Field-based investigation of low adhesion leaf layers: Assessing layer initiation and environmental influences

Leaf layer initiation

Due to the weather conditions over the week of testing (heavy rainfall) no leaf layers were observed developing over more than 24h, with early stage layers being removed through a combination of rain and train passes. Leaf layer initiation was observed through the week where leaves landed on or were crushed into the rail, a range of examples is shown in Figure 4.OPEN IN VIEWER

The black colouration typically observed in well established low adhesion leaf layers is seen here, developing first from the crushed or cut areas of the leaf. The crushing that has occurred on some (Figure 4(c), (e), (f), and (h)) of the leaves was from train passage, and they were pictured when accessing the rail the next day, approximately 17 hours later. For whole leaves that had fallen onto the rail (Figure 4(a) and (d)) the black colouration seemed to emanate from damage on the leaves (fungal spots or cracking). The leaf in Figure 4(g) was observed directly after a train pass and although the whole leaf was crushed, the black colouration only occurred in the moist areas of the leaf. This may have been due to the fact that at this point the railhead was extremely dry. This is explored further in the next section.

The mean value for the layer thickness during all of the testing was 11.6 µm and mean moisture level was 18.7%.

Timelapse imagery

Two main observations were made from the timelapse capture of leaf layers. During heavy rain, a well blackened section of contamination exhibited probable changes in surface energy and appeared hydrophobic (see Figure 5). The rail was “flooded”, but the area around the contamination appeared to repel the water to a large extent.OPEN IN VIEWER

The leaf layers being tracked at the time of these images exhibited a standard level of moisture ∼25% despite the rail being visibly flooded.

The second observation was that on a dry rail with no rain, leaf layer discolouration (to black) occurred only in the moist areas of the crushed leaf (see Figure 5). In the space of 1 h and 5 min the only area with significant discoloration is the thicker, less crushed region of the leaf. The railhead at this time was extremely dry (9.7% WME) and other areas of leaf may have exhibited discoloration if they had been able to absorb moisture. Unfortunately, this leaf section did not remain overnight for further examination.

Axle passes

Layers only survived train passes on the second day of testing. This was likely due to limited rainfall. When the rails were flooded on other days, combined with newly formed layers which were not well bonded, layers were removed by train passage beyond the initial crushing onto the rail. The drying effect of axle passes seen in other testing ³² was seen visually (see Figure 6).OPEN IN VIEWER

However, the changes in thickness and moisture observed in other testing were less apparent when measuring leaf layers on the railhead (see Figure 7). In leaf layer creation work^32,33 layers start at the thickness of the applied material and then rapidly decrease to a more stable thickness that survives for a large number of axle passes. The layers at the VoR struggle to survive one train pass and the layers that did (1 and 6) were already well homogenised at the time of testing, and therefore the thickness 4-40 µm may have been the stable thickness seen in other testing. Capturing the initial material thickness in the field is extremely challenging as the leaves are often drawn into the contact patch as the train passes. The expected moisture trends were likely confounded by intermittent rainfall and the fact that layers began at a limited level of moisture. In layer creation testing, layers began at between 25 and 50% moisture, whereas these “as found” layers began below this level. There were little to no observable changes in CoT for the leaf layer after train passes.OPEN IN VIEWER

Friction testing

CoT was recorded against a variety of measures for the layers observed at the VoR low adhesion site.

The trend of layer thickness and CoT (Figure 8(a)) resembles a log function with CoT rapidly increasing as the layer thickness increases between 5 and 15 µm before reaching a stable level. However, no layers thicker than 35 µm were observed and so additional data is required to strengthen this assertion.OPEN IN VIEWER

As moisture in in the layers increased the CoT reduced (see Figure 8(b)), this relationship (using linear regression) has an R² = 0.4578. The linear regression relationship improves when the values for when rain was falling are removed to R² = 0.5867. The blue values on Figure 8(b) are a reference reading taken at the test site for a cleaned, but flooded rail. There was no clear relationship between layer thickness and moisture in this testing. Moisture above 40% was recorded when the rail visually appeared flooded.

Due to the location and the weather, temperature was extremely stable at the site during testing (variation of 4°C) and no trends were observed related to change in air temperature unlike in testing with wider variation.^11,12 Relative humidity at the site was hard to measure due to the heavy rain inundating measurement equipment. The valid relative humidity values are shown against CoT in Figure 8(c). One layer that existed for a significant time (through multiple wheel passes) and was tested at humidity <90% (“Layer 2”) follows the trend that has been observed for clean rails in the field ¹² with CoT reducing as humidity increases (R² = 0.8713 for a linear regression). No rain fell during the testing of this layer, and as the humidity reduced the layer also dried out (moisture dropping from 32% to 14%).

The moisture in this layer was well correlated with the reduction in humidity throughout the day (R² = 0.8884), whereas the other layers (with associated humidities) were all measured during heavy rain (or extreme humidity before or after rain) where humidity was >90% and there was little relationship between RH and moisture in the layer (see Figure 8(d)).

The results all traction measurements of the layers were compared with the moisture levels and layer thickness at the time of testing (shown in Figure 9). The field data was fitted using the linear regression into a general equation for traction based on directly measuring thickness and moisture on the layer (Equation (1)):

CoT = 14161 100000 − 969 625000 M + 14577 10000000 T

(1)

OPEN IN VIEWER

Figure 9.

Contour plot for regression relationship between moisture levels, layer thickness and CoT.

Chemical testing

FTIR spectroscopy analysis was completed on 3 separate leaf layers found around the VoR test site. The results and images of the railhead are shown in Figure 10. Samples were scanned as found and then dried with lint free paper as there was a significant amount of water on the railhead. The “as found” spectrum is excluded as it was dominated by the water signal. There is a clear difference between the cleaned railhead (“Rail” in Figure 10(a)) and the assumed leaf layers, with a strong spectral response demonstrating the presence of material with organic bonds. Even when patted dry the layers exhibited a ∼3600 cm⁻¹ peak associated with the OH bonds in water, as the layers retain moisture. The peak at ∼1600 cm⁻¹ is associated with an aromatic COO⁻ bond and has been seen in other analysis of other organic (leaf) low adhesion layers.^26,28 Sample 1 and 2 were then collected after the FTIR scanning for solid-state, carbon-13 NMR analysis and XPS.OPEN IN VIEWER

The NMR analysis was unable to resolve a satisfactory spectrum from the rail scraping sample. Created layers in previous testing had 200-450 ppm iron, ²⁶ which was deemed low enough to attempt this technique. However, the rail surface at the VoR was likely significantly more oxidised than other testing sites and so the presence of more loose oxides in the sample may have interfered significantly with the process of resolution of the C13 solid state NMR. The nature of the field sample may also have meant it included a significant mass of non carbon material. Effective results from the NMR technique would have allowed for analysis of the carbon bonding in the leaf layer and facilitate comparison with the leaf material, providing further information about the transformation of leaves in to the bonded low adhesion leaf layer.

The scraped leaf sample was examined by XPS and the surface composition of the sample was determined (see Table 1) from the high resolution scans. Nitrogen was also seen in survey scans of the surface in a small concentration (1-2 at%). The differences between the positions in the sample is consistent with the visual observation of the scraped material, which appeared as a mixture of railhead oxide products and leaf material.

OPEN IN VIEWER

Table 1.

Surface composition (at%) determined from XPS high resolution scans.

Samples	Fe	O	C	K	Si
Position 1	4.4	42.4	50.4	<0.1	2.7
Position 2	10.5	50.6	31.4	0.2	7.3

Large amounts of carbon and oxygen were detected in the sample suggesting that the sample is made up of organic material. The spectra for the high resolution scans are shown in Figure 11.OPEN IN VIEWER

The pronounced peaks for the C spectra relate to C-C/C-H type carbons and the high proportion of C-O carbons aligns with the O1 data suggesting cellulose (or other biopolymers such as pectin or lignin) in the sample. A large proportion of the oxygen present is likely inorganic, appearing as oxides and hydroxides in the sample. The Fe fitting suggests Fe₂O₃ (iron (III) oxide) or potentially iron hydroxide. Silica is present and the analysis also suggests potentially bulk silicon, although this seems unlikely due to the samples and location.

Hydrophobia and water take up

The hydrophobia observed in the time lapse imagery (Figure 5(a)) is a phenomenon that has not previously been observed in research on leaf low adhesion layers. This phenomenon may be due to the fact that these are early stage layers and so the leaf may not have been broken down and homogenised. The outer layer of leaves are made up of hydrophobic long chain hydrocarbons (epicuticular wax)^37,38 which would resist water. The physical properties of the rain falling in large drops, rather than the fine mists or bulk water used in rig testing, may have also contributed. However, when the drops combined to be a bulk film on the railhead there was significant surface tension keeping the water on the flooded rail running onto the leaf layer. As previously stated there has been limited observation of layers in the field, and laboratory testing usually uses more homogeneous material (ground ³⁹ and/or soaked leaves^10,27) and so future testing should use freshly fallen leaves to be able to observe the hydrophobicity in controlled conditions.

Leaf layers are not completely hydrophobic, layers created in the laboratory ³⁹ absorb water well when dry, and in this testing reached high moisture levels (≥50% WME). The hydrophobicity seen in the captured images highlights that there is a delay between high humidity/rainfall and water take up. The moisture of the layer when rainfall is not imminent or happening is well correlated with humidity (Figure 8(d)) and testing on a created leaf layer strengthens this assertion (see Appendix II). The relationship seen between moisture and RH in Layer 2 is due to the diurnal change in RH, with the change being slow and gradual as the air temperature rises. However, at the higher humidities due to rainfall there was a significant range (14%–50% WME) in moistures of the layer. This highlighting a potential delay in water take up when RH rapidly rises due to rainfall. Future work should ascertain the speed at which the layer takes up water from the air. During times in the day when dew forms this will likely be quick due to condensation forming on the rail, but during the day rail temperatures are likely to higher and condensation will not occur, slowing the rate of absorption. Although RH is a significant factor in adhesion of leaf contaminated surfaces,^9,10,13 this is only as a proxy for the amount of water on the railhead/in the leaf layer. Past sensor testing has shown that high humidity alone does no correspond with a wetter railhead. ⁴⁰ The dewpoint sensor used did not register a “wet rail” (a signal above 0V) during all of the high humidity periods. Therefore, future work on forecasting adhesion in leaf layers should focus on the moisture in the layer as the driver of change rather than RH variation on its own.

Layer formation and low adhesion mechanism

Layers observed in this testing were early stage. They were made up of single leaves that had fallen onto the rail and begun to bond, while surviving one or two train passes. Due to the weather and early stage of development, the material seemed less well bonded than other layers observed in the field. Evolution was limited by abundant rain water and infrequent train passage preventing further material being deposited on the rail. It seems as if the material needs to be constantly agitated from the lineside by train passes ⁴¹ for the layer to become more established in wet conditions, even if the train passages also degrades the material on the railhead at this early stage. Leaf layer formation occurs when there is a convergence of lots of factors (weather, trees and site geography), and so is generally unlikely to occur. Observations from this testing demonstrate that even in a high risk area, with the “problem” trees and adverse geography, weather conditions prevent leaf layers from developing beyond early stage “seeding”. The limited field surveys on leaf layers have observed some developed layers.^4,14,34 These were made up of multiple leaves that have survived repeated train passes, and potentially many days and nights. Scaled testing on the lifecycle of leaf layers ¹³ has been completed, but the testing was completed with the layer experiencing 13-14 axle passes per minute. The layer development at the VoR was a more static situation, with much more available water and less axle passes.

Water appears extremely important for the chemical reactions that lead to leaf breakdown and to the characteristic black colouration as well as bonding to the rail. The time lapse imagery (see Figure 5(b) and (c)) showed the black coloration only appearing on the moist part of the rail. Static images of leaves that have fallen onto the rail show black colouration and degradation where the leaves have been broken and crushed, bypassing the outer cuticle layers of the leaves. Water allows the soluble iron oxides from the railhead to move freely and interact with the leaf material. The presence of water also provides a source of oxygen for further iron oxide formation. Iron oxides are an important to low adhesion layer formation^6,27,42 and the captured video demonstrates that the presence of moisture is important for the oxides to move freely.

Traction

In this testing there was no strong relationship between layer thickness and CoT of the layer as a single measure. The layers in the testing were thin compared to other UK rail surveys,^4,14 but past work has also found no correlation between layer thickness and railhead friction. ¹⁴

Shear strength of (created) layers is inversely proportional to moisture content ⁴³ and this aligns with the moisture-CoT relationship found in this testing, with CoT being inversely proportional to moisture levels in the layers. This finding supports the suggested shearing based mechanism for low adhesion.^6,13,44,45 The layers in this testing were assessed across a range of moisture levels, and the relationship between CoT and moisture appears linear, CoT dropping as moisture in the layer increases. This is the first time that layer moisture and CoT have been tracked for a naturally occurring layer, making it harder to compare with other data sets where moisture was not tracked. However, in this testing and the supplemental work (Appendix II) when rain is not falling moisture and RH have a positive correlation, and so the trends can be compared with RH standing in for moisture. The reduction in CoT as the layer moistens (with increasing RH or rainfall) is reflected in other surveys contaminated ¹² rails. Leaf low adhesion incident timing ²⁷ on national rail also peaks when humidity is highest across the day. Oxide caused low adhesion (“wet rail” ⁴⁶ ) does not have a linear relationship between presence of water in the layer and the low friction. Instead, low adhesion occurs in a small window of water volume.^47,48 However, this was not the case for the low friction layers at the VoR where it was a linear relationship, and a more moist layer had lower traction. Further, potentially rheological, testing could help to strengthen assertions about the moisture content of leaf layers and low adhesion conditions.

Limited relative humidity (RH) data was collected due to adverse conditions. The relationship seen between RH and CoT (in Layer (2) follows the expected pattern with reducing CoT as RH increases. The limited RH levels meant that it was not possible to observe if there is a discontinuity in the linear relationship between CoT and RH at lower RH as seen in scaled testing. ¹⁰

Combining moisture and layer thickness provided a stronger regression relationship to CoT compared to either individually (see Equation (1)). This relationship is weaker than rig based testing on leaf contamination ¹⁰ but stronger than field testing on clean rails ¹¹ (both looking at CoT changes due to relative humidity and air temperature). If this model can be validated through expansion, potentially using full scale testing, ³⁹ then it would be a useful tool as a proxy for adhesion using two directly measured values. As stated previous testing found no link between the thickness of a layer and the CoT, but if further research validates the relationship seen in this field survey this will help to develop the understanding of the low adhesion mechanism for leaf layers. It may be that thinner layers create lower friction when taking into account the moisture level of each layer (a measure not assessed in the last survey that tracked friction and layer thickness ¹⁴ ).

The leaf contamination observed in this testing was early stage. Even at this early stage, and in a transient state (easily removed by rain or wheel passage) it caused dangerous levels of friction, below the level required for safe breaking and traction. The observations from this testing showed that early stage leaf contamination could act like “wet-rail” – creating low friction (when wet) but also being destroyed by the passage of a train that experienced the low friction. ⁵ The initial material deposition can therefore create low friction in specific conditions or “seed” and begin the build up of more visible and well bonded layers seen in other testing.

Layer chemistry

The in situ FTIR analysis of the layers “as found” was dominated by the water in and on the layers, with a strong peak ∼3600 cm⁻¹ for the O-H bond. The wet layers were similar to others recorded from the field, ³⁹ due to both having extremely high water content. The layers, when dried, are similar to those created using the full-scale wheel/rail HAROLD rig. ³⁹ These laboratory based layers had been subjected to 150 axle passes with a wheel load of 57 kN, very different formation conditions to the layers at the VoR layers. The VoR layers were wet and exposed to the elements (UV radiation and oxygen in the air) for a long period, but still demonstrated a similar chemical change to the high axel pass layers. Leaf material therefore might not require a certain number of axle passes to become a low adhesion layer, there may be an alternative route where with high moisture levels and longer exposure times breakdown the leaves and the limited axle passes prevent the degradation and removal of the material.

The XPS analysis of the sample has a similar ratio of elements to that of a leaf layer on a branch line on the national rail network ⁴⁹ compared to a created leaf layer. As highlighted by Lanigan et al. ⁴⁹ the raised oxygen is likely due to the length of time that the layers are exposed to the atmosphere compared to created layers that experience a high number of traction events, but limited time on the rail. The national rail branch line where the comparison sample is from is a highly wooded area, with limited traffic is similar to the VoR location. This chemical analysis shows demonstrates that the layers tested at the VoR are, at least, chemically similar to low adhesion layers found on standard rail and not something specific to this narrow gauge rail.

The presence of iron oxide in the leaf layer aligns with the other tests of the mainline^26,49 and the raised iron (oxide) levels compared to more “mainline” rail ²⁶ is likely due to the low traffic volume on the line meaning limited axle passes and long time on the rail in wet conditions. Lanigan et al. ²⁶ found that the iron ppm increased significantly in a “full formed film”. The presence of a biopolymer (such as cellulose) confirms the testing completed on created layers. ²⁶ These materials are still present in the leaf layers when visually they have transitioned from leaf materials into the homogeneous, black leaf layer in naturally occurring conditions. The cellulose in the leaf layer may also explain the ability to absorb high levels water, as discussed in the section on the observed hydrophobicity. Cellulose has significant hydrogen bond networks ⁵⁰ increasing the wettability of the material.

The NMR data was unable to be resolved, likely due to high Fe (and oxide content), although no information can be drawn directly this is additional evidence that there is significant interaction between the organic material in the rail and the iron (and oxides) from the railhead in the formation of the low adhesion leaf layers. ²⁶ Future work will analyse leaf layers created using a full-scale rig at the University of Sheffield ³⁹ where there is more control over the conditions, and the level of loose iron oxides collected when sampling could be reduced.