Wear is usually associated with declining performance. Yet in the experiments discussed here, polishing an asphalt mixture initially increases its skid resistance. Understanding this result requires distinguishing two transformations: binder removal, which exposes the aggregates, and the subsequent wear of their asperities. Multiscale analysis helps track the transition.

An asphalt surface does not remain unchanged

An asphalt mixture combines aggregates, fine fractions and a bituminous binder. At its surface, tyre rubber does not encounter a mineral geometry that remains fixed throughout service. Aggregate exposure and asperity shape evolve under loading.

The question in these studies is therefore not simply which mixture has the highest initial friction level. It is how that level changes, and which texture transformations accompany the change.

The study published in Road Materials and Pavement Design examines seven mixtures. It extends the approach developed in my thesis, which includes a programme involving eight mixtures. These datasets are related, but they should not be merged as though they were one identical experimental series. [1, 2]

They share a two-stage behaviour: skid resistance first increases, reaches a maximum and then decreases. Explaining the two stages requires examining what polishing exposes as well as what it removes.

Observing change under controlled conditions

The experiments use a Wehner & Schulze machine, with separate units for polishing and friction measurement. Rubber cones work the specimens in the presence of water and abrasive silica. Friction is measured by braking a rotating head fitted with rubber pads. [1, 2]

The apparatus allows mixtures to be compared under a common protocol. It does not reproduce every condition found on a road, including vehicle diversity, weather, binder ageing and contaminants. A polishing-pass count is therefore a test variable, not a service duration.

In the seven-mixture publication, two specimens are tested per mixture. One is used, in particular, to identify the friction evolution; the other tracks that evolution while also allowing surface maps to be acquired. The reported friction values are averages of the two measurements. The mean coefficient of variation is 2%, with a maximum of 8%. These figures describe the repeatability observed within this programme, not universal accuracy or equivalence with field conditions. [1]

Stage one: polishing reveals asperities

In the article, the seven mixtures reach maximum skid resistance between 3,000 and 7,000 passes. The thesis reports maxima between 4,000 and 7,000 passes for its eight mixtures. Each range belongs to the results presented in its respective document. [1, 2]

The proposed explanation is progressive removal of binder and fine material from the surface. This exposes new mineral irregularities, so the rubber encounters a geometry different from the initial condition.

An increase in friction therefore does not indicate an absence of wear. A surface transformation can remove material while exposing asperities that contribute more strongly to contact and rubber deformation.

This initial stage is not identical for every mixture. The amount and nature of fines, aggregate grading and binder condition influence the observed changes. Saying only that “roughness increases” is too general: some sizes of surface feature may decrease while others are revealed.

Stage two: the asperities become smoother

After the maximum, the dominant mechanism changes. Wear of the exposed aggregates progressively reduces some of their relief and edges. Surface maps and texture parameters allow this flattening to be tracked.

Rubber is viscoelastic: its response depends on how it is deformed and on the rate of deformation. Part of friction is associated with energy dissipated during interaction with asperities. Changes in asperity height, slope or curvature therefore alter the conditions of that dissipation. [2]

This does not mean that every friction change can be attributed to one geometric parameter. It provides a mechanism for interpretation that must be compared with measurements. During the wear stage, decreasing texture parameters at small scales accompany decreasing friction. [1]

The friction maximum thus results from the succession and overlap of different transformations. Binder removal can bring mineral texture into contact more fully; subsequent polishing can reduce its effectiveness. A measurement taken at one instant cannot distinguish these trajectories.

Why a roughness average is not enough

The same average can describe differently organised surfaces. A small number of broad features does not deform rubber in the same way as many small asperities. Likewise, surfaces with similar height variation may have different slopes or curvatures.

The study follows several parameters: height variation through Sq, slopes through Sdq, summit curvature through Ssc and peak material volume through Vmp. Calculated over the whole surface, they describe aspects of its geometry. They do not directly identify which feature sizes carry the changes. [1]

This limitation becomes particularly important during binder removal. If very small features disappear while larger ones emerge, a global average can weaken or conceal both trends. We may then look for a simple relationship with friction where several mechanisms overlap.

The purpose is not to discard conventional parameters. It is to add scale information so that their meaning can be interpreted more effectively.

Looking at the same surface at different scales

The method used in the publication applies continuous wavelet decomposition to three-dimensional surface maps. It examines texture components associated with different feature sizes and recalculates parameters on those components. The two-dimensional analysis uses the Mexican hat wavelet. [1]

These are not new experiments on physically separated surfaces. They are different mathematical readings of the same measured topography. Their value is in locating changes within the scale range accessible to the measurement.

In the experimental programme, four 3 × 3 mm regions are mapped per mixture at several stages: initially, near maximum skid resistance, and after 90,000 and 180,000 passes. Repeated observations of the monitored regions help connect texture changes to friction evolution, although they do not eliminate specimen heterogeneity. [1]

During the initial stage, the authors relate the exposure of features between 300 and 3,000 µm to the removal of binder and fines. Responses vary with composition, and mixtures without sand do not evolve exactly like sand-rich mixtures. During wear, decreasing parameters at scales below approximately 1,000 µm are particularly associated with friction loss. [1]

These ranges describe the observations and analysis of this study. They do not define an optimum asperity size for every pavement. They also depend on map dimensions, resolution and processing: windows only a few millimetres across cannot describe the entire range of road macrotexture.

Aggregate grading changes the trajectory

Among the tested mixtures, higher friction levels are associated with certain compositions containing aggregates from 1 to 2 mm in size (the 1/2 fraction). Comparisons suggest that surface discontinuities and the geometry of these small elements influence rubber deformation. [1]

The result cannot be reduced to “smaller grains always mean better skid resistance”. Fine fractions and aggregates do not have identical shapes or necessarily the same resistance to polishing. Changing their proportions also changes surface organisation and the transformations that occur during binder removal.

Mixture comparisons therefore provide ways to understand and improve a formulation. They do not, on their own, establish a universal recipe: several characteristics change together and their effects must be interpreted jointly.

Binder choice can shift the friction maximum

The thesis provides a complementary example through mixtures F4 and F4 bis. They have comparable aggregate grading but different binders. F4 reaches maximum friction at 4,000 passes, whereas F4 bis reaches it at 7,000. Their reported maxima are 0.470 and 0.488, respectively. [2]

The difference in level, 0.018, must be interpreted in relation to test repeatability; it is not enough to claim a large, general improvement in skid resistance. The thesis discusses the shift in the position of the maximum in relation to the modified binder’s resistance to removal.

This case shows why complete curves are useful. Two mixtures may reach similar maxima while following different trajectories. Understanding their behaviour requires considering when the maximum occurs, not just its height.

The F4 and F4 bis labels used here are those of the thesis. This case is not added to the seven-mixture publication as though it belonged to the same dataset.

What these findings allow us to conclude

The experiments show that an initial increase in skid resistance is compatible with surface transformation through polishing. They also demonstrate the value of multiscale analysis in distinguishing the exposure of mineral relief from its subsequent flattening.

They do not directly establish the age at which a real pavement will reach its maximum. Nor can 180,000 machine passes be converted into kilometres or years without specific field validation.

Geometric correlations must also be distinguished from a complete causal explanation. Debris can be removed, reintroduced into contact or compacted, and binder removal is not uniform. The measured parameters summarise the effects of these processes without necessarily identifying each separately. [1]

Physical modelling can support a deeper interpretation. The thesis investigates contact and dissipation in rubber, but also recognises that an incomplete representation of the binder and its removal limits the model when applied to asphalt mixtures. Texture analysis and mechanical calculation are complementary; neither removes the need to represent the relevant mechanisms. [2]

Understanding performance as an evolving property

The main lesson is that the initial state is not enough to characterise a surface’s behaviour. What is exposed to contact changes, and so do the mechanisms governing that evolution.

For the tested asphalt mixtures, maximum skid resistance marks a transition between a stage in which mineral texture becomes exposed and one in which its wear becomes dominant. Multiscale analysis makes the transition clearer by showing which surface features change and how.

This knowledge supports more meaningful material comparisons. Alongside “how much friction does this surface provide?”, we should ask “why does that level change, at which scales and under what conditions?”.

Further reading

References

  1. Edjeou, W., Cerezo, V., Do, M.-T., Zahouani, H., Ropert, C. and Augris, P. (first published online in 2023). Multiscale analyse of the relation between skid resistance and pavements surfaces texture evolution with polishing. Road Materials and Pavement Design. Article and DOI.
  2. Edjeou, W. (2021). Analyse multiéchelle de la texture des chaussées - effet sur l’adhérence des revêtements routiers. Doctoral thesis, École centrale de Nantes, particularly Chapters 3 to 5. Manuscript on HAL.

This article presents laboratory findings and their interpretation. It is neither driving advice nor a pavement acceptance criterion.