Failure Mode, Effects, and Criticality Analysis (FMECA)-Based Fault Diagnosis of a High-Voltage Disconnect Switch

Abstract

High-voltage disconnect switches (HVDSs) are crucial for ensuring visible electrical isolation but often operate under challenging environmental conditions. While current monitoring methods can detect faults, simply identifying them is not enough for effective maintenance planning or efficient resource management. Electric utilities need tools or platforms that enable them to identify various failure modes and their impact on the safety and reliability of the system. This paper presents a failure mode and effects analysis of a HVDS and an assessment of its criticality (FMECA). For each failure mode identified by the FMECA, a cause-and-effect analysis is used to further investigate their root causes. Next, the criticality of each failure mode is evaluated using the Risk Priority Number (RPN). Finally, each failure mode is subjected to an analysis to prioritize the sensor used to measure the physical quantity associated with the failure. This sensor prioritization is performed by considering not only the RPN but also the sensor’s utility in terms of failure detection efficiency and ease of installation. The results provide a robust foundation for the development of predictive maintenance strategies customized for HVDS equipment, thereby contributing to enhanced reliability and resilience of electrical power systems.

1. Introduction

The reliability of electrical infrastructure is defined as the probability that an electrical component or system performs its intended function over a given period and is of critical importance for ensuring secure energy supply and grid stability [1]. Recent research has increasingly focused on reliability, safety, and energy management, whether for evaluating energy potential [2], optimizing operational models [3], or managing battery life cycles [4]. Ensuring the dependable performance of power system components is now central to enabling efficient integration of renewable resources, extending the lifespan of electrical assets, and supporting predictive maintenance strategies [1]. This growing emphasis on reliability reflects the need to address both technical and operational challenges associated with modern, decentralized power systems. Within power systems, high-voltage disconnect switches (HVDS) play a crucial role by providing visible electrical isolation, thereby reducing risks to personnel and equipment. Despite the apparent simplicity of their function, HVDSs are required to operate reliably under harsh conditions, including extreme temperatures, high humidity, mechanical stress, and environmental exposure. Ensuring their long-term effectiveness therefore requires addressing a range of technical and environmental challenges. Several critical failure modes have been identified in the literature [5]. Mechanical failures may result from the wear of components such as seals, pivots, and springs, leading to incomplete operations and abnormal vibrations [6]. Electrical contacts can degrade due to oxidation, contamination, or arcing, causing increased contact resistance and localized overheating [5,7]. Control system malfunctions may arise from failures in motors, brakes, or electronic devices, often due to moisture ingress, component aging, or insufficient environmental protection [8]. Under short-circuit conditions, electrodynamic forces can induce significant mechanical stress on the switchgear structure [9], while switching operations generate electromagnetic disturbances that may affect nearby equipment [10]. These failures create substantial operational challenges, including reduced substation safety, unintended circuit energization, accelerated equipment degradation, and increased maintenance costs [5]. A single observable symptom may originate from multiple fault mechanisms. For instance, abnormal contact heating may be caused by misalignment, insufficient contact pressure, contamination, or material degradation [5,7]. Similarly, incomplete switching operations can result from motor failure, mechanical linkage jamming, ice accumulation, electronic control board malfunctions, or defective limit switches. To mitigate these risks, utilities have implemented various condition monitoring strategies based on operational data [8]. Traditional approaches rely on monitoring motor current, vibration, and position signals during switching operations to assess equipment health. Deviations in motor current profiles can indicate increased mechanical resistance or faults in moving parts and contact interfaces [11]. Vibration analysis provides valuable information on dynamic behavior, impacts, and resonance phenomena associated with mechanical degradation [12], while torque measurements are effective for detecting structural defects in components such as porcelain columns [6]. More recent developments include wireless monitoring systems combined with support vector machines for fault diagnosis [13], multi-domain vibration signal analysis to improve fault discrimination [12], and deep learning-based models for disconnector fault classification [14,15]. Although significant progress has been made in fault detection and classification, the translation of diagnostic results into actionable maintenance decisions remains a major challenge [16]. Fault identification alone does not enable maintenance teams to determine the urgency of intervention or to optimize resource allocation. Utilities require clear guidance on which failure modes pose the greatest risks to system safety and reliability, which preventive actions should be prioritized under budgetary constraints, and, within a predictive maintenance framework, which sensors should be deployed, where they should be installed, and how the collected data should be interpreted in relation to actual equipment degradation. This gap between fault diagnosis and maintenance decision-making is particularly critical given the large number of disconnectors operated by utilities and the persistent limitations on maintenance budgets.

Failure Mode, Effects, and Criticality Analysis (FMECA) provides a structured methodology to address these challenges by systematically identifying potential failure modes, analyzing their causes and consequences, and assessing their criticality [17]. FMECA evaluates failures based on three key criteria: severity of consequences, frequency of occurrence, and detectability. Originally developed by the U.S. Department of Defense in 1949, the methodology gained widespread recognition following its successful application by NASA during the Apollo program in the 1960s. Today, it is extensively applied in sectors such as nuclear power, aerospace, automotive engineering, and electrical energy systems [18]. FMECA is particularly well suited to HVDS due to the strong interdependencies between mechanical, electrical, thermal, and environmental factors.

1.1. Literature Review

Recent research has mainly relied on data-driven approaches using measurements acquired from installed monitoring systems. Zhang et al. (2025) [19] proposed mechanical fault diagnosis framework based on vibration signal analysis using Variational Mode Decomposition (VMD), a black-winged kite optimization algorithm, and a dual-channel deep learning architecture. Their method successfully identifies faults such as mechanism jamming, mechanism looseness, and three-phase asynchrony. Similarly, Zhu et al. (2025) [12] developed a vibration-based diagnosis method capable of distinguishing six operating conditions, including opening and closing jams as well as incomplete operations. These studies demonstrate the effectiveness of advanced signal processing and machine learning techniques for fault classification. However, they focus primarily on predefined mechanical faults and assume that appropriate sensors have already been selected and installed. Other studies have investigated the condition assessment of disconnect switches. Obarcanin et al. (2023) [20] proposed a health-index-based methodology for evaluating the operational condition of disconnectors using operational and environmental data. Likewise, Westerlund et al. (2016) [21] focused on contact deterioration through temperature-current regression models based on infrared temperature measurements. While these approaches provide valuable information regarding equipment condition, they do not quantify the relative criticality of failure modes and therefore do not establish priorities for maintenance actions or monitoring investments. Several authors have proposed sensor-rich monitoring platforms for intelligent substations. For example, Chen et al. (2022) [22] integrated torque sensors, angle encoders, microswitches, temperature sensors, humidity sensors, and altitude sensors to achieve real-time monitoring of disconnect switch operation conditions. Although such systems considerably improve observability, sensor deployment is generally predefined based on engineering experience rather than being derived from a systematic risk analysis. Only a limited number of studies have applied structured reliability-analysis methods. Suwanasri et al. (2021) [23] combined FMECA with Analytical Hierarchy Process (AHP) to assess risks associated with transmission network assets. Their work demonstrated the effectiveness of FMECA for maintenance prioritization. However, the analysis was conducted at the transmission-line level and did not address disconnect switches or the selection of monitoring technologies. Similarly, Qiu et al. (2015) [11] investigated mechanical faults such as loose screws and transmission mechanism jamming using vibration and current measurement, but their work focused on fault recognition rather than systematic failure-mode identification and ranking.

The review reveals a significant research gap. Existing studies predominantly focus on fault diagnosis, condition monitoring, or fault classification after monitoring data become available. Very few investigations address the upstream problem of identifying all relevant failure modes, quantifying their criticality, and determining which sensors should be prioritized to maximize predictive maintenance effectiveness. Furthermore, no comprehensive framework specifically dedicated to high-voltage disconnect switches has been found that explicitly links failure-mode criticality assessment to sensor selection. A comparative summary of the main contributions reported in the literature is presented in

Table 1. Summary of recent studies on monitoring and fault diagnosis.

Articles	Failure Mode Analyzed	Methods	Sensors Used	Comments
Zhang et al. (2025) [19]	Mechanism jam, mechanism looseness, three-phase asynchrony	Variational Mode Decomposition (VMD), black-winged kite algorithm (BKA), gated recurrent unit (GRU).	Vibration sensors	Data-driven fault diagnosis approach focused on mechanical failure. Sensor is assumed and not derived from risk analysis.
Zhu et al. (2025) [12]	Jamming and incomplete opening and closing	Fusion of time-frequency domain energy features of vibration signal; particle swarm optimization (PSO) algorithm; SVM	Vibration sensor	Classification of six disconnector states. The article analyses only a few mechanical failure scenarios.
Obarcanin et al. (2023) [20]	Contact degradation, aging, mechanical deterioration	Condition assessment based on health indicators	Temperature, position and operational monitoring	The work focuses on a comprehensive assessment of the condition of the disconnectors. However, there is no quantitative prioritization of failure modes.
Chen et al. (2022) [22]	Incorrect positioning, mechanism failure, abnormal operating torque, environmental degradation	Online condition-monitoring platform	Torque sensors, angle encoders, micro-switches, attitude sensors, and temperature and humidity sensor	Sensor-rich monitoring architecture. However, sensor deployment is predefined and not justified through a systematic criticality analysis.
Suwanasri et al. (2021) [23]	Failure modes affecting transmission assets including conductors, insulators, and accessories.	FMECA combined with Analytical Hierarchy Process (AHP) and weighted-scoring methods. Criticality matrices are used to prioritize risks	No specific monitoring sensors considered	FMECA is used in the study. However, it focuses on transmission lies rather than disconnect switches and does not address sensor selection.
Westerlund et al. (2016) [21]	Contact deterioration, increased contact resistance, contact overheating	Temperature-current regression analysis for condition ranking	Infrared temperature sensors and current measurements	Focuses only on thermal degradation of contacts.
Qiu et al. (2015) [11]	Screw loose, transmission mechanism jammed	Variational Mode Decomposition, AdaBoost-SVM	Vibration, current sensors	The study develops a system for diagnosing mechanical faults. There is no criticality classification; the sensors are specified from the outset.

. This comparison highlights the lack of methodologies integrating failure-mode criticality analysis and sensor prioritization for predictive maintenance.

1.2. Research Gap and Contributions

Although significant progress has been achieved in fault diagnosis and condition monitoring of HVDS, existing studies mainly focus on fault detection after sensor deployment. A systematic methodology linking failure-mode identification, criticality assessment, and sensor prioritization remains largely unexplored. The present study addresses this gap by developing an FMECA-based framework that integrates failure analysis, criticality ranking, predictive maintenance requirement, and sensor selection into a unified decision-making process. To structure the identification of root causes and support systematic analysis, this study employs Ishikawa diagrams [24], which categorize potential causes into machinery, methods, materials, measurements, environment, and human factors, enabling comprehensive multidisciplinary evaluations [17,24].

Beyond failure classification, FMECA provides practical guidance for the selection of predictive maintenance sensors by linking each failure mode to appropriate monitoring measures [17]. By associating the Risk Priority Number (RPN) of each fault with two additional parameters: the usefulness of the sensor in terms of detection efficiency and the feasibility of its implementation considering cost, installation complexity, and environmental suitability, maintenance managers can make informed decisions regarding monitoring investments [25,26]. This structured approach is essential, as modern substations increasingly rely on diverse sensor technologies, and while additional sensors can improve diagnostic accuracy, they also increase costs, data-management complexity, and system vulnerability. In response to these needs, this study presents a comprehensive FMECA of a HVDS, systematically identifying and characterizing fifteen distinct failure modes. The criticality of each failure is assessed using the RPN framework, and root causes are analyzed through Ishikawa diagrams to support detailed cause–effect evaluation. For each failure mode, associated malfunctions, potential triggering factors, and numerical ratings for severity, occurrence probability, and detectability are defined in accordance with established FMECA guidelines [17,18,25]. This systematic analysis enables the development of a hierarchical sensor classification strategy for predictive maintenance, combining failure criticality with practical implementation considerations. The proposed multidimensional methodology provides a broader and more detailed assessment of the disconnector’s condition by accounting for parameters such as moisture, ice accumulation, and variations in contact pressure, which are frequently neglected despite their significant influence on failure development. The results provide a robust foundation for the development of predictive maintenance strategies tailored to high-voltage switching equipment, thereby contributing to enhanced reliability and resilience of electrical power systems.

2. Methodology

The methodology adopted in this study is structured into three main stages, as illustrated in the flowchart presented in Figure 1. The first stage involves the construction of the Ishikawa diagram [24,27,28,29,30], during which the problem under investigation is precisely defined and the potential causes of disconnector failures are identified and structured. The results of this analysis are subsequently utilized in the second stage, which is dedicated to the FMECA, to assess and prioritize the associated risks. Finally, the third stage aims to associate each prioritized failure mode with the corresponding measurable physical quantities, thereby enabling the selection and prioritization of sensors within a predictive maintenance framework for the disconnector. The different stages of this methodology are detailed in the following sections.

2.1. Construction of the Ishikawa Diagram

The construction of the Ishikawa diagram (Figure 2), also known as the cause-and-effect diagram [24,27,28,29,30], aims to exhaustively identify all potential causes that may lead to the failure modes observed in HVDS. This step constitutes the foundation of the FMECA and enables a structured examination of the technical, organizational, and environmental factors influencing the reliability of the disconnector. The first phase consisted of the precise definition of the undesired effect under investigation, which, in the context of this study, corresponds to a disconnector operating failure, whether of mechanical or electrical origin, or related to an insulation defect. A multidisciplinary expert panel was established to ensure a comprehensive understanding of disconnector operation and its potential failure mechanisms. The panel consisted of two engineers, two maintenance specialists with experience in the operation, inspection, and maintenance of high-voltage disconnectors and three researchers. The FMECA was developed using multiple sources of information, including historical intervention and maintenance records provided by an industrial partner involved in the project, documented failure cases, operational feedback from field personnel, and direct on-site observations. These records, accumulated since the company’s establishment in 2008, include several disconnector failure events and degradation cases that were analyzed to support the identification and assessment of failure modes. The occurrence ratings were assigned through expert consensus based on the frequency and recurrence of failure modes observed in these historical records, combined with operational experience accumulated over years of service. Because a complete statistical reliability database covering all disconnectors was not available, the occurrence scores should be interpreted as semi-quantitative estimates derived from expert judgment informed by historical operational evidence rather than as purely statistical probabilities. For each identified failure, the potential causes were classified according to the classical Ishikawa diagram categories: failure indicator, material, equipment, manpower, environment, and method. This classification improves the readability of the analysis and ensures coverage of all dimensions likely to affect disconnector reliability. The final Ishikawa diagram is developed following several brainstorming sessions, particularly aimed at identifying causes that are rarely documented but well known by field operators, thereby enhancing the relevance and completeness of the analysis. This step enabled a systematic structuring and comprehensive identification of the potential causes of disconnector failures.

2.2. Quantification, Prioritization, and Treatment of Causes Using FMECA for Predictive Maintenance

The FMECA represents the core stage of the proposed methodology. It transforms the causes identified through the Ishikawa diagram into structured failure modes that are evaluated and prioritized according to their impact on disconnector reliability. The primary objective is to obtain a quantitative and prioritized risk assessment, which is essential for effectively guiding the predictive maintenance strategy and sensor selection. Initially, the HVDS is decomposed into subsystems, each associated with an expected function. This functional decomposition allows for the precise definition of potential failure modes. Each failure mode is then linked to the causes identified in the previous step. According to the IEC 60812:2006-01 standard [31], the criticality of a failure mode is evaluated using the Risk Priority Number (RPN), defined as:

$$\mathrm{RPN} = \mathrm{S} \times \mathrm{O} \times \mathrm{D}$$

(1)

Parameter S (Severity) represents the seriousness of the failure effects on the system or the user. It is a dimensionless indicator that evaluates the intensity of the failure impact. The severity criteria defined in IEC 60812, widely applied in industries such as the automotive sector, are adopted in this study (

Table 2. Determination of severity parameter value (IEC 60812. STANDARD).

Severity	Criteria	Ranking
Very minor	No discernible effect.	1
Very minor	Fit and finish/squeak and rattle item do not conform. Defect noticed by discriminating customers (less than 25%).	2
Minor	Fit and finish/squeak and rattle item do not conform. Defect noticed by discriminating customers (less than 50%).	3
Very low	Fit and finish/squeak and rattle item do not conform. Defect noticed by discriminating customers (less than 75%).	4
Low	Item operable but comfort/convenience item(s) operable at a reduced level of performance. Customer somewhat dissatisfied.	5
Moderate	Item operable but comfort/convenience item(s) inoperable. Customer dissatisfied.	6
High	Item operable but at a reduced level of performance. Customer very dissatisfied.	7
Very High	Item inoperable (loss of primary function).	8
Hazardous With warning	Very high severity ranking when a potential failure mode affects safe operation and/or involves noncompliance with government regulation with warning.	9
Hazardous without warning	Very high severity ranking when a potential failure mode affects safe operation and/or involves noncompliance with government regulation without warning.	10

). In cases where failure modes have identical RPN values, those with the highest S value are treated as a priority [32].

Parameter O (Occurrence) corresponds to the probability of a failure mode occurring over a specified period. It is expressed as a ranking value rather than an actual probability.

Table 3. Determination of occurrence parameter value (IEC 60812 STD).

Occurrence	Frequency	Ranking
Remote: Failure is unlikely	<=0.01 per thousand items	1
Low: Relatively few failures	0.1 per thousand items	2
	0.5 per thousand items	3
Moderate: Occasional failures	1 per thousand items	4
	2 per thousand items	5
	5 per thousand items	6
High: Repeated failures	10 per thousand items	7
	20 per thousand items	8
Very high: Failure is almost inevitable	50 per thousand items	9
	>=100 per thousand items	10

presents the occurrence levels according to the estimated failure frequency.

Parameter D (Detection) represents the ability to detect a failure before it affects the system or the user. It is generally ranked inversely to severity and occurrence: the higher the D value, the lower the likelihood of detection, resulting in a higher RPN.

Table 4. Failure mode detection evaluation criteria (IEC 60812 STD).

Detection	Criteria: Likelihood of Detection by Design Control	Ranking
Almost certain	Design Control will almost certainly detect a potential cause/mechanism and subsequent failure mode	1
Very high	Very high chance the Design Control will detect a potential cause/mechanism and subsequent failure mode	2
High	High chance the Design Control will detect a potential cause/mechanism and subsequent failure mode	3
Moderately high	Moderately high chance the Design Control will detect a potential cause/mechanism and subsequent failure mode	4
Moderate	Moderate chance the Design Control will detect a potential cause/mechanism and subsequent failure mode	5
Low	Low chance the Design Control will detect a potential cause/mechanism and subsequent failure mode	6
Very low	Very low chance the Design Control will detect a potential cause/mechanism and subsequent failure mode	7
Remote	Remote chance the Design Control will detect a potential cause/mechanism and subsequent failure mode	8
Very remote	Very remote chance the Design Control will detect a potential cause/mechanism and subsequent failure mode	9
Absolutely uncertain	Design Control will not and/or cannot detect a potential cause/mechanism and subsequent failure mode; or there is no Design Control	10

presents the detection ranking scale used in accordance with IEC 60812. The final phase of this step consists of ranking the failure modes in descending order of criticality to identify the priority failures to be addressed.

2.3. Selection and Prioritization of Sensors for Predictive Maintenance

Following the objective prioritization of the most critical failure modes, the next step consists of selecting and prioritizing the sensors required for predictive maintenance. For each prioritized failure mode, the corresponding components are first mapped to the measurable physical quantities that reflect their degradation mechanisms, such as temperature, vibration, current, or mechanical position. Once these physical indicators are identified, suitable sensors capable of monitoring them are selected. The prioritization of sensor installation within the predictive maintenance framework is then determined by evaluating some criteria, including the RPN value of the associated failure mode, sensor selection constraints, and the expected usefulness of each sensor in supporting accurate and timely fault prediction. This ensures that sensing resources are allocated to the most influential parameters and that monitoring efforts are aligned with the system’s reliability needs. The relevance of such a structured prioritization approach is supported by numerous studies in the reliability and maintenance literature, which highlight the benefits of using composite priority indices that combine normalized RPN values with weighting or decision-based coefficients. For example, Braglia’s multi-attribute failure mode analysis demonstrated that normalizing risk indicators and applying weighting factors yields a more representative priority score, thereby improving the limitations of traditional RPN-based ranking [33]. Similarly, Huang et al. 2025 [34] proposed an enhanced FMEA risk-assessment method that integrates TOPSIS, multi-perspective weighting, while Anes et al. (2024) [35] showed that combining ROC-based weighting of FMEA criteria with the CoCoSo multi-criteria ranking method significantly improves the identification of critical risks. Their results confirm that utility-based and decision-based coefficients strengthen the discrimination of failure modes compared to conventional FMEA. These contributions collectively justify the use of a combined prioritization formula of the form:

$$\mathrm{Priority}={\displaystyle \frac{RPNi}{RPMmean}}\times \mathsf{\alpha }\times \mathsf{\beta }$$

(2)

where $\mathsf{\alpha }$ and $\mathsf{\beta }$ are the utility and choice coefficients, respectively. For the utility criterion, weighting coefficients of 1, 1.5, and 1.75 are chosen arbitrarily in this scenario, depending on whether the sensors are associated with one, two, or three components, respectively. For the choice criteria, a weighting coefficient of 1 is assigned when the sensor is affordable, easily accessible, and capable of providing accurate measurements. If one of these conditions is not met, the weighting is reduced to 0.75; if two conditions are not satisfied, a coefficient of 0.5 is applied. Finally, when none of the conditions are fulfilled, a minimum weight of 0.25 is used. These coefficients can be calibrated according to individual needs. The criteria for selecting the values of these coefficients are defined in

Table 5. Weighting of the selection criteria.

Utility
Choice	Weighting	Selection Criteria
1	1	Sensor for a single component
2	1.5	Sensor for two components
3	1.75	Sensor for three components
Choice
Choice	Weighting	Selection criteria (1) Affordable; (2) easily accessible and (3) accurate measurements.
1	1	All three criteria are satisfied
2	0.75	Two of three criteria are satisfied
3	0.5	One of three criteria is satisfied
4	0.25	None of the criteria are satisfied

.

Based on the methodology presented above, the results related to the construction of the Ishikawa diagram, the identification of failure modes and their criticality, as well as the selection and prioritization of sensors were obtained and are presented in the following section.

3. Results and Discussion

In this section, fifteen failure modes of a HVDS have been identified and characterized, with a detailed description of the observed effects and the associated potential causes for each. To assess their relative impact on equipment reliability, a classification was established based on the Risk Priority Number (RPN). This approach enables the ranking of failures according to their criticality level and, consequently, the prioritization of maintenance actions to be undertaken. Specifically, failure modes with an RPN greater than or equal to 80 are considered to present a high risk. Conversely, an RPN between 50 and 79 corresponds to a moderate risk, while an RPN below 50 indicates a low risk. This categorization represents a key step in guiding maintenance strategies and optimizing resource management. The failure modes belonging to each criticality class are analyzed in detail in the following sections, to highlight their occurrence mechanisms, functional consequences, and potential impacts on system performance. Furthermore, an analysis of the sensors deemed relevant for implementing a predictive maintenance strategy is also presented. This approach aims to enhance the monitoring of critical components, anticipate potential failures, and ultimately improve the overall availability and reliability of the high-voltage disconnector.

3.1. Detailed Analysis of Failure Modes by Criticality Level

The FMECA conducted on a HVDS enabled the identification of several failure modes exhibiting varying levels of criticality. Each mode was evaluated in terms of severity, occurrence frequency, and detectability, thereby allowing the calculation of the RPN. The results presented below highlight the most critical failures while emphasizing the relevant monitoring means for the implementation of a predictive maintenance strategy.

3.1.1. High-Critical Failure Modes

The most critical failure modes are not limited to mechanical causes; they may also result from electrical failures or exposure to harsh environmental conditions. The details of these different failure modes are presented below with their causes and effects summarized in

Table 6. High-critical failures root cause analysis.

Component	Function	Failure Mode	Potential Failure Effect(s)	Potential Causes
Copper contacts with silver coating	Transmission of the electric current through two mechanically separable elements	Contact burning due to inadequate jaw–blade contact	Disconnector becomes inoperable	Poor field fit during commissioning, contact spring stiffness, incomplete open/close, non-aligned contact
Sensors	Monitoring of the different physical quantities	Wrong indication	Erroneous data, aberrant data, lack of data, false control, communication problem	Poor sensor quality, improper installation, unfavorable environment, presence of external contamination (dust, etc.), presence of electromagnetic noise, connection and wiring issues between the sensor and the box, cyberattack
Cabinet	Protect the components of the motorized cabinet from external hazards and vice versa	Water infiltration, damaged structure/enclosure	Damage to electronic components, rusting of metal components	Gasket not designed for extreme environments, enclosure paint non-compliant/poorly applied, enclosure not properly closed, enclosure rusting
Disconnector pivot	Transfer the movement from the linkage to the rotary isolator	Blocked pivot	The disconnector is rendered inoperative due to a blocked blade	Water infiltration and other potential contaminants, presence of ice on the pivot, oxidized linkage, oxidized bearing, faulty seal
Heating element and thermostat	Maintain a good level of temperature and humidity for electrical components	The heating element has stopped working	Condensation formation inside the motorized cabinet	Premature rust on some components

.

• Silver-plated copper contacts

Silver-plated copper contacts ensure current conduction during the closing of the disconnect switch and maintain its electrical insulation when open. The silver coating provides low contact resistance and high resistance to aging, heat, and oxidation. However, several cases of contact burning have been reported, generally caused by poor contact between the switch jaws. This anomaly, which can render the device inoperative, may result from various factors such as improper leveling during installation, excessive spring stiffness, incomplete operating maneuvers, or contact misalignment. Although its occurrence is moderate (5/10), the high severity (9/10) and medium detectability (5/10) lead to a maximum RPN of 225, indicating a very high level of criticality.

• Sensor Failure

Sensors play a key role in monitoring the operation of the disconnect switch by measuring essential physical quantities. Incorrect reading can lead to inappropriate control actions, compromising the operational safety of the installation. Such failure may arise from poor sensor quality, installation errors, or adverse environmental conditions. With an RPN of 135, it is classified among high-criticality failures and highlights the need for regular verification and systematic calibration of sensors used in predictive maintenance.

• Pivot seizure

The pivot is a key mechanical component ensuring the rotational linkage between the fixed and movable parts of the disconnect switch. Its seizure may result from water or contaminant ingress, ice formation, oxidation phenomena, or seal failure. Although this failure is infrequent (2/10), its high severity (9/10) and medium detectability yield an RPN of 90, representing a major mechanical risk likely to impair the device’s maneuverability. Adequate lubrication, combined with enhanced protection against moisture ingress, is recommended to reduce the likelihood of this failure mode.

• Motorized enclosed and heating element

The motorized enclosure houses the control and actuation components of the disconnect switch. Despite being designed to withstand harsh environmental conditions, it remains vulnerable to water ingress, seal degradation, or improper closure. This failure, characterized by moderate severity (6/10), low occurrence (3/10), and limited detectability (6/10), results in an RPN of 108. To mitigate its effects, the enclosure is equipped with a heating element and thermostat to maintain stable internal temperature and humidity. However, in the event of heating element failure, condensation may occur, leading to premature corrosion of internal electronic components. The analyses conducted highlight the diversity of failure modes that may affect a HVDS, as well as the complexity of the interactions between their causes and effects. To structure these results and facilitate their interpretation, two summary tables are presented below.

Table 7. High-critical failures RPN.

Failure Mode	Severity	Occurrence	Detection	RPN
Contact burning due to inadequate jaw–blade contact	9	5	5	225
Wrong indication	9	5	3	135
Water infiltration, damaged structure/enclosure	6	3	6	108
Blocked pivot	9	2	5	90
The heating element has stopped working	5	4	4	80

presents the quantitative assessment of the risk associated with each failure mode according to the FMECA method. For each failure, it compiles the values assigned to occurrence frequency, detectability, and severity, along with the calculation of the RPN. This table serves as a decision-support tool, enabling the prioritization of risks, the identification of the most critical failures, and the ranking of corrective or preventive actions to be implemented.

In this synthesis, only failure modes with a high RPN (≥ 80) have been retained, due to their potential impact on the reliability and safety of the disconnect switch. These failures, considered as priority cases, will be subject to enhanced monitoring as part of a predictive maintenance strategy.

3.1.2. Moderate Critical Failures

Following the analysis of moderate risk failure modes, several additional failures were identified as exhibiting moderate criticality. Although their impact is lower compared to priority failures, their occurrence frequency and potential for disruption justify particular attention within the framework of a preventive and predictive maintenance strategy.

• Compensation spring, linkage and printed circuit board (PCB)

The compensation spring is designed to reduce the effort required to operate the disconnect switch. Its weakening caused by condensation, corrosion, or excessive stiffness can lead to breakage. Such a failure prevents proper blade compensation, rendering the switch inoperative. A forced operating attempt may additionally cause damage to the gearbox. Although less frequent than high RPN failures, this situation represents a significant risk to the overall mechanical integrity of the device. The mechanical and electronic control system is also at moderate risk of failure. A faulty printed circuit board (PCB) design, water ingress into the housing, an inadequate power supply or unauthorized external interference can disrupt the operation of the disconnect switch. The electronic boards may overheat or short-circuit. The presence of mechanical obstructions or ice on the pivots may damage the linkage and restrict the movement of the blades.

• Male and female arc horns

Poor contact between the arc horns, or the complete deterioration of one of them, leads to the formation of cavities on the copper contacts under the effect of the electric arc. This degradation gradually reduces the service life of the contacts. Ultimately, this can lead to more serious failures associated with high RPN values. Regular monitoring and maintenance of these components are therefore essential to limit serious faults. Attention is also paid to failure modes with an RPN between 50 and 79, as these can compromise the reliability of the disconnector if not properly anticipated.

Table 8. Moderate critical failures root cause analysis.

Component	Function	Failure Mode	Potential Failure Effect(s)	Potential Causes
Compensation springs	Reduce the effort required to operate the disconnector	The inner spring is broken	There will no longer be compensation on the blade and the disconnector becomes inoperable. If you force it, the gearbox can be damaged	Condensation and rust that weaken the spring over time, spring stiffness inside
Electronic board	Provide command and control of the entire disconnector	The electronic board is burnt or saturated	Operation of the disconnector disrupted, triggering of false alarms	Poor PCB design, cabinet infiltration (humidity, contaminants), blocked air vent, insufficient power supply, careless handling
Linkage	Transfer operator movement to all three phases	Damaged linkage	The disconnector becomes inoperable, asynchronization of the position of the three phases, blade blocked	Presence of an obstacle that prevents the blade from moving, presence of ice at the pivot, oxidized linkage
Arc Horn	Protect disconnector material	The male and female arc horns do not touch when in the closed position, or either one is completely burned; blade stuck	Creates cavities on copper contacts due to electric arcing and reduces their lifespan, prolonged arc duration	High residual current at the disconnector location in the substation that will cause arcing or misalignment in the field during installation, presence of an obstacle that prevents the blade from moving

compiles the information derived from the Ishikawa diagram, linking each relevant component to its main function, the identified failure mode, the observed effects, and the probable causes.

Table 9. Moderate critical failures RPN.

Failure Mode	Severity	Occurrence	Detection	RPN
The inner spring is broken	9	4	2	72
The electronic board is burnt or saturated	9	4	2	72
Damaged linkage	9	3	2	54
The male and female arc horns do not touch when in the closed position, or either one is completely burned; blade stuck	5	5	2	50

presents the quantitative assessment of risk for each moderate criticality failure, based on the FMECA criteria. It includes the values assigned to occurrence frequency, detectability, and severity, as well as the calculated RPN.

3.1.3. Low Critical Failures (RPN < 50)

In addition to the high and moderate criticality failures, several failure modes have been identified as presenting a low risk, characterized by a RPN below 50. Although their impact on the overall reliability of the disconnect switch is limited, their monitoring remains necessary to prevent progressive degradation or cumulative effects that could eventually lead to more severe failures.

• Electric Motor and Associated Control System

The motor actuator may experience malfunctions due to power supply loss, excessive inrush current, or short-circuiting between positive and negative terminals. These anomalies can cause the melting of protective fuses, thereby compromising the operation of the drive system.

• Brakes

The motor’s rotation is stopped by a braking system that may seize in case of insufficient clearance between pads or corrosion of mechanical components. Such blockage can prevent the motor from stopping properly, affecting the accuracy of the disconnect switch operations.

• Limit Switches

Limit switches play a crucial role in motor control by signaling to the control board the end of the operating stroke. A fault in these switches may cause incorrect stopping or even overtravel. This malfunction, combined with improper adjustment of the thermal relay, can lead to mechanical overload and damage to the gearbox.

• Ground Leakage

Insulation faults in the system can result in current leakage to ground, potentially causing unintentional tripping of adjacent circuit breakers. Although this type of failure is generally detectable and infrequent, it can disrupt the overall operation of the installation. To structure the results associated with this category and facilitate their interpretation, two summary tables are presented below.

Table 10. Low critical failures root cause analysis.

Component	Function	Failure Mode	Potential Failure Effect(s)	Potential Causes
Electric motor	Transfer electrical energy in rotary motion	Defective engine and not working	Disconnector becomes inoperative in electric mode, inconvenient opening/closing speed, blade blocked	Failure of the motor protection mechanism
Motor Control	Bringing power to the engine	Defective actuator and not functioning properly	Disconnector becomes inoperative in electric mode	High humidity, loss of power supply, presence of contaminants (dust, iron filings)
Fuses	Protect the motor’s electrical system	Blown fuse	Disconnector becomes inoperative in electric mode	Short-circuit to ground in the system, high starting current, presence of an obstacle preventing the blade from moving
Brakes	Stop the motor rotation	Blocked brake	Premature use of brake components, longer operating time or disconnect switch rendered inoperative in electric mode	Insufficient brake pad clearance, corrosion of brake components
Limit switches	Indicate to the control panel the moment when the motor should be stopped	Limit switch fault	The disconnector does not stop at the end of the opening or closing operation. Incomplete opening/closing	Fault in the limit switch and tripping of the thermal relay; the gearbox may break if the overload is not set correctly
Rotating Mounts & Isolators	Isolate the energized portion of the disconnector from the ground and support the energized portion of the disconnector	Flashover (ground fault)	Tripping of adjacent circuit breakers	High pollution area, accumulation of dust on the insulator sheds which, over time, can promote flashover, presence of ice on the sheds

compiles information derived from the Ishikawa diagram, linking each relevant component to its main function, the identified failure mode, the observed effects, and the probable causes. This approach provides a systemic perspective of failure mechanisms, even for low-criticality cases.

Table 11. Low critical failures RPN.

Failure Mode	Severity	Occurrence	Detection	RPN
Defective engine and not working	6	4	2	48
Defective actuator and not functioning properly	5	4	2	40
Blown fuse	5	4	2	40
Blocked brake	5	3	2	30
Limit switch fault	5	3	2	30
Flashover (ground fault)	7	2	2	28
Wrong indication	4	1	1	4

presents the quantitative risk assessment for each low failure according to the FMECA criteria. It includes the values assigned to occurrence frequency, detectability, and severity, as well as the calculated RPN. This table complements the overall risk hierarchy and identifies key areas of concern to be addressed in maintenance planning.

Following the identification and assessment of the failure modes of a HVDS using the Ishikawa diagram and FMECA, it is essential to examine the monitoring means that could enable early detection of such failures. In this context, the following section focuses on the analysis of sensors that could be integrated into the system as part of a predictive maintenance strategy.

3.2. Sensor-Based Monitoring for Predictive Maintenance of the High-Voltage Disconnect Switch

The failure modes of a HVDS together with their causes and effects have been identified and analyzed in the previous section; this part focuses on the selection and prioritization of sensors to be installed for the purpose of predictive maintenance. In accordance with the adopted methodology, the installation priority of each sensor is determined based on the RPN, weighted by the utility coefficient which reflects the sensor’s relevance for fault detection and the selection coefficient, which accounts for technical and economic constraints. The analysis of these parameters made it possible to classify the sensors according to their priority index. The results indicate that the current sensors exhibit a major level of criticality, with a priority approximately four times higher than the average. Its installation is therefore strongly recommended. For example, current sensors can be installed to measure leakage current, line current, or the current of the DC motor powering the HVDS. In

Table 12. Sensors prioritization.

Sensors	Components	Priority	Usefulness	Choice	RPN
Current sensors	DC Motor/Copper Contacts/Disconnector Pivot	5.7	3	1	225
	Copper Contacts/Network	3.3	1	1	225
	Leakage current	0.4	1	1	28
	Brakes	0.4	1	1	30
Voltage sensors	DC Motor	1.2	3	1	48
	Fuses	0.6	1	1	40
	Limit switch	0.3	1	2	30
Temperature sensors	External Temperature: Cabinet/Disconnector Pivot/Heating Element/Thermostat	2.7	3	1	108
	Copper Contacts/Arc Horn	2.4	1	2	225
	Internal Housing/Motor	1.6	1	1	108
	Electronic board	1.0	1	1	72
	Arc Horn	0.7	1	1	50
	Linkage	0.4	1	3	54
Humidity sensors	Cabinet	1.6	1	1	108
	Disconnector Pivot	1.0	1	2	90
Frost sensor	Copper Contacts/Linkage/Compensation Spring	2.2	2	1	100
Vibration sensors	Compensation Spring	1.0	1	1	72
	Linkage	0.8	1	1	54
Sound sensors	Copper Contacts	1.6	1	3	225
	Compensation Spring/Arc Horn	1.2	2	2	72
Wind sensor	Linkage/blades	1.4	1	1	100
Position sensors	Linkage	0.8	1	1	54
	Motor	0.7	1	1	48
Torque sensor	DC motor: at the output of the motorized control	0.7	1	1	48

, measuring the current of the DC motor powering the HVDS takes priority over measuring leakage current. Indeed, monitoring the motor current provides valuable insights into the overall system condition, particularly regarding the behavior of the motor, the copper contacts, and the pivot mechanism. An abnormally high current may, for instance, reveal a pivot blockage or a poor contact at the jaws. Moreover, the measurement of temperature and humidity, both inside and outside the motorized enclosure, is of particular importance. These sensors make it possible to verify the proper operation of the heating element inside the enclosure and to ensure that the electronic components operate under their optimal temperature and humidity conditions. A deviation in these parameters could indicate a malfunction of the heating system or premature aging of internal components. In addition, the installation of infrared temperature sensors directed toward the disconnector jaws provides an effective means of detecting abnormal temperature rises. Such increases are often associated with excessive contact pressure or progressive degradation of contact surfaces, which are early indicators of overheating and material aging in the conductive parts.

In cold or northern regions, where winter conditions are particularly severe, the presence of a frost sensor becomes essential. This sensor, whose priority index is approximately twice the average, enables the detection of ice formation on the jaws. This phenomenon can explain current peaks observed during the opening of the disconnector, caused by the increased mechanical resistance due to frost accumulation. The frost sensor thus complements the current sensor, contributing to a more accurate interpretation of electrical anomalies. Vibration and acoustic analysis also represent relevant approaches. Monitoring the vibration or acoustic signals during the opening and closing operations of the disconnector allows the assessment of the condition of the compensation spring and other mechanical components. Variations in the vibration or sound spectrum can indicate misalignment, spring relaxation, or mechanical wear. Finally, other sensors complete the supervision of the disconnector. The voltage sensor provides information on the presence and stability of the supply voltage, thereby helping to detect control anomalies. The position sensor, on the other hand, confirms the exact state of the disconnector (open, closed, or intermediate), ensuring both operational safety and reliability of control signals. Table 12 shows the different sensors with their priority level.

3.3. Robustness and Sensitivity Analysis of Sensor Prioritization Coefficients

A sensitivity analysis was conducted on the sensor prioritization coefficients α and $\beta$ to assess the robustness of the proposed model. Each parameter varied within the range −20% to +20% around its nominal value. The obtained results show that the output curves remain well separated and do not intersect, indicating a stable and monotonic influence of α on the model output. Similar behavior was observed for $\beta$, confirming that the proposed formulation is not sensitive to small variations in these coefficients. Therefore, the selected value of α and $\beta$ are justified as they ensure consistent system behavior without affecting the ranking structure of sensor prioritization. Figure 3 illustrates this phenomenon for current, voltage, and temperature sensors. For conciseness, only these three representative sensors are shown in the figure. However, the same sensitivity analysis was performed on all other sensors, and consistent trends were observed across all cases.

4. Conclusions

The analysis of FMECA reveals that contact-related failures exhibit the highest RPN values, confirming that both electrical conduction and mechanical stability are critical aspects of the high-voltage disconnector’s reliability. These results highlight that deterioration of contact surfaces, loss of alignment, or excessive mechanical stress at the jaws are among the most significant factors contributing to functional degradation. Consequently, improving contact integrity and monitoring parameters related to electrical and mechanical performance appear essential for preventing major failures. Building upon these findings, the ranking of sensors demonstrates a clear consistency with the identified dominant failure modes. In particular, the current sensor emerges as the most critical device, reflecting variations associated with contact resistance, mechanical blockage, or motor overload. This coherence between the FMECA outcomes and the sensor prioritization confirms the relevance of the adopted methodology, where the combination of the RPN, utility, and selection coefficients provides a robust basis for guiding sensor implementation strategies. Furthermore, this alignment supports the notion that monitoring the motor current can serve as a reliable proxy for assessing both electrical and mechanical health conditions within the disconnector. The integration of the identified sensors therefore represents a significant step toward the deployment of a monitoring strategy. By enabling continuous monitoring of key parameters such as current, temperature, humidity, and vibration, the system can detect early signs of degradation and schedule interventions only when necessary. This predictive approach minimizes unnecessary maintenance actions, reduces downtime, and ultimately enhances the availability and operational reliability of the high-voltage disconnector. In addition, the correlation between multi-domain signals (electrical, thermal, and mechanical) opens new possibilities for multi-sensor data fusion and advanced diagnostics. It is also noteworthy that, while previous HVDS fault detection has focused on electrical signatures, our FMECA-based sensor prioritization explicitly weights mechanical and environmental failure modes, which are underrepresented in existing monitoring strategies. This multidimensional framework offers a more comprehensive understanding of the disconnector’s condition by considering factors such as humidity, frost formation, or contact pressure variations, which are often overlooked but play a crucial role in failure propagation. Such integration extends the scope of predictive maintenance beyond conventional electrical monitoring and aligns with the current trends in smart grid asset management and industrial Internet of Things (IIoT)-based diagnostics.

5. Study Limits and Future Work

Although this study has successfully identified failures and prioritized critical sensors for the predictive maintenance of HVDS, the weighting of sensors using utility and selection coefficients remains partially subjective. Since it relies on expert judgment and literature data, these coefficients may need to be adapted to specific industrial environment. This limitation, however, provide a clear pathway for future work. A first step involves experimental validation of sensor priorities and their effectiveness in early fault detection, both on prototypes and in service. Furthermore, integrating multi-sensor data into a predictive maintenance framework using artificial intelligence represents a crucial advancement. By combining electrical, mechanical, and environmental signals through data-fusion techniques and machine learning algorithms, the predictive models can improve fault diagnosis accuracy and anticipate failures more reliably. Finally, future studies could expand the range of monitored parameters, adapt predictive models to different types of disconnectors and operational conditions, and develop optimized condition-based maintenance strategies, ultimately reducing unnecessary interventions, enhancing equipment availability, and improving the safety of high-voltage operations.