Riding on difficult road surfaces on an e-scooter: contact-patch physics on cobblestones, tram tracks, gravel, wet leaves, painted lines and expansion joints

19.05.2026 Scootify 14 min read

Among the guides already published, this one builds on braking technique, cornering and lean technique, emergency obstacle avoidance, carrying cargo and payload, and the complete weather axis — heat, winter, rain, night riding, wind, and fog. All of them assume a uniform road surface — familiar asphalt or concrete with a known dry μ ≈ 0.7–0.8. Reality in a Ukrainian or European city does not match that assumption: cobblestones in Lviv-Pidzamche, Kyiv-Podil, Kamianets-Podilskyi; tram rails in Kyiv, Lviv, Kharkiv, Dnipro, Odesa, Mariupol; gravel descents and broken slabs; in autumn, wet leaves and wet painted lines; patched repairs and expansion joints on bridges. Each of these surfaces alters the tire-road contact physics in a fundamentally different way than wet weather changes μ, which is why it deserves its own discipline — a surface axis on top of the already-covered weather axis and applied-physics circuit.

The prerequisite is understanding how friction force μN sets braking distance, how angular velocity and lean limit maximum cornering speed, how emergency obstacle avoidance depends on the available PIEV reaction time, how suspension and 8–10-inch wheels absorb irregularities, and how a wet road lowers μ. Surface is the fifth axis on top of weather, ride dynamics, gear/posture, and route planning, and in city conditions it often dominates the other four.

1. The contact patch as a common denominator — three failure modes

All surface effects are mediated through one object — the contact patch of the tire. For a typical e-scooter with 8–10-inch wheels under 80 kg rider + 20 kg scooter load, it is roughly 5–15 cm² per wheel (depending on pressure, weight, and tire construction). All forces are transmitted through this small area: braking, drive, lateral (for cornering), and vertical (weight). When the surface attacks the contact patch, everything breaks at once.

There are three physically distinct modes in which a surface attacks the contact patch, and they cannot be lumped together because the countermeasures differ:

Material μ failure — the surface material has an inherently low friction coefficient, or becomes so when wet. The tire remains in full contact with the surface, but the friction force μN drops 2–5×. This is the mode for wet leaves (μ ≈ 0.1), painted lines in rain (μ ≈ 0.2–0.3), metal manhole covers in rain (μ ≈ 0.1–0.2). The countermeasure is do not brake and do not turn on that surface; if possible, traverse it straight and upright.
Geometric trap-or-deflect — the surface contains a slot, joint, rail, or edge that either traps the wheel (wheel-slot trap) or makes it slip sideways (rail-deflect). Material μ may be perfectly normal, but geometry steals your contact patch: it drops into a joint or slides off the convex top of a rail. This is the mode for tram tracks, expansion joints on bridges, badly laid joints between slabs. The countermeasure is a crossing angle ≥45° and slower speed in the vicinity of such features.
Kinetic momentary contact loss — the surface is uneven enough that the tire periodically lifts off it (microseconds to tens of milliseconds). During those moments μN = 0, because there is no N. This is the mode for cobblestones, especially round ones (“river-rock cobbles”), where the wheel passes through peak-and-valley with 1–2 g of vertical acceleration. The countermeasure is lower tire pressure, active stance with soft knees and elbows for vibration absorption, and a smoother line.

Real surfaces often combine several modes: wet cobblestones = material μ ↓ + kinetic vibration; an expansion joint on a bridge after rain = geometric trap + material μ ↓. So the sections below each look at one scenario with a focus on “what exactly fails in the contact patch”.

2. Cobblestones — three stone types, two speed regimes

The general term “cobblestones” hides three different sub-categories, and their behaviour is very different:

Rectangular setts (Belgian setts, granite cubes) — laid in rows with narrow 5–10 mm joints. This is classic cobblestone pavement of historical European cities. The top of each block is flat and relatively large (10×10 cm), so the contact patch sits entirely on one block most of the time. The vibration has a fixed frequency: at 25 km/h on 10 cm blocks, that is 70 Hz joint-crossing, plus harmonics. This is precisely the canonical Paris-Roubaix pavé, where riders fit wider tires at lower pressure and ride in the middle of a block, parallel to the joints, rather than perpendicular to them.
Round river-rock cobbles (round cobbles, river-rock setts) — also blocks, but with rounded or dome-like tops. The contact patch jumps from peak to peak, periodically falling into the valley between stones. This is the worst surface: material μ may be normal, but kinetic contact loss makes the effective grip unpredictable. The worst sub-category is small, wet, round cobbles in courtyards and alleys.
Granite slabs and concrete pavers — large 30×30 cm or 50×50 cm slabs. The slabs themselves are flat, but the joints are 1–3 cm wide and often sunken. The surface gives both a long smooth section (on the slab) and a narrow sharp impact (at the joint). If a wheel enters a 3 cm wide joint at an angle, it can be trapped or deflected.

Dry vs wet. Dry granite and Belgian setts have μ ≈ 0.5–0.6 — below asphalt (0.7–0.8) but above damp concrete screed. Wet — μ drops to 0.3–0.4 on granite blocks and to 0.25–0.35 on round cobbles. Old polished granite slabs in historical centres, after centuries of foot traffic, become almost glass-like when wet: μ_wet ≈ 0.15–0.25. This is the region where an e-scooter behaves much like riding on ice.

The speed sweet-spot — the counter-intuitive part. On cobblestones, there are two risky speed zones:

Very slow (5–10 km/h): the wheel has time to enter every valley and climb out onto the next peak. The contact patch is constantly moving over the unevenness, producing maximum perceived vibration and low lateral stability (because at each moment the wheel is on a different irregularity).
Very fast (>30 km/h): the tire does not have time to deflect into the valleys, it “flies” from peak to peak. This feels smooth, but stopping distance grows (on low μN with kinetic contact loss), and an unexpected large irregularity (a chipped granite slab) produces a lateral impulse that the rider cannot react to in time.

The sweet spot for a typical e-scooter on cobblestones is 15–20 km/h on Belgian setts and 10–15 km/h on round cobbles. Slow enough to avoid serious shocks from large irregularities, fast enough that the contact patch is not crawling into every joint. On wet round cobbles — ≤10 km/h, treat like ice.

Line — parallel to joints, not perpendicular. Intuitively, riders want to “take the smoothest spot”, but that often means crossing joints at a 30–60° angle. Instead choose a line within one row of blocks, parallel to their long dimension (the joints). That means fewer “stepped” wheel transitions across joints; the wheel travels along the joint, which is either flat (Belgian setts) or has a constant depth (granite). If a joint must be crossed, cross it at 90°, like a tram rail.

Tire pressure — lower. On regular asphalt the typical pressure is 40–45 PSI (~2.8–3.1 bar). On cobblestones, dropping to 30–35 PSI (~2.1–2.4 bar) gives a larger contact patch, more tire flex over irregularities, and a smaller amplitude of kinetic contact loss. The trade-off is a higher risk of pinch-flat, when the tire is pinched between a sharp joint or chipped edge and the rim, producing two parallel cuts. Bead-lock or TPU tubes, or tubeless tires, help (though tubeless is rare on older scooters).

Stance — active, not passive. On smooth asphalt you can stand upright with straight legs and tire less. On cobblestones, straight legs pass every impact into the lower back and neck. An active stance with 5–10° bent knees and elbows, a slightly inclined foot (weight a bit rearward), a light grip — this adds 2–3 cm of vertical damping in the human body itself. Cycling and motorbike research shows: the difference between “rigidly locked knee” and “softly bent knee” on an uneven surface is a 30–50 % reduction in vertical shock transmitted to the spine.

3. Tram tracks — wheel-slot trap and convex rail head

Tram rails are the single most dangerous surface category for any narrow-wheeled vehicle (bicycle, monowheel, e-scooter). A series of academic studies from Edinburgh, Vienna, Zürich, and Toronto has shown: rails account for 17–35 % of all single-bicycle crashes in cities with tram systems. This is the highest single-cause risk in tram cities, higher than collisions with cars for non-motorised transport.

The geometry that creates the trap. Standard European gauge (UIC standard gauge 1435 mm) has two parallel rails. On open sections each rail sits separately on ties. In the city most rails are integrated into the road surface: the rail sits in a depression in the paved roadway, forming a groove 35–45 mm wide and 38–58 mm deep (per Vignole/Phoenix standards and grooved rail like type 60R2 — the most common in European trams). The groove is designed for the flange of a tram wheel.

An e-scooter wheel 50–90 mm wide does not usually drop fully into the groove, but that is small comfort: if the wheel meets the groove at an acute angle (≤30°), the front of the contact patch slides into the groove while the back stays on the road. At that moment, the surface reaction force becomes lateral instead of vertical, and the wheel snaps in the direction of the groove — simultaneously losing traction. This is the classic “tram trap”, 90 % understandable only after a first fall.

Convex rail head — the second failure mechanism. The very top of the rail is convex, with a radius of 200–500 mm and a flat width of just 50–60 mm. If the wheel crosses the rail exactly perpendicularly, the convexity permits brief contact. But if the crossing happens at a small angle, the tire slides sideways off the convex surface, because the effective μ of convex metal plus the lateral reaction component generate a sideways impulse. This is especially noticeable while braking: deceleration on rail metal is catastrophically low because μ is very small.

A wet rail — worse than ice. The wheel-rail interface is a separate branch of engineering literature (for trains). Measured μ for tire-on-metal contact in rain is 0.05–0.10. For comparison: μ of ice on metal is 0.10–0.15. So a wet tram rail is literally worse than ice. This is also not obvious by sight: frost or snow on ice is visually expected to be slippery, while a wet rail under drizzle looks just like a dry one.

Four failure modes — from most common to most extreme:

Front-wheel slip on wet rail head — the rider crosses the rail at a small angle in rain; the wheel slips sideways by 5–10 cm; the scooter goes down on its side. The most common mode, usually without serious injury at low speed (<20 km/h), but a thigh hit on the asphalt and torn wet jeans.
Wheel-slot trap at acute angle — the rider crosses the rail at an acute angle (<30°); the front wheel jams into the groove and rotates parallel to it; the rider is thrown onto the sidewalk or under a car. This is the classic mechanism of severe injury; documented in Edinburgh studies (Princes Street and York Place routes after the launch of Edinburgh Trams in 2014).
Parallel-rail glide — the rider travels parallel to the track at a close distance; a bump shifts them laterally by 3–5 cm and the front tire ends up on top of the rail; then a sideways slip. The worst variant, because the rider does not anticipate contact with the rail.
Cross-and-deflect at transition — the rider crosses the rail where it exits the paved surface (rail exit point, edge of a paved-over section); the wheel goes from paving to rail to paving again, and on one of the two transitions it slips. In the city, these are places where the line enters a depot, crosses a paved intersection, or changes surface type (from gravel to asphalt).

Defensive crossing technique. The canonical advice from cycling-safety literature and academic studies is: cross rails at an angle ≥45°, ideally 60–90°, on a dry rail, and slowly. In wet weather — dismount and walk across, especially at intersections with several rails in parallel. Specifically:

Plan a route that avoids rails where possible. OSM maps include railway=tram lines — they can be excluded from cycle routing. In tram cities (Kyiv, Lviv, Kharkiv, Dnipro, Odesa) this adds 5–15 % to the route, but drastically lowers the risk.
If a crossing is unavoidable — pick an angle as close to 90° as possible, even if that means moving into a car lane. A “half-angle” crossing is the trap mode.
Speed at the moment of crossing — ≤15 km/h dry, ≤8 km/h wet, or dismount.
Do not brake on the rail — reduce speed beforehand, cross coasting (free wheel), brake after the rail. Braking on the rail = immediate front-wheel slip.
Do not turn on the rail — turn before or after, not on it.
Do not ride parallel to the rail closer than 50 cm — the smallest lateral push (a pothole, wind, a pedestrian manoeuvre) puts the wheel onto the rail.

Edinburgh case study — real-world evolution. After the launch of Edinburgh Trams in 2014, the city saw a significant rise in bicycle injuries along the tram route; academic analysis found 191 serious injuries over 23 months of monitoring, 70 % with upper-body and head trauma. The city introduced: yellow “cyclist crossing zone” road markings at crossing points, rubber-insert groove repairs (rarely), and an information campaign. After that, injuries fell, but not to zero — the physics of the groove and the convex head remain.

4. Gravel, sand, dirt — two-layer dynamics and the plowing effect

Gravel and sand follow fundamentally different physics from a hard surface. Here it is not μN that determines behaviour but the interaction of the wheel with a loose layer above a hard base.

Two-layer model. The wheel rolls on the upper layer (gravel 5–30 mm, fine sand, compacted soil 1–10 mm fluffy) above a denser substrate (compacted earth, concrete, asphalt). The surface reaction force on the wheel has two components:

Resistance in the upper layer (rolling resistance) — proportional to the relative velocity of wheel and layer, the mass of material being pushed aside, and the depth of penetration. On gravel 5–10 mm thick this resistance is small (~20–40 % of asphalt); on 30–50 mm — significant (×2–3); on loose sand >50 mm deep the scooter gets stuck.
Drive and braking through the hard base — the wheel only partially reaches the firm substrate; its braking and drive force are split between the loose layer (unstable) and the substrate (normal μN). The effective braking μ is reduced by a factor of 1.5–3 compared with hard pavement.

Front-wheel plowing effect. In a corner the front wheel is supposed to generate a lateral force through slip angle (the difference between the steered direction and the actual velocity vector). On hard surfaces, that lateral force is proportional to μN. On gravel, part of the energy goes into shoving gravel sideways (like a plow displacing earth), and the effective lateral force is smaller. That means that at the same steering angle the scooter turns with a larger effective radius — the wheel does not “listen”, it slides forward until it catches the substrate.

Second consequence — on gravel and sand the scooter holds a straight line better than expected, but does not corner like on a hard surface. Forward drive — OK; cornering — weak. This is similar to riding on dry grass.

Rear wheel under power — under sudden throttle the rear wheel can spin in the upper layer (gravel spin), because motor torque exceeds the friction of the loose layer. On a scooter without traction control this produces a sudden sideways shift of the rear. On direct-drive (DD) hub motors it is especially pronounced. Countermeasure — feed in throttle smoothly, as on ice.

Front brake — on gravel it locks much sooner than on asphalt, because μN is lower. A locked front wheel on gravel = an instant face-plant, because friction disappears at once and momentum is transferred to the hands. Countermeasure — rear-brake priority; the front is used only as a backup at the very end of the stop.

Hard-to-loose transition — the most risky spot. Going from asphalt to a gravel exit at speed produces a sharp jump in rolling resistance and a simultaneous loss of part of the lateral force. If the scooter is turning at that moment (entering the exit at an angle), the front wheel slips because effective μN has just collapsed. Speed at the transition — ≤15 km/h, going straight, and only then turn on the gravel.

Tire pressure — raise, do not lower. Counter-intuitively: on gravel and sand pressure is raised to the upper limit (45–50 PSI), so that a smaller contact patch cuts through the loose layer and reaches the substrate faster. Low pressure floats (like a wider tire on snow).

5. Wet leaves, painted lines, metal covers — material μ failure

This is the most treacherous category, because the surface looks normal at a glance: the road is dry, the asphalt is fine, but where you plan to brake or turn there is a strip with μ 3–5× lower. You fall for no reason your eyes could have foreseen.

Wet leaves — μ ~0.1, like ice. In autumn, cobblestones and asphalt under a layer of wet leaves become slippery enough that police traffic divisions run dedicated annual campaigns about them (e.g. AAA in the US, AVD/ADAC in Germany). The reason — organic compounds of plant leaves (tannin, cellulose pectin) form a thin slippery film when mixed with water; a hydrodynamic underlayer appears between tire and road, the same as aquaplaning but already at very low speeds (from 3–5 km/h). Especially dangerous are wet leaf piles in shaded alleys that do not dry out all day and accumulate over several days. The wheel passes without noticing them, and the brake is applied after the leaves are already under the wheel.

Painted lines — μ_wet ↓ ×3. AASHTO and TRB (Transportation Research Board) have published extensive research on the skid resistance of road markings. The standard unit is the British Pendulum Number (BPN) — the British Pendulum Tester’s traction reading. Normal wet asphalt has BPN ~55–70. An old painted stripe (ordinary emulsion paint) when wet — BPN 15–30 (μ ≈ 0.15–0.30). That means the braking distance on a wet marking is 2–3× longer.

Modern markings can be better: profiled thermoplastic markings (raised reflective markings) have ribs that bite through the water film and give BPN 35–45 in the wet. But old paint on irregularly refreshed roads is globally worse.

Riding along the marking in rain — categorically avoid. Zebra crossings, turn lanes, perpendicular arrows — all of these in wet weather become miniature ice patches. The canonical advice from cycling-safety literature: cross markings straight and upright, do not turn or brake on them; if possible cross by the shortest path.

Metal manhole covers, drain grates, slab joints with a metal insert — the rarest in frequency but the highest in risk for a scooter. Dry metal covers have μ ~0.4–0.5, below asphalt but acceptable. Wet — μ ~0.1–0.2 (like wet leaves). And when ambient temperature is around 0 °C and it is raining, metal surfaces cool below the dew point through radiative loss and become coated in a thin frost film even while the asphalt is still dry (the same mechanism described in the fog guide for FZFG black ice).

Bridge expansion joints with a metal component — separately. Many city bridges have finger-type expansion joints with metal “fingers” 5–10 cm wide, spaced by a 1–3 cm gap. In rain, this surface combines all three failure types: parallel grooves (geometric, like a tram rail), wet metal (μ ~0.1), and frequently set at an angle to the direction of travel. The canonical answer is dismount and walk across, or ride in the very outermost lane where the finger components end.

Strategy in summary:

Anticipate — OSM has surface=metal_grating and barrier=stile/ford for metal surfaces; markings are visible on satellite imagery; leaves — by season and by the alley density of the route.
Line — go around wet leaves and markings if possible, or cross them at 90° and straight; do not brake, do not turn.
Speed — ≤15 km/h, 5 m before the crossing point.

6. Expansion joints and bad patch repairs

Large expansion joints on bridges and roads, and bad patch repairs (patched potholes), are the engineering prose of urban asphalt, and they often attack the scooter through geometric step-transitions.

Bridge expansion joints. Depending on AASHTO standards and national building codes, joints come in different widths (from 25 to 150 mm) and constructions: finger-type (metal fingers), modular (multi-cell with rubber/metal), strip seal (a single rubber strip). For everything except the last type, a bicycle/scooter encounters transverse parallel grooves that functionally duplicate a tram rail: 30–80 mm wide, 10–30 mm deep, crossed at any angle ≤30° = wheel-slot trap. Countermeasure — cross at 90°, slowly, without braking on the joint.

Patched potholes — step transitions. When a pothole is repaired, a typical result is the asphalt patch raised 1–3 cm above the surrounding pavement (with the opposite case — a sunken patch 1–2 cm down). The front tire crossing such a step-up gets an instantaneous vertical impulse which has a lateral component as well, if the step is not strictly perpendicular to travel. At slow speed (<10 km/h) this is just unpleasant; at 25 km/h + small 8-inch wheels = front-wheel deflect — the scooter is kicked sideways by 10–30 cm. Especially bad when several patches follow in a row: the rider compensates for the first and cannot react in time to the second.

“Tarmac jelly” — soft repair. Sometimes repair patches are made with fresh asphalt without proper compaction; for a few days that material stays as soft as gum-eraser. The wheel deforms through it: amplitudes are small, but timing is unpredictable, because the material reacts differently to different speeds and loads. In hot weather this is worst — heated tarmac takes on plasticine-like properties.

Sunken utility covers — recessed manhole lids. A very common type: a sewer or utility access with a cover, around which the asphalt has settled 3–7 cm. The result — the cover sticks out like a puck, with a 5 cm drop around it. At 25 km/h this is a serious hit: the front wheel drops in, the tire may get a pinch-flat, the rider — a sharp shock to the lower back. At night without bright lighting, these holes are invisible — described in the night-riding guide. Countermeasure — go around (line change of 30–50 cm), not “they’re tiny”.

Transverse slab joints (concrete slab joints) — on concrete roads, there are joints between slabs that degrade over time. If a wheel enters an open 2–4 cm wide joint at an angle, that is a wheel-slot trap, like on a tram rail.

Defensive driving line on bad roads. If a route runs through a stretch with patches every 5–10 metres — drop speed to 15–20 km/h and ride closer to the lane centre, where patches are more often absent (the wear-strip centre). At transitions between slabs/patches do not brake; brake force at the moment of crossing adds to the vertical impact and can knock the wheel sideways.

7. Tire pressure, stance, speed — the defensive cross-cut

Everything above shares a few defensive parameters that are set up before you reach the problem surface.

Tire pressure — flexible, not fixed. Normal asphalt dry: 40–45 PSI (~2.8–3.1 bar). For cobblestones, wet asphalt with leaves, gravel <10 mm thick: 30–35 PSI (~2.1–2.4 bar). For deep gravel or sand: raise to 45–50 PSI (to reduce sinking). For wet surfaces with painted lines: keep stock pressure (40–45), because lower pressure increases the contact patch but also raises the aquaplaning risk on the lines (μ_wet stays the same; pressure does not save you). The pinch-flat trade-off at low pressure is real; compensate with tubeless or TPU tubes, if those options are available for your scooter.

Stance — active with three degrees of freedom:

Knees bent 5–10°, never straight. That adds 2–3 cm of vertical damping at the joint alone.
Elbows likewise 5–10° bent, with a light grip — do not “strangle” the bars. On cobblestones a rigid elbow passes every shock to shoulder and neck.
Torso slightly forward (15–20° from vertical), weight distributed 60 % front / 40 % rear. This is the bumps mode; on floating surfaces (gravel, ice) — invert to 40/60 (weight rear) for front-wheel control.

Stance width — wider on rough surfaces. Feet set wider apart (at the edges of the deck) on cobblestones and gravel gives a larger support base and better lateral control. On smooth asphalt you can stand narrow (close together), but on rough surfaces that sacrifices stability.

Speed — 60–75 % of your normal. If a normal cruising speed on asphalt is 25 km/h, on cobblestones and wet markings it is 15–18 km/h. On gravel >10 mm — 12–15 km/h. On wet round cobbles or expansion joints — ≤10 km/h. This is not an arbitrary number — it is a compromise between lower kinetic energy (easier to stop) and enough momentum not to “stick” in the valleys.

Brake bias — prioritise the rear on slippery surfaces. On normal asphalt braking force is distributed 60–70 % front / 30–40 % rear, because the front wheel takes more load during deceleration. On a slippery surface (wet leaves, painted lines, wet rail, cobblestones) invert to 40 % front / 60 % rear. The front tire locks catastrophically (instant face-plant), the rear — only induces a sideways slip that an experienced rider compensates by setting the foot down. If the scooter has only a front brake (drum/hydraulic) plus regen rear — avoid extreme surfaces until the brake system is upgraded, or use only regen in poor conditions.

Look ahead 5–10 metres, not 1 metre. The reaction distance for an e-scooter at 25 km/h is 4–5 m (i.e. 0.5–0.7 s reaction × 7 m/s). On cobblestones and gravel, sensor overload can stretch reaction to 0.8–1.0 s (because the eyes are busy tracking the nearest pebble). Looking further ahead, the rider builds a preview map of the road and has time to plan a line instead of reacting to each individual pebble.

8. Route reading and seasonal/diurnal patterns

The best defence against problem surfaces is not to ride on them at all. A route can be planned to skirt the worst sections, if you know where they are.

Seasonal cycle and associated surfaces:

Spring (March–May) — thaw potholes (frost heave potholes). Through freeze-thaw cycles, water in micro-cracks expands and breaks up the asphalt. The first month after winter is the worst for pothole density. Plan the route around stretches that had potholes last year (they reappear in the same places). Cold pavement still has a thin morning ice film on metal for months.
Summer (June–August) — “tarmac jelly” in hot weather (>30 °C). Fresh repairs soften. Dry cobblestones have higher μ, but dust and sand on them from neighbouring construction lowers grip.
Autumn (September–November) — wet leaves, season #1 risk. There are days with +5…+10 °C and drizzle when asphalt and cobblestones are covered in organic sediment, invisible to the eye. Worst in park alleys, embankments, and yards with old trees.
Winter (December–February) — black ice on metal, FZFG freezing fog, wet rails at –1…+1 °C are most slippery (effective μ → 0.03–0.05). On cobblestones in wet snow, a salt slurry forms that lowers grip and makes wheels heavy.

Diurnal cycle (time of day):

Morning (before 9–10) — shaded streets that have not seen the sun stay damp even when neighbouring sunny ones have dried. Especially north-facing buildings, narrow alleys, riverside drops. Bridges in the morning are often wet with condensation.
Day (10–17) — optimal in dry weather; on hot cobblestones in summer — the day’s highest μ.
Evening (18–20) — as temperature falls, metal covers and joints cool first; dew can form even while the asphalt is still warm.
Night (20+) — visibility-of-potholes problems (see night riding); wet surfaces are harder to see; bad repair patches look like smooth asphalt instead of shadows.

Mapping and tag-based routing. OpenStreetMap has tags that directly describe surface quality:

surface=* — asphalt, concrete, paving_stones (regular sett-block pavement), cobblestone (historical irregular cobbles), gravel, sand, dirt, grass, metal_grating.
smoothness=* — excellent, good, intermediate, bad, very_bad, horrible, very_horrible, impassable. For scooter routes, aim for good and above.
railway=tram — tram lines (to avoid).
bridge=yes + expansion_joint=* — bridges with joints.

Cycle-routing services (BRouter, Komoot, Cycle.travel) can take these tags into account. Unfortunately, tagging coverage in Kyiv, Lviv, Kharkiv is far from complete, so personal knowledge of the route is the best supplement.

Alternative — a multimodal route. If the city centre is cobbled, the better option may be: tram/metro to the edge of the cobblestoned zone → scooter from there on asphalt. That is also a solution; a folding e-scooter docks well with public transport (transport guide).

9. Recap — 8 principles of the surface axis

The contact patch (5–15 cm²) is the common denominator of every surface discipline. Three failure types: material μ failure (low grip), geometric trap-or-deflect (rail/joint steals the wheel), kinetic momentary contact loss (vibration breaks contact).
Cobblestones — sweet-spot speed 15–20 km/h dry, 8–12 km/h wet; line parallel to joints; tire pressure 30–35 PSI; active stance with bent knees and elbows.
Tram tracks — #1 single-cause risk in tram cities. Crossing angle ≥45°, ideally 60–90°; never parallel within 50 cm; wet-rail μ ~0.05–0.10 — worse than ice; dismount in rain near intersections.
Gravel — two-layer dynamics: raise pressure to 45–50 PSI; front-wheel plowing reduces lateral force; rear-brake priority; feed throttle smoothly; hard→loose transition at ≤15 km/h straight.
Wet leaves μ ~0.1 = ice, painted lines μ_wet ↓ ×3 (BPN 15–30), metal covers and grates wet are worse still. Cross straight and upright, do not brake and do not turn on them.
Expansion joints functionally duplicate tram rails (parallel grooves) — cross at ≥45°; bad patch repairs produce step-transitions — go around or ≤10 km/h; sunken utility covers — go around.
Defensive cross-cut: tire pressure by surface (30–35 PSI cobbles, 40–45 stock, 45–50 gravel); active stance with three degrees of freedom; speed 60–75 % of normal; brake bias inverted on slippery (60 % rear, 40 % front).
Route reading — seasonal cycle (spring-potholes / autumn-leaves / winter-black-ice); diurnal cycle (morning shaded streets); OSM surface= + smoothness= tags for planning; alternative — multimodal route bypassing the worst section.

Road surface is the fastest-changing variable in an urban ride: between two blocks μ shifts from 0.7 to 0.1 without any warning. The other axes (weather, visibility, cargo) can be planned in advance; the surface axis demands continuous visual scanning 5–10 m ahead and readiness for instant technique adaptation. That is the essence of an experienced urban scooter rider: recognising the surface type 50 metres ahead and automatically switching into the right defensive mode.