Open research questions in Power System Reliability and Maintenance

What the literature leaves open

• The application of the proposed methodology to other industrial infrastructure. The development of more advanced machine learning-based predictive maintenance models.

  Reliability Modeling and Predictive Maintenance Integration for an Induced Draft Fan System Using Semi-Markov Process and Machine Learning-Based Predictive Models · 2026 · DOI

• The lack of effective predictive maintenance frameworks for induced draft fan systems in thermal power plants. The need for a methodology that combines reliability engineering and machine learning to improve system dependability and efficiency.

  Reliability Modeling and Predictive Maintenance Integration for an Induced Draft Fan System Using Semi-Markov Process and Machine Learning-Based Predictive Models · 2026 · DOI

• The increasing use of distributed energy sources, energy storage systems, and electric vehicles introduces new components to be modelled. The accuracy of calculation results depends strongly on the accuracy of input data. The paper needs to provide a measurement-based evaluation of distribution transformer losses characteristics.

  Measurement-Based Evaluation of Distribution Transformer Losses Characteristics for Sustainable Voltage Regulation · 2026 · DOI

• Future research should focus on the development of a digital planning environment for smart grid planning in industrial and commercial facilities. The research should investigate the application of ETAP-based analysis to other industries and regions. The paper suggests that future research should explore the integration of ETAP-based analysis with other technologies, such as IoT and AI.

  ETAP-Based Power System Analysis for Smart Grid Planning in Saudi Industrial and Commercial Facilities · 2026 · DOI

• The paper identifies a gap in the current literature on smart grid planning in Saudi industrial and commercial facilities. The gap is the lack of a digital planning environment that links the design of facilities, operational reliability, safety compliance, asset loading, and future distributed energy integration.

  ETAP-Based Power System Analysis for Smart Grid Planning in Saudi Industrial and Commercial Facilities · 2026 · DOI

• The reliability of the national distribution network remains inadequately quantified, hindering evidence-based policy interventions.

  A Multilevel Regression Analysis of Power-Distribution System Reliability in Ghana: A Policy Evaluation for Infrastructure Governance · 2026 · DOI

• Trade-offs between precision versus speed (physics-informed DTs requiring domain expertise versus data-driven models risking overfitting) and between centralized versus decentralized control architectures require further research and resolution.

  Impact of artificial intelligence-driven digital twins and lean six sigma-assisted power system asset management on long-term investment planning · 2026 · DOI

• A divide exists between traditional LSS practitioners who lack proficiency in AI tools and data scientists who undervalue LSS structured problem-solving frameworks, with differences of opinion stalling collaborative efforts.

  Impact of artificial intelligence-driven digital twins and lean six sigma-assisted power system asset management on long-term investment planning · 2026 · DOI

• Field validation on longitudinal operational data is lacking; most models are tested on simulated or partially anonymized datasets rather than actual critical infrastructure performance data, reducing empirical credibility and limiting deployment readiness for nuclear and grid system early failure detection.

  Data-Driven Approaches to Reliability Modeling in Critical Energy Infrastructure: Analytical Perspectives on Early Failure Detection · 2026 · DOI

• Federated learning and edge AI architectures for distributed early failure detection across geographically separated wind farms, renewable installations, and substations have not been systematically validated; real-time anomaly detection at device or substation-level without centralizing sensitive data requires operational implementation studies.

  Data-Driven Approaches to Reliability Modeling in Critical Energy Infrastructure: Analytical Perspectives on Early Failure Detection · 2026 · DOI

• Existing optimization methods have not adequately considered the impacts of planned and unplanned outages. There is a need for a new optimization method that explicitly incorporates these outages.

  Optimization for Battery Energy Storage Configuration Considering the Impact of Planned and Unplanned Outages · 2026 · DOI

• There is a need for a reliability assessment framework for evaluating the impact of photovoltaic integration on power networks. The study identifies a gap in the existing literature on the evaluation of PV-integrated grids.

  Reliability Assessment of a PV-Integrated 132 kV Power Network Using ETAP · 2026 · DOI

• The essential problem related to the maintenance of complex systems and structures is related to the challenges of predicting the failure behaviour of the components with due account of associated uncertainties. There is a need for a comprehensive assessment of statistical methods and reliability models for reliability function estimation in thermal power plants.

  Assessing statistical methods and reliability models for reliability function estimation in thermal power plants · 2026 · DOI

• The failures of individual components are assumed to be statistically independent. The failure rates are assumed to remain constant within the considered operating period. Common-cause failures and dynamic interactions between components are not explicitly considered.

  STUDY TO ASSESS ELECTRICITY GENERATION IN A THERMAL POWER PLANT USING FAULT TREE ANALYSIS MODEL · 2026 · DOI

• Further research can be conducted on the application of Fault Tree Analysis model for availability assessment of other types of power plants. The results can be used for developing more accurate models for predicting the availability of thermal power plants.

  STUDY TO ASSESS ELECTRICITY GENERATION IN A THERMAL POWER PLANT USING FAULT TREE ANALYSIS MODEL · 2026 · DOI

• The paper identifies a gap in the evaluation of real energy consumption by the population during power outages. The study highlights the need for a regression model to predict energy consumption limits.

  "Charge avalanche phenomenon" in conditions of stabilization shutdowns · 2026 · DOI

• Existing clustering methods, such as Z-score and k-means, have limitations. Prior studies used binary ramp event categorization or non-causative classification techniques.

  A novel hybrid clustering approach for robust ramp event characterization · 2026 · DOI

• Although the proposed ZK-means technique demonstrated effective performance in identifying and classifying wind power variations, several limitations should be acknowledged. The case study was conducted using aggregated national-level wind generation data from Belgium, which may smooth out local variability and extreme ramp events. Applying the method to higher-resolution or plant-level datasets could reveal more detailed ramp dynamics and enhance the robustness of the classification outcomes. In addition, the quality, completeness, and spatial representativeness of the available wind generation data directly influence the clustering results. Because the outputs of the ZK-means classification are spatially dependent, applying the method to a different location or dataset may yield variations in the relative frequency and distribution of clusters. Hence, the model’s transferability should be interpreted cautiously, and local calibration may be required. From a methodological perspective, the silhouette coefficient initially indicated two optimal clusters corresponding to low and high ramp events. To improve discrimination among ramp magnitudes and enhance operational interpretability and control. Also, to avoid the drawbacks of binary classification, this configuration was refined to four clusters. This refinement provided a more granular representation of ramp behavior without compromising consistency or stability across years. Furthermore, the current implementation relied solely on power time-series features, without incorporating meteorological variables such as wind speed, direction, or pressure gradients. Integrating such multi-source information could strengthen the physical interpretability of the derived clusters and improve their linkage to underlying weather patterns. Combining meteorological and generation-based features would also allow the method to distinguish between structurally different ramp causes (e.g., weather-driven vs. grid-induced). Although the method achieved reasonable computational efficiency, its scalability to large spatio-temporal datasets and real-time operation still requires further evaluation. Implementing parallel processing or streaming- based clustering architectures could help address this challenge. The algorithm’s hybrid structure, combining normalization with centroid-based partitioning, makes it promising for adaptation to real-time, incremental clustering frameworks. For future work, the ZK-means framework can be extended through the integration of unsupervised feature learning methods (e.g., autoencoders or transformers) to extract latent representations of ramp events. It can also be combined with probabilistic forecasting or operational decision-support modules to assist grid balancing and tested across different countries and renewable energy mixes to assess its generalization capability. Finally, incorporating uncertainty quantification into the ZK-means clustering process may provide confidence bounds for ramp classifications, enhancing its practical utility for grid operations.

  A novel hybrid clustering approach for robust ramp event characterization · 2026 · DOI

• The study identifies a gap in the existing literature on contingency analysis and ranking of power transmission lines. The study highlights the need for a more effective and efficient method for contingency analysis and ranking.

  Intelligent Contingency Ranking of Nigeria's 330 KV Transmission Network Using Artificial Neural Network · 2026 · DOI

• There is a lack of precise, longitudinal measurement of distribution system performance in many Sub-Saharan African nations. Robust, evidence-based policy frameworks are required for economic development.

  A Panel-Data Estimation of Power-Distribution System Reliability for Policy Formulation in Senegal (2000–2026) · 2026 · DOI

• The study identifies a gap in understanding the impact of physical climate risks on the electricity industrial chain. The study aims to investigate the dynamic risk linkage across the upstream, midstream, and downstream segments of China’s electricity industry chain.

  How Physical Climate Risks Impact the Risk Linkage of the Electricity Industrial Chain · 2026 · DOI

• Further investigation of the complex interactions between extreme weather events and high-renewable power systems. Application of the proposed framework to real-world power systems. Development of more advanced models to capture the complex temporal mappings between planning parameters and power surging.

  Risk analysis of power surging in novel power systems: a hybrid framework driven by accident chain and WPMixer · 2026 · DOI

• The limitations of traditional theories in characterizing cross-level and cross-regional transient power processes. The lack of a comprehensive risk analysis and planning methodology for power surging in novel power systems.

  Risk analysis of power surging in novel power systems: a hybrid framework driven by accident chain and WPMixer · 2026 · DOI

• The study identifies a gap in the current maintenance strategy, which is highly reactive and focused on corrective maintenance. The study shows that there is a need for a more proactive and preventive maintenance approach to improve boiler reliability and reduce forced outages.

  Machine learning enabled preventive maintenance strategy for improving boiler reliability of a thermal power station · 2026 · DOI

• • Implement the developed preventive maintenance strategy optimisation model at the Thermal Power Station Stage 2 as a decision-support tool for boiler maintenance planning. • Create a reliability review routine in which model predictions are compared with weekly plant outcomes, including boiler failures, availability, average generation, maintenance cost and mean time between maintenance. Strengthen condition monitoring by linking inspection records, operating data, boiler efficiency data and maintenance history into one controlled dataset for model updating. • • Use model outputs to prioritise preventive work orders, spares planning and outage preparation for high- risk boiler components. • Conduct a formal cost-benefit analysis after pilot implementation to quantify avoided failures, reduced maintenance cost, improved availability, and additional sent-out energy. • Retrain and validate the model periodically so that it remains aligned with changing plant conditions, maintenance practice and operating regimes.

  Machine learning enabled preventive maintenance strategy for improving boiler reliability of a thermal power station · 2026 · DOI

Most-cited papers in Power System Reliability and Maintenance

• A resilience assessment of an interdependent multi-energy system with microgrids · Sustainable and Resilient Infrastructure · 2020 · 39 citations
• Condition-based monitoring as a robust strategy towards sustainable and resilient multi-energy infrastructure systems · Sustainable and Resilient Infrastructure · 2022 · 26 citations
• Reliability evaluation of smart grid using various classic and metaheuristic clustering algorithms considering system uncertainties · International Transactions on Electrical Energy Systems · 2021 · 25 citations
• Photovoltaic distributed generation integrated electrical distribution system for development of sustainable energy using reliability assessment indices and levelized cost of electricity · Environmental Progress & Sustainable Energy · 2022 · 23 citations
• Risk-Based Reliability and Cost Analysis of Utility Poles Subjected to Tornado Hazard · Journal of Aerospace Engineering · 2019 · 22 citations
• Analysis and short-term predictions of non-technical loss of electric power based on mixed effects models · Socio-Economic Planning Sciences · 2020 · 20 citations
• Prioritizing CBs Maintenance and Identifying Mandatory Maintenance at Higher Priorities · International Transactions on Electrical Energy Systems · 2022 · 19 citations
• Unsupervised machine learning techniques applied to composite reliability assessment of power systems · International Transactions on Electrical Energy Systems · 2021 · 16 citations
• Probabilistic Quantification of Distribution System Resilience for an Extreme Event · International Transactions on Electrical Energy Systems · 2022 · 15 citations
• Critical Risk Indicators (CRIs) for the electric power grid: a survey and discussion of interconnected effects · Environment Systems & Decisions · 2021 · 13 citations

Most recent work

• Impact of artificial intelligence-driven digital twins and lean six sigma-assisted power system asset management on long-term investment planning · Scientific Reports · 2026
• A Methodological Evaluation of Power-Distribution System Efficiency in Senegal: A Difference-in-Differences Analysis (2000–2026) · Zenodo (CERN European Organization for Nuclear Research) · 2026
• AI-Enabled Predictive Maintenance and Asset-Health Analytics for MV/LV Distribution Substations Using Commissioning-Test Data and Condition Monitoring · Zenodo (CERN European Organization for Nuclear Research) · 2026
• Data-Driven Approaches to Reliability Modeling in Critical Energy Infrastructure: Analytical Perspectives on Early Failure Detection · International Journal of Innovative Research in Engineering & Multidisciplinary Physical Sciences · 2026
• STRUCTURE, SEGMENT AND GRID RELIABILITIES: A PROPOSAL · International Journal of Engineering Technologies and Management Research · 2026
• AI Based Predictive Maintenance and Fault Detection of Solar Based Microgrids · Zenodo (CERN European Organization for Nuclear Research) · 2026
• Optimal Recloser Placement in Rural and Urban 11 kV Distribution Feeders of Nepal Based on Reliability and Cost–Benefit Analysis · Journal of Advanced College of Engineering and Management · 2026
• Optimization for Battery Energy Storage Configuration Considering the Impact of Planned and Unplanned Outages · DOAJ (DOAJ: Directory of Open Access Journals) · 2026
• A Resilience‐Based Reward‐Penalty Scheme for Hardening Electricity Distribution Assets Against High‐Impact, Low‐Probability Disasters · Energy Science & Engineering · 2026
• Reliability Modeling and Predictive Maintenance Integration for an Induced Draft Fan System Using Semi-Markov Process and Machine Learning-Based Predictive Models · Journal of Data Science and Intelligent Systems · 2026

Related topics in Engineering