This article was submitted to Tribology, a section of the journal Frontiers in Mechanical Engineering
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
A surface texture can be subdivided into three categories based on the magnitude of its wavelengths, i.e., macro-geometrical form, waviness, and roughness (from largest to smallest). Together, these components define how a surface will interact with the opposing surface. In most ice tribology studies, <2% of the entire sample surface is topographically analyzed. Although such a small percentage of the entire surface area generally provides statistically relevant information, the missing information about the texture complexity on a larger scale might reduce the possibility of accurately explaining the resulting tribological behavior. The purpose of this study was to review the existing surface measurement methods related to ice tribology and to present a holistic approach towards surface topography measurements for ice tribology applications. With the holistic surface measurement approach, the entire sample surfaces are scanned, and the measured data is analyzed on different magnitude levels. The discussed approach was applied to sandblasted steel samples which were afterward tested on two different ice tribometers. The experimental results showed that additional information about the sample surface topography enabled a better understanding of the ice friction mechanisms and allowed for a more straightforward correlation between the sample surface topography and its ice friction response.
Motion between solid objects and ice is one of the most complex tribological systems. This is due to the many influencing factors, which define the properties of the liquid-like layer (LLL) on ice such as ambient temperature and humidity, sliding velocity, the contact area between the sliding object and ice, the roughness and wettability of the sliding object, the texture, and hardness of ice, etc. Scientists are continuously seeking a better understanding of this highly unstable process, but the various mutually related variables and demanding experimental execution make it a very challenging task.
Among the most influencing and at the same time least understood factors are the properties of the LLL. Its thickness is still largely unknown even for static systems, while for dynamic systems it is additionally influenced by the motion dynamics and sliding surface texture and thus even more difficult to estimate. Despite the mentioned obstacles, effective experimental work with various types of laboratory equipment has already been performed over a wide range of sliding velocities (
To improve our understanding of ice friction, reliable information about the interactions between the sample, ice and LLL are necessary. There is a general agreement among scientists that the topography of sliding body and ice play a significant role in the ice friction process (
Another issue in topography analysis is the selection of the most appropriate topography parameters. For example, it can be easily pictured how the surface asperities will penetrate the ice surface (
Schematic representation of an interaction between a sliding object and ice:
The purpose of this paper is to review the existing surface characterization methods, analyzed parameters, and equipment, used in the field of ice tribology. Along with this review, the authors provide some ideas on how ice tribology surfaces could be characterized in further research for more detailed information about surface topography and texture. The proposed measurement approach could help towards a better understanding of the complex ice friction process—not only in the specific experimental setups used in the present study but also in other applications related to ice and snow tribology.
The existing ice and snow friction studies provide different approaches to surface texture measurements and analyses. The most common are non-contact profilometry (
Limitations of the common surface topography measurement tools.
Form > 1 mm | Waviness > 10 µm | Roughness < 10 µm | Limitations | Typical measurement limits | |
---|---|---|---|---|---|
|
|||||
Profilometer-contact | + | + | + | Stylus tip diameter > 4 µm | Profile length: 50 mm |
Height: 2 mm | |||||
Optical microscope | X | ? | + | Surface reflection contrast | Area: 200 × 200 mm |
SEM | X | ? | + | Sample preparation | Area: 5 × 5 mm |
|
|||||
Coordinate measurement machine | + | ? | X | Stylus tip diameter > 300 µm | Produced in small and large sizes |
Contour measurement machine | + | + | ? | Stylus tip diameter > 4 µm | Profile length: 100 mm |
Height: 60 mm | |||||
Non-contact profilometer | + | + | + | Light absorption | Area: 150 mm × 200 mm |
Reflection | Height: 2 mm | ||||
Steep asperity slopes | |||||
Contact profilometer | + | + | + | Stylus tip diameter > 4 µm | Area: 50 mm × 100 mm |
Height: 2 mm | |||||
AFM | X | ? | + | Limited sample size | Area: 200 µm × 200 µm |
Height: 50 µm |
Symbols “+”, “?” and “X” stand for “measurement is possible”, “measurement is potentially possible”, and “measurement is not possible”, respectively.
Depending on the used instrument and data post-processing method, examined sample surface areas can be widely different, from 50 μm × 50 µm (
In some studies, only 2D profile data are used (
ISO 2D and 3D texture standards contain more than 30 different parameters from which only a few are ever used ice friction research (see
Texture parameters used in the existing ice tribology studies.
2D parameters | |
---|---|
Ra | Arithmetical mean deviation of the assessed profile. Defines surface asperity average height |
Rdq | Root mean square (RMS) slope of profile. Defines the steepness of the asperities |
Rsm | Mean width of the roughness profile elements |
|
|
Sa | Arithmetical mean deviation of the assessed surface |
Sq | RMS roughness |
Ssk | The skewness of the surface. Characterizes whether a sample has asperities on top of the flat surface or dimples/scratches below the flat surface |
Sku | Kurtosis of the surface. The measure of the asymmetry of the probability distribution of a real-valued random variable about its mean |
S10z | Ten-point height. Indicates surface height calculated using only 5 highest asperities and 5 lowest valleys. It gives better insight into texture asperity actual amplitude. Due to the involvement of the 5 highest asperities, this parameter might change rapidly if the sample starts to wear |
Sz | The maximum amplitude of the surface texture. Indicates height between surface highest asperity and deepest valley. As far as only the highest and lowest points are used, this parameter will start to change rapidly if sample starts to wear |
Sfd | Fractal dimension. Characterizes the complicity of the texture. If the parameter value aspires to number 2 surface is smooth and “simple”, if the parameter aspires to number 3 surface is more complex thus has a larger theoretical contact surface |
|
|
Β | Attack angle. The angle between sample surface which is considered as flat, and snow (ice) roughness asperity slope |
KK | The criterion of contact. It is calculated as Rsm/Sa ratio. Indicates the steepness of asperities, i.e., larger ratio represents smoother surfaces with low and wide asperities, but a smaller ratio represents high and densely packed asperities |
Besides the described parameters in
The final issue related to surface roughness measurements is texture filtration. In most studies, approaches described in ISO 4287 and ISO 2517 standards are used for the roughness measurements and post-processing. This leads to the filtration/neglection of the geometrical form and waviness from the measured surface profiles, thus changing the texture appearance and the calculated parameter values. For the calculation of the actual contact area, this approach might not be the most reliable since the filtered roughness profiles only provide a part of the information about the contact (
Schematic representation of an interaction between a solid surface and ice:
Examples of ice friction research texture measurements.
Sample type and dimensions (mm) | Measured lengths/areas (mm) | Equipment | Calculated parameters | Ref |
---|---|---|---|---|
Pin: Diameter = 3 | 0.4 × 2.8 | Confocal microscopy | Ra, Rq, Rsk, Rku |
|
Pin, dimensions: n/s | 0.5 × 0.5 | Stylus 3D profilometer | Ra, Rdq, Ssk, Sku, Sfd (D) |
|
Ring type slider: Outer diameter = 25.4; inner diameter = 23.4; H = 1 | Profiler: n/s | Noncontact profilometer, AFM, SEM | Ra, Microscale bump diameter |
|
AFM: n/s | ||||
SEM: 0.2 × 0.2 | ||||
Steel ski: L = 487.5; W = 30; H = 30 | 1.7 × 1.8 | Focus variation microscope | Sa, Sz, Ssk, Sku |
|
3.3 × 3.3 | ||||
8.2 × 7.8 | ||||
11.3 × 11.3 | ||||
Steel runner: L = 150; W = 8; H = 20; Runner transverse radius = 4 | 0.05 × 0.05 | AFM | Sa, St |
|
UHWMPE polymer samples: n/s | 0.4 × 0.5 | Interferometer, SEM | Attack angle |
|
UHWMPE ski sole on aluminum body: L = 65; W = 40; H = n/s | 0.5 × 1 | Confocal microscopy | Width of the ridges |
|
Laser textured skis with attachable metallic base plate: L = 200; W = 20; H = 0.5 | 0.6 × 0.6 | SEM, optical profilometer | Dimple diameter and depth |
|
Silicon carbide spheres: R = 0.75; 6.00 | 0.2 × 0.2 | Laser-scanning confocal microscopy | Sq |
|
Soda-lime glass spheres: R = 1.84 | ||||
Sapphire sphere: R = 1.59 | ||||
Model ice skate: R ≈22 | ||||
Steel block: L = 35; W = 18; H = 14 | 2 × 2 | Interferometer, contact type profilometer | Ra, Rsm, Rz, Rpk, Sa, KK, Sdq, Ssk, Sku |
|
20 × 10 | ||||
32 × 16 |
Abbreviation
In the existing literature on ice friction, very different topography measurement methods and approaches have been applied, and no standard methodology exists yet. The common trend is to analyze a small fraction of the full surface and neglect the waviness and geometrical form of the sliding surfaces. One might ask why a larger surface area should be measured at all. For example, let us imagine that a cylindrical pin with a flat tip is used on a tribometer test rig similarly as in studies (
In the present research, the authors investigate how full surface measurements of sliding samples can benefit the understanding of experimental data in the field of ice tribology. For this, samples with anisotropic surfaces were prepared and analyzed by using non-contact and contact 3D surface measurement equipment. The measured topography values were compared, and a method for sample contact area measurements was proposed. Furthermore, surface measurement data are compared to tribological results obtained on two different test rigs to verify whether more detailed topographical data can provide a better understanding of the ice friction process.
All samples were manufactured from the Uddeholm Ramax HH steel, which is a chromium alloyed, corrosion-resistant steel supplied in a high pre-hardened condition. Samples were cut as rectangular blocks with dimensions of 35 mm × 18 mm × 14 mm. All sides of the blocks were grounded simultaneously side by side in one batch to obtain as similar initial geometry as possible. Afterward, sample test surfaces were sandblasted to get isotropic texture. Sandblasting was followed by polishing to achieve three different surface roughness levels. Polishing times were set to 30, 150, and 240 s.
Closeup (500 μm × 500 µm) of sample surface 3D photo simulation obtained with contact type profilometer:
Sample texture measurement—full surface and small section measurement. Full surface measurements provide essential information about the geometrical form of the sample. Small section measurements have a higher measured point density, thus providing more details about the surfaces, e.g., asperity height, dimple size, pileup, and polished surface area. In both approaches, all surface parameters were calculated for the primary surface and for the filtered roughness profiles to evaluate the changes in the parameter values.
To compare contact and non-contact profilometry data, surface topography measurements were performed using a laser confocal microscope (VK-X250/260, Keyence International NV/SA, Belgium). Measurements were performed in the center of the sample surfaces, identically as in contact type small section measurements. The laser confocal microscope works on the principle of combined laser light and white light microscopy using a violet semiconductor laser with a 408 nm wavelength. It contains a 16-bit PMT (photomultiplier tube) color CCD (charge-coupled device) image sensor, the recording resolution is 3,072 × 2,304 pixels, i.e., there are more than 3 million measuring points in each scanning plane. Due to the high measurement point density, these measurements are also considered as high resolution (HR) measurements.
Contact type profilometry measurement post-processing was performed in Only roughness was considered for full surface and small section measurements. The primary surface (roughness and waviness) was considered for full surface and small section measurements. All surface levels (geometrical form, waviness, and roughness) were considered. Only full surface measurement was used without any applied filters.
During the first approach, geometrical form and waviness were filtered out according to ISO 2517 requirements for surface roughness characterization. This included the use of Gaussian filter, cut-off, and low pass filter according to Taylor Hobson Form Talysurf Intra 50 user manual and ISO 2517 guidelines. During the second approach, only the geometrical form was filtered out, while waviness was left unfiltered. In the third approach, no information was filtered out, i.e., the sample geometrical form, waviness, and roughness were left unmodified after the measurement.
For the evaluation of the influence of contact area between the sample and ice, a virtual sample slicing method was applied. Here, the highest peak on the surface was used as a reference point from which the sample surface was sliced with a virtual slicing plane at various depths between 1 and 12 µm below the highest peak.
The samples used in the present study have a noticeably curved shape, thus the contact area with ice will depend on how deep the curved surface will penetrate the ice during sliding.
Surfaces of test samples sliced at 4 µm depth under the highest surface peak. On the left side, 3D representations of the sliced surfaces are shown, and for comparison, on the right-side top view of contact area, images at the same slicing-depth are shown. Sample with the highest surface roughness (SP30) has the smallest contact area, which consists of fewer asperities than for other samples. Under loading conditions, fewer asperities can more easily penetrate the ice surface. Generally, the contact area of samples increased with polishing time since during polishing surfaces become smoother, and at the same time, the height of asperities is reduced.
Contact areas of the sliced surfaces were calculated using
Two different tribological test setups were used for the testing of the same samples. A schematic representation of both test setups is shown in
A schematic representation of the test setups used:
For velocity measurements, a 3 m long inclined plane tribometer was used (
The tribometer was located in a climate simulation chamber, which enables ambient temperature regulation in the range of +30°C down to −20°C and is equipped with an ice plane cooling system, allowing precise regulation of the ice temperature. Humidity and air temperature in the climate simulation chamber were measured using a P330 Temp thermometer (Dostmann electronic, Germany), while the ice temperature was measured with thermocouple TP-122-100-MT-K (Czaki, Poland) plugged into infrared thermometer Proscan 520 (Dostmann, Germany). Temperature and relative humidity measurements were documented after every 10th sliding test, and the final value was calculated as the average of all measurements of a single experimental session.
Before the sliding tests, the ice surface was leveled flat, and a small groove was embedded in ice in the movement direction to keep the samples in a straight trajectory during sliding. During the tests, the samples were slid down the ice track in a random order to prevent the eventual influence of fluctuation of ambient conditions on the experimental results of specific samples. The average sliding velocity for each sample was calculated from 40 individual sliding velocity measurements. During data post-processing five fastest and five slowest measurements were excluded from the calculation.
In oscillating tribometer tests, an ice rink with dimensions of 20 mm width, 80 mm length, and 5 mm depth was used. Before each test series, a new ice surface was prepared. For ice formation, 18 ml of distilled water was used to which 0.5 ml of tap water was added to accelerate the ice crystallization. In all tests, the ice temperature at the bottom of the ice bath was −10°C (at the surface, it was estimated to around −8°C under the applied ambient conditions).
Before tribology tests specially developed aluminum leveling tool with a contact area of 45 mm × 28 mm was inserted in the sample holder and rubbed against the ice surface to create an ice surface as flat and smooth as possible. The smoothing was performed at a normal force of 692 N and an average sliding velocity of 0.08 m/s until the height difference between the left and right sides of the ice surface was lower than 100 µm. The flatness of the ice surface was measured with a built-in tribometer dial gauge. After leveling, the leveling tool was replaced by an experimental sample, and the tribological test was conducted.
Experiments were carried out at a constant normal load of 52 N and a stroke of 24 mm. The contact pressure between the sample and the ice was ca. 0.084 N/mm2 if it assumed that the sample surface is perfectly flat and in full contact with ice. For each test, a run-in period of 60 s at 0.10 m/s was first performed to adjust the sample temperature to the ice temperature. Afterwards, experiments were carried out at 7 velocity levels (average sliding velocities of 0.02, 0.05, 0.10, 0.14, 0.19, 0.29, and 0.38 m/s). During each experiment, friction measurements at all velocity levels were performed twice—once at increasing and once at decreasing velocity. The change of sliding velocities during each experiment performed on the oscillating tribometer is shown in
Change of sliding velocities during each experiment performed on the oscillating tribometer. Tests were carried out at different velocities, first at increasing and afterward at decreasing velocity, to analyze the influence of both types of motion dynamics.
In
Surface roughness parameter Sa [µm] measured using contact and non-contact profilometers and processed according to the procedure described in Section of non-contact profilometry.
Non-contact measurement | Contact measurement | ||||
---|---|---|---|---|---|
(HR) | (HR) | (LR) | |||
Roughness | Primary | Roughness | Primary | Roughness | |
SP30 | 3.15 | 2.99 | 2.91 | 1.70 | 1.13 |
SP150 | 2.44 | 2.02 | 1.97 | 1.10 | 0.75 |
SP240 | 0.98 | 0.81 | 0.79 | 0.45 | 0.31 |
Abbreviations “HR” and “LR” stand for “High resolution” and “Low resolution” measurements, respectively.
The comparison of the Sa parameter shows that a higher measured point density on the surface (HR measurements) results in a higher Sa value. The highest Sa values were obtained with the non-contact laser confocal microscope, followed by small surface measurements with the contact type profilometer, while the lowest Sa values were obtained in large area roughness measurements with the contact type profilometer. Higher Sa values obtained with the non-contact method can be explained by its ability to measure deeper and narrower surface valleys than the contact type profilometer. This is because the 2 µm tip radius of the contact profilometer cannot reach small dimples or bores. It should also be pointed out that the cone-shaped profilometer tip tends to describe narrow and/or sharp edges as rounded ones.
Also, it was noticed that the unfiltered primary surface has a higher Sa value than the filtered roughness profile, which is because waviness enhances the overall amplitude, thus increasing the Sa value. Typically, in ice tribology studies, waviness is not considered, but it might play an important role. Therefore, for analyses of sample interactions with ice, the authors propose the use of the unfiltered primary surface instead of filtered roughness profiles.
The observations about the influence of point density and waviness on the measured Sa values are logical and do not provide any groundbreaking discoveries, but do provide a useful guideline: the question of which method for measurement and calculation of roughness parameters is the best and/or more accurate is very complex and difficult to answer, however, the roughness parameter values of different surfaces can be effectively compared as long as they are measured in the same way and by using the same device. This conclusion is supported by the data in
Virtual slicing of the sample surfaces was performed in 1 µm steps from the highest surface peak to 12 µm below it, as explained in “Non-contact Profilometry” section. The contact area values for sliced surfaces of all samples are shown in
In
Considering the calculated contact areas at the respective slicing depths, in
Inclined plane tests were conducted under the following conditions: ice temperature −7 ± 0.5°; air temperature −4 ± 0.5° and relative humidity 60 ± 3%. Sliding velocities were obtained at 4 different positions, and the final values shown in
Sliding velocities measured on the inclined plane tribometer. Sample velocity increases due to accelerated movement down the inclined plane. In the magnified closeups, the smoothest sample SP240 is faster than samples SP30 and SP150, which show similar velocities. Slightly different behavior between the latter is observed only in the first measurement point, where the roughest sample SP30 is faster than the intermediately rough sample SP150.
In
In measuring position V1, the sample behavior appears to be random, which is most likely because the samples start their movement from a steady state where an eventual stick-slip movement may randomly affect the samples’ ability to start sliding. In the steady state, the rougher sample has the advantage of a smaller contact area on which adhesive forces can work, possibly resulting in lower static friction. However, under higher applied loads this effect might not be observed because the asperities of the rough surface would penetrate the ice resulting in higher deformative friction, ploughing and/or mechanically interlocking with the ice surface.
As the samples move down the inclined plane, their velocity increases. When the samples have gained some inertia, i.e., in measuring positions V2, V3, and V4, the difference between the velocities of samples SP30 and SP150 decreased to approximately 1%, while the difference between the velocities of the slowest and the fastest sample, SP30 and SP240, respectively, were around 3–4%. Such difference may not seem much, but for winter sport athletes it could provide a major benefit.
Oscillating tribometer tests were conducted under the following conditions: ice temperature –8 ± 0.5°; air temperature 7 ± 1° and relative humidity 55 ± 2%. The measurements were conducted at various velocities as shown in
Coefficients of friction measured on the oscillating tribometer. The obtained friction curves show that samples with higher contact area (smoother ones) yield lower coefficient of friction values. At the same time, the influence of velocity on the coefficient of friction is noticeably higher for the roughest sample SP30 than for the smother samples SP150 and SP240.
In
The obtained coefficient of friction values and the general sample behavior correlate well with the data reported by (
In
In
The highest correlation between the contact pressure and the sliding velocity was observed at a slicing depth of 1–3 µm. This is because, at low contact pressure applied in the inclined plane tribometer tests (only sample’s weight), the samples did not penetrate deeply into the ice surface.
All analyzed surface texture parameters show a very high correlation (above 0.8) with the sliding velocity and indicate they can be used for the evaluation of sliding ability in ice tribology studies. The high correlation between the surface texture parameters and the sliding velocity may also be due to the low contact pressure applied in these tests. Since only the tip of the curved sample surface is in contact with the ice surface, texture roughness on the tip has a stronger influence on its sliding ability than at high contact pressures, where a higher proportion of the sample surface is in contact with ice and, therefore, the waviness and the geometrical form of the sample additionally influence its sliding ability.
For the oscillating tribometer setup where higher contact pressure was applied, the correlation results (
In
The present study shows that currently, researchers in the field of ice tribology measure surfaces in very different ways: by using different surface roughness parameters to describe the surfaces and post-processing the data differently. All these methods are efficient and provide a lot of useful information; however, for a better understanding of interactions between solid surfaces and ice, a unified surface measurement methodology is recommended so that in terms of surface characterization, all researchers in the field would “speak the same language.”
This specific study was conducted to highlight the importance of understanding the full surface geometry instead of focusing only on the micro-or nanoscale of the surface texture. The proposed surface analysis method is not in its final stage yet, however, it will be further developed and validated using different sample textures in future research. Since for theoretical calculations of coefficient of friction, information about the contact area is required as well, possible collaborations in this regard will be established to see if the proposed methodology can enable a more accurate match between experiments and theory. In this sense, applying data science and data mining methodologies (
The present research brings out several conclusions: The current surface measurement trends in ice tribology are focused on small surface area investigation, while the information about the sample macro geometry is rarely considered. This indicates that researchers may lack crucial information that could help them understand the sample behavior in contact with ice. The proposed surface measurement and contact area analysis approach using the virtual surface slicing technique provided important information about the sample macro geometry that helped understanding sample behavior under different measuring conditions applied in different experimental setups. The surface measurement approaches found in the available literature typically neglect the information about the sample form and waviness, providing only the roughness component of the surface. Such deficient information about the sample surfaces prevents from wholesome analyses and comparison of results from different studies. If the overall (primary) information about the test surface is presented, understanding of the sample behavior under different measuring conditions can be improved. Afterwards, additional filtering can be applied, and the obtained results analyzed in the context of roughness or waviness. The proposed surface analysis approach can be helpful not only for ice tribology but for other tribology studies, where two surfaces are in contact as well. A similar influence of surface texture and/or roughness on the coefficient of friction and sliding velocity (decrease with decreased surface roughness) was observed in different experimental setups (inclined plane tribometer and oscillating tribometer). The observed results correlate with observations reported by other scientists for similar experimental conditions.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Conceptualization, JL, EJ, and IV; investigation, JL, EJ, and IV; methodology, analysis and validation, JL, EJ, IV, IB, JV, TW; writing—original draft preparation, JL, EJ, and IV; writing—review and editing, IV, JV, TW, IB.
Parts of this work were funded by the Austrian COMET Program (Project InTribology, No. 872176) and carried out at the “Excellence Centre of Tribology” (AC2T research GmbH) in cooperation with V-Research GmbH and Riga Technical University. Financial support of Austrian Cooperative Research (ACR) is gratefully acknowledged. Parts of this work were also funded by the ERDF project “The quest for disclosing how surface characteristics affect slideability” (No. 1.1.1.1/16/A/129), which is being implemented in Riga Technical University. This research was partly supported by the Doctoral Grant Program of Riga Technical University. Riga Technical University Research Support Fund has supported the costs for open-access publishing of this article.
Authors IV, JV, and TW were employed by the company “VResearch GmbH”
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.
The Supplementary Material for this article can be found online at:
The proportion of variance (RSQ) between surface roughness/contact pressure and the measured sliding velocities on the inclined plane tribometer.
The proportion of variance (RSQ) between surface roughness/contact pressure and the measured coefficients of friction (COF) on the oscillating tribometer.