Source-linked AI summary
The Power Grid as a Complex Network: a Survey
Giuliano Andrea Pagani, Marco Aiello
TL;DR
The paper addresses how Complex Network Analysis can characterize Power Grid infrastructures and their reliability while complementing established electrical-engineering approaches. It surveys and compares studies using graph models, topological indicators, weighted representations, and resilience analyses, finding that CNA provides a general understanding while research moves toward richer network representations. The survey also emphasizes that CNA simplifies electrical complexity and depends on accurate Grid data.
Problem
Power Grid infrastructures require analysis of their topological properties and reliability, but their electrical complexity makes a common comparative view valuable.
Method
The paper surveys and compares Power Grid studies using Complex Network Analysis techniques, graph-related indicators, weighted representations, and resilience analyses.
Results
The survey finds that Complex Network Analysis provides a comprehensive general understanding of Power Grid properties, while studies increasingly represent network properties beyond a simple graph.
Takeaways & Limitations
CNA can help identify potentially critical Power Grid spots or aspects quickly before deeper analysis with traditional electrical-engineering tools.
Takeaways & Limitations
CNA is a high simplification of Power Systems complexity and requires precise, reliable information about Grid topology, components, energy flows, lines, and privately held infrastructure.
Abstract
from arXiv · showhide
The statistical tools of Complex Network Analysis are of great use to understand salient properties of complex systems, may these be natural or pertaining human engineered infrastructures. One of these that is receiving growing attention for its societal relevance is that of electricity distribution. In this paper, we present a survey of the most important scientific studies investigating the properties of different Power Grids infrastructures using Complex Network Analysis techniques and methodologies. We categorize and explore the most relevant literature works considering general topological properties, differences between the various graph-related indicators and reliability aspects.
1 Introduction
The paper frames the Power Grid as a socially essential complex infrastructure whose analysis spans multiple scientific disciplines. It surveys Complex Network Analysis studies to compare graph properties, indicators, and reliability aspects.
- Complex Network Analysis studies large networks that behave as complex systems and has been applied across many scientific and engineered domains.
- The Power Grid is studied because electricity transmission and distribution are essential to society, while Grid efficiency and operation affect the environment.
- The survey compares Power Grid studies using Complex Network Analysis techniques across parameters, graph metrics, and reliability-related properties.
2 Background and Survey Methodology
The background establishes graph concepts and Power Grid representations used throughout the survey, then motivates reliability analysis through node or edge disruptions. It also distinguishes topological descriptions from physical and electrical properties.
- The survey introduces graph definitions and uses a small example graph to illustrate each property’s application.
- A Power Grid graph represents substations, transformers, or consuming units as vertices, with edges denoting directly connected physical cables.
- Node degree, degree distributions, paths, distances, shortest paths, and betweenness provide measures of connectivity and node position in a graph.
- Power Grid degree distributions are typically exponential or power-law, differing in how quickly the probability of high-degree nodes decays.
- Topological centrality does not necessarily identify the nodes carrying the greatest electricity flow because real flows may not follow shortest paths.
- Reliability studies remove nodes or edges randomly or according to properties such as highest degree or highest betweenness to model failures and targeted attacks.
3 The Power Grid as a Complex Network
The survey reviews Complex Network Analysis studies of Power Grids, emphasizing the High Voltage Grid and cataloguing the principal works examined.
- The review focuses on the most important Complex Network Analysis studies of Power Grids, especially investigations of High Voltage networks.
3.1 Basic Power Grid characteristics
The surveyed studies cover real and synthetic Power Grid samples across voltage levels and geographic regions, using varied graph representations and dataset scales. Most studies analyze real infrastructures, while synthetic studies commonly use IEEE Bus models.
- The comparison records node and line counts, real versus synthetic samples, and whether studies examine transmission or distribution Grid layers.
- The studies use both unweighted and weighted graphs, with some weights representing physical line properties and others representing analytical quantities.
- The reviewed datasets span American, European, Chinese, Italian, Nordic, and Dutch Power Grid infrastructures, with both regional samples and broad interconnections.
- The survey includes High Voltage studies alongside Medium and Low Voltage analyses, including Dutch samples of about 4,200 and 700 nodes.
- Most surveyed data come from real electric infrastructures, whereas fewer studies use synthetic samples, usually based on IEEE Bus blueprints.
3.2 Statistical global graph properties
The survey compares global graph properties across Power Grid studies, including graph size, degree and path statistics, weighting, and geographic coverage. These studies commonly report low average degree and emphasize node-degree distributions, while weighted analyses and centrality measures remain less consistently used.
- Power Grid studies compare graph order, size, average degree, statistical analyses, weighting, and geographic or infrastructure coverage.Table 2 organizes these characteristics across the surveyed graph representations.
- Average node degree is generally between 2 and 3, reflecting physical, geographical, and economic constraints on substations and cables.
- Half of the studies examine node-degree distributions to identify the theoretical probability model associated with each Power Grid sample.
- Almost half of the studies analyze path length, usually as a step toward evaluating small-world characteristics.
- Topological studies generally use unweighted edges and node-degree distributions, whereas weighted studies usually omit degree-distribution statistics.The survey identifies centrality measures, especially in weighted graphs with flow or capacity information, as an area deserving more attention.
3.3 The small-world property
The surveyed studies do not reach a common conclusion about whether Power Grids are small-world networks. Findings vary by sample and voltage level, with some grids satisfying small-world criteria and others differing from random-graph expectations.
- Ten of thirty-two studies investigate the small-world property using path-length and clustering comparisons.
- The Nordic Grid has twice the random graph’s average path length but an almost one-order-of-magnitude larger clustering coefficient, supporting small-world classification.
- Most European Grid samples satisfy the small-world conditions when compared with random graphs.
- The Western United States Power Grid is reported as small-world despite sparse connectivity, a small clustering coefficient, and a large characteristic path length.About 80% of its edges are identified as shortcuts.
- Overall, small-world membership is sample-specific; Medium and Low Voltage networks appear farther from the property than many High Voltage samples.The survey also reports that some Chinese and High Voltage samples satisfy the conditions, while other tested grids do not.
3.4 Node degree distribution
The survey finds that Power Grid node-degree distributions do not fully agree across studies but are generally close to exponential decay, especially for High Voltage networks. Incorporating electrical impedance can change the apparent topology by emphasizing hubs.
- Node-degree probability distributions summarize the prevalence of highly connected nodes and are fitted to classes of curves in the surveyed studies.
- The surveyed results do not fully agree on one distribution, but Power Grid node-degree distributions are generally close to exponential decay.Figures 9 and 10 show fitted cumulative distributions from Table 3.
- The Western U.S. and Nordic Grids both appear to follow exponential node-degree distributions.
- In the IEEE 300-Bus blueprint, impedance-weighted shortest paths produce a graph dominated by few hubs, suggesting a scale-free topology under electrical connectivity.The weighted reconstruction uses the 411 node-to-node connections with smallest impedance.
- Node-degree centrality distributions in one study resemble a power-law, including when electrical parameters are incorporated.
- High Voltage studies generally report exponential or exponential-based degree distributions, while Medium and Low Voltage Grids remain comparatively understudied.The survey calls for deeper investigation across different geographic areas.
3.5 Betweenness distribution
Betweenness studies generally suggest a Power-law distribution, but evidence varies by voltage level and remains too limited for a definitive conclusion. Weighted-grid and power-flow analyses are identified as important gaps.
- Betweenness measures how many shortest paths traverse a node, indicating its importance in network path management.
- High Voltage networks tend toward Power-law betweenness distributions, whereas Medium and Low Voltage networks show mixed Power-law and exponential-decay patterns.
- Weighted analysis suggests that a few nodes support much of the weighted paths while most nodes are only slightly involved.
- 80% overlap appears among the 10 highest-betweenness nodes when comparing electrical-parameter and purely topological analyses.
- Power-law behavior appears dominant, but more studies are needed, especially for weighted Grid models and power-flow-based betweenness.
3.6 Resilience analysis
Resilience studies assess how Power Grid connectivity and service degrade under random failures or targeted attacks, often using topological and electrical metrics. Across surveyed works, grids generally tolerate random removals but are substantially more vulnerable when failures target highly connected, central, or heavily loaded components.
- Resilience analysis: Resilience analysis primarily measures connectivity or the ability to preserve efficient paths after nodes or edges are removed.Studies vary the attack target and reliability metric, and some also evaluate mitigation strategies.
- Node based attack analysis: Random failures usually produce gradual degradation, whereas targeted attacks can trigger threshold-like disruption after removing a small fraction of critical nodes.In one study, removing 2% of high-betweenness substations caused approximately 60% connectivity loss.
- Node based attack analysis: Highest-degree attacks fragment spanning or giant components sooner than random removals, with critical thresholds influenced by the network’s node-degree distribution.This pattern was reported across synthetic IEEE Bus models and the NYISO and WSCC real networks.
- Node and edge based attack analysis: Cascading-failure studies examine how overload redistribution and component tolerance affect load loss, power degradation, and the number of steps required to reach a new equilibrium.One experiment initiates redistribution by bringing the highest-load line near its trip point before overloading it.
- Node and edge based attack analysis: Several studies compare topological vulnerability measures with electrical or load-aware models that incorporate power flow, capacity, reactance, or physical line properties.These approaches include DC power-flow models, weighted centrality, electrical betweenness, efficiency, net-ability, overload, and power-degradation metrics.
3.7 Further studies
The survey highlights studies that extend Power Grid analysis beyond basic topology, examining system dynamics, reliability causes, weighted electrical properties, and cascading-failure mitigation. These works broaden the analysis toward physical behavior, operational data, and interventions.
- Additional analyses: Mei et al. model the electrical system as a self-organized critical Complex System, including blackout dynamics, power-flow redistribution, and High Voltage Grid evolution.The study extends beyond pure network analysis by addressing dynamical and growth processes.
- Additional analyses: About 5% of European Grid failures are attributed to line overload, while about 30% are classified as unknown.External causes account for almost 30% and other failures for almost 40%.
- Additional analyses: Wang et al. represent electrical properties through an admittance matrix combining network adjacency and line impedances.Line impedances show heavy-tailed behavior best fitted by a double Pareto lognormal distribution with exponential tail cutoff.
- Additional analyses: Weighted Medium and Low Voltage analysis distinguishes losses from reliability and redundancy, identifying samples more appealing for distributed energy exchange markets.The comparison reports more nodes traversed by weighted shortest paths than by unweighted paths.
- Additional analyses: Percolation-based cascading-failure modeling derives analytical conditions under a random geometric graph assumption without applying the model to real or synthetic Grids.The work is primarily theoretical rather than an empirical Grid study.
- Additional analyses: Pahwa et al. propose load reduction, targeted tree-based load reduction, and cluster isolation as mitigation strategies for cascading effects.The reported results are encouraging for preserving nodes outside the cascade when load is slightly reduced preemptively.
- Additional analyses: Other CNA studies use the Power Grid mainly as an example, including analyses of exponential degree decay and small-world behavior.These examples include the Southern California and Western States Power Grids.
4 Discussion and Conclusion
The survey concludes that CNA provides a broad graph-based view of Power Grid topology and reliability, while increasingly incorporating weights and electrical properties. Its usefulness is complementary rather than substitutive, and depends on broader geographic coverage and precise Grid data.
- Discussion and Conclusion: CNA offers a comprehensive, general understanding of Power Grid properties, but simplified graph models do not capture all electrical and electrotechnical complexities.The literature trends from unweighted graphs toward representations incorporating electrical properties.
- Discussion and Conclusion: CNA does not replace traditional Power Systems resilience and safety analysis, but can quickly identify critical spots for deeper electrical-engineering investigation.Its role is a simplified general screening perspective rather than a complete substitute.
- Discussion and Conclusion: CNA is emerging as a tool for designing and evolving Grid topologies, particularly for reshaping Medium and Low Voltage networks in future Smart Grids.The survey also connects this direction with topology optimization and distribution-cost analysis.
- Discussion and Conclusion: Most reviewed samples concern High Voltage networks in American, European, or Chinese Grids, while betweenness often suggests a few highly central nodes support many shortest paths.Reported conclusions about small-world behavior remain contrasting across studies.
- Discussion and Conclusion: CNA results show similarity to traditional electrical-engineering results, linking theoretical analyses with measured quantities in real environments.The survey presents this similarity as important for CNA accuracy.
- Discussion and Conclusion: More useful physically informed CNA models require precise and reliable information about Grid topology, components, energy flows, lines, and privately held infrastructure.Without adequate data, the survey states that actual vulnerability points cannot be discovered reliably.
- Discussion and Conclusion: Complex Network Analysis can support understanding and better design of natural and human-generated networked systems, including Power Grids.The paper frames CNA as a modeling technique providing analytical methods and metrics for complex systems.
- Discussion and Conclusion: More studies are needed across Asia, South America, and Medium and Low Voltage networks to broaden the evidence beyond the geographies and layers most often analyzed.Such studies may reveal different topologies and inform future Smart Grid design.