If only the engine had no friction, a Pulsar NS200 would deliver about 15% more power at full throttle and use roughly 13 to 30% less petrol for the same ride, depending on load. It would still waste about 60% of its fuel as heat, because thermodynamics, not friction, sets that limit. If friction vanished everywhere, the bike could not move, stop, turn or keep its nuts tight.
- ~11.5% of a car's fuel energy goes to engine friction, by the Holmberg global estimate.
- 13–30% fuel saving for a frictionless engine, because removing friction also shrinks exhaust and cooling losses.
- 51–62% is the Otto-cycle ceiling for an 11:1 engine (Bajaj manual, Otto-cycle formula, calculated in this post with a heat-capacity ratio of 1.3 to 1.4), so even a perfect engine stays well short of 100%.
- ~5% higher top speed from 15% more power, because drag power rises with speed cubed.
- 0 N of drive, braking or cornering force if friction disappears from the tyres too.
This sounds like a dream for anyone who has paid for a top-end rebuild. Friction eats piston rings, bearings and cam lobes, and it turns petrol into heat that the radiator throws away. So the question is fair: how much would we gain if it simply stopped existing?
I split it into two cases. First, a frictionless engine fitted to an otherwise normal bike. Second, a frictionless world, where the tyres, brakes, clutch and bolts lose friction too. The first is an engineering upgrade; the second is a physics disaster.
How much of a litre of petrol does friction actually eat?
About one-third of a car's fuel energy goes to overcoming friction of all kinds, and roughly 11.5% goes to friction inside the engine itself. That comes from the global study by Kenneth Holmberg of VTT Finland and Ali Erdemir of Argonne National Laboratory, which modelled an average car down to each rubbing contact.
Their average car turns fuel into roughly these shares: 33% out of the exhaust, 29% into the cooling system, and 38% into mechanical work. Of the friction losses, about 35% are in the engine, 35% in the tyres, 15% in the gearbox and 15% in the brakes. Multiply 35% by the one-third friction share and you get the 11.5% engine figure.
See the data as a table
| Share of fuel energy (%) | |
|---|---|
| Exhaust heat | 33 |
| Cooling system | 29 |
| Engine friction | 11.5 |
| Tyre rolling friction | 11.5 |
| Gearbox friction | 5 |
| Brake friction | 5 |
| Air drag | 5 |
Source: Holmberg, Andersson and Erdemir (2012), as reported by ScienceDaily; friction split applied to the one-third friction share · Average global passenger car, 60 km/h average speed. Friction rows are 35/35/15/15% of the 33% friction share.
In 1966 a British engineer named Peter Jost handed the UK government a report with an awkward title: Lubrication (Tribology): A report on the present position and industry's needs. His committee had gone factory by factory asking what friction and wear were costing the country. The answer was more than £500 million a year that better lubrication practice could save, which Holmberg and Erdemir later put at about £9,000 million in 2017 money. Jost also needed a word for the science of rubbing surfaces, so he coined one from the Greek tribos, rubbing. Sixty years later, every engine-oil spec sheet you read is a descendant of that report.
Holmberg and Erdemir: The impact of tribology on energy use and CO2 emission (OSTI open copy)Where is the friction inside an engine?
Nearly half of engine friction comes from the piston system: the skirt, the rings and the pin. Gore and colleagues at Loughborough, measuring a single-cylinder Honda motocross engine on a floating liner rig, put total engine friction at about 7–8% of combustion energy, and the top compression ring alone at about 3% of all engine losses.
The rest is spread across parts every rider knows:
- Crank and big-end bearings, sliding on a pressurised oil film.
- Valve train: cam lobes against followers, and the valve springs pushing back.
- Oil and water pumps, which are not strictly friction but are parasitic loads.
- Primary gears, the wet clutch and the gearbox, all churning through oil in a bike engine.

What would a frictionless engine gain? The worked solution
The trick is to separate the work the gas does on the piston (indicated work) from the work that reaches the crank (brake work). Friction is the difference. Engineers write this as mean effective pressures, which turn torque into an engine-size-independent pressure.
Assumptions
- Engine: BS6 Bajaj Pulsar NS200, 199.4 cc, bore 72 mm, stroke 49 mm, compression ratio 11:1, 24.5 PS at 9,750 rpm and 18.74 Nm at 8,000 rpm (Bajaj owner's manual).
- Indicated efficiency \( \eta_i = 38\% \) of fuel energy, the mechanical-energy share in Holmberg's car model.
- Friction mean effective pressure (FMEP) of 1.5–2.0 bar at 8,000 rpm. This is my estimate for a small high-revving single; I could not open a measured value for this engine.
- "Frictionless" means FMEP = 0. Pumping and heat losses stay.
Step-by-step
- The thermodynamic ceiling. The ideal Otto cycle efficiency depends only on compression ratio \( r \) and the gas's heat-capacity ratio \( \gamma \):\[ \eta_{Otto} = 1 - r^{\,1-\gamma} = 1 - 11^{-0.4} = 1 - 0.383 = 61.7\% \]With a more realistic \( \gamma = 1.3 \) for hot combustion gas, \( 1 - 11^{-0.3} = 51.3\% \). No amount of friction removal gets past this.
- Brake mean effective pressure at peak torque. For a four-stroke:\[ \text{BMEP} = \frac{4\pi T}{V_d} = \frac{4\pi \times 18.74}{199.4\times10^{-6}} = 1.18\ \text{MPa} = 11.8\ \text{bar} \]
- Mechanical efficiency at full load. With FMEP 1.75 bar, IMEP = 11.8 + 1.75 = 13.55 bar:\[ \eta_m = \frac{\text{BMEP}}{\text{IMEP}} = \frac{11.8}{13.55} = 0.87 \]
- Power lost to friction at 8,000 rpm. A four-stroke fires every two revolutions:\[ P_f = \text{FMEP}\cdot V_d \cdot \frac{N}{2\times60} = 1.75\times10^{5} \times 1.994\times10^{-4} \times 66.7 = 2.33\ \text{kW} \]The crank delivers \( 18.74 \times 837.8 = 15.7 \) kW there, so a frictionless engine makes about 18.0 kW at 8,000 rpm, a 15% gain. (Calculated in this post from the NS200's displacement and torque and the assumed FMEP above.)
- Fuel saved for the same ride. The work you need at the crank stays the same; fuel needed scales with \( 1/\eta_b \), and \( \eta_b = \eta_i - f \), where \( f \) is the friction share of fuel energy:\[ \text{saving} = 1 - \frac{\eta_i - f}{\eta_i} = \frac{f}{\eta_i} = \frac{11.5}{38} = 30\% \]
That last line is the surprise. Engine friction is only 11.5% of the fuel, yet deleting it saves 30% of the fuel. Holmberg and Erdemir make the same point: cutting friction also cuts the exhaust and cooling losses that came with burning the extra fuel, so the saving is close to three times the friction share.
A truly zero-friction engine is impossible, because any sliding contact dissipates some energy, and even the best lab result is a friction coefficient of about 0.000001 on a sub-millimetre contact. But an engine with near-zero internal friction would save 13% of fuel at full load and up to 30% at part load, and add about 15% power at 8,000 rpm. A bike with zero friction everywhere is impossible outright: Newton's laws need a friction force at the tyre to accelerate, brake or turn.
Why does friction matter more in city riding than on an open highway?
Because friction is roughly a fixed tax per revolution, while the useful work rises with throttle. FMEP depends mainly on engine speed, not load, so at small throttle openings the same friction is a much bigger share of a much smaller output.
| Riding case | Friction share of fuel energy | Mechanical efficiency | Fuel saved if frictionless | Basis |
|---|---|---|---|---|
| Full throttle, 8,000 rpm | about 5% | 0.87 | 13% | BMEP from Bajaj spec, FMEP estimate |
| Mixed riding | 7–8% | about 0.80 | about 20% | Gore et al. 2015 |
| Average car, 60 km/h | 11.5% | about 0.70 | 30% | Holmberg et al. 2012 |
So a frictionless engine helps most in exactly the riding that wastes the most: crawling through Bengaluru traffic at a quarter throttle. Use this to check what you just read with your own bike's numbers.
Your mileage with a frictionless engine
Would a frictionless engine make the bike much faster?
Only a little: about 5% more top speed. At top speed nearly all the power goes into pushing air, and aerodynamic drag power rises with the cube of speed:
Fifteen percent more power buys under five percent more speed. A bike that tops out near 130 km/h would gain about 6 km/h. The no-speed-limit what-if runs this cube law all the way to 600 km/h.
What breaks if friction disappears from the whole bike?
Almost everything that matters. Friction is not just a loss on a motorcycle; it is the working force at half a dozen places.
| Part | What friction does there | With zero friction |
|---|---|---|
| Rear tyre | Pushes the bike forward | Wheel spins, bike stays still |
| Front tyre | Supplies cornering force | No turn is possible; the bike goes straight |
| Brake pads | Turn motion into heat | No brake at all |
| Wet clutch | Couples engine to gearbox | Engine revs, nothing is driven |
| Bolts and nuts | Hold the thread from unwinding | Preload unscrews them |
| Chain and sprockets | Little; drive is by tooth shape | Still works |
The chain is the odd one out. It drives by positive engagement, teeth pushing rollers, so it does not need friction. The wet clutch is the opposite: it is nothing but friction plates.
How far would a moving bike coast?
About 1.2 km to slow from 100 to 20 km/h, if only air drag acts. With drag force \( \tfrac{1}{2}\rho C_dA v^2 \) on mass \( m \), define \( k = \rho C_dA/(2m) \). For 233 kg (an NS200 plus a 75 kg rider) and a drag area of 0.5 m², the upright-rider value measured in a University of Warwick wind tunnel:
The time is \( t = (1/v - 1/v_0)/k \approx 110 \) seconds. Compare a normal stop from 100 km/h with an assumed tyre friction coefficient of 0.8 (dry rubber on concrete is 0.6–0.85, Wikipedia): \( d = v^2/(2\mu g) = 27.8^2/15.7 \approx 49 \) m. Zero friction makes a 49 m stop into a 1.2 km drift with no steering.
Calculated in this post from the NS200's 158 kg kerb weight (Bajaj manual), an assumed 75 kg rider, and the 0.5 m² drag area from the Warwick study.
Why would the bolts unscrew themselves?
A thread is a ramp wrapped around a cylinder, and a tight bolt is a weight sitting on that ramp. Normally about 90% of tightening torque goes into friction, 50% under the head and 40% in the threads, according to Nord-Lock's bolting data. With zero friction, the clamp load pushes the nut down the ramp.
The unwinding torque is simply \( T = F p / 2\pi \). For an M10 bolt with 1.25 mm pitch and 20 kN of preload (my estimate for an axle-sized fastener):
That is a steady 4 N·m trying to spin every nut off, with nothing resisting it. Vibration would not even be needed.
Could a frictionless engine even run?
Yes, with three changes. The crank, pistons and valves need no friction to turn, so the core of the engine would run, and it would spin up faster. But three parts depend on friction:
- Replace the wet clutch with a dog clutch or a direct electric start-stop. A dog clutch locks by shape, like the gearbox.
- Keep the oil anyway. Oil also carries heat away from the piston crown and damps bearing loads. Friction gone does not mean cooling gone.
- Lock fasteners mechanically, with castle nuts and split pins or safety wire, as race bikes already do on critical bolts.
Engine braking would also change. Much of it comes from pumping losses against a closed throttle, explained in is engine braking bad for your bike, so it would weaken but not vanish.
Friction is the price of motion inside the engine and the source of motion outside it.The working above
How sensitive is the answer to the assumptions?
The fuel saving is the ratio \( f/\eta_i \), so it moves with both numbers:
- Lower indicated efficiency. At \( \eta_i = 32\% \) with 8% friction, the saving rises to 25%. Worse engines gain more.
- Higher friction. A cold engine runs thick oil; at 15% friction share and 38% indicated efficiency, the saving is 39%. This is why cold starts and short trips hurt mileage.
- Higher FMEP estimate. At 2.0 bar instead of 1.75, full-load mechanical efficiency drops to 0.855 and the frictionless power gain grows to 17%.
- Gearbox and chain. Holmberg counts another 5% of fuel energy in transmission friction. Add that and the frictionless drivetrain saves a further 13 points at part load.
Where the model breaks: it assumes heat and exhaust losses scale with fuel burned. A frictionless engine runs cooler at the bore, so its heat loss pattern would shift. Treat the 13–30% range as an estimate, not a measurement.
How close can real engines get to zero friction?
Closer every decade, but not to zero. Holmberg and Erdemir note that 25-year-old textbooks listed PTFE's 0.08 as the lowest sliding friction coefficient, while lab results now reach about 0.0005. In 2026 a Tsinghua University team reported friction coefficients as low as one-millionth on a sub-millimetre graphite contact, a state called structural superlubricity.
The catch is scale and dirt. Superlubricity needs atomically flat, defect-free, misaligned crystal layers; a piston ring is 72 mm round and lives in soot and fuel. Real engines get their gains from thinner oils, diamond-like carbon coatings, low-tension rings and roller cam followers. The oil change guide covers why grade matters.


What does the research say?
The authors model a global average car (75 kW, 1,500 kg, 8 l/100 km, 13,000 km a year) down to each friction contact, and compare it with a 2017 electric car. About 30% of the petrol car's fuel goes to friction and only 21.5% moves the car. Friction losses inside the car, excluding brakes and tyres, are 16.5% for the petrol car and 6% for the electric one. Worldwide, 20% of all energy (103 EJ) goes to friction. Limit: it is a modelled average car, not a measured motorcycle.
Read the paper (free)The team fitted a floating cylinder liner to a single-cylinder motocross race engine to measure in-cycle friction directly, under motored conditions, and matched it with an analytical model of the compression ring. They cite engine friction at about 7–8% of combustion energy, with nearly half from the piston system and about 3% of all engine losses from the top ring. Viscous friction dominated in their tests. Limit: motored, not fired, so peak combustion pressures are absent.
Read the paper (free)Using single-crystal graphite films with millimetre-sized grains, the authors built nearly defect-free sub-millimetre contacts and measured friction coefficients fluctuating around zero, as low as about 10⁻⁶, across loads from 1 mN to 0.5 N. Graphite on MoS₂ behaved similarly. Limit: 0.5 N is about the weight of 50 g; an engine bearing carries tens of thousands of newtons.
Read the paper (free)What's new
For the bigger picture of how an engine turns combustion into the torque you feel, read power and torque explained, and the Yantra page collects the rest of the engineering on this site.
What if next?
- How fast could a motorcycle go with only drag to stop it? The cube law that limited our frictionless gain sets a wall near 600 km/h.
- What if motorcycles were invented today? An electric drivetrain already loses about a fifth as much to friction.
- What if a bike were 3D-printed in one piece? No bolts means none to unscrew, but fatigue becomes the enemy.
Questions people ask
Would a frictionless engine be 100% efficient?
No. An engine is a heat engine, so its ceiling is set by thermodynamics. For an 11:1 compression ratio the ideal Otto cycle tops out at 51–62% (Otto-cycle formula with the NS200's 11:1 compression ratio, calculated in this post), and real combustion, heat loss and exhaust energy take it lower still.
How much power does engine friction take?
Typically 10–15% of indicated power at full throttle and a far larger share at light throttle. For an NS200 at 8,000 rpm, an estimated 1.75 bar FMEP costs about 2.3 kW.
Could a vehicle move with no friction at all?
Not by pushing on the road. Wheels need static friction to push the ground backwards. A vehicle on a frictionless surface needs a propeller, jet or magnetic track, which is the subject of what if tyres never touched the road.
Do low-friction oils actually improve mileage?
Slightly. Holmberg's group estimates that new tribology could cut car friction losses by 18% in 5–10 years, which saves fuel at about three times that friction share. On a bike, use the grade the manufacturer specifies, because the wet clutch needs some friction.
Sources
- Holmberg and Erdemir: The impact of tribology on energy use and CO2 emission globally and in combustion engine and electric cars, OSTI open copy (2019) (source for 30% friction share, 21.5% car efficiency, the 0.0005 friction coefficient and the Jost report figures)
- ScienceDaily: One third of car fuel consumption is due to friction loss (Holmberg, Andersson, Erdemir 2012) (source for the 33/29/38 split and the 35/35/15/15 friction breakdown)
- VTT: Global energy consumption due to friction in passenger cars (2012) (source for the 18% and 61% saving potentials)
- Gore, Rahmani, Rahnejat, King: Assessment of friction from compression ring conjunction (2015) (source for 7–8% engine friction and the piston-system share)
- Bajaj Auto: Pulsar NS200 owner's manual (source for bore, stroke, 11:1 compression, 24.5 PS, 18.74 Nm, 158 kg)
- Han et al.: Observation of robust macroscale structural superlubricity, arXiv / PRL (2026) (source for the 10⁻⁶ friction coefficient)
- Kirca and McGordon: Effect of rider position on the energy consumption of an electric motorcycle, University of Warwick (2024) (source for the 0.5 m² upright drag area)
- Nord-Lock: What is the relationship between torque, preload and friction? (source for the 50/40/10 torque split)
- Peter Jost, Wikipedia, citing the 1966 Jost report (source for the report title and the coining of "tribology")
- Phys.org: Physicists achieve near-zero friction on macroscopic scales (2026)
- EurekAlert!: Robust high-temperature macroscale superlubricity (2026)
- Physics Forums: Friction force on a car at constant velocity (reader question)
