Field demonstration of a robust humidity sensor system for condition monitoring of lubricating grease in axlebox bearings of rail wheelsets Open Access

Many people use rail vehicles every day, but only their manufacturers and operators know the demanding environmental conditions to which they are exposed during daily operation, such as large temperature fluctuations, high air humidity and mechanical loads such as vibrations and impacts. For this reason, all rail components, e.g. tapered roller wheelset bearings, must be able to withstand these conditions (Lenord + Bauer, 2018). To ensure the long-term safety performance of such bearings and, hence, for the entire vehicle, suitable lubrication of the axlebox bearing is of utmost importance, which is mostly provided by a lubricating grease, as this lubricant is generally used at locations where it is not practical or convenient to apply a lubricating oil as a standard solution. The main reasons for this are the high degree of immobility of the lubricating grease, resulting in an instantly forming lubricating film because machinery such as vehicles run intermittently or are stored for an extended period. However, rail machinery is not easily accessible for frequent lubrication because of the high safety requirements in the railway sector, in which any changes must be approved through costly processes (Pirro et al., 2001). Furthermore, lubricating greases act as sealants to prevent contamination with water or dust (Mang, 2014).

As rail applications use different types of lubricating grease, it is crucial to be aware of their conditions. In this context, the water content of a lubricating grease is one of the most important parameters because water is the most common source of liquid contamination, which has a destructive effect on both the performance of the lubricated components and lubricating grease (Karl and Bots, 2011). For example, only 1% water contamination in the lubricant can cause a 90% reduction in the life cycle of a bearing (Dittes, 2016), as water changes the lubricant characteristics both physically, such as the viscosity, lubricity and load-carrying characteristics, and chemically, comprising thermo-oxidative stability, tendency to form deposits and additive depletion. Furthermore, an increased water concentration can stimulate unwanted rust formation, corrosion, erosion and cavitation, all of which have a destructive impact on bearings (Day and Bauer, 2007).

Various techniques can be used to monitor the health status of lubricating greases. Initially, grease analyses were performed by taking several representative grease samples from different locations of the grease-lubricated rail component using a syringe and characterizing them afterwards in the lab by diverse benchmark methods, such as Karl–Fischer titration or Fourier transform infrared (FTIR) spectroscopy (Bots, 2013). As such procedures are time consuming and cannot provide real-time information about the grease condition, in recent decades, novel sensor concepts based on different principles have been developed without the need for any special laboratory techniques. One of these is a grease sensor (Fraunhofer ENAS, 2011) for rolling bearings based on optical infrared technology, which is currently an available online tool on the market for condition monitoring of lubrication greases. It can provide information regarding the grease condition with respect to its opacity, temperature, wear and water content. Consequently, it is possible to observe changes in the grease condition long before any degradation in the bearing occurs (Fraunhofer ENAS, 2011). The other onside concept uses a polymer-based capacitive humidity sensor to evaluate the water content of lubricating grease in bearings by measuring the emitted volatiles of the heated grease sample filled in a septum bottle. It can be applied directly to maintenance measures in axlebox bearings to determine the water content of the lubricating grease. As this technique shows high precision down to 10 ppm, it is comparable to the Karl–Fischer method (Pall et al., 2009; Rowe, 2016).

To date, entire onboard and wayside multisensor systems designed for the condition monitoring of rail vehicles typically include position measurements using the global positioning system (GPS), as well as temperature, vibration and acoustic emission sensors. Furthermore, complete sensor systems are integrated specifically into the bearing house within the axlebox bearing cover of rail vehicles for the monitoring of the health status of the lubricating grease; therefore, the lubricated bearings typically comprise temperature, vibration and rotational speed sensors (Bernal et al., 2019; Mirabadi et al., 1996; Vale et al., 2016). However, no sensor concepts based on electrical principles (e.g. capacitive) for the condition monitoring (e.g. water content) of the lubricating grease in rail bearings have yet been reported. The disadvantage of existing sensor approaches is that they require direct contact with lubricating grease (SKF, 2011). Consequently, the installation position is crucial for the result, especially as sensor installation directly in the relevant contact area is not possible.

As stated above, in the last few decades, condition-based and predictive maintenance has become important to meet the increasing demand for cost-efficient and environmentally friendly operations (Schneidhofer et al., 2023). However, their reliable use requires in-depth knowledge of the influence of the operating parameters on the service life of the components, including potential worst-case scenarios, such as sudden water entrainment. Furthermore, this knowledge is a prerequisite for the selection or development of sensors and the development of appropriate algorithms for data analysis. Basically, the definition of test parameters in the laboratory plays a crucial role, as more severe conditions to accelerate the aging processes are needed to shorten the test time. However, these stimulated aging mechanisms must be in agreement with field applications. Therefore, the so-called lab-to-field approach was used to establish a sensor development environment tailored to applications in railway vehicles. This concept enables the targeted development of both sensors and algorithms by laboratory experiments that mimic realistic single and mixed stresses such as temperature, humidity and mechanical loads (vibrations and shocks), as not only the corresponding rail vehicle components, but also the system-integrated sensors for condition monitoring are exposed to these environmental loadings. It contains several development steps (see Figure 1), during which the sensor functionality and hence the overall robustness can be validated gradually toward real conditions beginning from the lab (Schneidhofer et al., 2018) through validation of the relevant environment in the lab (Coronado and Kupferschmidt, 2014). Finally, the demonstration under field conditions is presented, which is the focus of this study. As reported in previous research, selected sensors could always provide reliable signals up to the level of field demonstration, through which it could be concluded that they are suitable for applications carried out under rail field conditions (Dubek et al., 2024).

In this study, a concept for monitoring the water content of a lubricating grease using commercially available robust humidity sensors placed in the atmosphere surrounding the lubricating grease within the axlebox bearing cover of a rail wheelset bearing is presented. In the event of an undesired water uptake of the lubrication grease, an increased air humidity close to the bearing should be measured by the Humidity Sensor in Axle Bearings (HSAB) system. After the robustness of the sensor system was successfully confirmed by several laboratory investigations, according to Dubek et al. (2024), its final field validation was conducted. Consequently, this study comprises the field evaluation of the HSAB system under different weather conditions installed at the wheelset bearing of a freight wagon. The wagon traveled on regular railway tracks in Austria to demonstrate the potential of this condition monitoring approach. Both the temperature and humidity signals of the HSAB system and a reference sensor mounted on the bogie frame were recorded and compared to each other, as well as to environmental data from the respective weather stations. Subsequently, the calculated characteristic curve for the grease–water content utilizing the algorithm developed under lab conditions considering the estimated grease temperature is presented. In summary, this concept offers several advantages over previous approaches, such as the compact structure of the sensor system, existing and approved technology, no necessary changes in the bearing structure used and comprehensive determination of the grease–water content by measuring the gaseous atmosphere around the grease, rather than investigating the grease only at one position inside the bearing (Dubek et al., 2024).

As stated in Section 1, this publication specifically addresses the final step of the lab-to-field approach concerning the verification of the robustness of both the sensor and the algorithm that was carried out during the field demonstration. For this purpose, the HSAB system used the concept illustrated in Figure 2 and described by Dubek et al. (2024). The main advantage of this approach is that the sensor is not in contact with the grease at only one specific position on the bearing, thus measuring the grease condition only at this location. Owing to the carefully chosen sensor position in the bearing atmosphere, a more holistic approach for detecting the water content of the grease was established. Furthermore, the given mounting position of the sensor does not require any change in the bearing construction itself because only the bearing cover is exchanged. As a result, the key innovation by using the concept of the HSAB system is the determination of the grease–water content based on monitoring the temperature as well as the humidity level of the atmosphere above the lubricating grease.

For the field test, three different sensors were used to elaborate on differences in functionality and robustness. Because the integration of only one sensor is intended, the sensor that is the most robust based on this final field evaluation would be selected for further application. All sensors applied within the HSAB system were commercially available temperature and humidity sensors. Figure 3 summarizes the specifications of the sensors used for the field test. The main aspects for choosing these sensors were commercial availability, size, electrical output and reported robustness according to the railway standard IEC 61373 (BSI, 2010), as published by Dubek et al. (2024).

According to the data sheets, the measurement accuracy of the temperature is ±0.4 °C for all sensors, while the relative humidity measurement tolerance range is ±1.5 RH% for sensor 1, ±2 RH% for sensor 2 and ±1.3 RH% for sensor 3 at the used temperature (from −10 °C to +70 °C) and humidity (from 0 RH% to 90 RH%) range, respectively. Outside these temperature and humidity ranges, higher errors were expected. The sensors used were calibrated by their manufacturers, and their functionality was confirmed with the corresponding calibration certificates. The transducer elements were protected against dust and dirt by using a customized coating. Furthermore, all soldering points were sealed against corrosion by the manufacturer. In addition, the sensors could also provide information on other physical parameters such as dew point, mixing ratio and absolute humidity (Dubek et al., 2024).

For the field demonstration of the HSAB system, an extended bearing cover from the company PJ Messtechnik GmbH equipped with selected sensors, as listed in Figure 3, was utilized, in which the sensors were mounted onto a customized sensor holder together with their cables for energy supply and communication as well as the clamping box, as shown in Figure 4(a). This arrangement ensured that the sensors were best exposed to the air atmosphere surrounding the lubricating grease inside the bearing. For the field demonstration, the sensors were additionally encapsulated into a two-component synthetic soft polymer resin for mounting in the bearing cover. During this process, the measuring part of the sensors (transducer elements) was not poured out to ensure contact with the surrounding air. Figure 4(a) represents the setup before encapsulation. The plug connection outside the bearing cover [Figure 4(b) on the right] was implemented using a socket fixed on the bearing cover and a plug on the cable from the Han 3HPR series from Harting with railway approval, according to the standard IEC 60529 (DIN, 2016). To protect the sensors from penetrating the grease directly into their sensing components, they were covered by a thin metal plate with a small hole on its edge (not depicted) to allow air exchange inside the bearing cover during operation.

During the field test, the HSAB system and other sensors were installed at selected points on the used Y25 bogie of the freight wagon, as depicted in Figure 5.

The customized bearing cover at position 1 was used to mount the HSAB system (Figure 4). The other tailored axlebox bearing cover at position 2 contained a generator unit from PJ Messtechnik GmbH that powered the measuring system components throughout all wagon movements. Both bearing covers met the IP69 protection requirements according to the standard IEC 60529 (DIN, 2016) and were sealed with a suitable sealing ring at their edges to the bogie frame. At the time of the standstill periods of the wagon, no power supply was generated and, therefore, the system was switched off. During the field demonstration, an additional temperature and humidity sensor (same type as sensor 2) used as a reference sensor was mounted on the bogie frame of the wagon (Figure 6, right). This was aimed at providing information about the relative humidity and temperature of the surroundings of the wagon and, therefore, information about the current weather situation and how this influences the water content of the lubricating grease in the axlebox bearing. On top of the bearing cover at axlebox no. 1, an acceleration sensor from PJ Messtechnik GmbH was integrated, which logged acceleration data in horizontal (a_y,Cover) and in vertical (a_z,Cover) directions during traveling.

Following measurement parameters were registered by mounted sensor systems, as depicted in Figure 6, for field validation and later evaluation of the field data gained:

• temperature and relative humidity inside the HSAB system;

• reference temperature and relative humidity outside the HSAB system directly on the bogie;

• horizontal (a_y,Cover) and vertical (a_z,Cover) accelerations on the top of the HSAB system;

• traveling speed;

• sea level;

• GPS coordinates of the wagon; and

• GPS time data.

For the power supply of the sensors listed above and for data acquisition, storage and transmission, the Wagon Tracker Advanced from PJ Messtechnik GmbH was utilized.

For the field demonstration, a wagon in normal operation from the company Steiermärkische Landesbahnen (StB) was selected. The wagon was operated on regular railway tracks in Austria both in winter and spring intervals for a period of 10 weeks (from 13 February to 28 April, 2022). Based on the calculation from the recorded velocity signals, a total mileage of approximately 15,000 km was traveled. Figure 7 shows all trips in the field test based on recorded GPS coordinate data.

In this section, the results of the field test regarding the recorded temperature and humidity signals of the HSAB system are discussed, which served as the basis for the calculation of the grease–water content (Dubek et al., 2024). The evaluation of the signals of all three humidity sensors provided nearly identical results; consequently, the signals of sensor 2 were selected for a detailed evaluation and comparison with the respective signals of the reference sensor mounted on the bogie frame. Because the algorithm to compute the grease–water content also requires the grease temperature as an input parameter that could not be measured during field evaluation, an approach to estimate the grease temperature from the sensor temperature was applied. At the end of this section, the results of the sensor functionality verification test after field evaluation are explained.

As discussed in the previous sections, the prevailing temperature and humidity signals were measured by the sensors of the HSAB system when the wagon was traveling during the field validation. In this case, measurements were taken every second, while, in standstill phases, a measurement was carried out every 6 h for 15 min to check the functionality of the entire measuring system. As an example, the sensor signals together with the associated velocity signals (both original and smoothed) for the 7th trip of the field test on the way from Linz Stadthafen to Hopfgarten on February 22, 2022, with three measurements in standstill phases at the beginning, are depicted in Figure 8. Here, the temperature and humidity signals of sensor 2 are marked as Sensor2-T and Sensor2-RH, while the reference sensor signals are labeled as Ref-T and Ref-RH, respectively. As can be seen, the temperature of sensor 2 increased at the beginning of the wagon movement due to the heat generated in the bearing. Furthermore, temperature changes in the surroundings (Ref-T) during the trip are also visible in the temperature signal of sensor 2, but delayed, as shown by the time-shifted decrease in temperature. Regarding the relative humidity of sensor 2, a correlation to the traveling speed can be established at enhanced speed, humidity rose, while at decreasing speed, it declined. This can be explained by the influence of grease temperature on speed. At higher speeds, the grease temperature also increased, resulting in a higher evaporation rate of water from the grease. This resulted in a higher humidity of the surrounding atmosphere, as measured by sensor 2. Consequently, knowledge of the grease temperature is important for the interpretation of the grease–water content. Considerations of the grease temperature are discussed in Section 3.3.

Similar to that, temperature and relative humidity signals for the entire field test are shown in Figure 9 in which bold lines (both the continuous for sensor 2 and the dashed for the reference sensor) illustrate the smoothed data to better show the resulting global trends. The minimum temperature was −7.5 °C and the minimum relative humidity was 9.3 RH%, while the maximum temperature was 37.2 °C and the maximum relative humidity was 88.2 RH% registered by the sensor 2 in the bearing cover, respectively. As can be seen, sensor 2 provided reliable signals during the entire time interval of the field test and the same could be concluded for the other two sensors of the HSAB system. Since the maximum recorded humidity value by sensor 2 is not close to 100 RH%, it indicates that no excessive water contamination occurred in the bearing cover during the field test. Otherwise, in the event of an unwanted water intake, sensor 2 would have shown at least 90 RH% for a longer period. Regarding the reference sensor, the minimum temperature was −9.4 °C and the minimum relative humidity was 7.5 RH%, while the maximum temperature was 35.9 °C and the maximum relative humidity was close to 100 RH%. As a result, based on the comparison of smoothed temperature signals in Figure 9(a), it is demonstrated that the characteristics of the sensor 2 over time and that of the reference sensor are very similar while the sensors in the bearing cover generally showed approximately 5 °C more. As the temperature difference between them remained approximately constant, it is assumed that the bearing did not overheat during the field test. This confirms a further advantage of the HSAB system as a significant deviation of the 5 °C difference to the reference sensor would indicate an occurring bearing damage. A similar correlation can be established also for the smoothed relative humidity signals of both sensors as illustrated in Figure 9(b). However, the humidity of the reference sensor was tendentially higher than that of the sensor 2 because the temperature in the bearing cap was generally higher resulting in a lower relative humidity. To represent the content of the temperature signals more expressively, they are also shown in form of histograms Figure A1. After subtracting these temperature signals from each other at every time stamp (Figure A2), the modal value of the temperature difference was 5.8 °C. The respective histograms of the relative humidity signals are depicted in Figure A3. The humidity values of the reference sensor covered a wider humidity range (from 10 RH% to 100 RH%) than the sensor in the HSAB system (from 15 RH% to 80 RH%). Since the relative humidity from sensor 2 is concentrated in the range of 20 RH% to 60 RH%, it can be determined that no water uptake took place indicating a save operation due to the absence of any water ingress during the entire field test.

By analyzing the weather data (Meteostat, 2015) from the field test recorded by the reference sensor, three main groups and accordingly three main types of weather conditions could be distinguished based on relative humidity data, as the associated temperatures covered a wider range, which were:

1. sunny (approximately Ref-RH < 50 RH%);

2. cloudy (approximately 50 RH% < Ref-RH < 90 RH%); and

3. rainy (approximately Ref-RH > 90 RH%).

The correlation between the data of the reference sensor and the prevailing weather conditions for both the temperature and relative humidity was verified by extracting weather data from weather stations geographically closest to the actual position of the wagon. Therefore, in comparison, the reference sensor reflected the predominant weather conditions.

In terms of mechanical loading on the sensors, the highest mechanical loads occurred vertically, as impacts and shocks mainly occurred in this direction in such field applications. The highest registered acceleration peak value closest to the sensors of the HSAB system on the top of the corresponding bearing cover (Section 2.2.3) was a_z,Cover = 334.7 m/s² (≈ 34.1 g_RMS) during the field test. Comparing this to the acceleration value of the shock test (a = 1,000 m/s² along all three axes) prescribed by the railway standard IEC 61373:2010, Category 3, axle mounted (BSI, 2010), it can be stated that the HSAB system was not exposed to such high acceleration values during the field validation than at the time of the laboratory investigations as discussed by Dubek et al. (2024). Consequently, the sensors were properly evaluated for field demonstration, which they successfully passed, as discussed in Section 3.4. Furthermore, acceleration data were evaluated to estimate the number of impacts that could be distinguished from the background noise of the acceleration signal. Therefore, an acceleration limit of 10 m/s² was applied, resulting in a total of approximately 4.4 million impacts exceeding this limit. This means around 300 impacts per kilometer traveled. In addition, 4,288 impacts with acceleration values higher than 100 m/s² and 84 impacts with acceleration values higher than 200 m/s² were counted. The latter vibration and shock acceleration peak values were significantly below those applied during lab testing (Dubek et al., 2024), as mentioned above, and served only to ensure the mechanical integrity of the HSAB system during even under these harsh environmental conditions field validation. This demonstrates that the HSAB system operated properly the vibration range without any degradation. But, in the scope of this work, the acceleration sensors were not integrated for predictive failure analysis.

In summary, all the sensors of the HSAB system provided reliable signals, and no unexpected behavior was detected, which indicates that the HSAB system can operate in the harsh environment of the bearing cover with regard to both environmental and mechanical loads.

In this section, the sensor signals obtained from the HSAB system are further evaluated using the algorithm developed under laboratory conditions. Therefore, an empirical formula was elaborated to correlate the water content in the lubricating grease (⁠$wtppm$⁠) and the relative humidity (⁠$RH%$⁠) at the sensor position in the HSAB system, as described by Dubek et al. (2024). Briefly, equation (1) was derived to properly calculate the grease–water content (⁠$wtppm$⁠) based on the sensor signal of the relative humidity (⁠$RH%$⁠) considering the temperature of the grease (⁠$TG°C$⁠) and the atmosphere at sensor place (⁠$TS°C$⁠) as follows:

The coefficients in this equation are the characteristic shape (K) and scale (b, c) parameters, which weigh the impact of temperature and relative humidity (Dubek et al., 2024).

The empirical model used to estimate the grease–water content is based on the principles of thermodynamic equilibrium. In accordance with Raoult’s law (Guggenheim, 1937), the partial vapor pressure of water in the gas phase above the lubricating grease correlates with its mole fraction in the grease structure and the saturation vapor pressure, which in turn follows the Antoine’s equation (Thomson, 1946). This approach implies that the grease–water content at a given temperature in the lubricating grease is approximately proportional to the relative humidity in the surrounding air enclosed by the bearing cover. The temperature dependence of the saturation vapor pressure is accounted for by exponential terms in the model (see $e-b·TS°C$ and $e-c·TG°C$ in equation (1)). This formulation is consistent with methods used in vapor–liquid equilibrium studies, such as those applied in membrane distillation modeling, as published by Alhathal Alanezi et al. (2021).

As described in equation (1), three input variables are necessary to compute the grease–water content, namely the temperature at the sensor position, the relative humidity in the air atmosphere of the bearing cover surrounding the grease and the temperature of the lubricating grease. As it was not possible to monitor the lubricating grease temperature in the axlebox bearing directly because of inaccessibility and safety reasons during the field test, the following approach (equation (2)) was used to estimate this parameter:

according to Tarawneh et al. (2009, 2012). For this approach, no additional information, apart from the operational and standstill phases, was required.

For the time intervals when the wagon was in standstill phases, it could be assumed that the grease temperature was the same as that of the sensor in the bearing cover. During the operation of the HSAB system, the estimation of the grease temperature adapted from Tarawneh et al. (2009, 2012) was used, in which the grease temperature under comparable operating conditions was approximately 30 °C higher than that in the immediate environment outside the bearing measured by the reference sensor. Considering the above-derived modal value for the difference between the sensor temperature in the bearing cover and the temperature of the reference sensor, which is around 5 °C (see Section 3.1), it could be assumed that the grease temperature in motion was approximately 25 °C higher than the sensor temperature in the HSAB system. Using this estimation, the grease–water content for the field test could be determined based on equation (1). The characteristic curve calculated for the grease–water content utilizing this approach is depicted in Figure 10. To better illustrate the resulting overall trends, the original curve was smoothed by the moving average algorithm (considering 100,000 data points for averaging at all working points representing a time frame of about one day), the result of which is shown in Figure 10 in red. A grease–water content value of around 800 ppm was calculated for both the beginning and the end of the field test. In addition, the main weather conditions prevailing during the individual periods, as described in Section 3.1, are shown in the background.

For a comparison, grease reference samples from different positions of the associated bearing were taken directly after the end of the field test. The grease–water content of these samples was measured later by the Karl–Fischer titration method (DIN, 1974) which resulted in 768 ppm. After comparing the value from the laboratory method and from the developed algorithm, a great agreement could be established as the sampling and the end of the field test took place on the same day. Since no grease samples could be taken during the field test for safety reasons, it is assumed that the grease–water content has changed according to the values indicated by the algorithm.

In the next step, it was examined to what extent the resulted curve for the estimated grease–water content from the field test correlates with the humidity signal of the reference sensor, as this was not considered in the algorithm. In Figure 11, the characteristic of the grease–water content using the first approach in red (the same curve as shown in Figure 10) is compared to the relative humidity signal measured by the reference sensor in blue (dashed bold line). In general, both curves exhibit similar time behavior. Therefore, it can be concluded that the grease interacts with the atmosphere and changes its water content, depending on the surrounding environmental conditions. During rainy periods (e.g. at the beginning of April 2022), a higher water content was present in the grease than in sunny dry weather conditions (e.g. between February 28 and March 10, 2022). This can be explained by the breathing effect of the bearing through the labyrinth shaft sealing on the inside of the axlebox bearing (Figure 2, right), which caused a time delay, particularly in the event of water transfer from the lubricating grease into the surrounding atmosphere in the bearing cover.

After the field evaluation, the HSAB system was tested in the laboratory under environmental conditions (temperature and humidity) according to the set-up test series used during laboratory investigations as described by Dubek et al. (2023). It covered all relevant combinations of temperature and relative humidity as the aim was to demonstrate the still existing correct functionality of the sensors used in the HSAB system, compared to a calibrated reference sensor, the type of which was the same as that of sensor 2.

The following states were examined as part of this test series:

• Normal: temperature and humidity at room conditions;

• Hot: T = 60 °C with low humidity;

• Wet: RH ≈ 100 RH% at room temperature; and

• Hot + Wet: T = 60−70 °C with high humidity.

To ensure stable environmental conditions at the sensor positions, the original metal plate was removed from the HSAB system (Section 2.2.2), and the top of the bearing cap was covered with another metal sheet in which holes were constructed to provide humid air and integrate the mentioned reference sensor. Next, the bearing cover was placed in a temperature chamber (Heraeus Vötsch VM 04 / 300) equipped with an additional self-made humidity control system that ensured the appropriate temperature and humidity conditions. During the verification test, all sensors of the HSAB system delivered approximately identical signals. Furthermore, the sensor signals deviated only slightly from the reference sensor signals in the equilibrated states (see Table 1) because all discrepancies were in the range of the measurement error of the sensors used (see Section 2.2.1). However, larger deviations occurred during the adjustment processes to set the respective stabilized test conditions for temperature and relative humidity.

The reason for these deviations is that the test sensors were encapsulated in a resin in the bearing cover (explained in Section 2.2.2), as the heat transfer to and from the sensor during the heating or cool-down phase was dependent on this material. In contrast, the chosen reference sensor was completely in air inside the bearing cover; therefore, the heat exchange with the surrounding atmosphere was not influenced by the encapsulation material.

The field evaluation of the customized HSAB system was carried out under varying environmental conditions with the aim of delivering important information about the condition of the lubricating grease in a wheelset axlebox bearing of a freight wagon. The results of the field test clearly demonstrated that the HSAB system provided reliable and stable signals. The sensor signals (temperature and relative humidity), along with the estimated grease temperature, were used to calculate the grease–water content using a specific algorithm. In addition, dry and wet weather periods were clearly detectable. The correct functionality of the sensors used in the HSAB system was also successfully confirmed by the assessment after functionality verification in the laboratory following the field test. Therefore, the HSAB system is considered suitable and sufficiently robust to operate under harsh rail vehicle operating conditions, thus making a valuable contribution to the reliable and safe operation of freight wagons.

The correlation between the grease–water content and the relative humidity of the environment, as recorded by the reference sensor on the bogie frame, also confirmed the applicability of the HSAB system. The grease–water content changed owing to the interaction with the surrounding atmosphere at the shaft seal, resulting in changes in the environmental conditions prevailing at the sensors in the bearing cover.

The presented approach with the HSAB system can also be applied in similar cases where grease-lubricated components operate within a relatively enclosed environment and a humidity sensor is integrated without direct contact with the lubricating grease. Examples include cars, utility vehicles, heavy-duty machinery, wind turbines and ships. Given that the chemical composition of the lubricating grease significantly influences its interaction with water, it is expected that the characteristic parameters of the developed algorithm will require adjustment for other types of lubricating greases. For future applications addressing different grease formulations, the characteristic parameters of the algorithm (see equation (1)) will need to be re-evaluated and modified accordingly to beneficially apply the HSAB system also to other grease types.

In the future, there will be more essential aspects to be considered. The first is the industrialization effort of the HSAB system for a higher technology development level and, as a result, a marketable product that can be installed on any freight wagon with minor adaptations. Concerning approaches to calculate the grease–water content, a universal temperature model to estimate the grease temperature should be developed based on the knowledge of the thermal balance model inside the bearing. This estimation can be applied if additional information (such as the traveling speed and load) is available. Here, the considerations between relative humidity and traveling speed in relation to the grease temperature, as exemplarily explained in the description of Figure 8, can be considered. Consequently, the selection of the respective approach to calculate the temperature of the lubricating grease used depends on the available information in addition to the HSAB system.

This work has received funding from the Shift2Rail Joint Undertaking (JU) under grant agreement no. 101004051. The JU receives support from the European Union’s Horizon 2020 Research and Innovation program and the Shift2Rail JU members other than the Union. The article reflects only the authors’ views and the Joint Undertaking is not responsible for any use that may be made of the information it contains.

Parts of the presented results (simulation of lubricant degradation) were acquired due to funding from the Austrian “COMET” program (InTribology2, project no. 906860) via the Austrian Research Promotion Agency (FFG) and the federal states of Niederösterreich and Vorarlberg and due to work carried out at the Excellence Center for Tribology (AC2T research GmbH). The team thanks PJ Messtechnik GmbH and Steiermärkische Landesbahnen (StB) for the support of the field demonstration. The authors acknowledge TU Wien Bibliothek for financial support through its Open Access Funding Program.