An Experimental Validation-Based Study of Airport Pavement Icing Mechanisms in Saline Environments and the Development of a Simplified Prediction Model

Abstract

Runway icing presents a significant challenge to aviation safety, especially in saline environments, where comprehending the icing mechanisms and predicting the icing onset are crucial for efficient airport operations. This study developed a specialized experimental apparatus to examine the mechanisms of airport pavement icing under controlled conditions. The apparatus, comprising an environmental chamber, a data acquisition system, and a scaled pavement structure, allowed for detailed simulations of various environmental factors. The experiments specifically examined the effects of the air temperature (−3 °C to −20 °C), wind speed (2 m/s to 6 m/s), and deicing salt concentration (0% to 80%) on the icing process. The results demonstrated that higher wind speeds and lower temperatures significantly accelerated the pavement surface cooling, leading to earlier icing onset. Under the most extreme conditions, the pavement reached critical icing temperatures within 15 min. In contrast, higher deicing salt concentrations delayed the icing onset by up to 67 min and 33 s at an 80% concentration, effectively lowering the pavement surface temperature. A comparison of the experimental data with the theoretical predictions showed initial consistency, although the discrepancies increased over time. This study culminated in the development of a simplified prediction model, which was validated against the experimental results, offering a practical tool for airport operators to manage runway safety during winter conditions.

1. Introduction

The increase in extreme weather events, largely driven by global climate change, has intensified the issue of airport pavement icing. This problem presents serious challenges to aviation safety, as icy runways increase the risk of flight delays, cancellations, and significant safety hazards for both passengers and crew. As a result, understanding the mechanisms of airport pavement icing and developing reliable prediction tools are essential for maintaining safety and operational efficiency in airport activities [1].

Unlike ordinary roads, airport pavements are more vulnerable to complex icing conditions due to their exposure to open areas. These conditions are significantly affected by factors such as the wind speed, temperature, and humidity. Moreover, the use of deicing agents alters the thermal properties of the pavement, making the icing process more unpredictable and further complicating the accuracy of existing prediction models. Given these complexities, research focusing on airport pavement icing is imperative [2]. The need for efficient and simplified icing prediction tools is particularly urgent due to the operational challenges airport operators face during adverse weather conditions. The existing tools often underperform in the saline environments typical of airports, especially due to the presence of deicing agents. This highlights the importance of rigorously validated, experimentally derived, simplified prediction models.

In recent years, data-driven approaches have shown promise in predicting road icing, particularly through the use of machine learning models that correlate meteorological factors—such as the temperature, humidity, precipitation, radiation, and wind speed—with the pavement temperature or icing conditions. Simple machine learning models were employed to establish this mapping relationship, generally based on basic models such as Multilayer Perceptron (MLP) [3,4] and Decision Tree [5]. However, these models often capture only the direct relationships between inputs and outputs, neglecting the temporal dependencies and sequential patterns that are crucial for accurate prediction [6]. The icing process, characterized by fluctuations in the surface temperature and ice thickness, clearly exhibits temporal dynamics. To address these issues, more advanced models such as Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks [7] have been widely adopted. These models can retain historical data within the network, enabling more effective processing of sequential data and a better understanding of temporal dependencies [8]. In recent years, enhanced variants of LSTM networks, such as Convolutional Long Short-Term Memory (Conv-LSTM) [9] and Sequence-to-Sequence (Seq2Seq) models [10], have gained popularity due to their ability to more effectively capture the spatial features of data. Despite these advances, the effectiveness of data-driven methods remains limited by the scarcity of high-quality data, particularly in diverse airport scenarios. These methods are more commonly applied to road icing prediction problems, as highways enable the collection of large amounts of data [11,12]. However, the reliance on extensive datasets—which are often unavailable for specific airport conditions—presents a significant challenge for accurate decision-making by airport operators [13].

On the other hand, theoretical analysis methods provide crucial insights into the physical mechanisms of icing through heat transfer and phase change theories. Basic models concentrate on the phase transitions between water and ice under specific meteorological conditions, typically using input data such as the meteorological conditions and the properties of road materials. The outputs, including the surface temperature and ice thickness, are often derived using finite element or finite difference methods [2,14]. An essential component of these models is the inclusion of latent heat during phase transitions, which reflects the impact of water–ice phase changes on the temperature field [15,16]. Some researchers have extended these models by incorporating the effects of moving vehicles or aircraft on the icing process, introducing factors such as wake-induced heat and water–ice dispersion [17,18]. However, these theoretical models often rely on idealized assumptions and typically lack validation and optimization using real-world data. Without experimental data obtained under controlled conditions, the accuracy and practical applicability of these models remain limited.

This study entailed the design and construction of an experimental apparatus to simulate the airport pavement icing process under actual saline environmental conditions. The theoretical models were validated and optimized using this experimental setup, enabling the analysis of the icing patterns under various environmental conditions. Based on these findings, a practical and simplified prediction formula was developed to assist airport operators in making prompt decisions during winter operations. The significance of this work lies in its potential to optimize the use of deicing salts and enhance aviation safety.

2. Design and Development of Experimental Apparatus

2.1. Environmental Chamber Design

The environmental chamber serves as the core component, which is specifically designed to control the temperature and humidity variations during the experimental process. It can precisely regulate the internal temperatures with fluctuations of less than ±0.5 °C and maintain the temperature uniformity within ±2.0 °C, while the humidity control remains within ±2.0%. Additionally, the chamber can precisely control the rate of temperature change and the duration for which the temperatures are maintained, effectively simulating the complex climatic conditions encountered in real-world environments. The sidewall of the chamber is equipped with sensor transmission interfaces, and a monitoring camera is installed overhead to observe the icing status in real time. This setup allows for accurate determination of the icing times without affecting the experimental temperature conditions. Figure 1 illustrates the icing simulation experimental apparatus system.

2.2. Data Acquisition Cabinet and Sensor System

The data acquisition cabinet, which is equipped with meteorological and temperature demodulators, serial ports, gateway interfaces, and power supplies, is responsible for collecting and processing the experimental data. It connects to the environmental chamber via data transmission cables, enabling unified collection of the surface temperature, humidity, wind speed, and other data from the structure within the chamber. The collected data are then transmitted to a computer through the gateway and stored in text format using serial port software. Figure 2 illustrates the data acquisition chassis.

2.3. Pavement Structural Body

The pavement structural body, a key component of the experiment, is designed as a full-scale cubic block. It represents a typical three-layer airport runway structure, based on commonly used design dimensions and materials for each layer. This structure, as shown in Figure 3, accurately reflects the multilayer pavement structure of airport runways. The structure consists of three layers: the upper part is a 480 mm × 480 mm × 380 mm C40 concrete slab, the middle layer is a 480 mm × 480 mm × 300 mm stabilized material layer, and the lower layer is a 480 mm × 480 mm × 200 mm soil base layer. To minimize the lateral temperature variations, the structural body is wrapped with a 50 mm thick insulation layer and a 10 mm thick PE board, as well as an additional 40 mm thick insulation board. A filter mesh, stainless steel plate, water tank, and heating film are installed beneath the soil base layer to simulate the thermal conduction environment under an actual airport pavement. To study the icing issues of other pavement structures, such as roads, bridges, or dock pavements, one only needs to create a corresponding structural body based on the specific dimensions and materials of those structures.

2.4. Environmental Simulation

To accurately simulate actual airport pavement conditions, the environmental simulation incorporates two key factors: wind speed and solar radiation.

Wind Speed Control: Adjustable wind speed fans are installed above the surface of the pavement structural body to replicate the effects of wind on the icing process. The wind speed can be varied from 0 to 10 m/s. The fans are equipped with transverse deflection plates to ensure an even distribution of the airflow, effectively simulating the natural convection conditions that influence the icing dynamics.

Solar Radiation Simulation: To simulate the effect of solar radiation, a set of infrared lamps is positioned above the pavement structural body. These lamps emit controlled radiation to mimic the sunlight exposure at different times of the day, allowing the experiment to consider the thermal impact of solar radiation on the icing process. The radiation intensity is adjustable to reflect varying solar conditions, ensuring a thorough analysis of the combined effects of the wind speed and solar radiation on pavement icing.

2.5. Data Collection and Processing

Typically, the detection and data collection in terms of pavement structures utilize embedded sensors [19]; however, our research primarily employs remote sensing methods. In our experiment, the data collection devices include platinum resistance temperature sensors, temperature and humidity sensors, pavement condition remote sensing sensors, and monitoring cameras. The sensors transmit the measured temperature, humidity, wind speed, and pavement condition data to the data center via the data acquisition cabinet, where they are processed into usable data by the demodulator. The processed data are then transmitted to a computer database via the network for subsequent analysis and processing. Figure 4 illustrates the environmental simulation equipment and the pavement surface condition acquisition devices.

2.6. Experimental Procedure and Control

The experiment uses a factorial design with 88 different combinations of independent variables—including the amount of deicing salt, wind speed, air temperature, and solar radiation—to simulate the icing process under various environmental conditions. The combinations are systematically varied to reveal the icing characteristics of airport pavements under saline conditions. Prior to the experiment, the pavement structural body is preconditioned in the environmental chamber to ensure that the surface temperature does not compromise the authenticity of the data. During the experiment, the temperature and humidity conditions within the chamber are carefully controlled, along with the simulation of the wind speed, solar radiation, and precipitation. The data collection process includes recording and analyzing the icing time, ice thickness, and temperature profiles across the different layers of the pavement.

2.7. Experimental Error Control

To ensure the reliability of the experimental data, several methods are employed to minimize experimental errors. These methods include using uniform materials, maintaining consistent cooling conditions within the chamber, having the same personnel assess the icing conditions, and conducting the experiment under standard atmospheric pressure.

3. Pavement Heat Transfer Theoretical Model and Solution

3.1. Theoretical Model Derivation

This study analyzes the heat transfer process in airport pavement structures in saline environments. As shown in Figure 5, the pavement system is considered an integrated heat transfer system, with the heat transfer process primarily comprising two parts: heat exchange between the pavement and the external environment, and internal heat conduction within the pavement.

① Heat Exchange between Pavement and External Environment: The heat exchange between the pavement surface and the surrounding environment occurs in three main forms: solar radiation, convective heat transfer, and latent heat of phase change. By deriving the heat balance equation, we can quantitatively describe the effects of these heat exchange mechanisms on the pavement temperature field. For instance, the solar radiation formula calculates the net radiative heat absorbed by the pavement, while the convective heat transfer formula accounts for the influence of the wind speed and air temperature on the pavement temperature. The latent heat of the phase change involves the heat released or absorbed during the water–ice–salt phase transitions, which significantly affects the temperature variation of the pavement surface. Drawing on the studies by Chen and Herb [14,20], the corresponding calculation formulas are as follows:

• The net solar radiation Q_sun received by the runway is calculated as Equation (1):

$$Q_{sun}=Q_{sun-in}-Q_{sun-out}=\left(1-{\alpha }_{ice}\right)\left(1-{\alpha }_p\right)q_{sun}-\epsilon \sigma T_s^4$$

(1)

• The convective heat transfer Q_wind is calculated as Equation (2):

$$Q_{wind}={\alpha }_{wind}\left(T_s-T_a\right)=\left(10.4v_{wind}^{0.7}+2.2\right)\left(T_s-T_a\right)$$

(2)

• The latent heat of phase change Q_vap is calculated as Equation (3):

$$Q_{vap}=q_{fr}+q_{hy}=m_{fr}{L}_{fr}+m_{hy}{L}_{hy}$$

(3)

Definitions of the variables used in all the formulas throughout this paper are provided in Appendix A.

② Internal Heat Conduction within the Pavement: The internal heat conduction model focuses on the heat transfer between the different material layers of the pavement structure. From top to bottom, the pavement consists of an ice–water–salt layer, a concrete pavement layer, a gravel base layer, and a soil base layer. To ensure consistency between the model and the experimental results, the thickness and thermal properties of each layer in the model are based on the specific dimensions and materials of the pavement structural body previously described. Thus, the model is constructed as a realistic representation of an actual airport runway. We assume that the temperature field varies significantly along the vertical (longitudinal) direction and use the one-dimensional heat conduction equation to describe the temperature distribution and variation. The one-dimensional heat conduction follows Fourier’s law, as shown in Equation (4).

$${\displaystyle \frac{\partial T}{\partial t}}=\alpha {\displaystyle \frac{{\partial }^{2}T}{\partial x^2}}$$

(4)

3.2. Numerical Solution Method

Based on the derivation of the theoretical model, this paper employs the Finite Difference Method (FDM) to numerically solve the heat conduction equation. By discretizing the time and space variables, a computational grid is constructed for the domain, and a recursive relationship is derived to solve the evolution of the temperature field over time, as shown in Equation (5).

$$T_j^{n+1}=T_j^n+\left[{\alpha }^2{\displaystyle \frac{T_{j+1}^n+T_{j-1}^n-2T_j^n}{∆x^2}}+f_j^n\right]∆t$$

(5)

Boundary conditions are crucial for the solution process. At the pavement surface, a mixed form of the second and third types of boundary conditions is used to account for the actual effects of the solar radiation and wind speed on the surface temperature. At the bottom of the soil base layer, the first type of boundary condition is applied by setting a fixed ground temperature to simulate the constant temperature conditions deep underground. The parameter values used in the model solution are provided in Appendix B.

3.3. Model Explanation

Although this study does not introduce innovations in theoretical modeling or numerical solution methods, by reproducing and validating existing heat transfer models, we ensure the accuracy and applicability of the model under experimental conditions. The model is validated by comparing its results with the experimental data, confirming that the calculations reasonably reflect the temperature changes and icing characteristics of the pavement under various environmental conditions.

4. Results and Discussion

4.1. Wind Speed and Air Temperature Effects

In this study, we systematically analyzed the effects of the wind speed and air temperature on the surface temperature of airport pavements and the onset of icing. The experiments were conducted under six different air temperature conditions (−3 °C, −5 °C, −7 °C, −10 °C, −15 °C, −20 °C) and three wind speed conditions (2 m/s, 4 m/s, 6 m/s).

• Effect of Wind Speed on Pavement Surface Temperature

At higher air temperatures (−3 °C to −5 °C), the pavement temperature decreased rapidly at first and then stabilized, as shown in Figure 6a,b. The cooling rate during the first 15 min was approximately 0.30–0.32 °C/min, after which it significantly slowed to nearly 0.01 °C/min. At lower air temperatures (−7 °C to −20 °C), the pavement surface temperature exhibited a three-phase decline: an initial rapid drop, a subsequent stable phase, and a final accelerated drop, as illustrated in Figure 6c–f. This temperature variation was most pronounced at a wind speed of 6 m/s, indicating that higher wind speeds accelerated the cooling process, ultimately resulting in temperatures that were 6.6 °C to 10.6 °C lower than the ambient air temperatures.

2.

Effect of Air Temperature on Pavement Surface Temperature

Under varying wind speed conditions, lower air temperatures led to faster and more pronounced decreases in the pavement surface temperature, as illustrated in Figure 7. At lower air temperatures (−10 °C to −20 °C), the temperature drop exhibited a linear trend, with the cooling rate increasing as the air temperature decreased. For instance, at a wind speed of 2 m/s, the cooling rates were 0.05 °C/min, 0.09 °C/min, and 0.1 °C/min at −10 °C, −15 °C, and −20 °C, respectively, as depicted in Figure 7a. This trend in cooling rate became more significant with an increasing wind speed.

3.

Effect of Wind Speed on Icing Onset Time

As shown in Figure 8, the impact of the wind speed on the icing onset time was most noticeable at higher air temperatures (−3 °C to −7 °C). As the wind speed increased from 2 m/s to 6 m/s, the icing onset time decreased significantly. For example, at −3 °C, raising the wind speed from 2 m/s to 4 m/s reduced the icing onset time by 29.4% to 35.7%, while increasing the wind speed further to 6 m/s shortened the time by an additional 8.3% to 22.2%. However, at lower air temperatures (−10 °C to −20 °C), the sensitivity of the icing onset time to variations in the wind speed diminished.

4.

Effect of Air Temperature on Icing Onset Time

As the air temperature decreased, the icing onset time was significantly reduced, as illustrated in Figure 9. For instance, at a wind speed of 2 m/s, when the air temperature dropped from −3 °C to −20 °C, the icing onset time decreased by 85.5% to 94.4%, with a maximum reduction of 25 min. This suggests that lower air temperatures enhance the temperature gradient between the water layer and the air, resulting in faster heat loss and a quicker onset of icing.

The experimental results clearly indicate that the wind speed and air temperature significantly affect the pavement icing process, which is consistent with the findings of Dan et al. [21]. As the wind speed increases, the rate of the temperature drop on the pavement surface accelerates, primarily due to the enhanced convective heat transfer between the pavement and the surrounding air. This effect is especially pronounced under conditions of a high wind speed and low air temperature, where the pavement surface cools more rapidly, leading to quicker icing.

Additionally, the air temperature significantly influences the icing process. Under lower air temperature conditions, icing occurs more rapidly. The experimental results indicate that as the air temperature approaches or falls below freezing, the time required for icing on the pavement surface is notably reduced. This finding is consistent with the study by Cui [22], which demonstrated that during repeated freeze–thaw cycles, the effects of salt solutions on the material performance are more pronounced under low-temperature conditions, particularly when both low temperatures and high salt concentrations are present.

4.2. Effect of Deicing Salt Concentration

To investigate the impact of the deicing salt concentration on the pavement surface temperature and the onset of icing, this study examined various combinations of wind speed and air temperature alongside deicing salt concentrations of 0%, 20%, 40%, 60%, and 80%. The deicing agent used in the experiment is sodium chloride, which is the most widely used deicing salt for roads due to its low cost, effective freezing point depression, and easy availability. In the experiment, the concentration of the deicing salt was defined by the ratio of the mass of salt applied to the surface to the mass of water present on the runway. The key findings are as follows.

• Effect of Deicing Salt Concentration on Pavement Surface Temperature

During the initial 10 min of the experiment, the effect of varying the deicing salt concentrations on the pavement temperature was minimal. However, as the time progressed—especially after 30 min—the pavement surface temperature with the deicing salt was significantly lower than without it, with the cooling effect becoming more pronounced at higher concentrations of deicing salt. For example, as shown in Figure 10a, at an air temperature of −10 °C and a wind speed of 2 m/s, the pavement surface temperature decreased by 1.8 °C and 4.63 °C for the 20% and 80% salt concentrations, respectively, compared to the absence of deicing salt. In contrast, Figure 10i indicates that at a lower air temperature of −20 °C and higher wind speed of 6 m/s, the temperature drops were 1.66 °C and 6.83 °C for the 20% and 80% salt concentrations, respectively.

2.

Effect of Deicing Salt Concentration on Icing Onset Time

Figure 11 illustrates the icing onset time under varying deicing salt concentrations and wind speeds, while Figure 12 depicts the icing onset time with different salt concentrations and air temperatures. The experiment revealed that increasing the concentration of deicing salt significantly delayed the icing onset time. This effect was especially pronounced at lower air temperatures (−20 °C) and higher wind speeds (6 m/s). For example, as shown in Figure 11a, at −3 °C, increasing the salt concentration from 0% to 20% delayed the icing onset time by 221% to 543%, extending it by up to 76 min. Similarly, as indicated in Figure 11f, at −20 °C, the delay ranged from 2252% to 6700%, extending the time by up to 67.55 min.

3.

Impact of Deicing Salt Concentration Under Different Conditions

This study found that when the air temperature was constant, lower wind speeds result in a more significant effect of the deicing salt concentration on the icing onset time. For instance, as shown in Figure 11a, at −3 °C and a wind speed of 2 m/s, increasing the salt concentration from 0% to 20% extended the icing onset time by 28 min and 13 s, representing a 256.6% increase. In contrast, at a wind speed of 6 m/s, the time was extended by 24 min and 18 s, reflecting a 303.7% increase. Under varying air temperatures, a higher deicing salt concentration also led to significant extensions in the icing onset time, as illustrated in Figure 12. Lower air temperatures exhibited a more pronounced effect, particularly at higher wind speeds. For example, at a wind speed of 2 m/s and an air temperature of −20 °C, increasing the salt concentration from 0% to 20% extended the icing onset time by 30 min and 15 s, leading to a remarkable 1008.3% increase.

The experimental results clearly demonstrate that the concentration of deicing salt significantly influences the cooling of airport pavement surfaces and delays the onset of icing. As the concentration of deicing salt increases, the pavement surface temperature decreases more markedly, and the icing onset time is notably prolonged. This phenomenon primarily occurs because the deicing salt lowers the freezing point of the surface solution. Consequently, as the concentration of deicing salt increases, the freezing point of the solution decreases, meaning that the pavement surface temperature must drop further to reach freezing conditions. This results in a lower surface temperature and a delayed onset of icing.

To further elucidate this effect, as the concentration of salt increases, the corresponding freezing point decreases. Under uniform environmental temperatures, surfaces with lower salt concentrations begin to freeze. This freezing process involves heat absorption, which slows the rate of temperature decline on the surface. Conversely, at higher salt concentrations, the lower freezing point prevents the onset of freezing, resulting in no heat absorption. Consequently, the surface temperature continues to decrease at a faster rate. This mechanism explains why surfaces treated with higher concentrations of salt exhibit more rapid cooling under the same environmental conditions, as they are less likely to reach the conditions necessary for icing to initiate.

However, the experiment also indicates that under higher wind speed conditions, the effectiveness of deicing salt may be compromised. This may occur because an increased wind speed can lead to an uneven distribution of the salt solution on the pavement surface. It may be challenging to directly study the impact of the wind speed as a variable, but similar studies have found that although the transport of salt by wind is minimal, in exposed areas, an increased wind speed can hinder the distribution of salt solution, thereby affecting its effectiveness in preventing ice formation [23].

Additionally, under extreme low-temperature conditions, while deicing salt can significantly delay the onset of icing, its effectiveness is somewhat diminished in higher wind speed scenarios. This observation aligns with findings by Fujimoto et al. [24], who developed a road surface freezing model based on the heat, water, and salt balance, which was validated through field experiments. Their study found that the distribution and dissolution rate of salt solutions on road surfaces significantly influence the icing process, particularly in situations where vehicle traffic causes the dispersion of the solution. These findings underscore the importance of considering the complex interactions between environmental conditions and deicing salt concentrations when developing deicing strategies for airport pavements. Selecting an appropriate deicing salt concentration based on real-time weather conditions, such as the wind speed and air temperature, can maximize the effectiveness of deicing efforts and ensure the safety and operational efficiency of airport pavements.

4.3. Comparison of Experimental Results and Theoretical Model

The comparison between the experimental results and the theoretical model predictions demonstrates general agreement in most cases. Figure 13 illustrates this comparison under various parameter conditions. For durations of less than 30 min, the theoretical results closely align with the experimental data, showing an average error of less than 24.4% and a maximum error of 36.3%. However, as the duration exceeds 30 min, the theoretical predictions begin to diverge from the experimental measurements, with the theoretical temperatures slightly higher than observed. This leads to an average error range of 28.7% to 63.7% and a maximum error of 90.2%. This discrepancy can be attributed to the simplified assumptions in the theoretical model and certain uncontrollable factors during the experiment.

The sources of the error can be analyzed from two perspectives. First, differences between the assumptions and parameters used in the theoretical model and the actual conditions may lead to discrepancies. The thermal conductivity of the materials used in the experiment, such as the soil base and cement concrete base layer, could differ from the values assumed in the model, resulting in variations in the heat transfer rates. Second, during the verification process, the boundary conditions of the theoretical model were adjusted based on real-time environmental and temperature data during the first 15 min of cooling, which resulted in a minimal error. However, as the air temperature stabilized at the set value and the external conditions ceased to change, the model’s ability to self-correct diminished, leading to an accumulation of errors over time, which further increased during the extended calculations.

This phenomenon aligns with the findings of Yang et al. [25], who identified similar issues in their study on the factors influencing asphalt pavement icing. They noted that while an increased wind speed and air temperature accelerated pavement cooling, these factors also introduced greater uncertainty, particularly during prolonged cooling periods, where the accuracy of the theoretical model predictions diminished. Additionally, Liu et al. [26] employed a numerical model based on heat and mass transfer to validate the snow melting process on heated pavements. They similarly found that while the theoretical model matched the experimental results well in the short term, the discrepancies became more pronounced over time. In conclusion, although the theoretical model effectively captures the overall trends in the pavement temperature changes, its predictions may diverge from the actual conditions during extended periods or under complex environmental scenarios. Therefore, when utilizing these models for engineering design and practical applications, it is essential to consider these potential errors and incorporate experimental data for necessary adjustments.

4.4. Simplified Prediction Formula Based on Experimental and Theoretical Models

To develop a practical and simplified prediction formula for estimating the onset time of pavement icing at airports, we first conducted a thorough analysis of the experimental results in comparison with the theoretical models. This analysis enabled us to identify the key factors that significantly impact the icing onset time. These factors include the pavement surface temperature (T_s), wind speed (v_wind), deicing salt concentration (r_salt), and air temperature (T_a).

The results indicate an exponential relationship between the pavement surface temperature and the onset time of icing. As the temperature decreases, the icing onset time occurs earlier, prompting us to adopt an exponential function to describe this relationship. Additionally, an increasing wind speed accelerates the heat loss from the pavement surface, thereby shortening the icing onset time; the experimental results demonstrate that the effect of the wind speed on the icing time is approximately linear. The concentration of deicing salt directly influences the freezing point of the salt solution, affecting the icing onset time, with a linear relationship observed between the salt concentration and the icing time. Moreover, the air temperature impacts the cooling rate of the pavement surface, thereby impacting the icing onset time; this relationship is also linear.

To determine the specific values of the fitting parameters, we employed the Least Squares Method (LSM) using the experimental data. This method minimizes the discrepancy between the predicted icing onset time and the observed experimental results, resulting in the most accurate fitting parameters. The Simplified Prediction Formula is presented in Equation (6), where b₁–b₅ are fitting parameters determined as follows: 0.8104, −7.3345, 6.4675, 2.422 × 10⁻⁸, 24.3328;

$$t_{ice-start}=b_1{v}_{wind}^2+b_2{v}_{wind}+b_3{r}_{salt}{e}^{{\displaystyle \frac{\left(T_s-T_a\right)}{b_4}}}+b_5$$

(6)

The proposed simplified prediction formula for the onset time of pavement icing demonstrates strong accuracy when compared to the experimental data. As illustrated in Figure 14, the predicted values closely align with the actual measured values, as evidenced by the data points clustering around the line y = x, which represents perfect prediction accuracy. The dashed lines indicating the boundaries of y = 0.8x and y = 1.2x show that the majority of the predicted values fall within a ±20% error range, confirming the model’s reliability for practical applications. The average error between the predicted and actual values is approximately 16.8%, with a maximum deviation of 22.4%, both of which are within acceptable limits for engineering purposes. This level of accuracy suggests that the simplified prediction model is not only effective but also practical for real-world scenarios, enabling airport operators to anticipate and respond to pavement icing conditions with a high degree of confidence. The model’s ability to account for critical factors such as the surface temperature, wind speed, salt concentration, and air temperature makes it a valuable tool for enhancing the safety and operational efficiency of airport runways during winter weather conditions.

5. Conclusions

This study investigated the critical issue of airport pavement icing in saline environments, which poses a significant threat to aviation safety. To address this challenge, an experimental setup was developed to replicate real-world conditions, enabling a comprehensive analysis of the effects of variables such as the air temperature, wind speed, and deicing salt concentration on the icing process.

Our findings reveal that lower air temperatures and higher wind speeds significantly accelerate pavement cooling. Specifically, wind speeds of 6 m/s decrease the icing time by 35.7%, while temperatures of −20 °C can reduce it by 94.4%. Under severe conditions (−20 °C, wind speed 6 m/s), pavements reach critical icing temperatures in just 15 min, underscoring the urgent need to manage the increased operational risks associated with rapid icing on runways during extreme weather. Conversely, high concentrations of deicing salt can delay the icing onset by up to 67 min and 33 s, representing a 6700% increase compared to conditions without salt, thereby effectively preventing icing and lowering pavement temperatures. However, the potential damage to pavements caused by deicing salt indicates that their optimal use warrants further study.

A comparison between our collected experimental data and theoretical predictions reveals an initial concordance, with the early and middle stages of icing exhibiting an average error margin below 30%. However, the observed discrepancies increased over time, likely due to the simplified assumptions inherent in our model. Based on these results, a simplified prediction model was developed and validated against the empirical data. This model serves as a practical tool for forecasting icing events and tailoring deicing strategies to the prevailing environmental conditions. By incorporating such predictions into operational planning, it is possible to significantly enhance runway safety and efficiency.

This study was conducted under controlled laboratory conditions, which may not fully capture the complexity of real-world environments. Future research will validate the prediction model in diverse operational scenarios and explore the effectiveness of deicing strategies under various climatic conditions. This will involve utilizing real environmental data collected via sensors on airport runways and applying artificial intelligence algorithms to refine the model’s accuracy and usability. Additionally, further studies will address airport operational risks and optimize the use of deicing salts to enhance the effectiveness and sustainability of these strategies.