A Review of the Conventional and Intelligent Multifunctions Electrical Generator Protection System, Challenges and Opportunities

Power systems are exposed to faults at many different time intervals, attributed to numerous factors; thus, it is imperative that these faults are anticipated and that appropriate safety measures are implemented within the power system [16].

The selection of protection devices is contingent upon the specifications and classification of the generators and associated equipment. The protection of generators consists of multiple components designed to detect abnormal operating conditions or lead to the instability of power systems [17], enabling the disconnection of the generator and mechanical systems before potential damage occurs. Two key protective elements related to synchronizing issues are reverse power protection (when the generator operates as an electric motor) and loss-of-excitation protection (failure of the excitation system) [18, 19]. The electrical generator protection systems are shown in Figure 1. In addition to the aforementioned relays, differential protection, field failure protection [20], protection for rotor ground, protection for reverse power, and protection for negative sequence phase are also employed for generator protection. Vibration signals can also be used to protect power system devices such as generators, transformers, etc. [21]. Different faults and their corresponding protective devices are discussed in the following sections.

01.png

Figure 1. Types of various electrical generator malfunctions and unusual operation circumstances [22-24]

Generator protection systems constitute assemblages that incorporate an array of protective measures for the generator. These protective measures are fundamentally predicated upon three principal factors: variations in load current magnitude, stator thermal conditions and impedance, and rotor rotational velocity [32].

A multitude of protection systems exist within generators, with the most prevalent among them being.

4.1 Overvoltage & undervoltage protection (Abnormal voltage protection)

In the absence of voltage rise, drop, or fluctuation from the rated value [27], this protection does not activate. It operates in two main scenarios: overvoltage conditions and undervoltage conditions [28]. Undervoltage conditions do not cause any negative impact or physical damage; however, there is a potential risk of overheating auxiliary motors. The importance of the overvoltage relay lies in preventing excessive dielectric stress, which could lead to insulation failure [29].

According to IEEE standards, voltage fluctuation ranges are classified, and acceptable limits are defined. The IEEE 1159-2019 standard categorizes voltage fluctuations into different classes, while IEEE 519-2022 provides guidelines on permissible limits. As per the IEEE C84.1-2020 standard, the utilization voltage (for end-user equipment) should remain within ±5% of the nominal voltage, while the service voltage (at the utility supply point) should stay within +5% to -2.5%.

Exceeding these limits can lead to significant damage, such as generator failure, reduced efficiency, increased electrical and mechanical faults, and decreased performance and equipment lifespan. Therefore, it is essential to adhere to these standards and take the necessary measures to ensure voltage stability and maintain equipment efficiency [30].

4.2 Earth fault and restricted earth fault protection (REF)

A restricted earth fault (REF) protection mechanism facilitates the precise identification of ground faults occurring in close proximity to the neutral point of a transformer or generator. The advantages derived from the sensitivity of the REF mechanism become particularly significant when the apparatus is grounded with low impedance, rendering the transformer phase differential element (87T) ineffective [31], Stator Earth Fault protection calculates the current passing through the neutral point of the generator or transformer without being restricted by any other value [32].

4.3 Overload & overcurrent protection

This protection detects any fault in the three phases of the generator. Any drop in the voltage of one of the phases may cause contact, even partial, with the ground or between the phases, which will lead to an increase in the main current of the generator due to the resulting short circuits. The relay begins to sense the current value once the voltage drops by a certain percentage from the rated value, as this protection depends on voltage. There are three options in this type, as follows.

1. Simple overcurrent mode.
2. Voltage control overcurrent mode.
3. Voltage restrained overcurrent mode.

In the first case, there is no need to adhere to the voltage, as this relay operates to detect an increase in current at the rated voltage, with a rise higher than the rated load current. The second case is used when the generating unit is directly connected to the grid without being constrained by the grid voltage. The third case, which is the one adopted in power generation stations, involves connecting generation stations indirectly through power transformers constrained by the grid voltage. Figure 3 illustrates the difference between the last two cases. The operation of this relay depends on time, with two types either fixed time (Definite Time) or inverse time, which depends on the curve of the relationship between current and voltage [33].

03.png

Figure 3. Overcurrent protection voltage control/voltage restrained

The inverse-time overcurrent relay Figure 4 is designed to isolate faults characterized by elevated current levels quickly.

The speed of fault clearance depends on the intensity of the

Various inverse-time relays present distinct time-current characteristics (TCCs), facilitating the selection of an appropriate relay tailored to the application's specific requirements. A selection of available options is enumerated below [34].

The defining equations pertinent to the previously mentioned categories of inverse-time overcurrent relays are articulated as follows [35].

04.png

Figure 4. Illustrates an increase in fault current, which results in a faster relay response

Standard inverse overcurrent relays:

$t_{o p}=\frac{0.14}{M^{0.02}-1} \times T M$            (1)

Very inverse overcurrent relays

$t_{o p}=\frac{13.5}{M-1} \times T M$           (2)

Extremely inverse overcurrent relays:

$t_{o p}=\frac{80}{M^2-1} \times T M$           (3)

where, t_op: operating time (s), TM: ratio of the input current to the pickup current, M: time multiplier or time dial setting thermal.

4.4 Over frequency & under frequency

The operational frequency of the system is maintained at 50 Hz during standard functioning. The frequency of the system is contingent upon the rotational velocity of the generator.

$n_s=\frac{120 \times f}{p}$           (4)

where, NS: synchronous speed [rpm], P: number of magnetic poles.

The difference between active power production and system requirements leads to an increase or decrease in the generator's rotational speed, resulting in the acceleration of the generator rotors and an overall increase in system frequency. The opposite occurs due to the sudden loss of significant generation capacity. Operating at an irregular frequency is inherently harmful, as it can negatively impact many aspects of the power system, jeopardizing the reliability and security of the entire generation facility [36]. When the frequency is irregular, it causes damage to electrical equipment. For example, the natural frequency (and its harmonics) of turbine blades is designed to be sufficiently separated from the nominal speed and its multiples to prevent mechanical resonance. If the turbine operates at speeds that differ from the specified parameters, the turbine blades may experience excessive mechanical stress, which could lead to the formation of cracks in the turbine structure [37].

4.5 Loss of excitation protection methods

There are two main concerns regarding Loss of Excitation (LOE) protection [38]:

1. Making sure that in the event of a loss-of-field condition, the relay will trip the generator fast enough to prevent damage to the equipment or detrimental effects on the system.
2. Making sure that Special Protection Systems (SPS) or temporary circumstances that don't harm the equipment won't cause the relay to trip the generator unnecessarily.

The most popular techniques for (LOE) protection are:

1. Mason

A single-phase offset mho relay product by Mason in 1949 [39], The distance relay was developed to avoid incorrect decisions between LOE protection and other abnormal operating conditions. The characteristics of the relay mho (diameter X_d and offset Xʹ_d).

2. Berdy

In 1975, Berdy [40] made a change to the protection system. (The first diameter is equal to 1.0 p.u without an external time delay, and the second unit has a diameter equal to X_d) This is to ride through transient conditions and undesirable operations [40].

Currently, Mason’s scheme is recommended for generators with X_d˂ 1.2 p.u., as well as Berdy’s scheme for X_d˃ 1.2 p.u. Figure 5 illustrates Berdy’ and Mason's methods.

05.png

Figure 5. The proposed protection characteristics (LOE) by: (a) Mason and (b) Berdy

3. Positive offset

Undervoltage, impedance, timers, and directional elements are considered components of this technique for protect generator and transmission networks [41, 42], as shown in Figure 6. Using the following formulas, the first relay (Z2), the positive offset mho element, is configured with a 10% margin above the steady-state stability limit.

Diameter = 1.1X_d + X_s          (5)

Offset = X_s                (6)

06.png

Figure 6. Positive offset scheme

4.6 Generator watt/var capability

Figure 6 shows a typical capability curve for a generator with a cylindrical rotor. The generator's operating limits under steady state (continuous) conditions are shown by this capacity curve. The generator capability curve is usually distributed at the rated voltage of the generator. In the under-excited area, salient pole generators show a slightly different feature.

The curve also shows how the Automatic Voltage Regulator (AVR) limits steady-state operation to stay within the generator's limits. Three separate curves the stator winding limit, the rotor heating limit, and the stator end iron limit combine to form the generator capability shown in Figure 7 [28, 43].

07.png

Figure 7. A cylindrical rotor generator's typical generating capability curve and operating restrictions

4.7 Over flux protection

This suggests that the per unit voltage is divided by the per unit frequency and can be attributed to various factors, including turbine overspeed, load rejection on the machine resulting from the disconnection of a load block, The malfunctioning of the Automatic Voltage Regulator (AVR) during the operational state of the machine, over-excitation during the machine's unloaded condition, situations where the machine operates at a terminal voltage that exceeds its rated capacity, and instances where the machine functions at its rated or lower voltage in conjunction with reduced frequency can lead to operational issues [44].

This phenomenon can be mitigated by implementing a relay that measures the ratio of voltage to frequency (V/Hz relay), adjusting the speed accordingly so that the generated voltage is proportionately regulated [22, 26].

The diverse modalities of safeguarding implemented for the generator can be described into two classifications:

1. Protective relays are designed to identify anomalies manifesting outside the generator.
2. Protection relays are established to recognize faults arising within the generator.

4.8 Other protective devices

Protective relays play a vital role in protecting generators from faults or abnormal operating conditions that could cause significant damage to the electrical generator, whether these faults originate internally or externally to the generator.

In addition to the power transformers and protection relays that are directly connected to the generator. Other devices include lightning arrestors, overspeed protection devices, instruments for measuring the temperature and vibration of shaft bearings, stator windings, transformer windings, transformer oil, and others, as well as monitoring lubrication systems such as oil flow, oil pressure, and oil temperature. Some of these protective arrangements are classified as non-trip types (Non-Trip Types) (provided that these protective arrangements do not cause damage to the electrical generator), meaning they are limited to issuing alarms in response to abnormal conditions. Table 1 provides a detailed enumeration of the additional protective relays used for safeguarding generators [33, 34].

Table 1. Generator protection elements

Phadke et al. [55] discussed the significant role of wide-area monitoring (WAM) in enhancing power system protection, particularly in the context of modern challenges faced by power systems.

It highlights the limitations of existing protection systems, which often rely on fixed relay characteristics that do not adapt to varying system conditions.

The authors emphasize the need for adaptive and intelligent protection schemes, as well as the importance of a robust communication infrastructure to support these advancements.

The paper aims to explore new protection concepts that can mitigate the risks of blackouts caused by relay maloperations and wide area disturbances [55].

Hasani et al. [56] discussed the investigation of the impact of electrode shapes on the performance of gas discharge arresters (GDAs), which are crucial for overvoltage protection in low-voltage circuits.

GDAs function as insulators under normal conditions but can discharge when voltage exceeds the gas breakdown threshold, leading to electric arcs.

The study emphasizes the significance of electrode shape in preventing the reignition of arcs after extinguishing, as the geometry of the electrodes influences the electric field strength.

The authors utilize a differential evolution optimization algorithm to compute optimal electrode shapes for improved electric-field distribution [56].

Dumitrescu [57] discussed the design and modelling of an intelligent multifunctional protection system for an electric generator. It proposes a fuzzy safety analysis model to evaluate fail-safe behaviour in automation and protection systems.

The electric generator's protection and automation system (PAS) is crucial for detecting and isolating faults to maintain the system's operational state. By utilizing fuzzy logic and event-tree analysis, the PAS can effectively respond to various dangerous events with imprecision and uncertainty in data. This innovative approach allows for qualitative evaluation of events using verbal statements for probabilities and consequences.

The work mentions using fuzzy logic to account for imprecision and uncertainty in data while employing event-tree analysis, indicating a limitation in traditional deterministic approaches that do not consider such uncertainties.

The very high, low, and moderate probabilities are verbal statements for probabilities and consequences permitted by fuzzy event-tree logic, highlighting a limitation in traditional numerical probability assessments that may not capture qualitative aspects effectively [57].

Taalab et al. [58] discussed and address the critical issue of stator winding faults in synchronous generators, which can lead to significant damage and maintenance costs.

It highlights the limitations of conventional differential relays in detecting ground faults, particularly under high-impedance grounding conditions.

The authors suggest an internal fault detector algorithm based on artificial neural networks (ANNs) that makes use of the average and difference between the currents flowing into and out of the generator windings.

This proposed method demonstrates high sensitivity and stability in fault discrimination, surpassing traditional detection methods [58].

Cui et al. [59] proposed a multifunction intelligent relay (IR) scheme that includes fault detection, fault ride through (FRT) selective blocking features, islanding detection, and fault type recognition for inverter-based distributed generators; this system presents a thorough IR modelling approach to lower computational resources and allows the integration of these functionalities. The approach outperforms standard detection time in a hardware-in-the-loop (HIL) context.

This innovative protection system addresses protection functions in terms of dependability, security, and the challenges posed by inverter-based distributed generators, enhancing overall grid reliability and performance.

The work highlights the challenges introduced by inverters in distributed generation systems, emphasizing the need for a sophisticated interconnection protection relay to address these challenges.

It mentions the importance of reducing offline computational resource consumption while integrating multiple functions in the same intelligent relay, indicating a limitation in traditional protection schemes [59].

Abdi-Khorsand and Vittal [60] proposed a new technique for identifying possibly malfunctioning relays during the planning stage with the goal of preventing unintended system separation and preventing these relays from tripping during erratic power swings. By lowering voltage and frequency fluctuations during disruptions like the triple line outage examined in the Western Electricity Coordinating Council data, the proper design and installation of out-of-step protective relays, identified by the suggested method, enhance the power system's dynamic performance.

The work does not explicitly mention any limitations of the proposed method for detecting mis-operating relays in time domain simulations.

It focuses more on the benefits and results of the novel method in improving the dynamic performance of power systems rather than discussing any potential drawbacks or constraints [60].

Li et al. [61] developed an efficient method for diagnosing faults in the elevator of fixed-wing unmanned aerial vehicles (UAVs) using deep learning techniques to improve fault detection accuracy and enhance UAV system reliability. The research relied on designing a model based on the Transformer architecture to analyze sensor data and detect faults in the UAV elevator. Deep learning techniques, particularly the Transformer architecture, were used to process time-series data and classify faults with high accuracy. The model demonstrated superior performance in diagnosing faults, outperforming traditional methods in detecting complex failures. However, some limitations exist, including the need for a large amount of training data and the difficulty of generalizing the model to other types of UAVs [61].

Hou et al. developed an effective method for diagnosing rolling bearing faults using an enhanced model based on the Transformer architecture, aimed at improving the accuracy and efficiency of diagnosis in industrial systems. A new model called "Diagnosisformer" was proposed, which relies on optimizing the Transformer architecture to identify faults in rolling bearings. The model was trained on a dataset containing vibrational signals collected from various bearings. Deep learning techniques were employed, with the Transformer architecture optimized for processing time-series signals, along with automatic feature extraction techniques.

The proposed model demonstrated high accuracy in fault diagnosis compared to traditional methods, with enhanced capability to handle large and complex datasets. However, among the limitations are the need for a large amount of data for training and the difficulty in interpreting the decisions made by the model [62].

Bashirov et al. [63] address the creation and investigation of an intelligent electrical network protection and control system, with an emphasis on relay protection in an active-adaptive network. To improve the protection system's capabilities, it incorporates supervised machine learning, artificial neural networks, and digital twins; while the focus is on relay protection within the broader electrical network, the methodologies and technologies discussed could potentially be adapted for designing and modelling an intelligent multifunction protection system for electrical generators, leveraging advanced technologies like neural networks for enhanced functionality and reliability.

Research was conducted to establish the limits on the application of artificial intelligence in an intelligent security system for electrical network relay protection.

In order to answer queries about the application of a neural network in relay protection, the authors examined the complex by adding digital twins, physical models of the power supply system, and an artificial neural network [63].

In general, many software programs have been used by researchers for protection systems, including MATLAB and LabVIEW [64].

Table 2 offers a sample of protection elements used in electrical generators, methods used, the circuit outcomes, conclusions, and features. This work presents a sum of works that used generator protection relays in power systems.

Table 2. Samples of protection elements, methods used, and circuit outcomes

The research studies mentioned in Table 2 [66-89] contain various constraints; however, they all share the limitation of not incorporating current, voltage, and temperature together to cover all protection aspects. These parameters have been integrated in this work, by the suggestion of a new intelligent multifunction protection relay system.

The main source of electrical power used in business settings is synchronous generators. Significant equipment damage, interruptions in electricity distribution, and financial consequences result from SG failure. Effective generator protection requires knowing the kind of fault, whether it is caused by rotor failures, stator problems, or unusual generator operating circumstances, and then implementing the proper mitigation techniques.

Many studies on generator protection have shown that implementing performance optimization methodologies within the system is crucial for maintenance cost reduction in addition to increasing the basics of protection systems targets such as speed, selectivity, reliability, and sensitivity. These studies' objectives and positive points were clarified, and their negative aspects were presented to identify and address them in the future.

All previous works did not include the incorporation of the capability curve and its constraints for the Intelligent relay or the effect of temperature on it, which can be added as an input-output training dataset.

08.png

Figure 8. A proposed model for the smart relay, including ANSI code numbers

With the integration of artificial intelligence (AI) into practical applications and the significant advancements in finding optimal solutions, a smart relay specifically designed for generating units can be developed. This smart relay would encompass all the protections mentioned in the paragraphs above, incorporating all constraints and functioning both during faults and in issuing warning signals. Furthermore, it would ensure that operation continues for non-critical faults that do not necessitate the shutdown of the generating unit. Figure 8 illustrates a new structure for the proposed smart relay model, including changing the capability curve with air temperature in addition to the other signal clarified as ANSI code number, which is explained in Table 4.

When the temperature increases, the efficiency of electric generators generally decreases, and the value of the stator and rotor winding limit on the capability curve, in addition to the machine stability margin, also decreases. This case can be a proposal for future research.

Table 4. ANSI code number with the equivalent protection type