Advancing Insights into Runway De-Icing: Combining Infrared Thermography and Raman Spectroscopy to Assess Ice Melt

Abstract

The “bare runway” principle aims to ensure passenger and employee safety by making runways more usable during winter conditions, allowing for easier removal of contaminants like snow and ice. Maintaining runway operations in winter is essential, but it involves considerable cost and environmental impacts. Greater knowledge about the de-icing and anti-icing performance of runway de-icing products (RDPs) optimizes operations. The ice melting test, as per the AS6170 standard, gauges the rate at which an RDP dissolves an ice mass to determine RDP effectiveness. Here, we introduce a novel integrated methodology for assessing RDP-related ice melting. We combine laboratory-based procedures with infrared thermography and Raman spectroscopy to monitor the condition of RDPs placed on ice. The plateau of maximum efficiency, marked by the most significant Raman peak intensity, corresponds to the peak minimum temperature, indicating optimal RDP performance. Beyond this point, RDP efficacy declines, and the system temperature, including melted contaminants and RDP, approaches ambient temperature. Effective RDP performance persists when the ambient temperature exceeds the mixture’s freezing point; otherwise, a freezing risk remains. The initial phases of RDP–ice contact involve exothermic reactions that generate brine, followed by heat exchange with surrounding ice to encourage melting. The final phase is complete ice melt, leaving only brine with reduced heat exchange on the surface. By quantifying these thermal and chemical changes, we gain a deeper understanding of RDP-related ice melting, and a more robust assessment can be provided to airports using RDPs.

1. Introduction

The presence of water, ice, or snow [1] on runways has led to numerous aircraft accidents [2]. Runway excursion is a major risk in the aviation industry [3,4,5], especially when aircraft must use icy runways, such as in Canada [6,7]. For example, on 8 February 2019, a thin layer of ice covering the runway at Bagotville airport (Saguenay, QC, Canada) led to an aircraft leaving the runway after landing [8]; fortunately, there were no injuries. Such accidents often result in significant property damage [9], and the frequency of incidents during winters negatively impacts airport operations at higher latitudes [10]. To ensure safe aircraft take-off and landing, airports remove winter contaminants (e.g., snow and ice) using mechanical tools and/or prevent their accumulation.

Standard runway maintenance operations in regions subject to freezing, e.g., Canada and the northern United States, involve applying thousands of tons [11] of de-icing products in liquid (aqueous solutions) or solid form. Despite the importance of winter conditions on airport operations and budgets, little effort has been made toward evaluating de-icing and anti-icing product performance or determining the optimal amounts of product to be applied to runways [12,13,14]. Product suppliers typically suggest application rates [15,16,17] and the duration of product effectiveness on the basis of their own experience and knowledge.

Snow accumulated on runways is removed using mechanical tools [18]. For ice, chemicals known as runway de-icing products (RDPs) are first applied to facilitate melting before its removal by mechanical means. Moreover, maintenance crews may initially spread RDPs to prevent ice from forming on the runway [19]. Three standardized tests are conducted jointly to determine RDP performance: ice melting [20], ice penetration [21], and ice undercutting [22]. Here, we focus on ice melting, which is the main property that interests maintenance managers when they assess the utility of an RDP.

Ice melting occurs when energy is added to the ice, causing it to melt. This added energy may be heat, a chemical, or another substance. The ice melting test estimates the amount of ice mass that an RDP will melt per minute, as described with the AS6170 test method standard [20] which was developed by an SAE G12 working group. However, the main shortcoming of this test is the absence of a failure value or minimum value, which leaves the results open to too wide of an interpretation. Determining the term “performance” is essential to understanding what is expected of the RDP. Performance can be expressed in terms of melting rate but also in terms of the product’s endurance, how long it reacts. As this notion is not addressed in the standard, it is difficult to estimate what performance is expected during the test. Ice melting is also influenced by various parameters inherent to airports: solar radiation, convective effect of wind, temperature variations, mixed icing conditions (snow, rain, ice, etc.), passing vehicles, and the presence of contaminants (aircraft fluid or fuel). Unfortunately, these parameters are not included in the test. So, ice melting is a comparative test, under controlled conditions, but needs further investigation. The output is a 3D measurement (mass or volume per minute) representing the speed at which a given product melts ice [12]. Much research has been conducted on liquid de-icers [23,24] but less on solids because of factors such as granulometry. Studies on solid RDPs often focus on NaCl and MgCl₂ [25,26,27]. Procedures for testing the melting capacity of ice rely on tools such as isothermal bottles [28] and ice cubes [29]; however, these methods require manipulation during testing, which increases the potential for error.

Here, we apply two methods to characterize ice melting through RDP use. The first, thermography, is commonly used in winter operations to measure the surface temperature of pavement (e.g., roads and airports) [30]. The second, Raman spectroscopy, reveals the vibrational and rotational states of the chemical bonds of the molecules in the RDP [31]. This non-destructive method details phase changes and RDP failure [14] and identifies the chemical in isolation or once it is dissolved. Our primary aim is to introduce a new approach applying infrared thermography and Raman spectroscopy to improve our understanding of ice melting using RDPs. By identifying and quantifying the RDP-related effects of ice melt, we can characterize the thermal and chemical properties of the phase changes. We base our laboratory analyses on field data collected from Montréal (QC, Canada) airports and the recommended amounts of RDP to be applied. Our study will assist airports in determining the efficacy of a specific RDP for melting ice.

2. Experimental Protocols

We conducted ice melting tests in controlled laboratory conditions at the International Laboratory of Anti-icing Materials (AMIL). The tests were carried out in a climate chamber measuring 9.1 m in height, 5.5 m in length, and 3.5 m in width. The chamber is designed to maintain a controlled air temperature ranging from 0 °C to −35 °C with a precision of 0.5 °C. The temperature inside the chamber for our tests was set at −10 °C. For all tests, we used commercially available solid-state sodium formate (NaFo) as the RDP. NaFo is commonly used for winter maintenance at airports, including the Montréal airports, because of its environmental and anti-corrosion properties and its cost-effectiveness. This RDP is a white granular solid with particles aggregated into non-cohesive flakes measuring 1–3 mm in diameter and whose freezing point is −16 °C for a mass percentage of 24.70% w/w [32]. Sodium formate is effective at preventing re-icing down to −16 °C. When NaFo is dissolved in water, it reduces the freezing point of the solution. As a result, the solution remains liquid at lower temperatures than pure water would [32]. It prevents ice reformation by maintaining a liquid state even in cold conditions. It also rapidly penetrates solid snow and ice layers, detaching them from asphalt or concrete surfaces.

Our protocol involved running three distinct tests: (i) ice melting assessments conducted according to standardized procedures using thermometric observations; (ii) ice melting evaluations conducted using the existing airport RDP application rate, supplemented with thermometric observations; and (iii) a detailed examination of ice melting phenomena, incorporating both thermometric and Raman spectroscopic observations.

2.1. Ice Melting Protocol Using the AS6170 Test

We followed the AS6170 test method standard [20] for the ice melting test (Figure 1).

First, a Petri dish was filled with 60 mL of ASTM D1193 type IV water and then cooled at −10 °C. After at least 8 h, the Petri dish and resulting ice were weighed using a calibrated scale (820 g max capacity, ±0.001 g accuracy). Then, 5 g of RDP was evenly deposited onto the ice surface. After 1, 2, 3, 4, 5, 10, and 30 min, the melted ice (brine) was removed using compressed air, and the Petri dish was reweighed after each removal to determine the mass of the melted ice. Each time interval was treated as a separate measurement and repeated three times.

To gain a deeper understanding of product performance, we used a thermal camera (Optris Pi450i, Optris, Berlin, Germany) with its corresponding software (Optris PIX Connect Rel.1.2.1030.0) to obtain an overall view of the Petri dish and ice. The camera had an uncooled 382 × 288-pixel microbolometer frame plane array detector and was equipped with an 18° × 14° field of view (FOV) Ge lens. The camera was positioned approximately 40 cm away from the target and was turned on at least 30 min before measurements to ensure thermal equilibrium. The camera had a Noise Equivalent Temperature Difference (NETD) of 60 mK without further specifications provided by the manufacturer.

Additionally, an aluminum foil mirror was placed on the ice surface in the camera’s FOV. The foil served to correct the radiative environment and allow for thermometric analysis using thermal images. Thermal images were recorded at 80 Hz in video mode, capturing temperature measurements during the ice-melting process. The entire test was repeated three times for reliability and consistency.

2.2. Airport Application Rate

In collaboration with Montréal Airports, we obtained access to a designated runway and observation of the use of maintenance tools to directly observe operational procedures. This unique opportunity allowed us to assess the effectiveness of RDPs in winter conditions.

According to the AS6170 test method standard [20], melting tests should be conducted using a 14 cm diameter Petri dish containing 5 g of product, which is equivalent to 280 g/m² of RDP. However, during the field test campaign, we noted that the actual application concentration (see Figure 2) was much lower than the specified standard. Note, however, that the spreading trucks lacked accurate monitors to confirm the precise amounts of applied RDP.

Photo analysis determined the distribution of RDP grains within a given surface area to approximate the application per square meter. The amount spread followed the manufacturer’s recommendations, which can also be adapted to the actual weather conditions. We considered the size and weight of each individual grain to determine an actual application concentration of slightly more than 25 g/m²; this coverage is more than 11× lower than that of the AS1670 standard test, leading to a possible overestimate of the RDP’s ability to melt ice when relying solely on the AS1670 standard. To align with actual airport use and conditions, we therefore used a reduced amount of product (0.5 g), corresponding to the actual estimated application rate for the ice melting test. Nevertheless, to maintain consistency with the standard and overestimated amount of RDP, the same test was also run using 5 g of product.

2.3. Thermometric Photography and Raman Spectroscopy

In the cold room at −10 °C, we then ran thermometric photography and Raman spectroscopy to observe the ice and RDP in real time (Figure 3). Building upon the initial ice melting test, we adapted these additional tests to be more detailed. For these tests, we used 0.05 g of crushed RDP (equivalent to one grain of RDP). The RDP was deposited onto an ice surface using a 0.05 mL semi-spherical spoon. The RDP was thus in direct contact with the ice—a contact surface of 28 mm², approximately the size of one salt grain and ensuring a consistent shape and surface area for analysis.

To measure the temperature distribution of the grain at the ice–RDP interface, we then used an Optris Pi640 thermal camera (Optris, Berlin, Germany) equipped with a microscopic lens and running Optris PIX Connect software (Rel.1.2.1030.0). This setup included a 640 × 480-pixel microbolometer frame plane array detector and a Ge lens, providing a 15° × 11° FOV for thermal imaging of the RDP on the ice. The camera had a NETD of 40 mK. The camera was positioned approximately 40 cm away from the target and was turned on at least 30 min before any measurements to ensure the system attained a thermal equilibrium. Additionally, a reflective crumpled aluminum foil mirror was placed on the ice surface in the camera’s FOV to correct for the radiative environment. Thermal images were captured in video mode at 30 Hz.

The second instrument was an i-Raman Plus portable spectrometer from BWTek (Plainsboro, NJ, USA), operating with a 785 nm laser at an output power of 340 mW. The Raman signal was collected using a confocal lens Raman probe, allowing for a contactless measurement at a distance of approximately 5 mm. The spectrometer was controlled using BWSpec 4.11 software, which managed the laser power (ranging from 1 to 100% of available power), integration time, and acquisition frequency. A portable spectrometer also permits in situ measurements of the Raman depth profile directly on the runway; this eliminates the need for sample collection and preparation as required for FTIR or DSC methods, thereby speeding up RDP analyses. The instrument provided Raman spectra within the 300–3400 cm⁻¹ range at a resolution of 4 cm⁻¹. During the ice melting experiment, the integration time was set at 20 s, and each produced spectrum was an average of four spectra. Each spectral measurement lasted 80 s, and Raman spectra were collected every 0.01 s. Raman spectra of the RDP or brine contained characteristic peaks that correspond to specific chemical bonds; peak intensity correlates with the concentration of the analyzed chemical compound. In this study, we focused specifically on the peak at 930 cm^–1.

3. Results

3.1. Ice Melting Test

We observed that the solid RDP melted rapidly into the ice (Figure 4). The RDP melting of ice was more pronounced using 5 g than using 0.5 g. In the first few minutes, the RDP-related melting rates were both elevated; however, these rates decreased markedly over the next 10 min (Figure 4). For 5 g of RDP, the melting rate was higher after 30 min, possibly because of the presence of pristine RDP still able to melt the ice and generate brine.

The 5 g ice melting rate was greater than that of the 0.5 g samples, although the rate was not proportional to the RDP amount. The standard deviation after one minute was greater for 5 g (±0.02) than for 0.5 g (±0.01), possibly because not all of the 5 g of RDP was initially in contact with the ice, resulting in only partial effectiveness. When using 0.5 g, proportionally much more area of the RDP grains is in contact with the ice, allowing for complete melting. After 10 min, the residual dry RDP in the 5 g samples, which was initially inactive, comes into contact with the water or brine, promoting further melting. Unfortunately, this delay in ice melting negatively affects the time, cost, and quantity required for winter runway maintenance. Therefore, using excessive amounts of RDP does not offer a suitable solution.

We then analyzed temperatures from the thermal video to understand RDP behavior in the Petri dish (Figure 5). Temperatures during the melting were not uniform or constant across the surfaces, ranging from −10 °C to just below −16 °C (Figure 5). There was an initial instantaneous temperature drop, an endothermic reaction from the salt dissolution [33]. The reaction can be described as follows: solid RDP + ice and then solid RDP + ice + brine, and it evolves quickly to a brine–ice interaction. The thermal images show that for 0.5 g of RDP, the grains were evenly distributed on the ice at 0 min. In contrast, for the 5 g samples, the grains were stacked. As the melting process continued, brine formed in both tests. We recorded a minimum temperature of −16 °C for the 5 g test, which corresponds to the peak effectiveness of the product. This temperature was reached after approximately 12 min and then increased until our final observation at 30 min. The minimum temperature of the 0.5 g RDP sample was approximately −13 °C at around 7 min. The minimum temperature was lower for the 5 g sample than for the 0.5 g samples, consistent with the difference in RDP amounts. The difference between the initial test temperature (−10 °C) and the lowest temperature observed on the thermal images was 3.4 °C for 0.5 g and 6.4 °C for 5 g. This temperature pattern can be explained by the endothermic dissolution of RDP and the melting ice related to poor thermal conduction and local convective effects in aqueous solutions below 0 °C [34]. The product temperature followed the same trend, with lower temperatures corresponding to higher melting rates. Thus, the correlation between the two methods supports the conclusion that higher melting rates resulted in lower temperatures. However, using 5 g to evaluate melting capacity, as recommended by the standard, may produce false results by exaggerating the endothermal reaction and causing an excessive temperature decrease.

3.2. Raman Spectroscopy

We then focused strictly on the RDP–ice interface. For the 0.5 g samples, each RDP grain reacted independently. RDP grains are polyhedral-shaped, resulting in suboptimal contact with the ice. Therefore, the RDP grains were roughly crushed and molded into a semi-spherical measurement spoon to obtain 0.05 mL of RDP, representing a single grain. Raman spectroscopy specifically targeted the interface between the 0.05 mL of RDP and the ice surface. We collected 62 spectra. Each spectrum was smoothed to facilitate analysis, and the baseline was corrected to obtain spectra similar to those depicted in Figure 6. The corresponding peaks are indicated in the figure for informational purposes only.

These spectra show five significant peaks representing O-C=O in-plane deformation, C-C stretch, pyridine or isothiocyanate salt, CH₃ symmetric deformation in CH₃CO⁻, carboxylate ions, and CH₃ stretch. The most prominent peak is observed at 930 cm⁻¹, which likely corresponds to C-C stretch, typically found in NaFo [35]. It could also correspond to pyridine or isothiocyanate, used to prevent salt aggregation. Monitoring this peak would be useful during analysis, as the peak intensity relates to the concentration of the associated chemical bond. Continuous monitoring of this peak provides insight into the kinetics of ice melting and its variations. A weaker peak intensity indicates greater RDP dilution and increased ice occurrence. (Note that the omission of detailed peak frequency indexed to RDP chemical bonds is to maintain conciseness and clarity in the text.)

Thermal video (Figure 7) recorded how RDP temperatures evolved. The Raman probe and the aluminum foil are also visible in these images.

The RDP begins to dissolve and melt the ice, resulting in a high temperature (Figure 7A). The top of the RDP dome is not yet in solution, indicating a highly effective product, as it gradually dissolves, leading to a strong melting reaction (Figure 7B). When fully dissolved, the product remains effective (Figure 7C). Finally, after almost 20 min (Figure 7D), the reaction between RDP and ice (or between brine and ice) is no longer thermally visible.

4. Coupled Thermographic and Spectral Analyses

The Raman spectra of the RDP or the brine contain peaks characteristic of the chemical bonds constituting them and their concentration. In this study, we focused on the peak at 930 cm⁻¹, directly related to RDP. We monitored its intensity over the course of the experiment to better understand its development. We therefore traced the evolution of the minimum temperature reached by the RDP over time. To achieve this, however, we required a radiative correction of the thermal data by applying Equation (1) [36].

$$T_{\mathit{measured}}^4={\epsilon }_{\mathit{sample}}\times T_{\mathit{sample}}^4+\left(1-{\epsilon }_{\mathit{sample}}\right)\times T_{\mathit{environment}}^4$$

(1)

where $T_{\mathit{measured}}$ corresponds to the corrected thermal value, ${\epsilon }_{\mathit{sample}}$ is the ice emissivity (0.97) [37], $T_{\mathit{sample}}$ is the measured temperature of the sample during the test, and $T_{\mathit{environment}}$ is the mirror temperature during the test.

The top curve in Figure 8A shows the evolution of the significant peak of the RDP at 930 cm⁻¹ over time, and the bottom curve presents the evolution of the minimum temperature of the RDP over time.

Inflections in both Raman intensity and temperature occurred simultaneously. Between 0 and 188 s, the temperature and Raman intensity increased as ice melted and concentrated brine formed. Between 188 and 503 s, the 930 cm⁻¹ peak intensity plateaued, whereas the temperature slowly decreased because of a locally saturated brine and latent dissolution heat. After 503 s, the Raman peak of the RDP reached maximum efficacy, coinciding with the minimum temperature. All available RDP was used, allowing for additional ice melting and diluting of the brine and causing a decrease in Raman peak intensity. Subsequently, the efficiency decreased, and the minimum temperature rose, although the product remained effective throughout the test.

Note that the minimum temperature rises at the start of this test, whereas it falls in the ice melting test (Figure 5). This can be explained by the camera angle and the number of grains. In the ice melting test, the camera is positioned above the Petri dish and there are several grains present in the test. As a result, the minimum temperature observed can be a very precise point in a fairly large area. In the Raman spectrometer test, the camera is tilted to observe the interface between the RDP and the ice, and focuses on a single grain, so that the minimum temperature observed is more precise and the measurement area much smaller.

To visually understand this phenomenon without spectroscopic support, we ran an observational test at −10 °C. Solid RDP powder was mixed with rhodamine B, which is used in ice penetration [21] and ice undercutting [22] tests and does not alter RDP properties. Unfortunately, compaction of RDP was not possible when mixed with rhodamine B; however, this did not affect the visual observations made to understand the phenomenon.

The RDP started to create brine 3 min after being placed on the ice surface (displayed as a sharper pink color; Figure 8B). This intense melting state continued until approximately 8 min and then gradually diminished (turning faded pink). From 0 to 200 s, the RDP produced an exothermic reaction and melted to form brine (Figure 8C), a similar pattern to that observed previously [12]. Between 200 and 500 s, heat exchange occurred between the RDP and the ice, resulting in the dissolution of the RDP. From 500 s until the end of the test, only brine remained, which had limited heat exchange with the ice.

5. Discussion

Our approach involved continuously monitoring ice melt using different analytical techniques to provide specific visual, thermal, and chemical information. The collected information is consistent and allows us to quantify the effectiveness and kinetics of applied RDP, data that are valuable for product developers. Our tests also provide information on the concentration of RDP brine, which, if too diluted, depending on the temperature, may cause the system to freeze again, posing a skidding risk to aircraft. The use of a portable Raman spectrometer would help with taking data directly from the runway to be more effective in RDP analysis with a lot of parameters, such as wind.

The integration of advanced technological tools such as thermal cameras, Raman spectrometry, and visual cameras has great potential for enhancing the efficiency and safety of winter runway operations. When combined with artificial intelligence (AI) programs, these tools significantly improve decision-making processes [38]. By applying data-mining techniques to analyze data from thermal cameras, Raman spectrometry, and visual cameras, a comprehensive understanding of RDP behavior in dynamic environments can be achieved. Neural networks can process these data along with inputs such as air and ground temperatures, temperature variations, and precipitation intensity (e.g., freezing rain or snow). Operators can then make well-informed decisions regarding the optimal timing of RDP replacement. Moreover, AI-driven systems can forecast runway conditions and slippage probabilities [39], offering valuable support to runway operations teams. This sophisticated approach represents a notable advancement in runway management, particularly considering the rapid changes in weather conditions that require prompt and accurate decision-making [40]. By leveraging AI and integrating data-mining techniques, runway operations can be carried out with greater precision, efficiency, and safety to ensure smooth air traffic flow, even in challenging winter environments.

6. Conclusions

We used thermal and chemical analyses to comprehensively characterize the interaction between runway de-icing products (RDP) and ice during ice melting tests. Existing standard practice results in excessive product usage. By applying Montréal Airports’ current application concentration, we observed that each salt grain reacts individually with the ice. Correlating results obtained from infrared thermography and Raman spectroscopy provided a better understanding of the dynamic exchange between RDP and ice, shedding light on ice melting kinetics and variations. Notably, the intensity of observed peaks in Raman spectroscopy decreased as RDP was diluted, indicating increased ice melting.

This spectroscopic approach allowed real-time measurements directly on the runway, eliminating the need for sample collection and preparation typically associated with other methods like FTIR or DSC. Visual tests confirmed the observed reactions within a specific time frame, demonstrating the effectiveness of our approach.

These insights significantly contribute to our understanding of RDP–ice interactions in a laboratory environment which can be transposed to natural airport conditions. Future research should use an airport traffic simulator to study the mechanical activation of solid RDP since the standard ice melting test omits this parameter, simulating interactions with spreader truck wheels and passing aircraft. Such simulations will help us understand how traffic accelerates the dissolution process, thereby offering ideas on how to further improve the efficacy of the applied de-icing product, but may also enable us to look at the anti-icing properties of the products used following different winter contaminations. Additionally, further exploration of Raman spectroscopy could provide deeper insights into the molecular dynamics of RDP–ice interactions, aiding the development of more sophisticated AI-driven tools for improved runway de-icing strategies.