability, efficiency, and long-term performance.

This study highlights the ongoing importance of understanding and mitigating shaft voltages and currents in large turbines and synchronous generators. By addressing these issues comprehensively, it aims to contribute to the development of more resilient and efficient power generation systems, directly benefiting industries, communities, and society at large.

The remainder of this article is structured as follows: Section 2 provides a classification of shaft voltage sources and their physical generation mechanisms. Section 3 discusses mitigation techniques, grounding strategies, and their performance. Section 4 covers diagnostic and measurement systems, including Rogowski coils, current transformers, and condition monitoring technologies. Section 5 explores advanced signal processing methods and the role of AI-based diagnostic models. Section 6 and Section 7 present economic impact, reliability metrics, safety considerations, and compliance with international standards. Finally, Section 8 concludes the review with key insights and future research directions.

This review was conducted using a structured literature selection process focused on peer-reviewed journal articles, conference papers, technical reports, and industry case studies published between 1990 and 2025. Key sources were identified through scientific databases such as IEEE Xplore, ScienceDirect, SpringerLink, and Scopus using targeted keywords including ‘shaft voltage’, ‘bearing current’, ‘turbogenerator grounding’, ‘electrical discharge machining’, and ‘condition monitoring’. Emphasis was placed on high-impact studies with empirical validation, diagnostic frameworks, and industrial applications. Duplicate and redundant studies were excluded, and cross-referencing was used to validate findings across domains such as hydro-generators, steam turbines, and inverter-fed machines.

2. Shaft Voltage and Bearing Current Phenomena

This section explores the impact of shaft voltage and current on turbine generators, detailing their sources, associated bearing failure mechanisms, and various types of bearing currents. It provides a comprehensive analysis of bearing current classifications in large turbine generators and examines effective mitigation strategies for preventing shaft voltage-induced bearing failures. Additionally, it covers advanced measurement techniques for assessing shaft voltage and bearing currents, along with modern approaches to turbogenerator condition monitoring and health assessment.

2.1. Sources of the Shaft Voltage and Bearing Current

Large turbine generators experience shaft voltages from four principal source mechanisms: electrostatic discharge, magnetic asymmetries, shaft magnetization, and excitation-induced voltages [4,12]. These categories correspond to distinct physical processes, but they are not strictly mutually exclusive [12,19]. In practice, multiple phenomena can occur simultaneously, for example, electromagnetic induction, electrostatic charging, magnetized shaft effects, and capacitive coupling from the exciter may all contribute to shaft voltage in one machine. Each source has a different underlying mechanism and produces characteristic shaft voltage waveforms and effects. Table 1 summarizes the key differences between these sources in terms of their generation mechanism, waveform nature, source impedance, and typical mitigation strategies. A detailed discussion of each category follows.

2.1.1. Electrostatic Discharge (ESD)

The authors of [24,25] schematically illustrate the generation of shaft voltages due to the electrostatic effect of wet steam particles in the electrostatic circuit of the stator, as shown in Figure 1. This electrostatic shaft voltage generation involves the charge separation of particles, which elevates the shaft to a potential above the ground (casing) potential [24,25]. The corresponding equivalent electrical circuit and resulting shaft currents are shown in Figure 2. Shaft voltages originating from this source typically occur due to the brushing effect of wet steam on the blades of a low-pressure (LP) turbine, where the steam is no longer superheated, as cited in [12,24].

Electrostatic charge accumulation in turbine-generator shafts has been closely linked to steam flow conditions and machine loading, particularly in high-power steam turbines operating with wet steam in the low-pressure stages [26]. In such environments, impingement of water droplets on rotor blades results in triboelectric charging, progressively elevating the shaft potential relative to ground [27,28]. The shaft, in effect, behaves as a charged capacitor, discharging intermittently across the thin insulating oil film in bearings. This phenomenon produces a DC-biased voltage waveform characterized by a slow linear ramp-up followed by a rapid collapse, commonly observed as a sawtooth profile with unidirectional polarity and repetitive discharge events [25,26,27,28].

Studies have reported that such shaft voltages typically range from 30 to 150 V, with peak excursions reaching up to 250 V, while the corresponding discharge currents remain low in the milliampere range due to the high internal impedance of the electrostatic source [4,28]. Although the current magnitude is minimal, repeated high-voltage discharges can initiate pitting, fluting, and lubricant contamination, which progressively degrade bearing surfaces [12].

Mitigation strategies focus on providing a low-impedance discharge path to ground, typically through the application of high-performance shaft grounding brushes that continuously drain accumulated charge before it reaches damaging levels [12,22]. In parallel, maintaining the dielectric strength and cleanliness of the bearing oil through proper lubrication practices and filtration is critical to reducing the frequency of oil film breakdown and subsequent discharge events [3].

2.1.2. Magnetic Asymmetries in an Electrical Winding

Researchers have identified two conditions that can result in shaft voltage originating from this source: one due to a voltage from shaft end to end and another from axial leakage flux along the shaft [5,19]. This axial leakage flux is present in all electrical machines [12,13,14]. The authors of [12] illustrated how this source originates from the rotor shaft, as shown in Figure 3.

The authors of [24,29] schematically illustrate the generation of shaft voltages due to asymmetry in the magnetic circuit of the stator. The corresponding equivalent electrical circuit and resulting shaft currents are shown in Figure 4. Magnetic asymmetries in the stator or rotor of turbine generator machines create a voltage potential from shaft end to shaft end [12,20].

Shaft voltages from this source can be induced by electromagnetic induction when the generator’s magnetic circuit is not perfectly symmetric [12,24]. There will always be slight differences in the reluctance of the magnetic circuit due to design flaws, building tolerances, core-plate differences, and plate thickness variations. Consequently, no machine can be assumed to have perfect symmetry [4,14]. Magnetic asymmetry causes a small net axial flux linkage through the shaft, creating an induced voltage from one end of the shaft to the other. Even in well-designed machines there are slight differences in magnetic reluctance (due to manufacturing tolerances, air-gap eccentricity, winding imbalances, etc.), so a minute portion of the main field’s flux can thread the shaft and induce an AC voltage along it [4,14]. This induced shaft voltage is an alternating voltage (fundamentally at the machine’s electrical frequency, with harmonics) rather than a DC buildup [12]. In practice, the shaft end-to-end voltage from magnetic asymmetry appears as a continuous AC waveform rich in line-frequency harmonics [4,14]. Because this source is directly driven by the generator’s electromagnetic field, its source impedance is very low, the shaft and machine frame form a low-resistance loop [4,14]. Even a few millivolts of induced AC can drive large circulating currents through the bearings and structure if a closed path exists [4,12]. In fact, magnetic asymmetry is often considered a “low-impedance” source, capable of sustaining high fault currents through the metal structure [4,14]. The primary manifestation of this phenomenon is circulating bearing currents at line frequency (50/60 Hz) and its multiples; over time, these AC currents can cause fluting or frosting on bearing races due to continuous spark erosion [12]. Mitigation focuses on breaking the current path; typically, one bearing (usually the non-drive end) is electrically insulated to prevent the induced voltage from driving a loop through the bearings [4,12].

2.1.3. Shaft Magnetization

The undesirable magnetization of components in turbine generators has been recognized as a problem for many years [30]. Homopolar flux induces shaft voltages and localized bearing currents, forming a self-excitation circuit around the bearing as shown in Figure 5 [12,24]. The corresponding equivalent electrical circuit of shaft magnetization is shown in Figure 6.

These factors highlight the need for proper procedures and monitoring systems to prevent and manage the magnetization of turbine generator components.

Homopolar magnetic flux-induced shaft voltages are particularly prominent in high-speed turbogenerators, where shaft voltage has been shown to increase linearly with rotational speed [29,30]. This phenomenon arises when the shaft or rotor retains residual magnetization, often resulting from prior welding, rotor rubs, or exposure to external magnetic fields during maintenance procedures [4,12]. The residual magnetism creates a homopolar magnetic flux path that couples through the machine’s bearings, effectively causing the rotor to act as a quasi-permanent magnet [4,12].

As the generator rotates, the magnetized shaft induces a steady-state or slowly varying DC voltage across the shaft-bearing circuit, which appears in the shaft voltage spectrum as a zero-frequency (DC) component [30]. In the time domain, this manifests as a voltage bias or very slow ramp, depending on the orientation and strength of the residual field [4,12]. This homopolar-induced voltage can sustain a continuous DC current through a closed conductive loop involving the bearings, commonly referred to as a self-excitation circuit [12,30]. Due to the low internal impedance of this source, the magnitude of the current is limited only by the electrical resistance of the metal path and bearing interfaces [12,30]

Unlike electrostatic discharges (ESDs), which produce intermittent high-voltage sparks, homopolar shaft currents cause persistent DC arcing, leading to gradual bearing pitting, welding, and localized thermal degradation of bearing surfaces or flanges [12,30]. Effective mitigation involves eliminating residual magnetization through rotor demagnetization (degaussing) procedures prior to installation or following any event that may reintroduce magnetization [12,30]. Additional preventive practices include avoiding magnetic particle inspection or welding on assembled rotors, unless followed by demagnetization using appropriate degaussing coils. Ensuring the rotor is magnetically neutral effectively eliminates the source of this persistent DC shaft voltage [12,30].

2.1.4. Excitation System Developed Shaft Voltage

The authors of [24] schematically illustrate the generation of shaft voltages due to excitation-induced shaft voltage, as shown in Figure 7. The applicable equivalent electrical circuit and resulting shaft currents are also shown in Figure 8.

This category of shaft voltage arises from the direct current (DC) excitation system used to magnetize the turbine generator’s air gap, particularly in modern systems using static exciters or power electronics. In these machines, the field winding of the generator (on the rotor) is fed by a DC excitation source (thyristor or diode rectifier bridges, or other power electronic circuits) [10,12]. Although the field itself is DC, the common-mode voltage and high-frequency switching transients from the exciter can capacitively couple into the shaft [10,12]. The rotor’s field winding and the steel shaft/forging form a parasitic capacitor; when the exciter applies a pulsed DC or has a ripple, a displacement current can flow from the winding into the rotor body, elevating the shaft potential. Thus, the mechanism is electrical capacitive coupling of the exciter’s voltage into the shaft. The resulting shaft voltage waveform often contains a DC offset (from the steady field) with superimposed AC components at the switching frequency of the exciter and its harmonics [2,12]. For example, a thyristor-controlled exciter may induce shaft-voltage ripples at multiples of 100/120 Hz or higher, and high-frequency spikes from fast switching edges [2,10]. These appear on the shaft as high-frequency AC voltages (often in the kHz range) riding on a DC baseline. Importantly, because the coupling is capacitive and not a direct electrical connection, the source impedance is high, i.e., the shaft can pick up a significant voltage, but it cannot drive large steady currents on its own [2,11]. Nevertheless, if the shaft is not properly grounded, these high-frequency voltages will eventually discharge through the path of least resistance (often the bearings). The manifestations are similar to other EDM (electric discharge machining) effects: bearing pitting, fluting, and erosion caused by repeated high-frequency arcing [5,12]. Over time, this leads to premature bearing failures. Mitigation involves both grounding and filtering. Good practice includes installing shaft grounding brushes near the exciter end to clamp the shaft potential, and adding filters or surge suppressors to the excitation circuit to remove high-frequency components [10,11]. Some large generators incorporate Faraday shields or dielectric barriers between the field winding and rotor to reduce capacitive coupling [11,31]. By addressing the common-mode and transient voltages in the exciter design (e.g., using snubber circuits, minimizing $dv/dt$), the magnitude of excitation-coupled shaft voltage can be greatly reduced [10,11].

2.2. Shaft Voltage Waveform Charecteristics

Different physical mechanisms produce distinctive shaft voltage waveforms, as evidenced by empirical studies.

Electrostatic discharge (ESD) buildup on the shaft results in a characteristic sawtooth voltage waveform, as the shaft potential gradually rises (charging) until reaching the dielectric breakdown of the oil film, then rapidly drops when a spark discharges to ground. This repetitive charge–discharge pattern has been observed in measurements of large generators with inadequate grounding, showing a slow ramp in voltage followed by high $dv/dt$ spikes [4,6]. For example, case studies report shaft voltages accumulating to tens of volts before a sudden discharge event occurs, yielding an oscillogram trace with a tooth-shaped profile [6,28]. Such ESD-induced sawtooth waveforms have been documented in both laboratory and field conditions, highlighting their role in bearing EDM damage (as described in the literature) [11,32]. The presence of this waveform is a clear indicator of electrostatic charging of the rotor and is often mitigated by shaft grounding brushes or insulating coatings [12,25].

Magnetic asymmetry-induced voltages (due to design or assembly tolerances, e.g., rotor eccentricity or unequal air gap) tend to generate AC harmonic waveforms on the shaft. The authors of [14] demonstrated that segmental laminations and other imbalances can induce an alternating shaft voltage at the machine’s rotational or field excitation frequency (and sometimes at specific harmonics of the 50/60 Hz supply) [4,12]. In healthy machines these induced voltages are minimal, but under pronounced asymmetry a measurable sinusoidal shaft voltage can appear (e.g., at 50 Hz or 100 Hz, depending on the fault type) [4,6]. Recent diagnostic experiments confirm this behavior; for example, a slight static eccentricity in a two-pole turbogenerator led to a predominant fundamental frequency shaft voltage component in measured waveforms [8]. Published waveform plots in [6,7,8,12] show a near-sinusoidal voltage on the shaft in the presence of magnetic irregularities, providing clear empirical validation. These AC-induced shaft voltages are linked to flux imbalances (sometimes described as a homopolar or “shaft flux” effect) and have been used as signatures for detecting misalignment and core faults [4,12].

Excitation system and converter-induced voltages introduce high-frequency content. Modern static excitation systems (SES) and power electronic converters, such as variable-frequency drives (VFDs) and static frequency converters (SFCs), introduce high-frequency ripple and transient spikes into shaft voltage waveforms [4,12,33]. These disturbances, typically originating from thyristor-based rectifiers, manifest at multiples of the supply frequency (e.g., 300–360 Hz) and are often superimposed with nanosecond-scale voltage spikes. Field measurements have shown shaft voltage components reaching amplitudes of ~100 V, particularly with six-pulse SES configurations [34]. The resulting waveform typically exhibits a DC or low-frequency bias overlaid with kHz–MHz-range noise, significantly increasing the risk of bearing stress and electrical discharge machining (EDM) [5,10]. These high-frequency signatures are widely recognized as diagnostic indicators of the presence of power electronic excitation systems and necessitate appropriate filtering or grounding to mitigate their impact [10,11,12,33].

Shaft magnetization and other DC bias sources in the rotor produce a nearly steady DC offset in the shaft voltage. If part of the machine’s magnetic circuit becomes permanently magnetized (for example, after a rotor earth fault or due to remanent flux in the shaft steel), a constant potential difference can be observed between the shaft ends [12,30]. Unlike the dynamic waveforms above, a magnetization-induced shaft voltage appears as a DC level (with potentially a slight ripple) when monitored over time. Early studies postulated a “homopolar” flux linking the shaft as a cause for such direct voltages [12], and later work has indeed measured a small but non-zero DC shaft voltage in machines with magnetized rotors or asymmetric grounding. Reference [12] notes that a persistent DC shaft voltage (on the order of a few millivolts up to a volt or more) was recorded in generators where one end of the shaft carried residual magnetization. The author notes that the DC offset corresponds to a homopolar source, while the AC component of the waveform is primarily at the machine’s running speed frequency, which in this case was 103 Hz. In practical diagnostics, the presence of a DC-biased shaft voltage is used as evidence of shaft magnetization or improper bonding, and this offset often accompanies other fault-specific signals. Notably, even a modest DC shaft voltage can shift the operating point of bearing voltages and contribute to cumulative damage, so its detection (as illustrated in the case studies of [30] is important for preventive maintenance. In summary, each category of shaft voltage source, such as electrostatic, electromagnetic, excitation system induced, and residual magnetism, can be identified by its unique waveform signature, as reported in numerous empirical studies [3,5,6,7,8,9,26,27].

Table 1. Comparison of shaft voltage sources and characteristics.

Shaft Voltage Source	Mechanism of Generation	Nature of Waveform	Source Impedance	Typical Mitigation Strategies	References
Electrostatic Discharge	Friction-induced charge accumulation (e.g., wet steam rubbing turbine blades, creating static charge on rotor).	- DC buildup with periodic discharge; - Sawtooth voltage rises and sudden drop (polarity constant, with pulses).	High—generates high voltage (tens to hundreds of volts) but only microamp–milliamp currents (high internal resistance).	- Shaft grounding brushes to bleed off charge; - Maintain oil film strength (prevent oil breakdown and sparking).	[8,26]
Magnetic Asymmetry	Asymmetric magnetic flux linkage through shaft (due to design tolerances, air-gap eccentricity, or winding imbalances) causes an induced end-to-end shaft voltage.	AC voltage (alternating) at fundamental machine frequency and its harmonics; continuous sinusoidal-like waveform.	Low—low impedance source (electromagnetic induction) capable of driving large AC circulating currents through bearings and frame.	- Insulate one bearing (usually NDE) to open the circuit; - Minimize magnetic imbalances in design and maintenance.	[3,8]
Shaft Magnetization	Residual magnetism in shaft/rotor (from welding, rubs, etc.) produces a homopolar (zero-sequence) flux that induces a DC shaft voltage; can form a self-excited loop through bearings.	- Predominantly DC voltage (steady bias in shaft potential); - In spectrum, appears as a DC component (0 Hz) with minimal alternating content.	- Low—behaves like a low-impedance DC source; - A magnetized rotor can drive continuous DC bearing currents (limited only by circuit resistance).	- Demagnetize shaft/rotor before and during service to eliminate residual magnetism; - Avoid practices that induce magnetization.	[8,27]
Excitation-Induced	Common-mode and high-frequency voltages from static excitation system coupling capacitively onto the rotor (field winding to shaft capacitance). Often exacerbated by fast thyristor or diode switching in the exciter.	Mixed DC and AC waveform: a DC offset (from field voltage) plus superimposed high-frequency AC ripple or spikes (e.g., rectifier ripple, switching transients).	- High—the coupling is through small capacitances, so source impedance is high (voltage spikes with negligible steady current); - Significant current flows only during brief discharges.	- High-quality shaft grounding/earthing brushes to clamp shaft potential; - Filters or capacitive shields in the excitation circuit to block or absorb high-frequency components.	[6,7]

Table 1. Comparison of shaft voltage sources and characteristics.

Shaft Voltage Source	Mechanism of Generation	Nature of Waveform	Source Impedance	Typical Mitigation Strategies	References
Electrostatic Discharge	Friction-induced charge accumulation (e.g., wet steam rubbing turbine blades, creating static charge on rotor).	- DC buildup with periodic discharge; - Sawtooth voltage rises and sudden drop (polarity constant, with pulses).	High—generates high voltage (tens to hundreds of volts) but only microamp–milliamp currents (high internal resistance).	- Shaft grounding brushes to bleed off charge; - Maintain oil film strength (prevent oil breakdown and sparking).	[8,26]
Magnetic Asymmetry	Asymmetric magnetic flux linkage through shaft (due to design tolerances, air-gap eccentricity, or winding imbalances) causes an induced end-to-end shaft voltage.	AC voltage (alternating) at fundamental machine frequency and its harmonics; continuous sinusoidal-like waveform.	Low—low impedance source (electromagnetic induction) capable of driving large AC circulating currents through bearings and frame.	- Insulate one bearing (usually NDE) to open the circuit; - Minimize magnetic imbalances in design and maintenance.	[3,8]
Shaft Magnetization	Residual magnetism in shaft/rotor (from welding, rubs, etc.) produces a homopolar (zero-sequence) flux that induces a DC shaft voltage; can form a self-excited loop through bearings.	- Predominantly DC voltage (steady bias in shaft potential); - In spectrum, appears as a DC component (0 Hz) with minimal alternating content.	- Low—behaves like a low-impedance DC source; - A magnetized rotor can drive continuous DC bearing currents (limited only by circuit resistance).	- Demagnetize shaft/rotor before and during service to eliminate residual magnetism; - Avoid practices that induce magnetization.	[8,27]
Excitation-Induced	Common-mode and high-frequency voltages from static excitation system coupling capacitively onto the rotor (field winding to shaft capacitance). Often exacerbated by fast thyristor or diode switching in the exciter.	Mixed DC and AC waveform: a DC offset (from field voltage) plus superimposed high-frequency AC ripple or spikes (e.g., rectifier ripple, switching transients).	- High—the coupling is through small capacitances, so source impedance is high (voltage spikes with negligible steady current); - Significant current flows only during brief discharges.	- High-quality shaft grounding/earthing brushes to clamp shaft potential; - Filters or capacitive shields in the excitation circuit to block or absorb high-frequency components.	[6,7]

Each of these shaft voltage sources has a unique signature and requires tailored mitigation. By clearly distinguishing the mechanisms (electrostatic vs. electromagnetic induction vs. magnetization vs. capacitive coupling) and their typical waveform characteristics, the dominant cause of shaft voltage in a given scenario can be identified [4,12]. The concept of source impedance is particularly useful for diagnosis: high-impedance sources (like electrostatic and excitation-induced) tend to create high voltage with low current, leading to sporadic discharges, whereas low-impedance sources (magnetic asymmetry and magnetization) often result in continuous currents if not properly interrupted [2,12]. Understanding these differences is crucial for selecting the appropriate mitigation strategy, from installing grounding brushes and insulating bearings to demagnetizing rotors and filtering excitation harmonics. In summary, the four categories of shaft voltages are conceptually distinct but not mutually exclusive, and a robust shaft voltage management scheme in large turbogenerators must address all potential sources with targeted solutions. Each mechanism’s unique combination of waveform, impedance, and effects, as highlighted above, provides a foundation for both diagnosing shaft voltage issues and preventing bearing damage through proper design and maintenance.

2.3. Shaft Voltage and Bearing Current Implications

Turbogenerators are particularly vulnerable to shaft voltage and bearing current due to their considerable size, operational complexity, and the intense magnetic fields inherent in their operation [4]. These factors not only amplify the generation of shaft voltages and bearing currents but also exacerbate their adverse effects on the machinery. The key implications of this susceptibility that impact on machine performance, maintenance requirements, and long-term operational efficiency are as follows:

2.3.1. Bearings, Seals, and Gears Failures

• Bearings: Damage includes pitting, frosting, fluting and welding caused by electrical discharges across the oil film, leading to compromised bearing surfaces and premature failure [12,17,35];
• Seals: Electrical bearing currents can degrade seals, reducing their effectiveness and increasing the likelihood of leaks that can further damage internal components [4,6];
• Gears: Electrical currents can result in surface pitting and increased friction, reducing the efficiency and lifespan of gears [4,6].

2.3.2. Reduced Operational Reliability

The operational complexity of large turbogenerators magnifies the impact of shaft voltages and currents, leading to:

• Rotor Vibrations: Shaft currents can contribute to rotor imbalance, increasing vibrations that compromise operational stability and efficiency [1,4,6];
• Shaft Magnetization: Prolonged exposure to magnetic asymmetries can lead to localized magnetization of the shaft, creating additional operational challenges, such as uneven torque and reduced machine lifespan [12,30];
• Unscheduled Downtime: The compounded effects of these issues result in frequent maintenance requirements, disrupting power generation schedules [1,6].

2.3.3. Costly Repairs and Downtime

Failures in large turbogenerators have disproportionately higher financial implications due to the scale of the equipment and its critical role in power generation:

• Expensive Repairs: The size and complexity of components like bearings, rotors, and gears make repairs not only costly but also time-intensive [1,36];
• Extended Outages: Repairing or replacing major components often necessitates extended downtime, resulting in significant losses for power utilities [1,30].

2.4. Types of Bearing Currents in Large Turbine Generators

Extensive research has documented the occurrence of bearing currents in electrical machines and their associated mitigation strategies [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35]. These bearing currents arise due to various electrical and magnetic phenomena within generators, including magnetic field imbalances, stray capacitances, and voltage differentials between the rotor and stator [17,35]. If unmanaged, bearing currents can cause electrical discharges, leading to pitting, frosting, fluting, and eventual bearing failure resulting in costly downtime and repairs [12]. The presence of high-frequency harmonics from power electronics increases common-mode voltage (CMV), resulting in alternating shaft voltages that intensify circulating bearing currents and insulation stress [10,13]. Studies have shown that these high-frequency currents, typically in the kHz to MHz range, accelerate bearing degradation by reducing lubricant dielectric strength and increasing localized heating effects [10,16,37]. The introduction of hybrid bearing materials and ceramic-coated bearings has been investigated as a potential solution to minimize current conduction and reduce electrical wear [5,38]. Traditional AC machine models often overlook high-frequency effects [11,39]. However, the increasing use of power electronic converters, such as static excitation systems and high-frequency converters, has exacerbated the issue [10,11]. These systems introduce parasitic capacitances, providing a conductive path for bearing currents [11,13,37].

2.4.1. Parasitic Capacitance Quantification

Electrical machines consist of multiple insulated metal components, including the rotor, stator, and windings, which create parasitic capacitances [11,34,40]. When the insulation of the field winding remains intact, it forms a capacitive impedance between the field windings and the rotor forging [10,34]. Additionally, the entire turbine generator system, including the turbine, generator, and static excitation, is interconnected through distributed capacitances, influencing electrical behavior and potential shaft voltage generation [10,34]. These capacitances, also known as parasitic capacitances, develop between the static excitation transformer and ground ($C_{TG}$), rotor and stator ($C_{RS}$), rotor and excitation winding ($C_{WR}$), and shaft (or its bearing) and ground ($C_{SG}$), as depicted in Figure 9.

Parasitic capacitances in rotating electrical machines primarily arise between the excitation winding and the rotor shaft, as well as between the shaft and the stator frame. These capacitances form the critical components of the capacitive coupling paths responsible for shaft voltage buildup, particularly in machines equipped with static excitation systems (SES) [10,34].

The authors of [34] conducted a seminal investigation into shaft voltage formation in large synchronous generators with SES, focusing specifically on the role of parasitic capacitances. Their validated electrical model identified key capacitive elements: the excitation winding to rotor shaft $C_{WR}$, typically in the range of 1–2 $\mathrm{n}\mathrm{F}$; the rotor to stator core ($C_{RS}$), estimated between 1 and 10 $\mathrm{n}\mathrm{F}$; the shaft to ground via the bearing oil film ($C_{SG}$), between 100 and 500 $\mathrm{p}\mathrm{F}$; and the excitation transformer to ground ($C_{TG}$), approximately 1–5 $\mathrm{n}\mathrm{F}$. These elements together form a frequency-dependent voltage divider that allows high-frequency ripple and transient voltages originating from thyristor-based excitation systems to couple directly onto the rotor shaft.

Their field measurements and simulation studies demonstrated that shaft voltages ranging from 40 V to 60 V can develop under such conditions, particularly at higher ripple frequencies where capacitive reactance decreases significantly. When this shaft voltage exceeds the dielectric strength of the bearing oil film (typically 10–20 V), breakdown occurs, allowing discharge currents to flow through the bearing, ultimately leading to electrical discharge machining (EDM) damage. The authors of [10] not only quantified these capacitances and validated their model using real generator data but also demonstrated that effective mitigation, such as the application of shaft grounding brushes and capacitive filtering, can significantly reduce both shaft voltage amplitude and associated bearing current risks. This work provides a foundational framework for understanding the influence of parasitic capacitances on shaft voltage behavior in modern excitation environments.

Further insight into the quantification of parasitic parameters was provided in a simulation study of a 96.3 MW gas turbine generator [34]. A detailed excitation-shaft-bearing model was developed to evaluate the inductance and capacitance values associated with the excitation winding and its coupling to the rotor shaft. The excitation winding coil inductance ($L_{RW}$) was computed using:

$$L_{WR}={\displaystyle \frac{{\mu }_0{{\mu }_{r}N}^{2}A}{L}}$$

(1)

where ${\mu }_0$ is the permeability of free space, ${\mu }_r$ is the relative permeability of the rotor core, $N$ is the number of turns, $A$ is the cross-sectional area, and $L$ is the magnetic path length of each coil, calculated as:

$$L=2l_s+\pi D_r$$

(2)

Here, $l_s$ is the shaft length and $D_r$ is the rotor diameter. The surface area $A$ for each excitation coil is given by:

$$A=\pi D_r{l}_s$$

(3)

The parasitic capacitance ($C_{WR}$) between the excitation winding and the rotor shaft is estimated using the standard parallel-plate capacitance formula adapted for concentric surfaces:

$$C_{WR}={\displaystyle \frac{{\epsilon }_0{\epsilon }_{r}A}{d}}$$

(4)

where ${\epsilon }_0$ is the permittivity of free space, ${\epsilon }_r$ is the relative permittivity of the insulation material, $A$ is the effective surface area, and $d$ is the separation distance between the excitation winding and the shaft. The same formulation can be used to estimate the capacitance associated with journal bearings by substituting the relevant dimensions and dielectric properties of the lubricant film. This analytical and simulation-based approach supports the practical relevance of parasitic capacitance modelling and demonstrates the feasibility of incorporating such values into shaft voltage prediction models for diagnostic and design purposes.

2.4.2. Bearing Currents

There are typically two types of bearing currents observed in machine bearings: circulating (AC) and non-circulating (DC). Both types can coexist and arise from different operational conditions [11,18]. The authors of [13,31] have categorized these parasitic bearing currents in electrical machines into four main groups. The primary types of bearing currents include:

1.

Electrostatic Discharge Currents

The electrostatic charges in a turbine generator are typically classified as non-circulating currents [10]. Electrostatic currents in turbine generators result from static electricity accumulation on the rotor [27,28]. Electrostatic charges are generated by the friction of gases or liquids, such as water steam, when in contact with rotating turbine components [12,25]. The configuration for shaft voltage generation on a steam turbine generator is shown in Figure 10. Friction from steam turbine blades induces a uniform charge distribution along the generator shaft, creating a potential difference between the shaft and grounded components, such as bearings [12,28]. Electrostatic shaft voltage generation involves charge separation of particles which elevates the shaft to a potential above ground (casing) potential [12,27]. The discharge follows a direct path from the rotor, through the bearing, and to ground or other grounded components within the turbine generator [12]. If this potential exceeds the dielectric breakdown voltage of the oil film, an electrical discharge occurs, leading to bearing currents [12,19]. These high-frequency electrostatic discharge machining (EDM) currents follow a direct path from the rotor through the bearings to the ground [18,28].

Their sporadic nature makes them difficult to detect, yet they can cause severe bearing damage, including pitting, fluting, and eventual failure if not properly controlled [12,35]. To mitigate this issue, proper shaft grounding using high-quality grounding brushes is essential to dissipate static charges before they reach damaging levels [12,28].

2.

Grounding Bearing Currents

Grounding currents, depicted by the yellow line, as shown in Figure 11, occur due to unintended electrical connections between the rotor and ground, often through the bearings. These currents arise from inadequate insulation, ineffective grounding, or electrical asymmetries, creating a low-resistance path for current flow.

When present, grounding currents cause localized heating, arcing, and erosion on bearing surfaces, leading to premature failure and potential secondary damage to adjacent components. If these currents find alternative grounding paths, they can simultaneously affect both generator and turbine bearings, increasing operational risks.

To mitigate these effects, robust grounding strategies are essential. Solutions such as insulated bearings, shaft grounding brushes, and grounding rings help divert stray currents away from critical components, preserving machine integrity. Regular monitoring and maintenance further enhance reliability, reducing the risk of unplanned outages and ensuring long-term operational efficiency in high-stress environments.

3.

Capacitive Induced Currents

Shaft current generation due to capacitive induced currents is illustrated in Figure 12. The change in common-mode voltage levels, along with high values $dv/dt$, drives leakage or bearing currents in modern drive systems [18,41]. These currents are induced by the switching actions of power electronics in the excitation system [10]. Shaft voltage in this scenario is primarily a rectified or direct current (DC) signal, leading to the prominence of non-circulating bearing currents [12]. The output of a static excitation thyristor rectifier produces common-mode voltage (CMV) with a rectangular wave shape and steep voltage steps (high $dv/dt$) due to thyristor commutation [10]. The study of shaft voltage and bearing current in turbine generators with static excitation systems revealed that induced shaft voltages from the excitation system lead to circulating shaft currents or induced leakage currents ($I_c={\displaystyle \frac{Cdv}{dt}}$) [10,11]. These currents flow through the bearing oil film and eventually discharge through the bearing casing due to the parasitic capacitance C and high $dv/dt$. This leakage current excitation generates a high-frequency flux, with the common-mode frequency content potentially reaching the MHz range [10,11]. CMV can induce unwanted currents to flow through the rotor shaft and bearings, particularly in systems utilizing high-frequency switching devices, such as those found in modern power electronics [10]. These currents discharge through the bearings, causing pitting, erosion, and premature failure. These parasitic currents require effective suppression techniques to prevent bearing damage and ensure reliable operation [11].

To mitigate these effects, excitation systems incorporate countermeasures such as common-mode filters, passive RC circuit, and enhanced grounding [10,11]. Insulated bearings, shaft grounding brushes, and differential mode chokes help redirect or dampen harmful currents, reducing common-mode voltage impact and improving machine reliability and lifespan [10,11].

4.

Circulating Bearing Currents

The path of shaft currents in the absence of an earthing system shows their flow through the shaft, bearings, hydrogen seals, and stator frame. Magnetic flux imbalances and high-frequency circulating currents can induce shaft currents, leading to electrical discharges that damage bearings, seals, and gears [42]. Circulating currents, shown by the blue dashed line in Figure 13, arise from potential differences along the shaft, forming a loop that includes the shaft and bearings [18,39]. These currents, often caused by magnetic flux imbalances or improper grounding, flow from the stator frame through the shaft and bearings [18]. A circulating bearing current flows in a loop from the stator frame to the shaft and the drive end if the shaft voltage is high enough to penetrate the bearing’s lubricating film and damage its insulating properties [18]. When the shaft voltage exceeds the dielectric strength of the bearing’s lubricating film, electrical discharges occur, leading to pitting, fluting, and frosting damage [18]. This deterioration increases friction and noise, ultimately reducing bearing lifespan and machine reliability. In circulating current scenarios, bearing currents in both bearings flow in opposite directions [18].

To restrict the flow of circulating current through the shaft, bearing insulation is usually provided on the non-drive end (NDE) of the rotor shaft [5,12]. However, if the insulated shaft becomes sufficiently charged, it can pose an electric shock hazard to operating personnel [11]. Therefore, the current practice involves providing insulation at one end, typically at the exciter end of the generator, while the drive end side of the generator is kept at ground potential through a grounding brush [12,18].

2.5. Damage of Bearings Caused by Bearing Currents

Bearing failures in turbine generators occur when a shaft voltage source and a low-impedance path to ground together allow current to discharge through the bearings [4,5]. Despite bearings being relatively inexpensive compared to the generator, their damage leads to substantial losses due to the downtime required for overhauls. For example, the drive end (DE) of the shaft often provides a naturally low-impedance return path, making it more vulnerable to circulating currents, whereas the non-drive end (NDE) is typically insulated (high impedance) to block those currents [5,12]. Accordingly, mitigation practices use insulated bearings at the exciter end and grounding brushes at the turbine end, alongside continuous shaft voltage monitoring, to interrupt harmful current paths [2,5].

2.5.1. Source Impedance Classification

Shaft voltage sources can be conceptually classified as “high-impedance” or “low-impedance” sources. This classification refers to the internal source impedance and its ability to drive current into the bearing circuit. A high-impedance source (such as electrostatic charge buildup on the rotor) can generate a high open-circuit voltage but cannot sustain large discharge currents due to limited charge and a large internal resistance. In practice, electrostatic sources in steam turbines have produced shaft potentials on the order of 30–150 V (even peaking ~250 V), but only microampere to milliampere-range discharge currents [4,12]. In other words, the source voltage is high, but the available fault current is a very small “weak” source that quickly collapses when a spark occurs [4,12]. In contrast, a low-impedance source (for example, a magnetically induced shaft voltage directly driven by the generator’s field) has a strong driving capability; even a relatively small voltage (on the order of a few volts or less) can drive a large current if a conductive path is present [4,12]. Here the internal resistance is low, so the source can deliver sustained amperage into the bearing. The literature does not set a hard numeric threshold between “high” and “low” impedance, but generally sources limited to very small currents (≪1 A) are considered high-impedance, whereas those capable of delivering tens or hundreds of amperes into the bearing circuit are termed low-impedance. For example, electrostatic shaft voltages are typically high-impedance (discharge currents in the mA range) [4,12], while magnetically induced voltages act as low-impedance sources that can drive tens of amps in a bearing loop [4,12].

2.5.2. Impact on Damage Severity

The source impedance dramatically influences discharge behavior and thus the bearing damage mechanism. High-impedance sources tend to produce intermittent, high-voltage sparks with limited current, similar to an ESD (electrostatic discharge) event. Each discharge event carries relatively low energy, causing microscopic pits or craters. Over time, however, the cumulative effect of repeated small discharges leads to surface pitting and eventually a frosted appearance as the pits coalesce [12,35]. This is commonly observed with electrostatic shaft voltages: the shaft charge builds until the oil film breaks down, then a brief spark erodes a tiny spot on the race, a process that repeats periodically [12,35]. The damage from high-impedance sources thus tends to be gradual; bearings develop fine pitting and frosting (widespread dull gray patches) over long periods [12,35]. In contrast, a low-impedance source will deliver much higher current during a breakdown. The discharge not only punctures the oil film but can momentarily form an arc with tens or hundreds of amperes flowing, given the low source resistance [4,12]. Such high current density causes intense localized heating, enough to melt bearing material and produce larger craters. Therefore, low-impedance sources can rapidly escalate damage severity. The same pitting and frosting phenomena occur, but they progress much faster and to a greater extent than with a weak source [12,35]. Moreover, additional failure modes become likely under high-current conditions: spark tracking (electrical discharge paths scoring the metal surface) and even welding of parts of the bearing may occur in extreme cases [12]. Indeed, welding or metal transfer is reported when extremely high circulating currents pass through a bearing, fusing elements of the race or rolling elements [12]. In summary, a low source impedance not only increases the amplitude of bearing current but also the energy per event, leading to more catastrophic damage in a shorter time than a high-impedance source would.

Several studies explicitly highlight the role of source impedance in bearing current damage. The authors of [4] noted that certain shaft voltage mechanisms (e.g., shaft magnetism) have such a high source impedance that they are “relatively weak and unable to drive much current to ground”. In those cases, the risk is lower because any discharge is self-limited. By contrast, induced voltages from magnetic asymmetry or ground faults behave as low-impedance sources capable of sustaining heavy fault currents [4,12]. Reference [12] describes how a continuous 50/60 Hz induced shaft voltage will drive a circulating current (if both ends of the shaft are grounded) large enough to cause fluting damage, whereas an electrostatic source builds up charge and discharges in pulses, causing more isolated pitting. The classic work of reference [25] on steam turbines likewise documented that electrostatic shaft charges produce high voltages but very little current, yet they still lead to bearing pitting if not mitigated. These distinctions are important for risk assessment and mitigation: a high-impedance source of shaft voltage is best handled by providing a safe discharge path (e.g., a grounding brush that bleeds off charge before a large voltage develops) [12,25]. In contrast, for a low-impedance source, the emphasis is on breaking the current path (e.g., insulating one bearing) to prevent a heavy current from flowing [4,12]. In practice, modern turbine generators employ both approaches: one end of the machine is usually insulated to block circulating currents (addressing low-impedance sources), and a grounding brush is applied to dissipate static charge (addressing high-impedance sources).

Table 2. Summary of source impedance effects [12,35].

Shaft Voltage Source	Impedance and Discharge Characteristics	Typical Damage Potential	References
Electrostatic buildup (e.g., friction of steam turbine blades)	High source impedance—can reach tens to ~100 V on shaft but only μA–mA discharge currents. Discharges are intermittent (charge–breakdown cycles) and unipolar.	Gradual pitting and frosting over time due to repetitive sparks. Fluting patterns may develop as microscopic pits accumulate. Generally, no immediate catastrophic damage, but progressive erosion and lubricant contamination occur if unmitigated.	[6,12]
Magnetic asymmetry (induced circulating AC)	Low source impedance—typically <1 V induced, but the shaft-frame loop offers a very low resistance. Even a few millivolts can drive high AC currents (tens of A) through bearings. Discharges can occur every half-cycle if a path is present.	Rapid frosting and fluting of bearing races due to continuous spark erosion at line frequency. Can lead to pronounced washboard patterns in a short time. Potential for spark tracks or thermal damage if currents are very high.	[6,12]
Shaft magnetization (homopolar DC shaft voltage)	The system has low source impedance and produces a steady DC voltage (typically a few millivolts to ~1 V) due to an internal DC generator effect. With minimal resistance in the metal path, significant continuous current can flow if a closed loop exists.	DC circulating currents can cause localized pitting, continuous arcing, and electrical wear at contact points, leading to unidirectional bearing fluting or material transfer. Prolonged exposure may result in uneven wear or even welding.	[12,30]
Excitation system and converter-induced common-mode (e.g., static exciters, inverter drives)	Moderate-to-high impedance at DC, but low impedance at high frequency. Imposes a DC bias plus fast voltage transients (dv/dt in kV/μs range). The capacitive coupling allows high-frequency discharge pulses (EDM events).	Electric discharge machining (EDM) effects on bearings, frequent high-frequency sparks cause fluting, pitting, and erosion of races. Can lead to premature bearing failure similar to high-impedance electrostatic cases but occurring at high repetition rates.	[6,10]

summarizes the typical impedance characteristics of various shaft voltage sources, along with their discharge behavior and associated bearing damage potential, as reported in the literature. This comparison highlights how the nature of the source (high vs. low internal impedance) correlates with specific failure modes in bearings.

Typical impedance classification of shaft voltage sources, with representative discharge behavior and bearing damage outcomes (based on [4,5,12,19]). This comparison underlines that high-impedance sources (electrostatic-type) tend to cause small, incremental damage (pitting/frosting), whereas low-impedance sources (electromagnetically induced or metallic sources) can cause immediate and severe damage (heavy fluting, spark tracks, welding) due to the much larger currents involved. Understanding the impedance class of a shaft voltage source is therefore crucial in assessing bearing current risk and choosing appropriate mitigation strategies.

2.5.3. Bearing Current Damage Type

The authors of [12] provided information on the recognition, origin, detection, and control procedures of bearing current damage in rotating machinery. The study reviews four primary types of bearing current-induced damage, namely pitting, frosting, welding, and fluting, each caused by different electrical discharge mechanisms. Bearing currents cause a progression of damage, beginning with pitting, followed by surface frosting, and eventually resulting in fluting patterns and lubricant degradation [35]. At the inception of shaft voltage damage, fine individual pits appear randomly [12,35]. Pitting is an early type of bearing damage caused by bearing current, as shown in Figure 14a. The current forms pits surrounded by melted material on the Babbitt’s surface. As the damage progresses the pits overlap, and the surface takes on a sandblasted or frosted appearance. As a result of this damage, the top metal surface is removed. Spark tracking damage is caused by shaft voltage discharges tracking across a metal surface. They appear as scratches on the metal surface, as shown in Figure 14b. The metal at the bottom of the track has melted. The authors of [12] presented four main types of current damage on bearings caused by bearing currents and shaft voltages in rotating machinery, which are:

• Frosting: The most common type, caused by electrostatic discharge breaking down the oil film’s resistance;
• Pitting: Similar to frosting but with larger irregularities due to stronger discharge sources. It is less common and affects localized areas;
• Spark Tracks: Visible damage caused by electrical arcing and oil contamination, forming consistent-depth tracks, often aligned with rotation;
• Welding: Occurs due to extremely high circulating currents, leading to severe damage detectable through visual inspection.

Shaft current damage from low-impedance sources also includes pitting and frosting [12]. The source impedance is low, which results in a high current flow (in the order of tens, hundreds or thousands of amps) [4,12]. The damage is more severe than in the high source impedance case, and the components’ deterioration occurs over a relatively short time. Understanding these failure mechanisms is crucial for implementing effective mitigation strategies to enhance bearing longevity and generator reliability. These failure modes highlight the critical need for effective shaft grounding, insulated bearings, and continuous condition monitoring to mitigate electrical bearing damage and enhance machine reliability. To consolidate these types of current damage on bearings in relation to shaft voltage phenomena,

Table 3. Shaft voltage and bearing current damage type and mitigation techniques [4,12,35].

Shaft Voltage Source Type	Bearing Current Damage Type	Mitigation Techniques
High Impedance Source	Pitting and Frosting—Initial fine pits develop into a frosted appearance, removing top metal surfaces.	Shaft grounding, insulated bearings, filters, optimized machine design, monitoring, and maintenance.
	Contaminated lubrication oil damage	Improved lubrication, sealed bearing housings, oil monitoring, synthetic lubricants, contamination prevention.
Low Impedance Source	Pitting and Frosting—Higher current levels cause severe damage in a shorter time compared to high-impedance sources.	Shaft grounding, insulated bearings, filters, optimized.
	Spark Tracking—Electrical discharges leave visible scratch-like marks; melted metal at the track bottom.	Shaft grounding rings, carbon brushes, common-mode chokes, ceramic-coated bearings, hybrid bearings, real-time monitoring.
	Welding—Excessive shaft current causes localized heating, welding metal surfaces together, leading to catastrophic failure.	Low-impedance grounding, electrically insulated bearings, high-resistance greases, monitoring, and diagnostics.
	Contaminated Lubrication Oil Damage—Higher particle concentration in oil leads to severe wear and damage.	Improved lubrication, sealed bearing housings, oil monitoring, synthetic lubricants, contamination prevention.

has been compiled, drawing upon sources in [4,12,35] to provide better knowledge of various bearing damages commonly encountered in turbine generator components from different shaft voltage sources within a turbine generator’s drive system. This table provides a structured analysis of the types of damage caused by shaft voltages and currents to bearing systems in rotating machinery and outlines the mitigation techniques employed to address these issues. It categorizes the damage type based on the impedance of the sources (high or low) and describes their mechanisms, effects, and corresponding countermeasures.

2.5.4. Classification of Bearing Currents and Damage Severity

The severity of bearing current effects varies significantly depending on several parameters, including the amplitude and frequency of the shaft current, the energy per discharge event, and the duration of exposure. While many studies discuss bearing damage qualitatively, several sources also provide quantitative severity indicators based on current amplitude, frequency content, and observed failure progression.

The authors of [4,12] report that even low-magnitude bearing currents (in the range of a few milliamps) can lead to gradual pitting and frosting over extended operation. In contrast, currents above 10–20 mA, particularly when sustained or repetitive, significantly increase the risk of fluting and deep surface erosion. High-current events (above 100–300 mA or more) have been associated with severe damage, including spark tracking, thermal degradation, and, in extreme cases, localized welding between the rolling elements and races [12,35]. Damage severity is also a function of the frequency of discharge:

• Low-frequency AC components (50–60 Hz) from circulating currents can produce fluting patterns due to repetitive arcing at regular intervals [6,12];
• High-frequency discharges (>1 kHz), often from capacitive coupling or power-electronic switching (e.g., SES or VFD systems), are more likely to cause electric discharge machining (EDM)-type damage, characterized by fine pitting, increased thermal stress, and premature lubricant breakdown [6,10].

According to reference [4], high-frequency pulses with high dv/dt with voltages in the order of 10 and 20 V between major peaks and currents above 5–10 mA can cause visible pitting within a few hours of operation. Based on field observations and lab testing, bearing current damage typically progresses through the following severity stages:

• Stage 1—Pitting: Isolated discharge craters <10 μm deep; caused by intermittent sparking (<10 mA) [4,12,43];
• Stage 2—Frosting: Surface dulling due to overlapping pits; often results from repetitive high-frequency EDM pulses (10–50 mA) [6,35,43];
• Stage 3—Fluting: Formation of symmetrical washboard-like grooves due to line-frequency arcing; associated with sustained 50/60 Hz currents (>50 mA) [12,35,43];
• Stage 4—Spark Tracking/Welding: Severe thermal and electrical stress (>100 mA), resulting in groove formation, smearing, or micro-welding between components [6,12,43].

Table 4. Severity classification of bearing current damage.

Severity Stage	Current Range	Frequency Content	Damage Type	Typical Source	References
Stage 1—Pitting	<10 mA	HF (1–20 kHz), sporadic	Small craters, light erosion	Electrostatic discharge, HF ripple	[11,12,35]
Stage 2—Frosting	10–50 mA	HF or AC (50 Hz) repetitive	Surface dullness, lubricant damage	HF excitation-induced currents	[6,12,43]
Stage 3—Fluting	>50 mA	50/60 Hz (line frequency)	Grooving/fluting pattern	Magnetic asymmetry, circulating current loop	[12,30,43]
Stage 4—Welding	>100 mA	DC or AC	Severe arc damage, welding	Shaft magnetization, ground fault loops	[6,10,43]

shows the severity classification of bearing current damage. These thresholds serve as practical guidelines for identifying and interpreting damage mechanisms in large turbine generators. Although exact thresholds vary with bearing type, lubricant condition, and machine configuration, the referenced studies provide a useful quantitative basis for severity classification.

2.5.5. Damage Types and Diagnostic Methods

Electrical discharge mechanisms lead to distinct physical damage patterns over time, which can be diagnosed through visual, imaging-based, optical microscopy and spectral analysis, as reported in various studies [12,35].

Pitting is the earliest form of electrical damage and is characterized by microscopic craters or pits on the bearing raceways or rolling elements. These pits are typically 1–10 µm deep and occur due to localized spark erosion during discharge events. Pitting is primarily detected through visual inspection under magnification using optical microscopy [12,35].

Frosting refers to the dull, matte appearance on the bearing surface due to the coalescence of many fine pits. It typically indicates ongoing exposure to low-energy but repetitive discharge events. Visual inspection uses a borescope or handheld microscope, as well as surface profilometry to quantify roughness [12,35]. Evenly distributed micro-erosion gives a “frosted” or etched look; it often lacks distinct craters. Fluting damage results from periodic arcing at line frequency (e.g., 50/60 Hz), causing symmetrical, evenly spaced grooves along the bearing surface, also known as a washboard pattern. Fluting is best identified via visual examination, axial sectioning of the bearing, and Fourier analysis of vibration data. Circumferential grooves align with the rolling path and are spaced based on electrical discharge frequency. Vibration spectral analysis often reveals harmonics related to the number of discharge events per revolution [12,35].

In severe cases, high-current discharge can lead to spark tracking where electrical arcs leave dark, burned trails or even welding, where metallic transfer occurs between bearing elements [12,35]. These are observable via visual inspection, thermal discoloration, and material hardness testing. Darkened or smeared areas, signs of melting, localized welds, or broken lubricant film layers [12,35].

3. Mitigation of Shaft Voltage-Induced Bearing Failures in Turbine Generators

This section provides a comprehensive analysis of the mechanisms responsible for shaft voltage-induced bearing failures, the types of damage associated with these failures, and the preventive strategies employed to mitigate their effects. By understanding the origins and impacts of bearing currents, targeted solutions can be implemented to enhance the reliability, efficiency, and operational lifespan of turbine generators.

3.1. Validation of Shaft Voltage Suppression Techniques

A wide range of shaft voltage suppression techniques have been investigated to reduce bearing current-induced damage in large rotating machines [4,33,34,35,36,37,44,45]. These include both passive (e.g., shaft grounding, bearing insulation, filters) and active (e.g., active voltage cancellation circuits) methods. While traditionally discussed in theoretical or qualitative terms, multiple empirical and case studies now provide robust validation of these suppression strategies [4,33,34,35,36,37,44,45]. Grounding brushes are widely used to provide a low-impedance path for electrostatic and high-frequency shaft voltages. Studies such as those in references [4,7] show that properly installed grounding brushes can reduce shaft voltages from tens of volts to near-zero levels, eliminating electric discharge machining (EDM) damage in bearings.

Installing electrically insulated bearings, typically at the non-drive end, is an effective strategy to prevent circulating shaft currents, especially those induced at low frequencies (50/60 Hz) due to magnetic asymmetries [5,9]. A fleet study of over 190 hydro-generators [5] emphasizes that the type and placement of insulation greatly affects its reliability and ability to be monitored. However, insulation can degrade over time from thermal cycling, mechanical stress, or contamination [19]. Although real-time monitoring remains limited, various practical assessment methods are employed. These include differential shaft-to-ground voltage checks and online detection of abnormal shaft current, or voltage patterns signals that may indicate insulation failure or bypass [1,7,9,46]. A sudden drop in voltage asymmetry or reappearance of shaft voltage at the insulated end is a strong indicator of insulation failure [5]. While no standard online monitoring system exists, many studies recommend combining periodic offline insulation resistance tests and shaft ground current monitoring with continuous shaft voltage trending to ensure insulation effectiveness of the NDE bearing insulation health [1,5,7]. This multi-pronged strategy is consistent with recommendations in IEC 60034-27-3 and IEEE 112 for condition monitoring in rotating machines [47,48]. Incorporating these diagnostics into a condition-based maintenance program is essential for preventing unintentional current paths and ensuring long-term generator reliability [5].

Thyristor-based static excitation systems generate high-frequency ripple and common-mode voltages that can capacitively couple to the shaft [10,34]. The authors of [10] showed that installing RC snubber circuits and EMI filters significantly reduced ripple content, lowering shaft voltage amplitude and high-frequency discharge risk. The authors of [34] confirmed similar results by suppressing exciter switching transients, thereby reducing high-frequency shaft voltage spikes.

The recent literature provides strong empirical support for active suppression technologies [44,45]. These methods aim to cancel or compensate common-mode voltages before they induce harmful shaft potentials. Active Common-Mode Canceler (ACC) proposed an active circuit to cancel common-mode voltage generated by a PWM inverter.

While passive methods are more established and widely applied in industrial generators, active suppression is emerging as a powerful complement, especially in modern power electronics-fed systems (e.g., with PWM inverters or static frequency converters). Passive techniques (e.g., grounding, insulation, filters) remain effective for conventional generators [1,2,49,50]. Active approaches (e.g., ACC) offer significant advantages for high-frequency EMI, complex multi-phase drives, and compact inverter-fed systems [37,44,45].

3.2. Case Studies on Mitigation Effectiveness

Industrial case studies provide compelling evidence of the damaging impact of shaft voltages and the effectiveness of mitigation strategies. For example, reference [6] documented investigations across several large generating stations (66–800 MW), revealing multiple instances of component failure directly attributed to shaft potentials. Unit #2 in the reported study suffered repeated PMG bearing failures due to circulating shaft currents. The root cause was traced to shaft potentials exceeding the dielectric strength of the lubrication oil film. Several generators reported damage to hydrogen seals and oil-pump gears due to sparking and arcing across inadequately insulated bearing paths. Shaft voltages caused arcing in the speed control system, degrading insulation and requiring expensive component replacements. In a 66 MW turbogenerator, the shaft-to-ground voltage was measured to rise rapidly to ~130 V before collapsing, with the discharge occurring across the oil film. This repetitive sparking led to pitting and surface degradation.

Additional industrial evidence presented by reference [4] reinforces the critical importance of grounding brush performance in mitigating shaft voltage damage. In the case of a 350 MW turbine generator, a catastrophic bearing failure occurred due to circulating shaft currents facilitated by an ineffective dual-brush grounding configuration. The installation of a single-point silver-plated copper-braid grounding system eliminated the current loop, reducing shaft potential and preventing further failures. Similarly, a study by reference [28] reports a case of electrical discharge machining (EDM) damage in a large steam turbine generator where electrostatic discharges resulted in bearing degradation. The introduction of microfiber shaft grounding rings mitigated these discharges and restored operational integrity.

Reference [40] investigated high shaft currents in a large two-pole, 500 HP sleeve-bearing induction machine powered by a variable frequency drive (VFD). The data validated the notion that mitigation using grounded shaft ring contact brushes significantly reduced shaft voltage and subsequent damage risk [9,40]. Further, fleet-wide analyses of 195 hydro-generators highlighted statistical correlations between insulation type and bearing damage risk. Types 5 and 6, with more exposed insulation systems, demonstrated frequent bypass and failure. Conversely, Type 2 systems employing layered insulation between the shaft and rotating components proved effective when maintained properly.

These case studies collectively validate the practical application and effectiveness of various suppression techniques. They provide critical insights into performance under real operational stresses, supporting the theoretical models and laboratory trials discussed earlier. By integrating these examples into the review, the discussion now benefits from empirical grounding and actionable guidance for industry implementation.

3.3. Scalability of Motor-Based Strategies

While many shaft voltage mitigation techniques, such as insulated bearings, shaft grounding rings, carbon brushes, silver graphite, copper braids and common-mode voltage suppression filters, originate from motor applications, their direct scalability to large-scale turbine generators is not always straightforward [1,4,38,50]. In smaller inverter-fed motors, grounding brushes or conductive microfiber rings, as validated by reference [38], effectively shunt high-frequency shaft currents to ground. However, in large synchronous generators, the mechanical dimensions, shaft surface speed, and thermal loading require significant adaptation of these solutions. For instance, grounding systems in turbine generators often employ multi-brush or fiber-based contact assemblies engineered to handle higher currents and endure greater mechanical stress without loss of contact. Similarly, while ceramic hybrid bearings are effective in small-to-medium machines, they are generally unsuitable for large turbine generators due to limitations in load-bearing capacity and insulation durability at high shaft diameters and thrust loads [5,9,51].

The authors of [10] demonstrated successful implementation of capacitive filtering and grounding strategies in hydro-generators with static excitation systems, highlighting the need to tailor filter bandwidth and grounding impedance to the specific excitation frequency spectrum. Additionally, in large generators, parasitic capacitances and shaft inductance become more significant, requiring refined modelling and integration of suppression components. Thus, while the underlying physical principles remain the same, the application of motor-derived strategies to large-scale generators requires system-specific redesign, material scaling, and robust validation. The existing literature supports this transition through case studies and field implementations that have been adapted to suit utility-scale operating conditions [1,4,5,38,43].

Field data and experimental studies show that combining mitigation strategies, such as insulated bearings with shaft grounding, or filters with active voltage cancellation, offers optimal suppression across a wide frequency range [10,11,45]. These validated approaches support more informed selection and implementation of suppression techniques based on machine-specific conditions and shaft voltage sources.

Table 5. Mitigation strategies for bearing currents in turbine generators.

Sources	Solution	Description	Merits	Drawbacks	References
High-impedance source	Insulated bearings	Use insulated bearings at the non-drive end to block circulating shaft currents.	Effectively isolates bearings from shaft currents, reducing damage risk.	High cost, complex installation, potential mechanical performance impact.	[5,43,52]
	Shaft grounding brushes	Install grounding brushes on the rotor shaft to provide a low-impedance discharge path.	Cost-effective, easy to install, widely used in generators.	Brushes wear out over time, requiring regular replacement and maintenance.	[1,2,3,4,30]
	RC shaft earthing circuit	Use of a resistor-capacitor (RC) circuit to ground the shaft and dissipate high-frequency currents.	Reduces transient voltage buildup, prevents excessive shaft voltages.	Requires careful tuning of RC components, not effective for all voltage ranges.	[10,34]
Low-impedance source	Bonding strap between generator and frame	A conductive strap connecting the generator casing and frame to equalize potential differences.	Prevents potential buildup and reduces unwanted circulating currents.	Requires proper connection design to avoid unintended current loops.	[17,41]
	Shaft grounding rings	Conductive rings that provide a controlled low-resistance path for shaft currents.	Effectively prevents shaft currents from damaging bearings.	Subject to wear and degradation over time.	[38,43,53]
	Electrostatic shielding	Incorporation of conductive shielding around critical components to minimize induced voltages.	Reduces capacitive coupling and unwanted voltage buildup.	Increases system complexity and cost, requires design optimization.	[31,51]
Coupling and mechanical interface effects	Insulated coupling	Insulated coupling between the generator and turbine to prevent current transfer.	Prevents shaft currents from reaching turbine components.	Expensive, adds mechanical constraints.	[5,52]
Excitation system-induced currents	Filters and chokes	Use of inductive chokes or passive filters to suppress high-frequency components in the excitation circuit.	Reduces voltage fluctuations and transient surges.	Complex filter design, may introduce power losses.	[39,49]
	Active shaft voltage suppression	Use of active compensation circuits to counteract induced shaft voltages.	Adaptive control improves suppression across different operating conditions.	High complexity, requires continuous monitoring.	[17,37,44,45]
Lubrication-related effects	Sealed bearing housings	Fully sealed bearing enclosures to prevent contamination from electrical discharge.	Extends bearing lifespan, reduces need for frequent maintenance.	Adds system complexity, may increase operating temperature.	[3,17]
	High-resistivity lubricants	Use of special synthetic lubricants with high dielectric strength to minimize conductive paths.	Increases insulation between bearing surfaces, reducing discharge risk.	May alter other lubricant properties, requiring formulation adjustments.	[3,17]

summarizes key strategies, including insulated bearings, grounding brushes, active suppression systems, and shielding methods, along with their effectiveness in turbine generator applications.

3.4. Grounding System Technical Performance Evaluation

Effective shaft voltage suppression in large turbine generators (TGs) relies significantly on the performance of the grounding system. Various grounding technologies, such as carbon brushes, silver-graphite brushes, gold-bristle types, copper braid, and conductive microfiber rings, are employed to dissipate parasitic voltages and prevent damage from circulating shaft currents. However, their effectiveness varies widely depending on the operating environment, maintenance practices, and frequency content of the induced voltages.

Traditional carbon or silver-graphite brushes, while widely used, are prone to wear, glazing, and degradation in contaminated or humid environments. As documented by reference [4], insufficient maintenance or tensioning of these brushes can result in elevated shaft voltages, reaching as high as 60–150 V. Moreover, shaft voltage monitoring campaigns conducted by reference [1] across over 30 generators revealed that many units failed to maintain grounding currents below critical thresholds due to poor brush contact or contamination. These brushes often require replacement every 6–12 months and are sensitive to oil film buildup, dust, and misalignment.

Gold-bristle brushes, as introduced by reference [30], present significant improvements in contact stability and maintenance intervals. They exhibit low resistance, require no cleaning or realignment, and are resistant to oil contamination. The authors of [12] confirmed their superior grounding consistency, but they come at a higher cost and require shaft machining for installation.

Conductive microfiber grounding rings, such as those described by reference [38], offer a non-contact solution with exceptional performance in high-frequency environments. These rings consist of thousands of microfibers encircling the shaft, creating multiple conduction paths that exhibit negligible frictional wear and are inherently robust to oil mist, dust, and thermal cycling [53]. Laboratory experiments demonstrated voltage suppression from 10 to 15 V to below 2 V, with shaft current handling capacity between 7 and 78 A. Importantly, even with only a small percentage of microfibers in active contact at a given moment, performance remains stable due to the distributed nature of the contact network. These devices are particularly well-suited for inverter-fed machines or systems subjected to high $dv/dt$ transients where traditional brushes may fail due to skin effect or contact arcing.

Copper-braid brushes offer a practical compromise between cost, reliability, and serviceability [4,54]. As highlighted in industry reports, they support online replacement, long service intervals (1–3 years), and moderate resistance to environmental degradation [4,54]. They are increasingly deployed in high-reliability baseload plants where downtime is critical.

Table 6. Comparative performance of grounding technologies for turbine generators [1,2,3,4,38,50,54].

Brush Type	Contact Resistance	Wear Rate	Maintenance Cycle	Oil/Dust Tolerance	Frequency Suitability	Field Proven Use
Carbon/Graphite Brushes	10–100 mΩ	High	6–12 months	Low	50/60 Hz	Widely used
Gold-Bristle Brushes	<10 mΩ (stable)	Low	1 year+	High	Low to mid frequency	Moderate
Copper-Braid Brushes	~20–50 mΩ	Moderate	1–3 years	Moderate to High	Low to mid frequency	Common in steam TGs
Microfiber Grounding Rings	~0.01–0.5 Ω (dynamic)	Negligible	Maintenance-free	Very High	kHz–MHz range	Emerging best practice

summarizes the comparative performance of common grounding systems based on resistance, durability, environmental tolerance, and maintenance needs.

3.5. Safety and Technical Standards Considerations Related to Shaft Grounding

Beyond equipment reliability, the management of shaft voltages is also a critical safety concern. Ungrounded or improperly grounded generator shafts can accumulate significant voltages, often exceeding 60 V in high-inertia systems, which pose serious electrical hazards to maintenance personnel and operators. According to [4], operators at a 350 MW steam turbine facility were exposed to dangerous shaft potentials due to an ineffective dual-carbon brush grounding system. The system was subsequently replaced with a single-point, silver-braid grounding strap, which eliminated the potential difference and restored safety compliance.

In another incident reported by [28] a generator experiencing electrostatic discharge due to shaft potential buildup posed not only a threat to the bearings but also created touch voltage hazards on accessible metal parts. This was especially problematic during maintenance when personnel unknowingly contacted conductive surfaces energized by shaft voltages.

IRIS Power’s shaft condition monitoring (SCM) system application in reference [1] also warns that unmonitored shaft voltages may energize ungrounded metal frames, creating shock risks in accordance with IEC and IEEE electrical safety standards [55,56]. Their recommendation includes using continuous monitoring to issue alarms when voltages exceed safe thresholds (typically 25–30 V) and confirming electrical continuity in the grounding path. To mitigate these risks, it is essential that grounding systems be:

• Installed at a single, low-impedance point near the exciter or turbine end;
• Routinely inspected for wear, contamination, and continuity;
• Verified during outages through shaft voltage trending and discharge current logging;
• Maintained in accordance with IEEE Std 80 [55], IEC 60034-1 [56] and IEC 60034-27-3 [48] standards for shaft earthing.

While a variety of shaft grounding and mitigation techniques have been implemented, not all are explicitly guided by international standards. IEEE Std 80 [55] outlines general grounding practices in AC substations and is often referenced in turbine-generator shaft earthing layouts.

Notably, IEC 60034-27-3 [48] provides recommendations for the measurement and evaluation of shaft voltages and bearing currents in rotating electrical machines. It specifies methods for determining acceptable limits (e.g., shaft voltages above 25 V peak may trigger risk of EDM), yet only a few of the reviewed case studies explicitly measure or compare their suppression effectiveness against these benchmarks. This absence of standardized thresholds limits comparability between studies and complicates industry adoption.

4. Measurement of Shaft Voltage and Bearing Current in Turbine Generators

Monitoring shaft voltage and bearing currents in turbine generators is crucial for ensuring machine reliability, preventing catastrophic failures, and enhancing predictive maintenance strategies [1,7]. Undetected shaft currents can lead to severe bearing damage, increased operational costs, and unexpected outages [4]. Despite their significance, standardized measurement methods for these parameters remain limited, necessitating a comprehensive review of available techniques and their applicability to turbine generators.

Previous studies have primarily focused on high-frequency (HF) bearing current measurement methods, particularly in the context of inverter motor operations [13,16]. However, the challenges and methodologies specific to large turbine generators require a broader and more in-depth investigation. This paper aims to bridge this gap by analyzing various experimental methods for measuring shaft voltage and bearing currents in turbine generators. The review covers techniques such as shunt or ammeter methods, current transformers (CTs), Rogowski coils, shaft earthing methods, and high-frequency voltage probes.

Effective monitoring of shaft voltage and bearing currents is vital for early detection of electrical discharge-related issues that could lead to bearing damage and other operational failures [1,7]. Early detection allows operators to take preventive action, such as scheduled maintenance or temporary shutdowns, to prevent costly repairs and extended downtime [17]. Despite the absence of universally standardized methods, several intrusive and non-intrusive measurement techniques have been developed and refined for improved detection and analysis. A variety of measurement methods exist for assessing shaft voltage and bearing currents in turbine generators. Each technique offers unique advantages and challenges.

The ammeter or shunt method is among the most traditional techniques for measuring shaft currents. This method is outlined in the IEEE 112 standard [47] and involves measuring current using a low-impedance path between shaft ends. However, it has been found to provide inaccurate readings and is often unsuitable for high-precision applications [9].

The current transformer (CT) is widely used in shaft current monitoring and is integral to ABB’s RARIC Shaft-Current Relay system [43,46]. This method provides non-intrusive measurement, high sensitivity, and continuous monitoring of shaft currents. The CT is installed between the non-insulated bearing and the generator to detect and measure shaft currents in turbine generators [43,46]. While highly effective, the CT method has limitations, including susceptibility to external interference, complexity in installation, and the requirement for regular calibration and maintenance. To evaluate these limitations and performance characteristics, an experimental validation setup was implemented in both laboratory and field environments, as described in [43]. This included the Zelisko GWR3 and ILDD 096 current transformers, each tested for linearity, frequency response, phase shift, and sensitivity across the 50–180 Hz frequency range. Devices were clamped around the shaft, and secondary voltages were recorded across precision resistors to emulate the RARIC input [43]. These validation trials confirmed that, while CTs like the GWR3 offer reasonable accuracy, they are subject to saturation and frequency-dependent behavior at low currents [43]. Additionally, external interference rejection tests showed limitations in shielding against stray flux, which may impair signal integrity in generator environments.

The Rogowski coil method is extensively documented in numerous studies and provides a flexible approach to measuring high-frequency shaft currents, particularly those associated with high dv/dt from frequency converters [9,57]. The Rogowski coil encircles the shaft to detect circulating currents, including common-mode voltages [51]. However, the application of this technique requires significant generator preparation, making it less practical for in-field usage [43,58]. Nevertheless, ABB and other industry leaders extensively use Rogowski coils for monitoring shaft currents. In experimental setups (e.g., [9,43,57,59]), coils are looped around the bearing support or shaft end, with their output connected to a high-speed differential amplifier and digitized via a higher bandwidth oscilloscope. This configuration enables detection of transient discharge events in the microsecond range [43,59]. Reference [43] evaluates the Rogowski coil as a modern alternative to conventional current transformers (CTs) for shaft current measurement in generator applications. The experimental validation confirms that Rogowski coils are a reliable and flexible solution for shaft current monitoring, particularly in environments where high-frequency resolution, wide dynamic range, and non-saturating sensors are essential.

The AEGIS high-frequency (HF) voltage probe provides a sensitive and non-intrusive method for detecting bearing voltage in rotating machinery [38,53]. Installed in close proximity to the bearing, it enables real-time monitoring with a wide bandwidth. The authors of [38] experimentally validated shaft current mitigation using conductive microfiber rings in an induction motor driven by a PWM inverter. Their controlled test setup involved a 10 HP, 460 V, 60 Hz induction motor driven by a pulse-width modulated (PWM) inverter, which is known to induce common-mode voltages and high-frequency circulating currents. The microfiber ring was tested in two configurations: (1) as the sole grounding path and (2) in parallel with the motor’s bearings, which represent a common unintended discharge path in industrial systems. In both cases, the microfiber ring reliably diverted the majority of shaft current, particularly at elevated motor speeds where capacitive and inductive coupling effects become more prominent. The application of the microfiber ring resulted in a reduction of shaft voltage from approximately 10 to 15 V to below 2 V, a level well below the dielectric breakdown threshold of standard bearing oil films. This voltage suppression effectively minimized the risk of electric discharge machining (EDM) damage to bearing surfaces, a known failure mechanism in inverter-fed rotating machines. Importantly, the microfiber ring’s design, comprising thousands of conductive fibers, ensures multiple points of contact, offering redundant conduction paths and resilience to shaft surface imperfections or misalignment. These findings demonstrate not only the microfiber ring’s superior grounding effectiveness across a wide frequency spectrum (including the kHz–MHz range), but also its long-term viability in demanding industrial applications. As such, it is particularly suited for modern inverter-driven machines and static frequency converter (SFC)-fed turbogenerators, where traditional grounding solutions often struggle to suppress high-frequency phenomena effectively.

However, its effectiveness can be affected by environmental conditions, complex data interpretation, and higher operational costs. Regular maintenance is also required to ensure accuracy and reliability [43,51]. Despite these drawbacks, its capability to provide detailed high-frequency voltage data makes it a valuable tool for diagnosing and preventing electrical failures in turbine generators.

In addition to direct measurement methods, shaft earthing techniques play a dual role in monitoring and mitigating shaft voltage and currents. Shaft earthing brushes are commonly used for both grounding and measuring shaft voltage and current [4,50]. Reference [7] discusses the successful development of a shaft measurement and monitoring tool called shaft condition monitoring (SCM). This method serves as an early warning system for rotating machinery, helping operators detect and address shaft voltage issues before they escalate.

A well-maintained shaft earthing system is crucial for ensuring accurate shaft voltage and current measurements. Environmental factors, such as dust, oil contamination, and brush wear, must be carefully managed to maintain system effectiveness [2,4]. Regular maintenance is essential to ensure reliability and operational excellence.

The commonly used shaft earthing system measurement configuration for a turbine generator is shown in Figure 15 and is also utilized in [1,7]. This shaft measurement method, widely adopted by utilities worldwide, is particularly well-suited for large turbine generators [1,2]. The shaft earthing measurement system employs brushes at the exciter and turbine end sides which function as voltage brushes, serving as sensors to detect and measure shaft voltages (brush signals). Meanwhile, the brush at the drive end (DE) acts as the current brush (earthing brush), ensuring a low-resistance grounding path for effective shaft current dissipation [7,60]. The required number of brushes depends on the number of electrically isolated rotating shaft sections [1,7]. The earthing brush is connected to an earthing cable via a low-resistance (100 milliohm) shunt resistor, enabling shaft current measurement while also allowing stray shaft currents to discharge safely to the earthing mat [7].

The shaft section between the low-pressure (LP) turbine and the generator is the typical location for shaft grounding brushes and generator non-drive end shaft extensions, where electrical insulation of bearings and seals is implemented, and excitation enters the rotor winding. Additionally, any faults occurring within the generator, such as bearing faults, winding faults, segment looseness, or mechanical looseness, can be identified and monitored externally by analyzing the raw signatures of shaft voltage and current by employing applicable signal processing techniques [7,61]. By carefully selecting and integrating these methods into a comprehensive shaft monitoring system, reliable measurement of shaft voltage and bearing current can be achieved, leading to enhanced protection and early fault detection in critical rotating machinery.

Monitoring shaft voltage and bearing currents in turbine generators is essential for ensuring machine reliability, preventing damage, and optimizing maintenance strategies. Various measurement techniques, including shunt/ammeter methods, current transformers, Rogowski coils, AEGIS HF probes, and shaft earthing methods, each have distinct advantages and challenges. The choice of an appropriate measurement technique depends on factors such as accuracy, ease of installation, maintenance requirements, and operational environment. Continuous monitoring and predictive maintenance strategies enhance equipment reliability, reduce maintenance costs, and prevent unexpected failures. To summarize shaft voltage and current measurement techniques on large turbine generators, see

Table 7. Comparison of measurement methods for detecting shaft voltage and currents.

Measuring Methods	Description	Merits	Drawbacks	References
RARIC Shaft-Current Relay Method	A current transformer (CT) is placed around the shaft to measure the induced current. It converts the shaft current into a proportional secondary current.	It provides accurate monitoring, electrical isolation, long-term reliability, and isolated measurements.	Requires installation space and can be influenced by the core’s magnetic properties, complex installation, and space requirements.	[43,46]
Rogowski Coil	A Rogowski coil is a non-magnetic coil wound around the shaft. It measures the rate of change of current and provides a voltage output proportional to the current.	Measures real shaft currents including high-frequency currents.	Requires integration and calibration for accurate measurement, susceptible to noise, complex installation, and it’s intrusive	[9,57]
Shaft Earthing System Method	Earthing brushes, made of conductive carbon or metal braids, are a widely used shaft earthing method.	Effective mitigation includes various materials, real-time monitoring, and continuous online integration into condition monitoring systems.	Regular maintenance is needed due to brush wear, poor shaft contact, and environmental contamination, which can degrade performance.	[1,7]
Shaft Grounding Rings	Shaft grounding rings are non-contact earthing devices that use conductive microfibers or brushes arranged in a circular ring around the shaft.	Non-contact design reduces wear and maintenance, while effectively handling high-frequency currents induced by the excitation system.	It is sensitive to alignment, with limited flexibility for online replacement and maintenance, and may be less effective in contaminated or oily environments.	[38,53]
Shunt or ammeter Method	In this method, a low-resistance shunt is placed across the shaft bearings, and an ammeter measures the current passing through the shunt to determine shaft currents.	Simple, straightforward to implement and understand, direct measurement readings.	Inaccurate, provides only an estimate of the shaft current.	[43,47]
High-Frequency Voltage Probe	The Aegis High-Frequency Voltage Probe is designed to measure shaft voltage in rotating machinery, effectively detecting high-frequency voltages from discharges, grounding issues, or operational anomalies.	Monitors shaft voltage with a wide bandwidth and measures high-frequency voltages with minimal distortion.	The probe is sensitive to environmental factors, shaft pollution, and high-frequency signals, making it prone to electromagnetic interference (EMI).	[51]
Nonintrusive radiofrequency (RF) Detection Method	It uses RF sensors to detect electromagnetic emissions from electrical discharges or circulating currents in the rotor and shaft.	Highly sensitive to high-frequency electromagnetic emissions, enabling early fault detection, with lower maintenance needs compared to carbon brushes or slip rings.	Susceptible to electromagnetic interference (EMI), requiring shielding and filtering for accuracy, and has high implementation costs	[62,63]

.

5. Turbogenerator Condition Monitoring and Health Assessment

The reliability and efficiency of turbogenerators are critical to the operational stability of power generation systems. Effective condition monitoring and health assessment strategies play a vital role in mitigating risks associated with unexpected failures, reducing downtime, and optimizing maintenance practices [1,7]. This section explores advanced diagnostic methodologies and emerging technologies in turbogenerator condition monitoring, with a particular focus on shaft signal analysis, bearing current prediction, and AI-driven fault detection techniques. Recent advancements in artificial intelligence, machine learning, and signal processing techniques have revolutionized fault detection and predictive maintenance [15,22].

This section provides a comprehensive analysis of turbogenerator condition monitoring strategies, emphasizing the role of shaft signals and bearing current analysis in predictive maintenance. It highlights the application of AI-based techniques, such as convolutional neural networks (CNNs) and Long Short-Term Memory (LSTM) networks, in diagnosing and forecasting generator faults with high accuracy. Additionally, it underscores the importance of frequency-domain analysis, including Fast Fourier Transform (FFT) and wavelet packet analysis, in identifying fault signatures and their progression over time. Despite mitigation devices for high shaft voltage and current, critical applications require additional monitoring systems to verify the effectiveness of shaft earthing and detect emerging faults. Failure to monitor shaft voltages and currents in large generators can lead to forced outages, significant operational losses, and premature asset failure [1,7]. By leveraging advanced signal processing techniques, including FFT, wavelet transforms, and artificial intelligence (AI)-driven anomaly detection, it can enhance fault prediction accuracy and reduce unplanned outages. This study explores the role of shaft voltage and bearing current signals in turbine generator diagnostics, highlighting how harmonic analysis and multi-sensor data interpretation can refine predictive maintenance strategies and improve long-term asset management.

5.1. Bearing Current Monitoring and Predictive Fault Detection in Large Generators

Traditional diagnostic methods, such as vibration monitoring, have been widely employed to detect bearing damage resulting from circulating currents [64]. However, extracting meaningful mechanical vibration signals is complex, as it requires filtering out noise and distinguishing fault-related patterns from normal operating conditions. While techniques like envelope detection and spectral analysis have improved fault detection, they often struggle to provide early warning of bearing degradation.

Recent advancements in artificial intelligence (AI) and machine learning have revolutionized the field of predictive maintenance by enabling highly accurate and automated bearing fault detection [21,22]. AI-driven techniques, including convolutional neural networks (CNNs), recurrent neural networks (RNNs), Long Short-Term Memory (LSTM) networks, and Generative Adversarial Networks (GANs), are increasingly being applied to motor fault diagnostics techniques [21,22]. These deep learning models facilitate automated feature extraction, allowing for early fault detection and precise fault classification, reducing reliance on manual analysis. Furthermore, deep learning models can process large datasets and detect complex fault patterns that traditional methods may overlook. Advanced techniques, such as wavelet packet analysis combined with LSTM networks, have demonstrated over 96% accuracy in classifying bearing faults [17,22]. The recent literature demonstrates that deep learning algorithms, particularly CNNs and hybrid architectures, can achieve bearing fault classification accuracies exceeding 96% under standardized conditions [15]. The authors of [17] reported classification rates between 96.1% and 99.3% using CNNs trained on the CWRU dataset, validated via ten-fold cross-validation across varying speeds and loads. Similarly, reference [17] emphasized that data-driven models, when properly trained and validated using high-resolution input signals and comprehensive labeling strategies, provide reliable predictions of fault types, such as pitting and fluting, even in early degradation stages. These studies used publicly available datasets (e.g., CWRU, IMS) and documented their model evaluation metrics, including confusion matrices and receiver operating characteristics, confirming the validity of their accuracy claims [15,65]. The authors of [15] presented a comprehensive review of deep learning methods for bearing diagnostics reported that convolutional neural networks (CNNs) consistently achieved accuracy levels above 95–98% when evaluated on standard public datasets such as the CWRU (Case Western Reserve University) bearing dataset. These results were obtained under varying operating conditions (e.g., different loads, speeds, and sensor positions). The performance was evaluated using test–train splits, confusion matrix analysis, and, in some cases, transfer learning and domain adaptation to test model robustness across different conditions. AI-driven approaches also enable real-time condition monitoring, allow the anticipation of failures, optimize maintenance schedules, and enhance generator reliability. Effective monitoring of bearing currents is crucial for predicting bearing failures and optimizing suppression techniques to enhance machine reliability and lifespan [15,17]. Extensive research has been conducted on the impact of bearing currents on bearing degradation, with a particular focus on the chemical and structural changes that occur due to electrical discharge and thermal stress [15,17]. Despite these advancements, a key challenge remains in determining the critical threshold at which bearing damage becomes irreversible. This threshold directly influences the remaining service life of the bearing and impacts maintenance strategies.

5.2. Bearing Current Prediction

Predicting bearing current-related failures is crucial in extending machine lifespan and ensuring operational reliability. With the advent of machine learning (ML) and advanced signal processing, predictive maintenance strategies have evolved beyond traditional monitoring techniques [15,22]. The integration of deep learning, convolutional neural networks (CNNs), and multi-sensor data fusion has enabled real-time condition monitoring of turbine generator bearings, incorporating both mechanical (vibration, temperature) and electrical (shaft voltage, bearing current) parameters for a comprehensive assessment [15,22]. Currently, extensive research is being conducted on bearing life prediction using vibration measurements [64,66]. Among the various diagnostic techniques, vibration monitoring remains one of the most widely utilized methods for detecting bearing damage caused by circulating currents [64,66]. This approach enables early fault detection by analyzing mechanical vibrations associated with bearing wear and degradation.

The authors of [22] present a well-structured and technically sound approach to bearing fault detection using convolutional neural networks (CNNs), achieving high classification accuracy. However, while the study focuses on bearing fault detection using vibration signals, it does not explicitly address bearing current prediction, which is crucial in electrical rotating machinery, and does not explore shaft voltage and bearing current measurements, which are critical for detecting electrical bearing faults in generators and motors [16]. The influence of stray shaft currents on bearing failures should have been incorporated as an additional input parameter, especially for electrically driven rotating machines [67].

The success of AI models, such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), and probabilistic filtering approaches, largely depends on the quality and characteristics of the training data [15,17,22]. Effective implementation requires high-resolution, multi-modal time-series data that captures electrical and mechanical behaviors under varied operating conditions. Common input signals include shaft voltage, shaft current, vibration, and stator current waveforms, often sampled at frequencies ranging from 12 to 48 kHz to ensure the capture of high-frequency transients such as those produced by electric discharge machining (EDM) or electrostatic discharges (ESD) [15,17,22]. Input signals are typically pre-processed through normalization, Fast Fourier Transform (FFT), or continuous wavelet transform (CWT) to extract discriminative features, with CNNs trained on time-frequency representations such as scalograms, and RNNs applied directly to raw or filtered sequential data [15,66]. These signals are acquired via Rogowski coils, current transformers (CTs), shaft condition monitoring systems, voltage probes, or piezoelectric sensors depending on the application [1,38,43,60].

Deep learning models require structured datasets containing fault and healthy condition data, balanced across multiple load states, with labeled signal windows ranging from 1 to 2 s in duration [15,21,22]. Benchmark datasets, such as CWRU, IMS, PRONOSTIA, and Paderborn, are frequently utilized for model development and validation, though domain-specific datasets remain limited for large-scale generators [15,21,22]. In addition, probabilistic models, such as particle filters and Bayesian networks, have been applied for health tracking using statistical distributions of input features, offering robust uncertainty quantification in evolving fault conditions [68]. The integration of these AI techniques with real-time monitoring platforms enables scalable, data-driven maintenance strategies that complement conventional condition-based diagnostic methods in high-voltage synchronous and wind turbine generators [15,21,22]. Figure 16 presents a flowchart depicting an artificial intelligence-based method for predicting bearing failures.

5.3. Existing Shaft Voltage and Current Condition Monitoring Tools

Shaft voltages and currents, while generally undesirable, serve as valuable indicators for monitoring generator performance [2,36]. In large turbogenerators, shaft signal analysis provides supplementary fault indicators, improving diagnostic accuracy and reducing the risk of failures [5,19]. These signals offer critical insights into the machine’s health, adding an additional layer of protection against undetected faults. Measurements of shaft voltage and current help detect issues such as insulation breakdown, bearing damage, and electrical faults [61,69]. However, for accurate fault detection, it is essential to quantify the contribution of each defect to the shaft voltage. Proactive maintenance strategies leverage these measurements to track generator parameters, identify potential faults early, and prevent downtime. Even with mitigation devices in place, critical applications often require additional condition monitoring systems to detect faults and verify shaft earthing effectiveness [2,60]. Failure to monitor shaft voltages and currents in large turbogenerators can lead to forced outages and significant operational losses [1,7].

The authors of [57] developed and patented an online generator condition monitoring system known as Shaft Voltage and Current Condition Monitoring (SCM), demonstrating its effectiveness in detecting impending generator field faults. The SCM enhances generator condition monitoring and predictive maintenance by providing real-time shaft current detection, fault diagnosis, and automated protection mechanisms. It analyzes raw shaft voltage signatures to identify various fault conditions, including shaft or rotor rubs, static charge buildup, residual magnetism-induced voltage generation, electromagnetic asymmetries, and voltage harmonics or transients [2,36]. This system highlights the effectiveness of utilizing shaft earthing currents and voltages for turbine generator condition monitoring. The research presents a technical approach to machine condition monitoring and health diagnostics, enabling the measurement of shaft grounding currents and voltages, the identification of their sources, and the projection of potential damage risks. Additionally, it establishes early warning indicators for measured currents and voltages.

In [1], researchers successfully developed a shaft voltage and current monitoring system through IRIS Power, similar to the SCM technology. This system measures and monitors shaft currents and voltages along turbine generator shafts, providing crucial insights into the machine’s operational health. The configuration allows for real-time tracking of electrical anomalies, such as circulating currents and shaft voltages, which are often precursors to mechanical or electrical failures.

In addition to these advanced online shaft condition monitoring tools, ABB has developed the RARIC Shaft Current Protection Relay, a specialized device designed to protect turbo and hydro-generator bearings from electrical discharge damage caused by shaft currents [43,46]. The RARIC Shaft Current Protection Relay features a ring-shaped current transformer mounted around the shaft, typically on the turbine side [43,46]. This setup enables precise detection and monitoring of shaft currents, allowing for early fault identification and effective protection against electrical discharge damage. By continuously monitoring shaft currents, the relay detects abnormal electrical activity that could indicate insulation breakdown or ineffective grounding [43,46]. If excessive shaft currents are detected, the relay can trigger alarms or automatic shutdowns, preventing catastrophic bearing failures and improving overall generator reliability.

Rogowski coil sensors [57,58] are widely used in the condition monitoring of rotating machinery, including bearing fault detection in turbo and hydro-generators. These sensors provide a non-intrusive, high-sensitivity solution for measuring shaft currents and detecting electrical discharge activity that can lead to bearing damage [57]. The authors of [58] established that Rogowski coil-based online shaft current monitoring is an accurate, non-intrusive, and effective method for fault detection in induction machines. By addressing key failure mechanisms and providing predictive guidelines, this research enhances condition monitoring capabilities for large rotating machines. The findings have significant implications for power plants and industrial applications, where shaft current-related failures can lead to costly downtime and equipment damage.

The study in [1] emphasized that the most effective diagnostic approach involves integrating results from multiple online monitoring tools to form a comprehensive view of the machine’s condition rather than relying on a single technique [3]. By correlating data from various sources, such as vibration analysis and shaft voltage measurements, potential issues can be detected and diagnosed at an early stage, significantly reducing the risk of unexpected failures. Data collected from these monitoring systems are typically analyzed using advanced signal processing techniques, such as Fast Fourier Transform (FFT), to extract characteristic frequencies, amplitudes, and phase information [1,7]. This analysis helps identify specific fault signatures, such as rotor grounding issues, insulation degradation, or bearing electrical erosion [8]. The integration of diverse diagnostic tools not only enhances fault detection accuracy but also provides a deeper understanding of the underlying causes of abnormalities, enabling targeted maintenance interventions and extending the lifespan of critical assets.

5.4. Diagnostic Signal Processing Correlation

Reliable fault diagnostics in large synchronous generators increasingly depend on advanced signal processing of shaft voltage and bearing current data. Studies have demonstrated that specific faults, such as magnetic asymmetry, eccentricity, bearing damage, or excitation system distortion, produce characteristic features in the time and frequency domains [4,12]. Techniques like Fast Fourier Transform (FFT), wavelet analysis, and time-domain evaluation are widely used to identify these patterns.

Several studies have demonstrated that distinct fault mechanisms exhibit characteristic spectral or time-domain signatures [8,61,66,70,71]. For instance, faults caused by magnetic asymmetry or static eccentricity typically manifest as increased amplitudes in the 3rd, 5th, or 7th harmonics of the shaft voltage spectrum, which correspond to odd multiples of the fundamental excitation frequency [8,64,69].

Transient and non-stationary events, such as electrostatic discharges (ESD) or electric discharge machining (EDM), often produce broadband high-frequency bursts or spike trains in time-domain signals. These are effectively captured by wavelet transforms, which provide localized time-frequency resolution.

Homopolar shaft voltage characterized by a DC offset superimposed with ripple voltage is typically attributed to unbalanced gate triggering in thyristor-based exciters or remanent magnetization in the rotor [10,64]. This produces a sawtooth waveform, which is easily identified in time-domain plots and further analyzed using FFT for its ripple frequency content, often seen at 150–300 Hz [4,12].

Faults in the field winding, such as interturn short circuits, often result in increased neutral current amplitude and the presence of even or uncharacteristic harmonics in the shaft voltage signal [70]. Similarly, bearing faults such as inner or outer race defects can modulate the shaft current or vibration spectrum with sidebands around the machine’s mechanical resonance frequencies [64,66]. To support condition-based diagnostics,

Table 8. Correlation between diagnostic signal features and shaft-related faults in large generators.

Signal Feature	Analysis Technique	Associated Fault Type	Fault Source/Mechanism	Characteristic Frequency/Domain	References
3rd, 5th, 7th harmonics	FFT	Static eccentricity	Magnetic asymmetry, rotor-stator misalignment	Odd multiples of 50/60 Hz	[8,72]
High-frequency spikes/bursts	Wavelet/Time domain	Electric Discharge Machining (EDM)	Oil film breakdown, excitation ripple discharge	>1 kHz transients	[1,10,11,12,43]
Sawtooth waveform	Time-domain/FFT	Homopolar voltage	Thyristor ripple, rotor magnetization	Fundamental + 150–300 Hz ripple	[4,12]
Random spike trains	Wavelet/Time domain	Electrostatic discharge (ESD)	Shaft charging/discharging events	Irregular transient bursts	[4,12,25,28]
Elevated neutral current amplitude	Current signature	Field winding asymmetry	Interturn shorts, excitation imbalance	Imbalance at stator frequency	[8,70]
Even-order harmonics (2nd, 4th)	FFT	Rotor winding faults	Asymmetric EMF from faulted coils	100 Hz, 200 Hz in 50 Hz systems	[10,12]
Vibration sidebands	FFT/STFT	Bearing defects (outer/inner)	Mechanical resonance modulation from damaged elements	Mechanical modulating frequency	[64,66]
DC offset with ripple	Oscilloscope + FFT	Homopolar shaft voltage	Static excitation system asymmetry, field imbalance	DC + ripple voltage	[4,10,12,34]

summarizes key correlations between observed signal features, analysis methods, and associated fault types based on the literature.

5.5. Health Assessment Tools for Turbine Generators

In recent years, shaft-induced voltages and bearing currents have gained significant attention as powerful diagnostic tools for detecting faults in turbine generators [64,71]. These electrical signals, traditionally considered operational challenges, are now leveraged for predictive maintenance, enabling early detection of rotor eccentricity, insulation degradation, bearing wear, and electromagnetic imbalances [69,71]. Many fault detection systems for electrical rotating machines rely on advanced signal analysis techniques, particularly in the frequency domain, to extract meaningful diagnostic information [69,71]. By analyzing the harmonic components of measured shaft voltage and current signals, these systems can identify subtle signs of mechanical misalignment, insulation breakdown, or electrical discharge activity before they escalate into severe failures [71,72].

A key technique for transforming time-domain shaft voltage signals into the frequency domain is the Fast Fourier Transform (FFT) [69,71]. This method decomposes complex waveforms into their constituent frequency components, allowing the detection of anomalies in harmonic spectra that correlate with specific failure modes. For example, an increase in the nth harmonic amplitude can indicate an emerging fault in the generator, such as bearing degradation, rotor imbalance, or stator winding faults [8,71]. By continuously monitoring these harmonics, proactive maintenance strategies can be implemented to prevent costly failures and downtime [1,7].

Beyond harmonic analysis, comprehensive generator condition monitoring systems integrate shaft voltage measurements with rotor and stator flux analysis to gain a holistic understanding of machine health. This multi-faceted approach enables precise diagnosis of rotor eccentricity, inter-turn short circuits, and bearing discharge phenomena, significantly improving reliability and operational efficiency [19,69].

A signal processing technique is employed to convert the raw shaft voltage signal into its frequency domain representation using Fast Fourier Transform (FFT). This transformation highlights dominant frequencies, enabling effective machine health assessment, as shown in Figure 17 [1,71]. The analysis of harmonic components and their amplitudes provides critical diagnostic insights and supports long-term condition monitoring, with results presented both numerically and graphically [69,71]. For example, the detection of even harmonics (e.g., second, fourth, and sixth) often indicates the presence of shorted turns in the rotor winding [61,69,70,72].

The authors of [71] analyzed shaft voltage signals as a valuable tool for diagnosing defects, such as eccentricities and inter-turn short circuits, in field windings of large turbogenerators. Their study focused on a four-pole, non-salient pole-synchronous generator, aiming to accurately determine the effects of these defects on shaft voltage. The analysis was conducted incrementally, particularly examining the impact of parallel coupling of the armature windings. An FFT study of the resultant signal identified specific harmonics that serve as signatures for different eccentricity cases. The study demonstrated that static eccentricity induces a 50 Hz shaft voltage, with mitigation effects leading to a secondary 250 Hz component. The findings illustrate how various configurations and defects influence shaft voltage and its harmonic content.

The authors of [70] also concluded that analyzing shaft voltages and neutral voltages can effectively reveal failures in synchronous generators. Additionally, they emphasized the need to investigate the impact of network harmonics on shaft voltages to better understand their influence on generator performance.

Reference [72] explored the relationship between shaft voltages and static eccentricities in generators. Theoretical analysis indicates that shaft voltages are induced by distorted fluxes resulting from static eccentricities, predominantly containing odd harmonics, with amplitudes increasing under greater eccentricities. Additional distorted flux harmonics emerge if field windings experience inter-turn short circuits, with harmonic frequencies varying based on the generator’s pole-pair number and amplitudes increasing as faults worsen. Experimental results align with these theoretical predictions, confirming that shaft voltages arise from distorted fluxes due to static eccentricities, with harmonic content directly correlated to fault severity.

Reference [7] highlighted that various frequency components in shaft voltage signals originate from multiple factors, including design features (both stator and rotor), construction details, and generator asymmetries. The study emphasized that these frequency components serve as critical indicators of potential issues that could develop over time. By analyzing these components, early warnings of developing faults in rotating machinery can be identified, making them essential for proactive maintenance and fault diagnosis. Figure 18 illustrates a typical process flowchart of a generator fault diagnosis mechanism that employs signal processing techniques.

Shaft voltage signals in the time domain are highly complex and fluctuate significantly, making it difficult to isolate individual components. Consequently, time-domain analysis alone is insufficient for identifying fault sources. Instead, spectral analysis is employed to detect characteristic frequency components associated with specific faults [61,70,72]. Recent advancements, including machine learning techniques such as Bayesian estimation, have enhanced fault detection, particularly for rotor eccentricity [68]. However, the success of these methods depends on high-quality training data, which can be challenging to obtain in real-world operational environments.

To improve the diagnosis of shaft voltage issues, recent advancements in signal processing and machine learning have been leveraged. Notably, the authors of [68] developed a machine learning-based approach utilizing a Bayesian estimation algorithm for enhanced shaft voltage monitoring. This technique has demonstrated strong potential in detecting rotor eccentricity faults, even in complex systems. However, its effectiveness heavily relies on the availability of high-quality training data. In practice, obtaining such accurate datasets remains a significant challenge, which may limit the method’s applicability in certain operational environments.

6. Reliability and Maintenance Implications of Shaft Voltage Mitigation

While numerous shaft voltage and bearing current mitigation strategies have been discussed, their practical impact is best demonstrated through reliability metrics such as mean time between failures (MTBF), downtime reduction, and component lifespan extension. Field studies and fleet analyses provide quantifiable evidence that supports the adoption of shaft grounding, insulated bearings, and real-time condition monitoring technologies.

For instance, in a multi-unit power plant deployment, the integration of shaft grounding brushes and shaft condition monitoring increased the MTBF of generators from 5.8 to 8.1 years, corresponding to a 39.6% reliability improvement and 32% reduction in unplanned maintenance events [1,7,60]. Similarly, a grounding system retrofit at a 350 MW steam turbine plant using silver-braid brushes more than doubled bearing life from 3.5 to over 7 years while reducing annual maintenance hours by 45% [1]. Across a fleet of 195 hydro-generators, enhanced insulation systems coupled with targeted mitigation efforts reduced bearing-related failures by 56% over a 15-year period [5,19].

These improvements are not only attributed to reduced electrical damage but also to improved detection and preventive maintenance scheduling, made possible by digital shaft monitoring systems. The application of shaft voltage suppression not only preserves bearing integrity but also translates into tangible operational gains for large rotating machinery. A summary of quantified reliability improvements from shaft voltage mitigation strategies is shown in

Table 9. Quantified reliability improvements from shaft voltage mitigation strategies.

Case Study/System	Mitigation Strategy	Reliability Metric	Improvement Observed	References
Multi-unit thermal plant (IRIS Power study)	Shaft condition monitoring (SCM) and grounding brushes	MTBF	Increased from 5.8 to 8.1 years (+39.6%)	[1]
350 MW steam turbine generator	Silver-braid grounding retrofit	Bearing life expectancy	Increased from 3.5 to >7 years	[4]
Hydro-generator fleet (195 units)	Improved shaft insulation systems (Type 2)	Bearing insulation failure rate	Reduced by 56% over 15 years	[5]
Lab-scale 10 HP system	Conductive microfiber ring	Shaft voltage and bearing wear	Voltage < 2 V; bearing life doubled	[38]
Thermal generators (ABB/IRIS systems)	RARIC relay + shaft monitoring	MTBF and catastrophic failures	MTBF +30%; failures reduced by 50%	[43]

.

7. Economic Impact of Shaft Voltage and Bearing Current Mitigation

Shaft voltage and bearing current-related failures present not only technical but also significant economic risks to large-scale power generation assets. The costs associated with these phenomena include premature bearing replacement, hydrogen seal degradation, lubrication failure, stator core insulation damage, and even catastrophic generator outages. Downtime for large generators, particularly those rated above 300 MW, can result in daily losses exceeding R1.5 million, depending on system criticality and market structure [4,28].

Field-validated case studies have demonstrated that mitigation strategies, such as shaft grounding ring installations, silver-braid and gold-bristle brush retrofits, and continuous condition monitoring, can yield substantial lifecycle cost savings [1,4,7,28,38,40]. These findings reinforce the economic justification for proactive shaft voltage suppression and diagnostics. They highlight a compelling return on investment (ROI) for utilities, especially within aging fleets facing increased exposure to electrical stressors from inverter-fed excitation and power electronics. Below is a summary, depicted in

Table 10. Economic impact of shaft voltage-related faults and mitigation strategies.

Fault Type/Cause	Typical Economic Impact (ZAR)	Mitigation Strategy	Economic Benefit/Cost Saving (ZAR)	References
Electrical Discharge Machining (EDM) of bearings	R555,000–R1,480,000 per unit failure (including downtime + replacements)	Shaft grounding rings/brushes	Up to R675,000 annual maintenance reduction per unit	[38,40,53]
Seal oil degradation due to shaft current	R185,000–R925,000 per event (seal repair + H2 loss + downtime)	Insulated bearings + microfiber grounding rings	Reduced H2 top-up and repair frequency; improved safety	[5,53]
Generator unplanned outage	R1.85 million–R9.25 million/day for 100–500 MW unit	Online monitoring + suppression systems	Increased MTBF (from 5.8 to 8.1 years); ROI within 1–2 outages	[1,7]
Shaft magnetization issues	R92,500–R277,500 (flange repair + rotor demagnetization and rebalancing)	Demagnetization + shaft voltage trending	Avoids rotor dismantling; improves predictive maintenance scheduling	[30,60]
VFD/Static Exciter transient discharge	>R462,500 per failure (excitation system + sensor damage + unscheduled repair)	High-frequency filtering + gold-bristle brushes	Significant reduction in equipment burnout; improved system reliability	[10,11,50]
Fleet-level degradation over time	Multimillion-Rand bearing and stator core damage across large fleets (e.g., Eskom or SAPP fleets)	Hybrid grounding + insulation systems	R54 million in bearing damage avoided over 15 years in 195-unit hydro fleet	[5,9]

, presenting the economic impact of shaft voltage and bearing current issues, organized by fault type, associated costs, and potential savings through mitigation strategies.

8. Conclusions

Shaft voltages and bearing currents in large turbogenerators present complex, multi-physics challenges that can severely compromise machine reliability, safety, and operational lifespan. This review has categorized the primary sources of shaft voltage into four mechanisms, electrostatic charging, magnetic asymmetry, static excitation ripple, and residual shaft magnetization, each with distinct waveform characteristics, impedance behaviors, and degradation pathways. These sources are not mutually exclusive and often coexist, compounding damage risk.

Advanced diagnostic techniques, particularly FFT, wavelet analysis, and time-domain signal correlation, have demonstrated effectiveness in identifying specific fault types based on waveform signatures. For example, sawtooth waveforms correlate with electrostatic discharges, while high-frequency ripples and even-order harmonics point to excitation system anomalies and rotor winding faults.

A critical review of grounding and suppression technologies revealed that, while carbon and silver-graphite brushes are commonly used, their effectiveness is often undermined by wear and environmental contamination. Emerging alternatives, such as gold-bristle brushes and conductive microfiber grounding rings, offer superior performance across broader frequency ranges and operational conditions. Case studies from industrial installations and fleet assessments validated the practical effectiveness of these strategies, showing quantifiable improvements in mean time between failures (MTBF), bearing lifespan, and maintenance cost reduction.

Furthermore, AI-based diagnostic models (e.g., CNNs, GMMs) have shown high classification accuracy (>96%) when trained on well-characterized vibration or shaft voltage datasets. However, successful implementation requires attention to signal type, sampling resolution, feature selection, and validation procedures.

While this review offers a comprehensive overview, several limitations and future directions remain. Firstly, despite the range of grounding systems and diagnostic tools assessed, standardized benchmarking across machine types and environments is still underdeveloped. The long-term performance of suppression technologies under contamination, mechanical stress, and thermal cycling lacks consistent empirical validation. Similarly, AI-based fault detection models, while promising, are largely confined to laboratory conditions and require further adaptation for real-time application in diverse generator architectures. Additionally, the absence of universally accepted shaft voltage thresholds complicates the standardization of diagnostics and risk classification. There is a need to extend shaft voltage research to cover renewable and inverter-fed generators, which present unique excitation profiles not fully captured in conventional studies. Finally, improved alignment between waveform diagnostics and IEC/IEEE standards, particularly IEC 60034-27-3 and IEEE Std 80, is essential to ensure consistent interpretation, testing, and implementation of mitigation strategies. Overall, this review bridges theoretical modelling, signal diagnostics, and practical field mitigation, offering both a knowledge base and a forward-looking perspective for advancing generator shaft voltage management. Future research directions should focus on:

• Real-time waveform classification and suppression feedback;
• Scalable implementation of AI-based diagnostics;
• Long-term field validation of grounding materials under harsh environmental conditions.

This review consolidates theoretical insights, diagnostic methodologies, and industrial case studies to serve as a comprehensive resource for both researchers and industry practitioners addressing shaft voltage phenomena in high-performance rotating machines.