Time-Resolved On-Board Measurements of TRWP Using Distributed Particle Sensor Systems

Abstract

The focus of this article is on the measurement of tire and road wear particles (TRWPs) during vehicle operation. The long-term objective is to determine the sources of particulate matter. Consequently, the development of sustainable tires can be supported in the future by identifying factors influencing the concentration of particulate matter in vehicle-based tire tests. In an initial campaign, a test vehicle was equipped with a total of seven low-cost sensors (LCSs) for measurement campaigns on an isolated outdoor test track. The purpose of this was to evaluate the particle measurements in combination with GNSS data and driving data such as acceleration and speed. The potential observed in the initial investigation led to further investigations with an advanced, interconnectable modular particle and environmental sensor system (iMPES), which was developed in-house. The iMPES records measurement data for PM10 via the PMS7003 and PM100 via the SDS198 at 1 Hz over a period of up to 6 h, using a mobile power supply. The findings of the study indicate a robust characterization of the particle concentrations over the temporal and local course of the campaign drives. The results demonstrate the potential of the method to be part of a methodology to differentiate the particle sources and to derive influencing factors on the particulate matter concentration. The paper proposes a methodology for the mapping and analysis of lap-based data on a normalized route. Consequently, an inquiry into the local and driving-dependent dynamics is conducted, alongside a comparison with driving data.

1. Introduction

Road traffic inevitably leads to mechanical abrasion of tires and road surfaces. As a result of these interactions, particles are produced that are released into the air through direct abrasion or the resuspension of materials. These emissions contribute to particulate pollution and are increasingly the subject of health and environmental research.

Depending on the driving conditions, around 0.1–10% of tire abrasion is airborne by mass [1]. This equates to specific emissions of between 0.00093 and 11 mg/vehicle/km, with a median value of 1.1 mg/vehicle/km and an average value of 2.7 mg/vehicle/km [2]. These values only include tire abrasion and resuspension, excluding brake or road wear. The total amount of tire abrasion is approximately 110 mg/vehicle/km and 68 mg/vehicle/km/t, respectively. This implies a direct correlation of emissions to vehicle mass [3]. In the European Union, annual emissions from tire abrasion are estimated at 1,327,000 t [4].

In recent decades, regulatory measures have played a crucial role in significantly reducing particulate matter from vehicle emissions. A further contributing factor has been the growing electrification of vehicle fleets, which has led to a marked decline in exhaust-related emissions. However, this shift has also introduced new challenges: electric vehicles tend to be heavier due to battery weight, which in turn increases non-exhaust emissions such as those from tire and road wear. Compounding this issue is the global expansion of the vehicle fleet, which continues to drive up overall emission levels despite technological advances [5]. A number of studies have confirmed the ecotoxicological relevance of tire abrasion, particularly with regard to its impact on aquatic organisms [6,7]. However, the health relevance of airborne tire and road wear particles (TRWP) requires further clarification [8]. From an environmental policy perspective, a reduction in anthropogenic particle emissions is imperative. A central aim of emissions research is to differentiate the origin of measured particles. For example, particles may originate from tires, brakes, exhaust gases, or even biogenic sources such as pollen. The development of suitable methods for source identification and the characterization of particle emissions across different temporal and spatial scales is therefore essential.

A number of experimental approaches have already been developed to measure particle emissions directly at the vehicle. One such study [9] employed an electric low-pressure impactor (ELPI) alongside a high-volume sampler, both installed on a test vehicle and operated under real-world conditions. Building on this, another setup [10] positioned two ELPIs and two optical particle counters (OPCs) on the front wheels, with measurement points—15 cm behind the tire and 7 cm above the road surface—strategically chosen based on CFD simulations. In a separate investigation [11], researchers used a vacuum nozzle to extract particles behind the front wheel, complemented by an encapsulated brake system for isolated analysis. Notably, this setup revealed a correlation between PM10 emissions and the transmitted friction force, although the findings were obtained under controlled test conditions. Chatbouillot et al. [12] employed an extraction system comprising three nozzles on the left rear wheel to ascertain the emission factor of tire abrasion. The left rear wheel was selected due to the likelihood of greater external contamination on the right side, attributable to its proximity to the roadside. For the purpose of isokinetic extraction, a sampling line was integrated in order to facilitate analysis with an ELPI+. An aerodynamic study was conducted with the objective of ascertaining the optimum extraction positions. In [13], Schmerwitz et al. employed an extraction nozzle system positioned behind the left rear wheel, with the collection of particles being conducted within a HEPA filter. Titanium dioxide (TiO₂) was used as a marker in the test tire, with the objective of distinguishing the primary particles from foreign particles. A detailed investigation was conducted into the size distribution and shape of the particles. Subsequently, a comparison was made between the experimental results and the emission results from a test rig. It was determined that the test bench exhibited a propensity to measure larger particles. The modal value was determined to be 110 µm in comparison to 90 µm in the vehicle-based test.

Löber et al. [14] conducted a series of particle measurements on a chassis dynamometer utilizing a rudimentary driving cycle comprising acceleration phases, high-speed phases, and braking events. The findings of the study demonstrated that, as had been anticipated, particle emissions occur during braking. Furthermore, the vehicle’s longitudinal acceleration leads to increased particle emissions due to tire–road contact. Nevertheless, it has also been demonstrated that elevated vehicle velocities are associated with increased concentrations of particles.

This paper introduces a novel, distributed sensor system designed for mobile, time-resolved measurement of particle concentrations on vehicles. The PM10 sensors used in this study are based on low-cost sensor (LCS) technology, as described in [15], and were calibrated against a Palas Promo 1000 in accordance with [16]. In addition to our calibration, ref. [17] also shows that the sensors produce consistent results. This allows relative comparative studies to be carried out in particular. LCS have previously been applied in traffic-related studies to capture particle emissions with both spatial and temporal resolution [15,16]. Their high inter-sensor comparability allows for the derivation of reliable, relative measurements of particle distribution.

To complement the sensor system, the test vehicle is equipped with a tire particle collection device mounted on the left rear wheel, enabling fractionated and partially online analytical detection, based on [13]. While LCS have so far been predominantly employed to monitor local air quality or assess pollution from different modes of transport [18,19,20,21] the distributed sensor system presented here opens new avenues for investigating the formation and dispersion of particles generated directly by the vehicle.

This concept was first tested during a preceding research project, which forms the foundation for the present work. The following section outlines the preliminary studies conducted on a closed test track, aimed at examining TRWP emissions under controlled conditions and evaluating the performance of the distributed sensor system.

As part of this study, a methodology for normalizing lap-based measurement runs is being implemented, which improves location- and driving situation-related evaluation when the same driving cycle is performed multiple times.

Our study focuses particularly on local and driving-related influences on emission behavior. Therefore, we only use one type of tire. As expected, the literature shows that the type of tire influences emission behavior, as the emission level [22] and the influence of the load direction [23] vary. The methodology developed herein provides a basis for the investigation of such questions in the future. First, we will present the results of preliminary investigations that motivated to our current study. Then, we will introduce the iMPES that we developed, as well as the setup of the measurements considered here, followed by a presentation of the results and their discussion.

2. Preliminary Work

In the course of a preceding research project, the authors conducted measurement campaigns with a test vehicle on a closed and previously cleaned test track. The objective of this study was to investigate the emissions of TRWP under controlled conditions and to evaluate the functionality of distributed sensor systems for the purpose of particle time-resolved recording. The ensuing sections will present the test setup and the significant results of the preliminary measurement campaign. The test vehicle was equipped with seven LCS in order to measure PM10 and PM2.5 concentrations (Figure 1).

In addition to the sensors installed on-board, stationary sensors were also utilized along the test route in order to investigate particle emission. To detect particle transport along the route, additional 30 LCSs were installed at the roadsides (see Figure 2). These recorded the PM10 concentration. In addition, a PM100 sensor was installed in the “M” curve area.

To simplify the referencing of sections of the test track, these were numbered with letters in the direction of travel.

The ensuing measurement results are derived from the combined analysis of both mobile and stationary sensors, and are presented in the subsequent section. The concentrations of PM2.5, as determined by the on-board measurement technology, exhibited significant variations depending on the position of the sensor (see Figure 3). The highest concentrations were detected at position 1 (above the nozzle), while position 6 (roof) and position 7 (door) exhibited the lowest values. The values of PM2.5 and PM10 exhibited a significant correlation with an r-value greater than 0.935 and with the p-value near zero.

The measurements acquired from the stationary PM10 sensors along the designated route did not reveal any significant alterations in concentration during the course of the entire campaign that could be attributed to the test vehicle’s journey. Conversely, the PM100 sensor in measurement section “M” documented substantial maxima when the vehicle passed by (see Figure 4). Increases in concentration were observed, particularly at short vehicle distances, such as, for example, at time $t_A$. These increases in concentration persisted for approximately 30 s, a duration attributable to the sensor’s operational characteristics. No increase was observed at greater distances from the vehicle, such as, for example, at time $t_B$. The difference between variants $t_A$ and $t_B$ is due to the roadway used in each case and the different direction of travel.

Moreover, additional correlations were investigated between the measured concentration values and the recorded driving dynamics data. Furthermore, acceleration data in three spatial directions, vehicle speed and GNSS positions were recorded. An evaluation showed that high PM10 concentrations for sensor position 1 occurred particularly in the acceleration range (section “D”) and at increased lateral acceleration in curves (sections “F” and “M”) (Figure 5).

In addition to the LCSs, we collected particles on the rear left wheel using a vacuum nozzle for ex situ analysis. Concurrently, we sampled on the rear right wheel with an TSI Mini-MOUDI impactor with 8 stages (see Figure 6). This enables offline analysis of the chemical composition and particle size distribution.

The tire utilized in this study contained titanium dioxide (TiO₂) as a tracer material with a mass fraction of 6% [13,24]. The analysis of three samples using energy dispersive X-ray spectroscopy (EDX) clearly demonstrated the presence of titanium (Ti). The results of the analysis, as illustrated in Figure 7, demonstrate that the titanium content initially decreases with decreasing step size and exhibits peaks at 1.8 µm, 0.32 µm and 0.18 µm. The latter is consistent with the primary particle size of TiO₂ (X₅₀ = 0.385 µm; log-normally distributed). An analysis of trends in mass fraction indicate that the majority of tire abrasion particles are found above 5 µm.

Furthermore, the analysis encompassed the components of tires and brake disks, namely iron, in addition to barium, an additive employed in brake pads. The maximum iron content detected was at the nominal cut point 0.32 µm. The barium distribution exhibited a bimodal curve, with peaks at 3.2 µm and 0.32 µm.

The following statements can be deduced from the preliminary investigations carried out. The results show the basic feasibility of distributed LCS in terms of near-emission particle detection on vehicles. The integration of a PM100 sensor exhibited significant potential for the detection of large TRWPs, given that a substantial proportion of TRWPs are within the PM100 range [25]. The observed correlations with acceleration phases and cornering indicate a dynamic-dependent emission behavior. In addition, the on-board particle collection provides new perspectives for element analysis and origin assignment, particularly when using suitable tracer materials, such as TiO₂.

3. Measurement and Sensor System Architecture Concept

In the OLRAP project (Online Analysis of Airborne Tire Abrasion Particles During Formation and Differentiation from Particles of Other Origins), a measurement and analysis concept was developed to detect airborne tire abrasion particles under real-world driving conditions. The concept was implemented in a test vehicle and is based on two complementary approaches. Distributed measurement is achieved through the utilization of multiple modular particle and environmental sensor (iMPES) units, building on previous LCS development work [15]. The implementation of targeted sampling, in conjunction with online analysis, is facilitated by a flow-optimized extraction system. The overarching aim of the system is to ultimately integrate both methods into a unified, real-time capable analysis framework.

As illustrated in Figure 8, the comprehensive measurement architecture onboard the vehicle is represented schematically.

A total of seven iMPES units were mounted in strategic locations (see Figure 9). These included the “Roof” unit, which was used to measure the ambient background concentration. Moreover, one iMPES unit was mounted on the driver’s door (“LD”) and another to the passenger’s door (“RD”), with the purpose of measuring the local concentration at mid-height. Furthermore, two iMPES units were mounted on each rear wheel, each with an upper and lower sensor (e.g., “RRU”, “RRD”, “LRU”, and “LRD”), which allowed for the profiling of particle concentration in close proximity to the wheels.

The online particle sampling system consists of a rear suction nozzle (S), which is located behind the left rear tire, and a front suction nozzle (T), which is located behind the left front tire. The front unit is designed to minimize parasitic effects and allows the non-driven wheel to be examined. The rear extraction unit is mounted on the driven wheel and is used for online and offline analysis of the emissions in terms of their composition and particle size distribution. For this purpose, the extracted volumes are directed to the online analysis module (A), where they are analyzed and collected. The analysis module is currently equipped with an ELPI system, and with the addition of sensors such as iMPES is possible. Furthermore, a multisampler was installed in the analysis module, which is used to collect particles under event-specific driving conditions in order to subsequently analyze them offline in the laboratory. The power supply is provided by the power module (P). The data collected on board is then analyzed by the central data logger (D). With regard to the online particle collection system, reference is made to the corresponding results in the publication [26].

4. The iMPES and Experimental Setup

The iMPES has been designed as a modular, cost-effective, and adaptable platform for real-time particulate and environmental monitoring.

The utilization of an ESP32-S3 microcontroller was necessitated by the requirement to ensure the maintenance of adequate power reserves in the future, in addition to the availability of a substantial number of connection options. The base board is equipped with components that facilitate the expeditious provision of a data-recording device. These components are permanently installed on the base board. This includes an SD card connection that is connected in 4-bit MMC mode, which enables sufficient write rates for the expected data rates. Furthermore, a 3.3 V switching regulator has been integrated to provide the operating voltage for the microcontroller from a 5 V supply voltage. Two LEDs of different colors are available to indicate operating states, which can be used to display error states, for example.

The Grove ecosystem from Seeedstudio [27] was utilized to connect the actuator and sensor peripherals. This enables components from this ecosystem to be integrated quickly. Due to the relatively wide distribution, components and accessories are easily available and also allow higher quantities due to the low price. The base board supplies each Grove port with 5 V.

The PMS7003 from Plantower [28] is used as the PM10 sensor. This sensor has been successfully used and calibrated in the preliminary work [16]. A separate adapter board has been developed for the connection via Grove connector, which converts the IDC10 connector to the Grove port. As in the preliminary work, the PM100 sensor [29] is the SDS198 from NovaSensor. The connection to the sensor connector is also achieved using a specially developed adapter board. To provide reference times and GNSS position, an Ublox SAM-M8Q [30] on an Arduino MKR GPS shield is utilized in the iMPES configuration in this work. A Bosch BME680 [31] sensor has been integrated as a Grove module to collect environmental data such as temperature, humidity, and barometric pressure. A separate power module is used to charge an 18650 lithium battery and provide 5 V power.

The firmware was further developed on the basis of the LCS firmware to enable connectivity via the MQTT protocol. The use of the MQTT protocol allows an easy and flexible adaption to different application scenarios. In this project, the central data acquisition of all sensors was realized using the MQTT protocol in a time series database. The firmware was continuously advanced and evaluated during the project and optimized in particular with regard to the cycle times and the accuracy of the cycle rates. Table 1 lists the components and characteristics of the iMPES.

The implementation of the iMPES on the vehicle was achieved through the design and development of a robust housing, as shown in Figure 10. The requirement was for a straightforward and secure installation on the vehicle. The iMPES is mounted using Velcro tape and secured by means of a wire rope which is attached to the housing via a cable tie. The particle inlets and the air intake for the environmental sensors are located on the (transparent) side of the cover. In addition, the GNSS antenna is located close to the lid, thereby ensuring the best possible signal strength and thus the highest possible GNSS accuracy in this configuration.

As in the preliminary work, the measurements were carried out on the same 1100 m test section to minimize interference from external emissions. The track was pressure washed and vacuumed at the start of each day of measurements. The track’s configuration enabled high-speed driving of up to approximately 80 km/h, as well as dynamic driving maneuvers including acceleration, cornering and braking. As illustrated in Figure 11, the driving profiles that were measured along the designated track are presented.

A four-day testing period was successfully completed. Table 2 provides a concise overview of the environmental and dynamic conditions under investigation.

The intra-day measurement results presented in Section 5 are derived from a designated measurement day; the general characteristics of the iMPES measurement results are also comparable across other measurement days. The weather conditions were characterized by clear skies with occasional cloud cover. As indicated in Table 2, the temperatures and humidity levels presented are derived from the official measurements recorded by the German Weather Service (Deutscher Wetterdienst, DWD) [32].

5. Results

In the following, we first look at the measurement results within one day using the example of measurement day 2. This day was selected as it provides a complete database, the same applies to measurement day 1. On measurement days 3 and 4, individual sensors failed due to insufficiently charged batteries. The periods in which data is available on these days are comparable with day 2. As illustrated in Figure A1, the number of evaluable laps per sensor is shown in relation to the total number of laps driven. Figure A2 demonstrates the dynamic progression over the normalized distance, with the RLU sensor serving as a representative example. Despite a partial failure of RLU on day 4, its dynamics remain comparable to those of the other days. The evaluation of the daily measurement data is carried out selectively for individual sensors due to the high volume of data. After analyzing the concentration dynamics within the measurement day, a comparison is made of the dynamics across all sensors and for one sensor across all days.

As illustrated in Figure 12, the first result is that the in-house-developed iMPES shows consistent reliability when performing time-synchronized measurements over the entire measurement day. The dynamic behavior of the individual PM10 concentration curves, as shown in Figure 12, depends on the driving mode of the test vehicle on the test track. The Y-axis in Figure 12 is scaled equally for each sensor in order to be able to better assess the influence of the sensor installation position. A clear difference in the level of PM10 concentration can be observed between the rear left and right wheel housings, the doors and the roof. Nevertheless, similar particle concentration dynamics have been observed at the doors and in the wheel housing, suggesting potential similarities in their behavior. The particle concentration measurement on the roof does not indicate any driving-dependent dynamics in this observation. The assumption that the roof sensor essentially provides information about the background concentration is supported by this observation. However, it is apparent that the concentration in the left wheel housing (“LRD”,”LRU”) is higher than in the right wheel housing (“RRD”,”RRU”), even if the sampling via extraction for further analysis being carried out on the left side. It is possible that the extraction process will result in unwanted turbulence or a variation in the detachment behavior of the particles from the tire.

On the left-hand side, the sensor mounted higher up “LRU” in the wheel housing has a higher concentration than the sensor at the bottom “LRD” (see Figure 9 for positioning). On the right-hand side, the behavior was reversed—statistical comparison can be found in Table A1. The differences between the top (“LRU”, “RRU”) and bottom (“LRD”, “RRD”) are relatively small, so this should also be investigated further in the future, e.g., by installing a second iMPES at each of the corresponding positions and calculating an average value for each of the related sensors.

A break between two measurement blocks is evident on this particular measurement day, occurring at approximately 9:10 and 9:20. The particle concentrations in the left and right wheel housings are at a comparable level during passive phases, i.e., when the vehicle is not being driven.

The profile of the PM100 concentration also demonstrates a driving-dependent dynamic over the entire measurement day. Figure 13 shows a section of the measurement day; the start of each lap after leaving section A is also marked (see Figure 2). It should be noted that the Y-axes are scaled logarithmically here due to the large differences in the absolute value.

It can be seen that the concentrations are significantly higher in the arches of the wheels (“LRU”, “LRD”, “RRU”, “RRD”) in comparison to the concentrations observed at the doors (“LD”, “RD”) or on the roof. In contrast to the PM10 signal, the PM100 signal for the “roof” sensor demonstrates a clear dynamic response that is dependent on driving.

For the measured values PM1, PM2.5 and PM10, the drive-dependent dynamic is only clear for the other sensors, but not for the sensor on the roof.

However, the 30 s integrating behavior of the PM100 sensor must be taken into account (see Section 2, Figure 4). The dynamics at the roof sensor are not identical to the dynamics at the other sensors. While the PM100 curve of the other sensors shows the maximum values at the beginning and end of a lap as well as in the middle of a lap, the “Roof” sensor only shows the maximum range in the middle of a lap. In Figure 13, the beginning and end of a lap are indicated by the vertical lines (At the end of section A); therefore, the midpoint of the lap is located between these two vertical lines.

By recording the driving data including the GNSS position and a matching this data with the particulate matter measurement data using the precise GNSS-supported time stamp, it is possible to analyze the particle concentration along the vehicle route. Figure 14 shows the PM10 concentration along the route for the “Roof” sensor.

It can be seen that there is a geographical influence component for the PM10 particulate matter concentration, which we define as the background concentration in the case of the “Roof” sensor. As is evident from the data, there is a high degree of correlation between the particle concentrations and the velocities, particularly in sections “E”/”F” and “L”/”M”. However, a direct assignment to the speed profile or to the acceleration values is not feasible. It has been demonstrated that the particulate matter concentration for the “Roof” sensor increases at the start of the acceleration phases (section “D”) and only falls to the minimum at the turning points (“A” and “H”). The narrow bandwidth of the concentrations makes local allocation more difficult. Furthermore, in this representation variant it must be assumed that the measured values in each round are similar to each other. As the absolute values are less relevant than the relative course for the investigation of the local influencing variable, we normalize the concentration course for each round to the value range 0 to 1 based on the maximum and minimum values of the respective round.

To do this, we first sort each data point ($y_i$) to the corresponding round, thereby forming a group ($Y_r$). For the normalization, we proceed each $y_i$ with Equation (1) to ${y^{\prime }}_i$.

(1)${y^{\prime }}_i={\displaystyle \frac{y_i-\mathit{min}{Y}_r }{\mathit{max}{Y}_r-\mathit{min}{Y}_r}}, y_i\in Y_r$

Furthermore, we also map the traveled distance in each lap to the value range from 0 to 1. The start of a lap is defined as leaving the western turning point (Crossing from section “A” to section “B”) and is determined using geofencing. For all subsequent data points in the round, the geometric distance to the previous data point is calculated using the Haversine formula (Equation (2)). The symbol $R$ denotes the radius of the earth and the index $i$ or $i-1$ the two consecutive data points. In addition, the letters $lat$ and $lon$ are used as acronyms for the corresponding latitude and longitude coordinates. A cumulative sum is then used to determine the distance traveled in the lap for each data point. The distance is assigned 0 to the starting point of a lap determined using geofencing.

(2)$$\begin{gathered} \Delta \phi =la{t}_i-la{t}_{i-1} \\ \Delta \lambda =lo{n}_i-lo{n}_{i-1} \\ a=si{n}^{2\left({\displaystyle \frac{\Delta \phi }{2}}\right)}+cos(la{t}_{i-1})\times cos(la{t}_i)\times si{n}^{2\left({\displaystyle \frac{\Delta \lambda }{2}}\right)} \\ {d}_i=R\times 2\times {\mathit{tan}}^{-1}{\displaystyle \frac{\sqrt{a}}{\sqrt{1-a}}} \end{gathered}$$

Normalization of the track length to the value range 0 to 1 is carried out on a lap basis by division with the track length value of the last data point of the corresponding lap. As demonstrated in Figure 15, the route length has been calculated and the resultant normalized progress along the route has been visualized.

To enable comparison of the measurement data over the traveled distance progress of each round, it is possible to plot all data points over their corresponding distance progress. Figure 16 shows the raw PM10 concentration over the normalized distance progress for the “Roof” sensor on the left-hand side. On the right side, the normalized PM10 concentration over the normalized distance progress is shown. A polynomial fit is also shown to describe the mean characteristic of the measurement data curve over the distance progress and thus over the geometric position. The fitting function

(3)$f\left(x\right)=\sum _{k=1}^{n}a\times cos\left(2\times \pi \times k\times x\right)+b\times sin(2\times \pi \times k\times x)$

was used to represent the periodic progression resulting from the repeated execution of identical laps. In this particular context, the factor n is used to determine the maximum degree or maximum frequency via the running variable k. The parameters a and b are adjusted out according to the function fit.

In the present context, the degree or frequency of the polynomial was set to 20 to reproduce the complex concentration curve of the examined data (in the following the ELPI+ data) as accurately as possible. To ensure the consistency in the polynomial fits for curves 0 and 1, the measurement data was set three times in succession. The corresponding X-values were adjusted to the ranges −1 to 0, 0 to 1 and 1 to 2, while the Y-values remained unchanged.

Using this normalization, we can better represent the dynamics along the route. Figure 17 again shows the PM10 concentration curve for the “Roof” sensor, this time using the normalization. In the following, only the representation using the fitted polynomial is used to investigate the dynamics over the position.

If we compare the course of the particle concentration at the “RRD” sensor for the same day (Figure 18), we also see that the particulate concentration falls back to the minimum range at the turning points (“A”/”H”). However, in contrast to the “Roof” sensor, the significant increase in particle concentration only occurs later in the braking phases in the westbound direction (section “L”), shortly before the bend and in the eastbound direction in the bend itself (“E”/”F”). Additionally, no direct correlation has been identified between this sensor and any of the driving data.

As already shown with the schematic measurement and analysis setup on-board in Figure 8, in addition to the iMPES, a nozzle, connected to two Kärcher NT30/1 Tact H industrial vacuum cleaner [24] was also installed on the left side of the vehicle, positioned behind the rear wheel. An ELPI+ was integrated into the suction line. A comparison of the PM10 concentration measured by the ELPI+ along the route for the same measurement day (see Figure 19) reveals that the dynamics are significantly more variable over the course of the route. The minimum values are again attained in the areas of the turning points. In contrast to the sensors shown above, the increase in concentration is found in the acceleration phase in the eastbound direction (section “D”) and also a significant increase after the eastbound bend (section “G”). The scatter plot of the PM10 concentration versus the traveled distance progress shows further increases.

The data from the preliminary project of the “Above Nozzle” sensor, shown in Figure 20, was analyzed identically (see Figure 5 in Section 2). The analysis of these data shows a significant increase in the PM10 concentration along the vehicle route in the eastern direction of travel. This increase is particularly noticeable in the acceleration phase (section “D”) as well as before and in the eastern curve (section “G”/”F”). The observed correlation between the ELPI+ measurement data and the aforementioned increase substantiates the validity of the findings. This implies that the iMPES is able to produce qualitative measurement results that are comparable to those of the reference device. However, it should also be noted that high concentrations are detected in other sections of the route, which correlate with the measurement results of the “RRD” and “Roof” sensors, for example. It can be concluded that, in addition to the pure TRWPs, other particles, such as brake particles, can also be measured at the time of formation. Consequently, positioning is a critical parameter when measuring the particle concentration on the vehicle and assigning the potential emission source. To measure mainly the tire-related particles, the positioning of the sensor should be very close to the contact zone of the tire and road surface. Bej et al. determined that the most suitable position was 15 cm behind the wheel and 7 cm above the road [10]. Brandt et al. consider the difference between sampling directly behind the tire and at a later point after extraction [26].

Figure 21 shows the regression polynomials for each sensor on day 2, with the data for PM10 on the left and the data for PM100 on the right. Thereby, a similar local dynamic can be observed for each sensor. The fluctuation range of the sensors “LD”, “RD” and “Roof” is smaller compared to the other sensors. This implies that the concentration varies less along the route.

An analysis of the available measurement data from the PM100 sensors reveals that the dynamic range of the “Roof” sensor is not as pronounced as the other sensors. The deviating dynamics of this sensor can be explained by a combination of two factors. On the one hand, this is the low number of particles measured, whereby each particle has a comparatively high weight. The absolute mass concentration is 1 to 2 orders of magnitude lower than with the other sensors (see Figure 13). On the other hand, the 30 s storage behavior of the sensor must be taken into account (see Figure 4, Section 2). The comparison of PM10 and PM100 data reveals comparable local dynamics.

The comparison in Figure 22 shows the same local dynamics over all measurement days for both PM10 and PM100 using the “LRD” sensor as an example. The results for “LRD” are shown here because failures were recorded for the “RRD” sensor on the PM100 channel. The dynamics of PM10 and PM100 show some differences. The concentration of PM100 shows an earlier increase than that of PM10. In addition, PM10 concentrations tend to be higher in the second half of the route than in the first half. The concentration of PM100 particles, on the other hand, shows a tendency towards higher values in the eastern direction of travel, i.e., in the first half of the route (section 0 to 0.5), compared to the second half of the route. The reason for this must be investigated further, for example, by means of split collection of the particles in the direction of travel and subsequent analysis.

The research results presented in this document can be used as a foundation for further analyses using the methodology described therein. The normalization of the measured values and the normalization of the reference measuring distance makes it possible to compare measured data that would otherwise be difficult to compare due to their characteristics and behavior, which are strongly influenced by external factors.

As illustrated in Figure 23, the vehicle dynamics data is presented using the normalized traveled distance method, akin to the approach employed in Figure 10. It is important to note that, in this context, speed and altitude are normalized, while lateral and longitudinal acceleration are not. This deviation from the norm is attributed to the significance of the direction of action in this particular domain.

A visual comparison of the progression characteristics reveals a clear similarity between the concentration progressions and the velocity progressions. The local shift between the velocity and concentration curves may be indicative of a lag time between the change in velocity and the increase in concentration. However, it is that this shift may be associated with the increase in velocity, such as increased turbulence, braking events after the end of the high velocity or more pronounced lateral accelerations. A direct comparison of the acceleration values and the concentration curves indicates an increase in the concentration curves within the range of 30 to 40% of the distance traveled and within the range of 80 to 90%. In these ranges, significant acceleration events are also observed in the longitudinal direction of the vehicle as well as significant lateral vehicle accelerations. The methodology outlined in this paper demonstrates that a comparison of the driving dynamics values and the concentration values on the short-term scale is not very meaningful for each time point individually. Furthermore, this comparison does not result in any correlations. The subsequent phase of research will include comparative analyses on an average time scale. That is to say, the execution of an integral observation per lap is to be carried out. Despite the objective of ensuring comparable weather conditions on all measurement days, external factors could still influence particle emission levels and size distribution. After performing relative comparisons, it was determined that the influence on the results is negligible. However, further studies should investigate the influence of environmental conditions in more detail.

6. Conclusions

The present study has demonstrated the robustness of the iMPES and has proven its effectiveness in analyzing particle concentrations and the influence of driving behavior with local resolution. The findings suggest that the precise measurement positioning relative to the tire is essential for ensuring the quality of the data, particularly in the context of evaluating the factors influencing the formation of tire–road wear particles (TRWPs). When positioned with the same sampling location, the iMPES achieved results comparable to those of the ELPI+ reference system in terms of concentration dynamics, thereby underscoring its reliability. The use of GNSS-based time facilitated effective distributed measurement, followed by synchronization of the recorded driving dynamics to derive correlations. The method of normalizing lap-based data to the distance traveled and the particle emission level within a lap enabled the analysis of fluctuating particle levels. A correlation was observed between driving dynamics, in particular speed, and particle concentration. Nevertheless, a direct correlation between driving dynamics and particle emissions remains elusive. In consideration of the elevated complexity inherent in driving behavior and emission dynamics, the utilization of short-term comparisons is unfeasible and does not appear to offer any significant utility. The variability of absolute concentration values complicates the correlation of these values with local influences. It is recommended that future studies concentrate on medium time scales, for example, on a lap basis, or a section-by-section analysis in order to better investigate the effects of driving dynamics on particle emissions. The results highlight the significance of context-specific analyses and the potential of the presented methodology for enhancing the interpretation of driving behavior and emissions. The methodology developed facilitates the investigation of further influencing factors using open-field tests, thereby enhancing the efficiency and efficacy of the process. The factors that are pertinent to this issue include the sharpness of the turns within individual sections, as well as the influence of tire type.

Author Contributions

Conceptualization, G.L., S.B., F.S., B.O., G.-P.O. and C.S.; data curation, G.L.; formal analysis, G.L.; funding acquisition, G.-P.O. and C.S.; investigation, G.L., S.B., N.A. and M.H.; methodology, G.L.; project administration, F.S. and B.O.; resources, B.O. and C.S.; software, G.L.; supervision, F.S., B.O. and C.S.; validation, G.L.; visualization, G.L.; writing—original draft, G.L.; writing—review and editing, F.S., B.O., G.-P.O. and C.S. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The datasets presented in this article are not readily available because the data are part of an ongoing study. Requests to access the datasets should be directed to [email protected].

Acknowledgments

We would like to extend our gratitude to the test drivers and other employees of Continental GmbH for their active support in data collection, the practical implementation of the tests and the valuable feedback that contributed to the success of this work.

Conflicts of Interest

Benjamin Oelze, Nadine Aschenbrenner, and Malte Hothan are employees of Continental Reifen Deutschland GmbH. The paper reflects the views of the scientists and not the company. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1 LCS on-board at seven locations, #1: above nozzle, #2: beside nozzle, #3: wheelhouse rear left, #4: mudguard rear right, #5: wheelhouse front right, #6: roof, #7: door front right. In the lower section of the figure, the detailed views are referenced by colored dots, thereby creating a link to the views in the upper section.

Figure 2 LCS at the edge of the test track. The different sections of the route are distinguished by the use of letters. The position of the PM10 sensors is indicated by yellow dots, while the PM100 sensor is indicated by a red dot and arrow in section “M”.

Figure 3 PM2.5 concentration at the different positions on-board. The highest concentrations are measured above the nozzle, the lowest on the roof and at the door.

Figure 4 PM100 concentration (blue) and distance (orange) between the sensor in section “M” and the test vehicle. The distances determined from the vehicle to the sensor are overestimated in this case, which is due to measurement inaccuracies of the GNSS sensor.

Figure 5 PM10 concentration at position 1 along the test track. Marked sections with high concentrations. The letters are used to distinguish between different sections of the route.

Figure 6 Mounting the impactor for particle collection on the rear right wheel.

Figure 7 Comparison of the mass contents of barium and titanium as well as iron plotted over the separation limits of the different impactor stages with moderate operation.

Figure 8 Schematic measurement and analysis setup on-board. The positioning of the iMPESs is indicated by the purple boxes. The “S” and “T” structures represent extraction devices located behind the two tires on the left-hand side of the vehicle. The function of the “A” unit is to collect and analyze the particles present in the extraction stream. The data is collated centrally in “D”. All systems are supplied by the power supply “P”.

Figure 9 Test vehicle with iMPES units at different positions. “RRD”, “RRU”, “LRD”, and “LRU” = rear right/left, down/up; “LD”, “RD” = driver/passenger door; “Roof” = ambient measurement.

Figure 10 The iMPES used in the OLRAP Project. The temperature sensor is a BME680, GNSS Sensor is a UBlox SAM M8Q, PM10 Sensor is an PMS7003 and the PM100 Sensor is a SDS198.

Figure 11 Driving dynamics on the test track: (a) Speed profile (0–80 km/h), (b) Altitude profile (107–123 m above sea level), (c) Lateral acceleration (±0.6 g), (d) Longitudinal acceleration (−0.25 to +0.42 g). All with labeled sections.

Figure 12 PM10 over time profile on measurement day 2.

Figure 13 PM100 over time profile on measurement day 2 with lap starts shown as vertical lines.

Figure 14 PM10 Measurement of the “Roof” sensor over the course of the track on measurement day 2. The letters are used to distinguish between different sections of the route.

Figure 15 Visualization of the normalized traveled distance along the route.

Figure 16 PM10 concentration vs. traveled distance progress on measurement day 2 of the iMPES “Roof”. (a) Raw PM10 Data (b) Round based normalized.

Figure 17 PM10 along vehicle route for sensor “Roof” on measurement day 2 (a) Round based normalized (b) Fitted polynomial based of the normalized concentration; Labeled sections.

Figure 18 PM10 sensor “RRD” on measurement day 2 (a) Fitted polynomial based of the normalized concentration along route. (b) Concentration vs. traveled distance progress.

Figure 19 PM10 sensor “ELPI+” on measurement day 2 (a) Fitted polynomial based of the normalized concentration. (b) Concentration vs. traveled distance progress.

Figure 20 PM10 sensor “Above nozzle” from preliminary work (see Figure 1) (a) Fitted polynomial based of the normalized concentration. (b) Concentration vs. traveled distance progress.

Figure 21 Fitted polynomials for each sensor on measurement day 2 (a) normalized PM10 (b) normalized PM100.

Figure 22 Fitted polynomials for sensor “LRD” on each measurement day (a) normalized PM10 (b) normalized PM100.

Figure 23 Fitted polynomials of driving characteristics versus distance traveled with daily comparison. (a) Polynomial fit of normalized speed; dashed lines indicate overshoots and undershoots, (b) Polynomial fit of normalized speed, (c) Polynomial fit of lateral acceleration, (d) Polynomial fit of longitudinal acceleration polynomial fit; positive values are decelerations.

Table 1

Components and characteristics of the iMPES.

Component
Microcontroller	ESP32-S3-WROOM-1U
PM10, PM2.5, PM1 sensor	PMS7003
PM100 sensor	SDS198
Environmental sensor	BME680
GNSS Sensor	Ublox SAM-M8Q (Arduino MKR GPS)
Characteristic
Sampling rate	1 Hz
Connectivity	WiFi, MQTT, Bluetooth, SD-Interface, USB/Serial
Connector	Grove with 5 V Power rail
Response Time of particle sensors	<1 s

Table 2

Environmental and driving conditions during four test days.

Parameter	Day 1	Day 2	Day 3	Day 4
Temperature (°C)	13.5–25.8	20.4–25.8	18.6–21.3	19.3–23.5
Humidity (%)	34.9–49.0	45.0–48.5	67.9–70.8	65.8–85.0
Wind Speed (m/s)	2.3 ± 0.52	7.7 ± 0.89	4.7 ± 0.67	3.8 ± 0.67
Mean Wind Direction (°)	67	110	259	210
Lat. Acc (m/s2)	0.164 ± 1.78	0.297 ± 1.71	0.541 ± 0.85	0.486 ± 0.89
Long. Acc (m/s2)	0.105 ± 1.00	0.065 ± 0.78	0.017 ± 0.58	0.052 ± 0.49
Mean Speed (km/h)	27.92 ± 23.04	19.33 ± 23.18	12.51 ± 17.41	10.35 ± 16.93
Max Speed (km/h)	75.97	82.42	84.58	82.46
Rounds (#)	73	67	46	46

Appendix A

Table A1

Statistical sensor comparison of PM10 (µg/m³) data for day 2.

Sensor	Mean	Median	Min	Max	Std
LRU	22.21	15	4	124	16.95
LRD	14.25	9	0	96	14.80
LD	7.75	7	2	36	3.05
RRU	14.84	11	4	56	8.67
RRD	18.30	14	4	83	11.88
RD	8.45	8	3	31	2.35
Roof	9.42	9	4	19	2.10

Figure A1 Number of rounds evaluated for each sensor per day; the expected number of rounds to be evaluated is highlighted in gray.

Figure A2 Daily comparison of PM10 emission dynamics over the distance traveled for sensor LRU.

References

1. Joint Research CentreInstitute for Energy, Transport Grigoratos, T.; Martini, G. Non-Exhaust Traffic Re-Lated Emissions—Brake and Tyre Wear PM—Literature Review; Publications Office of the European Union: Brussels, Belgium, 2014.

2. Unice, K.M.; Weeber, M.P.; Abramson, M.M.; Reid, R.C.D.; Van Gils, J.A.G.; Markus, A.A.; Vethaak, A.D.; Panko, J.M. Characterizing export of land-based microplastics to the estuary—Part I: Application of integrated geospatial microplastic transport models to assess tire and road wear particles in the Seine watershed. Sci. Total Environ.; 2019; 646, pp. 1639-1649. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2018.07.368] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30115451]

3. Giechaskiel, B.; Grigoratos, T.; Mathissen, M.; Quik, J.; Tromp, P.; Gustafsson, M.; Franco, V.; Dilara, P. Contribution of road vehicle tyre wear to microplastics and ambient air pollution. Sustainability; 2024; 16, 522. [DOI: https://dx.doi.org/10.3390/su16020522]

4. Wagner, S.; Hüffer, T.; Klöckner, P.; Wehrhahn, M.; Hofmann, T.; Reemtsma, T. Tire wear particles in the aquatic environment-a review on generation, analysis, occurrence, fate and effects. Water Res.; 2018; 139, pp. 83-100. [DOI: https://dx.doi.org/10.1016/j.watres.2018.03.051]

5. Eurostat. Parc de Véhicules par Catégorie par Région NUTS 2; Eurostat: Luxembourg, 2025.

6. Chibwe, L.; Parrott, J.L.; Shires, K.; Khan, H.; Clarence, S.; Lavalle, C.; Sullivan, C.; O’Brien, A.M.; De Silva, A.O.; Rochman, C.M. A deep dive into the complex chemical mixture and toxicity of tire wear particle leachate in fathead minnow. Environ. Toxicol. Chem.; 2022; 41, pp. 1144-1153. [DOI: https://dx.doi.org/10.1002/etc.5140] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34125977]

7. McIntyre, J.K.; Prat, J.; Cameron, J.; Wetzel, J.; Mudrock, E.; Peter, K.T.; Tian, Z.; Mackenzie, C.; Lundin, J.; Scholz, N.L. Treading water: Tire wear particle leachate recreates an urban runoff mortality syndrome in coho but not chum salmon. Environ. Sci. Technol.; 2021; 55, pp. 11767-11774. [DOI: https://dx.doi.org/10.1021/acs.est.1c03569]

8. Baensch-Baltruschat, B.; Kocher, B.; Stock, F.; Reifferscheid, G. Tyre and road wear particles (TRWP)—A review of generation, properties, emissions, human health risk, ecotoxicity, and fate in the environment. Sci. Total Environ.; 2020; 733, 137823. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2020.137823] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32422457]

9. Truong, X.T.; Muresan, B.; Lumière, L.; Liu, Y.; Cerezo, V. Emission dynamics of tire and road contact particles emitted at the rear of the wheel of a passenger vehicle using on-road experiments on different routes. Transp. Res. Procedia; 2023; 72, pp. 2277-2285. [DOI: https://dx.doi.org/10.1016/j.trpro.2023.11.717]

10. Beji, A.; Deboudt, K.; Khardi, S.; Muresan, B.; Lumière, L. Determinants of rear-of-wheel and tire-road wear particle emissions by light-duty vehicles using on-road and test track experiments. Atmos. Pollut. Res.; 2021; 12, pp. 278-291. [DOI: https://dx.doi.org/10.1016/j.apr.2020.12.014]

11. Feißel, T.; Kunze, M.; Hesse, D.; Ivanov, V.; Augsburg, K.; Gramstat, S. On-road vehicle measurement of tire wear particle emissions and approach for emission prediction. Tire Sci. Technol.; 2024; 52, pp. 2-14. [DOI: https://dx.doi.org/10.2346/tire.22.21024]

12. Reifen—Fahrwerk—Fahrbahn; VDI Verlag GmbH: Düsseldorf, Germany, 2019.

13. Schmerwitz, F.; Wieting, S.; Aschenbrenner, N.; Topp, A.; Wies, B. Characterization of Tire Road Wear Par-ticles in the Field and at Laboratory Scale. Proceedings of the 12th International Munich Chassis Symposium 2021: Chas-sis.tech plus; Online, 29–30 June 2021; Pfeffer, P. Springer: Berlin/Heidelberg, Germany, 2022; pp. 729-743.

14. Löber, M.; Bondorf, L.; Grein, T.; Reiland, S.; Wieser, S.; Epple, F.; Philipss, F.; Schripp, T. Investigations of airborne tire and brake wear particles using a novel vehicle design. Environ. Sci. Pollut. Res.; 2024; 31, pp. 53521-53531. [DOI: https://dx.doi.org/10.1007/s11356-024-34543-9] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39192151]

15. Ostermeyer, G.P.; Sandgaard, M.; Otto, J. Usage of Low-Cost Particulate Matter Sensors for Characterization of Emission Dynamics at a Traffic Intersection and for the Description of Pollution in Urban Areas. SAE Int. J. Adv. Curr. Pract. Mobil.; 2020; 3, pp. 988-993. [DOI: https://dx.doi.org/10.4271/2020-01-1636]

16. Ostermeyer, G.P.; Kwade, A.; Sandgaard, M.; Schilde, C. Measurement of Dynamic Signature of Brake Emissions on a Pin-On-Disc Tribotester with Low-Cost Particulate Sensors; EuroBrake Barcelona: Barcelona, Spain, 2020.

17. Badura, M.; Batog, P.; Drzeniecka-Osiadacz, A.; Modzel, P. Evaluation of Low-Cost sensors for ambient PM_2.5 monitoring. J. Sens.; 2018; 2018, 5096540. [DOI: https://dx.doi.org/10.1155/2018/5096540]

18. Gozzi, F.; Della Ventura, G.; Marcelli, A. Mobile monitoring of particulate matter: State of art and perspectives. Atmos. Pollut. Res.; 2016; 7, pp. 228-234. [DOI: https://dx.doi.org/10.1016/j.apr.2015.09.007]

19. Masic, A.; Pikula, B.; Bibic, D. Mobile Measurements of Particulate Matter Concentrations in Urban Area. Proceedings of the 28th International DAAAM Symposium 2017; Zadar, Croatia, 8–11 November 2017; Katalinic, B. DAAAM International: Vienna, Austria, 2017; pp. 452-456.

20. Aix, M.L.; Claitte, M.; Bicout, D.J. A Low-Cost Sensors Study Measuring Exposure to Particulate Matter in Mobility Situations. Proceedings of the 12th International Conference on Sensor Networks; Online, 23–24 February 2023; pp. 32-41.

21. Ramel-Delobel, M.; Peruzzi, C.; Coudon, T.; De Vito, S.; Fattoruso, G.; Praud, D.; Fervers, B.; Salizzoni, P. Exposure to airborne particulate matter during commuting using portable sensors: Effects of transport modes in a French metropolis study case. J. Environ. Manag.; 2024; 365, 121400. [DOI: https://dx.doi.org/10.1016/j.jenvman.2024.121400] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38936028]

22. Woo, S.H.; Jang, H.; Mun, S.H.; Lim, Y.; Lee, S. Effect of treadwear grade on the generation of tire PM emissions in laboratory and real-world driving conditions. Sci. Total Environ.; 2022; 838, 156548. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2022.156548] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35688251]

23. Schläfle, S.; Unrau, H.J.; Gauterin, F. Influence of load condition, tire type, and ambient temperature on the emission of tire–road particulate matter. Atmosphere; 2023; 14, 1095. [DOI: https://dx.doi.org/10.3390/atmos14071095]

24. Aschenbrenner, N.; Oelze, B.; Schmerwitz, F.; Bomkamp, M.; Brandt, S.; Lehne, G.; Schiefer, F.; Ostermeyer, G.-P.; Schilde, C. Tire Wear Particle Evaluation: Outdoor and Indoor. Proceedings of the 15th International Munich Chassis Symposium 2024; Munich, Germany, 4–5 June 2024; Pfeffer, P.E. Springer: Berlin/Heidelberg, Germany, 2025; pp. 43-57.

25. Kreider, M.L.; Panko, J.M.; McAtee, B.L.; Sweet, L.I.; Finley, B.L. Physical and chemical characterization of tire-related particles: Comparison of particles generated using different methodologies. Sci. Total Environ.; 2010; 408, pp. 652-659. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2009.10.016] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19896165]

26. Brandt, S.; Lehne, G.; Schiefer, F.; Oelze, B.; Bomkamp, M.; Schmerwitz, F.; Ostermeyer, G.P.; Schilde, C. Linking Vehicle Driving Maneuvers and Tire Originated Particle Release; [email protected]: Braunschweig, Germany, 2025; (manuscript in preparation; to be submitted)

27. Seeed Studio. Grove Ecosystem Introduction. Available online: https://wiki.seeedstudio.com/Grove_System (accessed on 4 September 2025).

28. Plantower PMS7003 Series Data Manual (Version 2.5). 2016; Available online: https://www.espruino.com/datasheets/PMS7003.pdf (accessed on 18 July 2025).

29. Nova Fitness Co., Ltd. Laser PM100 Sensor Specification (Version 1.2). 2015; Available online: http://www.inovafitness.com:80/software/SDS198%20laser%20PM100%20sensor%20specification-V1.2.pdf (accessed on 23 October 2016).

30. Ublox. SAM-M8Q Datasheet. 2024; Available online: https://content.u-blox.com/sites/default/files/documents/SAM-M8Q_DataSheet_UBX-16012619.pdf (accessed on 24 May 2025).

31. Bosch Sensortec. BME680—Datasheet. 2019; Available online: https://www.mouser.co.uk/datasheet/2/783/BST_BME680_DS001-1509608.pdf (accessed on 23 May 2023).

32. Deutscher Wetterdienst, Deutscher Wetterdienst—Climate Data Center (CDC). Available online: https://opendata.dwd.de/climate_environment/CDC (accessed on 4 September 2025).