Ice and Wind Are Breaking Power Grids: New Review Maps the Fight to Predict Tower Failures

When a severe ice storm sweeps across a mountain ridge, the tall lattice towers that carry high-voltage transmission lines become some of the most vulnerable structures in the modern world. Ice accumulates on conductors, adding enormous weight and changing the aerodynamic shape of the cables, while strong winds push against the iced surfaces with forces that can exceed anything the towers were designed to resist. When a single tower collapses, the sudden release of tension can drag down neighboring towers in a cascading domino effect, cutting electricity to entire regions. A new review article by Bingzhao Qiu, Yiqi Liu, and Mingfei Ban of Northeast Forestry University in Harbin, China, published as a preprint under discussion in the journal Mechanical Sciences, takes stock of the scientific toolbox engineers use to understand and predict these failures, and it arrives at a striking conclusion: no single method is sufficient on its own.

The review focuses on transmission tower-line systems, often abbreviated TTLS, which are not simple structures to analyze. A transmission line is a coupled system in which slender, flexible cables interact with comparatively stiff lattice towers through insulator strings and fittings. When ice forms unevenly on a conductor, or when wind causes the cables to oscillate, the resulting forces are transmitted dynamically to the towers, which can vibrate, buckle, or shed their loads suddenly. Dynamic response analysis, the discipline concerned with calculating how these structures move, deform, and fail under time-varying loads, is therefore indispensable for explaining failure mechanisms and for guiding disaster mitigation strategies. The authors organize the field into four broad families of methods: analytical and mechanics-based approaches, numerical simulation methods, stochastic and statistical techniques, and emerging methods built on artificial intelligence.

Analytical and mechanics-based methods represent the classical foundation of the field. These approaches rely on simplified physical models, such as treating a conductor as a catenary cable or representing a tower as a beam or truss framework, and then deriving equations of motion that can be solved to estimate stresses, deflections, and natural frequencies. Their great strength is transparency: because every assumption is explicit, engineers can trace exactly why a structure fails and how design changes alter its behavior. They are also computationally cheap, making them suitable for rapid design iterations and code-based verification. The review, however, emphasizes their limitations as well. Real ice accretion is irregular, wind fields are turbulent and spatially varying, and the coupling between towers and lines introduces nonlinearities that simplified hand calculations struggle to capture. When the geometry of ice shedding or the burst of force from a sudden cable release must be modeled in detail, analytical formulas alone run out of explanatory power.

Numerical simulation methods fill that gap. Finite element models can represent a full tower-line system in three dimensions, capturing the nonlinear behavior of cables, the plastic deformation of steel members, and the contact mechanics of ice breaking away from a conductor. The review highlights how such models allow researchers to reproduce catastrophic scenarios in a virtual laboratory, testing how a system responds to combinations of ice thickness, wind speed, and direction that would be dangerous or impossible to stage physically. Detailed analyses of the fundamental assumptions, solution procedures, and uncertainty modeling of these methods form a central part of the paper. Yet numerical simulation carries its own burdens. High-fidelity models demand enormous computational resources, and their predictions are only as good as the inputs they receive: the true shape of the ice, the statistical properties of the wind, and the condition of aging steel all introduce uncertainty that a deterministic model cannot resolve by itself.

This is where stochastic and statistical methods enter the picture. Rather than computing a single deterministic answer, these approaches treat loads, material properties, and structural parameters as random variables and estimate the probability of failure. Wind and ice are inherently stochastic phenomena, and historical records of storms, ice accretion events, and structural damage provide the empirical basis for probabilistic risk assessment. The review examines how these methods quantify uncertainty explicitly, allowing engineers to move beyond the question of whether a tower survives a nominal design load toward the more realistic question of how likely a failure is over the lifetime of the grid. The authors also note the methodological limitations of purely statistical techniques: they depend on the availability and quality of long-term data, and they may struggle to extrapolate to unprecedented storm conditions, precisely the situations that cause the most severe damage.

The fourth family, methods based on artificial intelligence, is the fastest-moving part of the field and a major reason this review is timely. Machine learning models trained on simulation outputs or monitoring data can predict structural responses far faster than a full finite element run, opening the door to near-real-time risk assessment during an unfolding storm. The authors describe AI approaches as showing increasing potential to enhance predictive performance and computational efficiency, a claim that resonates with a broader trend in structural engineering where surrogate models and data-driven classifiers complement physics-based simulation. Still, the review is careful to point out the caveats. AI models can behave as black boxes, their training data may not cover rare extreme events, and their predictions lack the physical interpretability that engineers need to certify safety-critical decisions. The authors treat AI as a powerful complement to mechanics-based understanding, not a replacement for it.

Perhaps the most consequential message of the review is the argument for hybrid strategies. Because each method family illuminates a different facet of the problem, combining them yields assessments that none could deliver alone. An analytical model can provide the physical grounding for a machine learning surrogate; a numerical simulation can generate the training data that a statistical model needs; a probabilistic framework can wrap the whole pipeline in a rigorous treatment of uncertainty. The authors show that this complementarity is not a luxury but a necessity for improving risk assessment, resilience, and future decision-making. In an era when climate change appears to be intensifying both ice storms and extreme wind events in many regions, the cost of relying on any single method is measured in blackouts and collapsed infrastructure.

Beyond cataloging methods, the review identifies key challenges in modeling, computation, and uncertainty quantification that the field has yet to solve. Modeling challenges include representing the complex, time-dependent process of ice accretion and shedding, and capturing the coupled dynamics of long spans with many towers. Computational challenges stem from the sheer scale of full-line simulations, which can involve thousands of degrees of freedom and highly nonlinear behavior over long durations. Uncertainty quantification remains perhaps the deepest difficulty, because the inputs that matter most, such as extreme ice loads and turbulent wind fields, are the hardest to measure and characterize. The authors propose future research directions aimed at intelligent and resilience-oriented strategies, envisioning systems that not only predict failures but support sustainable development of transmission infrastructure under increasingly hostile atmospheric conditions.

For grid operators, regulators, and the public, the stakes of this research are concrete. Power outages triggered by ice-wind hazards cascade into disruptions of heating, hospitals, communications, and water systems, and the economic losses can reach into the billions. The work of Qiu, Liu, and Ban provides a map of where the science stands and where it must go: toward methods that fuse physical insight with data-driven speed, quantify uncertainty honestly, and design grids that bend rather than break. As the discussion of this preprint opens to the research community, the review stands as both a synthesis of decades of effort and a blueprint for the resilient, intelligent power infrastructure the coming decades will demand.

Subject of Research: Dynamic response analysis methods for transmission tower-line systems under combined ice and wind hazards

Article Title: Review article: A review of dynamic response analysis methods for transmission tower-line systems under ice-wind hazards

Article References: Qiu, B., Liu, Y., & Ban, M. (2026). Review article: A review of dynamic response analysis methods for transmission tower-line systems under ice-wind hazards. https://doi.org/10.5194/ms-2026-169

Image Credits: AI Generated

DOI: 10.5194/ms-2026-169

Keywords: transmission towers, ice-wind hazards, dynamic response analysis, structural engineering, finite element simulation, uncertainty quantification, artificial intelligence, power grid resilience, risk assessment, conductor icing, wind loads, Mechanical Sciences