Black ice remains a difficult pavement hazard because it is thin, transient and visually elusive, yet it can rapidly compromise surface friction. This study develops a microencapsulated phase change asphalt mixture (MPCAM) for delaying black ice formation and, more importantly, establishes electrical conductivity monitoring as a quantitative means of tracking the freezing of thin surface water films. A microencapsulated phase change material (MPCM), composed of an n-tetradecane core and SiO₂ shell, was synthesised, characterised and incorporated into SMA-13 asphalt mixture at 0%, 1.0%, 1.5% and 2.0% by mass of the total mixture. The thermal response, freezing behaviour, ice-pavement adhesion strength and skid resistance of the mixtures were evaluated. During cooling from 20 °C to −15 °C, the mixture containing 2.0% MPCM maintained an internal temperature up to 5.66 °C higher than the control, indicating effective latent-heat release during phase change. The conductivity response captured the abrupt loss of liquid-water continuity during freezing, enabling both the onset and completion of black ice formation to be identified. The ability to distinguish these two freezing stages is an important methodological feature of the study. The maximum freezing delay reached 95 min for a 1 mm water film and 100 min for a 2 mm water film. MPCM also reduced ice-pavement adhesion and improved skid resistance under icy conditions. These findings show that MPCAM is best understood not as a material for preventing sustained icing, but as a passive pavement technology for delaying short-duration black ice formation around the freezing point.
Ice lens development remains a fundamental problem in frost-heave research, yet most existing criteria treat initiation and growth as a single coupled process and rarely account for the distinct boundary conditions under which soil freezes. This study addresses this gap by explicitly distinguishing between closed and open freezing modes. Under closed freezing conditions, high-pressure freezing experiments show that the relationship between pressure and freezing temperature is consistent with the generalized Clausius-Clapeyron equation when the system is close to equilibrium. Under open freezing conditions, the coupling between the liquid-water-fraction gradient within the frozen fringe and the pore water pressure associated with phase change yields a heterogeneous interfacial body force. Integration of this body force across the interfacial transition zone yields an equivalent interfacial stress pwΔn, which is determined by the pore water pressure pw and the interface asymmetry Δn. Ice lens initiation occurs when this equivalent interfacial stress exceeds the overburden pressure. A relaxation-based framework further defines the applicability of the criterion in terms of the dimensionless ratio tres/τ, where tres is the characteristic residence time within the frozen fringe and τ is the characteristic relaxation time for interface-asymmetry development. This ratio delineates two freezing regimes: through-going ice lens formation and kinetically suppressed interface development. The proposed framework decouples ice lens initiation from growth and treats them as two mechanistically distinct sub-processes. Model predictions are in good agreement with experimental data, supporting the proposed distinction between ice lens initiation and growth.
The lining of water-conveyance canals in northern China is vulnerable to frost heave in winter, and the deformation of frozen foundation soil significantly affects the stability and mechanical behavior of the lining structure. To overcome the limitations of the Winkler frozen-soil beam model, which neglects the interaction between adjacent soil springs and requires assumptions regarding tangential freezing forces— this study introduces a Pasternak shear layer to characterize the interaction between frozen soil and canal lining. A Pasternak frozen-soil foundation beam model is developed for trapezoidal arched-bottom canals. Using an arched-bottom trapezoidal section in the Jingtai irrigation area (Gansu Province) as a prototype, frost heave displacements along the lining were calculated and compared with those obtained from the engineering mechanics model, the Winkler foundation beam model, and field monitoring. Key findings include: (1) when tangential and normal frost heave and normal freezing are jointly considered, the Pasternak model proves more suitable for practical frost-resistant design in safely configured canals; (2) rising groundwater levels increase frost heave loads, causing stress and displacement along the slope slab to first increase, then decrease, and rise again toward the canal bed, with peak effects occurring at the slope slab center and the arch-bottom lining; (3) the field validation results indicate that the Winkler model overestimated maximum frost heave by 56.5% relative to monitoring, the Pasternak model underestimated it by 4.6%, and engineering-mechanics calculations underestimated it by 63.7%. Overall, the Pasternak model aligns best with observed deformations, providing a theoretical foundation for frost-resistant canal design in cold regions.
Freeze-thaw cycles (FTC) progressively degrade rock mechanical properties and reduce dynamic strength, posing challenges for the safe design of geotechnical structures in cold regions. However, limited datasets constrain the reliability of machine learning (ML) applications in such conditions. This study proposes a data-centric probabilistic ML framework for predicting the dynamic strength of sandstone under FTC. Four data augmentation techniques-SMOTE, Copula-based Latin hypercube sampling, Gaussian noise injection, and polynomial perturbation together with bootstrap resampling are employed to expand a 216-sample dataset to 1000 samples, while a correlation-based RMSE criterion ensures preservation of geomechanical relationships. Gaussian process regression with automatic relevance determination kernels and Bayesian optimization is used for probabilistic prediction. Results show that bootstrap (RMSE = 0.00439) and Gaussian noise (RMSE = 0.00546) best preserve inter-feature dependencies and achieve superior performance, with test R2 of 0.9919 and 0.9891, respectively. Interpretability analyses indicate that stress variables (impact pressure and confining pressure) dominate rock strength, while FTC and hydraulic properties govern damage evolution. The results demonstrate a data-centric and uncertainty-aware modeling strategy for predicting sandstone behavior under data-scarce conditions.
Electrically heated concrete pavements offer an active approach for snow/ice removal in cold regions, yet practical deployment requires heating networks that are both energy efficient and able to mitigate cold spots. This study investigates electrically heated concrete pavements incorporating a carbon-glass fiber geogrid, focusing on how carbon-fiber bundle-activation layouts govern surface warming, temperature uniformity, and energy demand. Two nominally identical 30 × 30 × 10 cm pavement prototype slabs were fabricated with the geogrid at a 5-cm depth. Six activation layouts (0-skip to 5-skip) were designed to produce nominal energized-bundle spacings of 1.9–11.4 cm. Temperature-rise tests were conducted at room temperature under a similar initial power density of around 500 W/m2 to quantify the heating rate and time, the maximum transverse temperature difference, and the cumulative electrical energy to reach target average temperature rises (ΔT¯) of 5 °C and 10 °C. Results show that heating performance is strongly layout-dependent: moderate spacing accelerated early-stage warming and reduced energy-to-target relative to full activation (0-skip), whereas large spacing increased hot-cold segregation and cold-spot risk. The strictly periodic 3-skip zone (7.6-cm spacing) yielded zone-based cumulative electrical energy per unit area values that were 46.3% and 51.4% lower than those of the 0-skip global baseline at ΔT¯= 5 °C and 10 °C, respectively. For the 3-skip layout, a locally enlarged gap affected global temperature uniformity and energy-to-target metrics, highlighting the maximum effective spacing as a key design control. An outdoor test demonstrated snow-melting feasibility: with the 3-skip layout, a 275 g (or 5 cm thick, assuming 0.06 g/cm3) snow layer was fully melted within 3.25 h under ambient air temperatures of −7.8 °C to −2.8 °C, 0 m/s wind, and predominantly cloudy skies without direct solar radiation. Across all tests, the electrical response was stable, with the power density varying by less than 10.9% over time during each constant-voltage heating run. These findings provide layout-oriented guidance for energy-efficient winter roadway maintenance.
Galloping of ice-coated transmission lines is a large-amplitude, self-excited oscillation that threatens grid reliability. This paper proposes a novel inerter-based hybrid damper that distinguishes itself from conventional devices through a synergistic integration of a tuned mass damper (TMD) and an inerter-based tuned viscous mass damper (TVMD) spacer. Unlike existing solutions that address either absolute or relative motion separately, this dual-action design simultaneously suppresses both the overall vibration of the conductors and their inter-phase differential motion, thereby achieving superior control performance. A comprehensive nonlinear aeroelastic-mechanical coupled model is first established, capturing the vertical-horizontal-torsional coupling three-degree-of-freedom dynamics of the conductors and the nonlinearities from both aerodynamic forces and damper kinematics. The optimal parameters of the hybrid damper are then derived analytically and validated numerically via a linearized framework, providing explicit design formulae. The Harmonic Balance Method (HBM) is subsequently employed to characterize the nonlinear vertical galloping response, confirming that the linearly optimized damper effectively suppresses bifurcation and limits oscillation amplitudes. Finally, fully coupled nonlinear time-history analyses demonstrate the damper's robust performance across varying wind speeds and span lengths, showing its capability to mitigate not only simple periodic galloping but also complex, multi-modal oscillations, including incipient chaotic motions in long-span configurations. The study concludes that the proposed hybrid damper offers a theoretically sound and practically viable solution for superior galloping control, establishing a solid foundation for the development of next-generation anti-galloping devices with enhanced adaptability and reliability for modern power transmission infrastructure.
This study investigates the straight-line resistance of a following ship navigating in an icebreaker-created, pre-formed narrow brash ice channel. Field observations and ice-basin tests provided channel-condition data and resistance measurements for assessing the numerical model. In the ice basin, a model icebreaker first generated the narrow brash ice channel, after which the resistance of a following ship was measured within the pre-formed channel. A corresponding numerical ice basin was then developed using a truncated ship model and DEM-based brash ice particles, and two-way coupled CFD–DEM simulations were conducted to reproduce the ship–ice–water interaction process. The simulations captured the main observed brash-ice motion patterns, including bow accumulation, particle rotation, and lateral clearing. The predicted resistance showed reasonable agreement with the ice-basin measurements within the tested speed range, indicating that the model can reproduce the dominant resistance trend under the present experimental conditions. The assessed model was further used to examine brash-ice motion, channel contraction, and sectional hull loads. Finally, a comparative analysis of narrow brash ice channel and open brash ice area was conducted to evalute the influence of confinement on the ice resistance and performance of following ships. The proposed experimental–numerical framework provides physical evidence and mechanistic insight for future studies, particularly those concerned with resistance estimation and hull-load assessment of following ships operating in Arctic ice-covered waters.
Snow depth is a key variable for characterizing snowpack conditions and cold-region hydrological processes. Passive microwave remote sensing provides an important basis for continuous large-scale snow depth monitoring in the Northern Hemisphere. Complex land-surface conditions, variations in snowpack state, and microwave signal saturation under deep-snow conditions create nonlinear and heterogeneous brightness-temperature–snow-depth relationships. Conventional machine-learning models can improve global nonlinear fitting relative to empirical algorithms, but they generally learn a unified mapping shared across all samples. Such models therefore do not explicitly establish specialized local mappings for heterogeneous sample subspaces. To address this limitation, this study develops an AMSR2 snow depth retrieval model based on a Mixture-of-Experts (MoE) network. The model uses multi-frequency dual-polarization brightness temperatures as primary inputs, integrates topography, forest fraction, and snow density, and dynamically assigns expert weights through a routing network. Based on GHCN-Daily station observations from 2015 to 2024, station-independent validation was conducted against RF, XGBoost, MLP, and the JAXA AMSR2 snow depth product. The MoE model achieved an overall RMSE of 12.56 cm, MAE of 8.44 cm, R of 0.83, and Bias of 0.07 cm. Compared with RF, XGBoost, MLP, and the JAXA product, it reduced RMSE by 12.2%, 8.7%, 3.5%, and 47.8%, respectively. MoE also improved error control in the 50–100 cm and >100 cm snow depth ranges and under typical land-surface conditions. These results indicate that MoE improves retrieval adaptability under spatial heterogeneity, although deep-snow underestimation remains.
Natural snowflakes exhibit diverse morphologies (e.g., plate, aggregate, rosette) that coexist during snowfall with dynamically varying proportions. Current numerical simulations of roof snow loads typically simplify snow particles as homogeneous spheres, thereby neglecting the gradation effect of snow crystal morphology and leading to considerable prediction errors. To address this limitation, this paper proposes a graded snow-phase modeling approach within the Eulerian-Eulerian framework. The snow phase is divided into multiple sub-phases, each assigned a distinct drag coefficient corresponding to a specific snowflake morphology. Coupling among sub-phases is achieved through volume fraction constraints and momentum exchange with the air phase. The proposed model is validated against field measurements on a stepped roof, where four snowflake morphologies and their proportions were recorded. Results show that the model predictions agree well with the measurements, whereas the conventional spherical assumption significantly overestimates snow depth in the wake region. Parametric analysis reveals that the mean snow depth is most sensitive to the proportion of plate-shaped particles (20.6% change per 25% variation), followed by aggregate (16.5%) and rosette (8.4%). To reduce computational complexity, equivalent spherical diameters are derived based on the principle of equal terminal settling velocity (plate: 0.264 mm, aggregate: 0.300 mm, rosette: 0.380 mm). Simulations employing these equivalent diameters yield snow depth distributions consistent with those from the graded model. This study provides a physically refined yet computationally efficient method for predicting roof snow loads.
Underground garage shafts are recognized as vital for the alleviation of surface space constraints in megacities; however, their construction in water-rich soft soils faces challenges from water inflow and ground deformation. Substantial potential for the resolution of these issues is offered by artificial ground freezing, whereby a temporary impermeable frozen curtain is formed. However, the thermal effects of the surface environment, underwater convection, cement hydration heat, and freezing pipe layout across the complete construction cycle remain unexplored. A three-dimensional temperature field model for circular and square garage shafts is developed within this study, wherein these four thermal factors are comprehensively integrated throughout the freezing, excavation, and pouring stages. The reliability of the proposed model is subsequently verified through the execution of mesh independence tests and validation against classic analytical solutions. Results show that the shallow temperature field is influenced by surface thermal effects to about 2 m, after which temperature variation sharply decreases. Due to convective heat transfer, underwater excavation deteriorates the frozen wall, with significant influence ranges of 1.16 m (circular) and 1.7 m (square). During concrete pouring, a greater melting range of the frozen wall in the square garage shaft was observed, and the melting peak was delayed. In the circular shaft, hydration heat reaches 79.7 °C, melting 0.056 m of frozen wall. After pouring, lateral frozen walls connect fully within 11 days, forming a closed curtain wall in 22 days. This work provides a theoretical basis for freezing design and temperature field prediction in saturated soft soils.
Recent discourses point to ultra-low-density lightweight cellular concrete (LCC) as a potential subbase alternative for pavements constructed in cold regions. However, limited field-validated evidence exists regarding its hydraulic transport behavior and freeze-thaw durability under realistic exposure conditions. This study investigates the hydraulic and durability performance of ultra-low-density LCC (400, 475, and 600 kg/m3) through a coordinated laboratory and field program in Ontario, Canada, where pavements experience approximately 120–140 annual freeze-thaw cycles. Laboratory testing quantified water absorption, sorptivity, permeability, and resistance to freeze-thaw cycling, while field-instrumented test sections monitored in-situ soil water potential over up to three years. Results show that freeze-thaw resistance and moisture transport are strongly governed by density. The 400 kg/m3 mixture exhibited high absorption (up to 33%) and permeability (K = 0.035 cm/s) and failed durability criteria within 45 cycles. In contrast, 475 and 600 kg/m3 mixtures maintained relative dynamic modulus values above 70% after 180 cycles and demonstrated substantially reduced permeability (K ≤ 0.012 cm/s). Field monitoring revealed a density-dependent hydraulic regime. Higher-density LCC functioned as a vertical moisture barrier, maintaining drier subgrade conditions but retaining moisture within upper layers, whereas lower-density LCC facilitated drainage but provided reduced subgrade protection. Spring thaw was identified as the critical moisture event governing long-term saturation. The findings suggest a consistent relationship between laboratory-measured hydraulic properties and in-service moisture behavior, highlighting a fundamental design trade-off between drainage capacity and subgrade protection. These findings support performance-based selection of LCC density for pavement applications in cold climates, while noting that the results are based on materials from a single supplier and specific mix designs, warranting cautious generalization to other LCC formulations.
Underwater vehicles experience complicated interactions with fluid and ice as well as vehement nonlinear dynamic responses while breaking ice. To address these issues, a numerical model was developed to simulate underwater vehicles breaking ice, utilizing an adaptive coupling algorithm that integrates the finite element method (FEM) with smoothed particle hydrodynamics (SPH). Simulations of ice ball impact and wedge body water entry were conducted to validate the computational method's effectiveness for key values. An experimental setup was developed to evaluate vehicles with hemispherical, 60° spherical cone, and 120° spherical cone heads breaking ice at high speeds under varying ejection pressures. The study employed dimensional analysis to develop a dimensionless formula for ice-breaking by underwater vehicles, and identified the critical geometric, material and physical dimensionless parameters that govern the ice-breaking performance of vehicles. The study examined how dimensionless shell thickness and ice-breaking angle affect the vehicle's motion and load characteristics. As shown in the results, the initial velocity of the vehicle went up but the velocity loss rate became lower with the increase of ejection pressure. For a vehicle with a 120° spherical cone head, the velocity loss rates at ejection pressures of 3 MPa, 4 MPa, and 5 MPa were 30.79%, 22.55%, and 18.18%, respectively. The vehicle's velocity loss rate inversely correlated with both the dimensionless shell thickness and the ice-breaking angle. The stress extremum was negatively correlated with the dimensionless shell thickness. Increasing the dimensionless shell thickness from 0.001 to 0.009 reduced the vehicle's stress extremum from 2.46 × 108 Pa to 3.91 × 107 Pa.
