Wankel Rotary Engine

Why anybody ever built a rotary engine

The reciprocating piston engine is, in mechanical terms, an absurdity. A column of metal accelerates from zero to 25 m/s and back to zero, twice per revolution, while a connecting rod, crankshaft, balance shafts, valves, springs, cams, lifters, and pushrods conspire to translate that lurching one-dimensional motion into smooth shaft rotation. At 7,000 rpm a small piston changes direction 14,000 times per minute and the inertia loads can exceed the combustion loads. Every gram of reciprocating mass costs an exponential price in vibration and bearing wear.

Felix Wankel's insight in the early 1950s — first realised in the DKM 54 of 1957 — was a combustion engine with no reciprocating parts at all. A single triangular rotor turns inside a peanut-shaped housing. Three independent combustion chambers, formed between the three rotor faces and the housing wall, each go through intake, compression, power, and exhaust simultaneously. There is no valvetrain. There is no connecting rod. The only moving parts are the rotor, the eccentric shaft, the apex seals, and the side seals. A single-rotor Wankel has about a dozen moving components total; a single-cylinder piston engine has fifty.

The price for that simplicity is paid in three currencies. First, sealing — each rotor apex must drag a 3 mm steel strip past the housing wall at 16 m/s while keeping combustion gas in the right chamber. Second, fuel economy — the long, thin combustion chamber has a high surface-to-volume ratio at TDC and the flame must travel far. Third, emissions — slow combustion plus partial quench near the housing walls means high HC. The rest of this article is the engineering of that trade.

The epitrochoid and the 3:2 internal gearing

The housing's inner profile is an epitrochoid — the curve traced by a point on a circle of radius r rolling around the outside of a fixed circle of radius R, with R = 2r. Mathematically:

x(t) = (R + r) cos(t) − e cos((R + r)/r × t)
y(t) = (R + r) sin(t) − e sin((R + r)/r × t)

with R = 2r,  (R + r)/r = 3:  the housing has 2 lobes
and the rotor must have 3 vertices  to seal against 2 lobes.

Eccentricity e:  the offset between rotor centre and shaft centre.
Generating radius R:  determines the housing's "long radius".
On a Mazda 13B:  R ≈ 105 mm, e ≈ 15 mm.

The rotor itself rides on an eccentric — a lobe of radius e ground onto the central output shaft, offset from the shaft centreline. A large internal-tooth gear inside the rotor meshes with a stationary external pinion bolted to the side housing. The tooth ratio is exactly 3:2 — there are 3 teeth on the rotor's internal gear for every 2 on the stationary pinion (typically 24:16 or 30:20 in practice). That ratio mechanically constrains the rotor to turn exactly one-third of a revolution about its own axis for every full revolution of the eccentric shaft.

So in one full shaft turn: the eccentric makes one full orbit, dragging the rotor's centre around with it; the rotor rotates 120° about its own axis; one of the three rotor faces completes one full intake-compression-power-exhaust cycle. After three shaft turns the rotor has completed one full revolution and all three faces have each fired once. Power output rate: one combustion event per shaft revolution — comparable to a two-cylinder four-stroke piston engine.

The four phases — same as a four-stroke, but in parallel

Each of the three rotor faces continuously runs through the same four phases as a piston engine, but staggered 120° apart. At any given instant, all four phases are happening at once on different faces. The phases:

• Intake. A rotor face passes the intake port; the trailing apex uncovers the port; the trapped volume grows; fresh mixture rushes in. The intake port is cut into the side housing (later Mazda Renesis) or the peripheral housing (earlier 12A and 13B-REW). Port geometry determines the engine's character: side ports give clean idle and low emissions but limit top-end; peripheral ports give explosive top-end but bad idle.
• Compression. Both the leading and trailing apexes of that face are now between ports; the trapped volume shrinks as the rotor advances; pressure rises 8 to 10 times. Compression ratio is typically 9:1 on a naturally aspirated 13B, 8.5:1 on the turbocharged 13B-REW.
• Power. The face passes the spark plug (often two: leading and trailing); combustion sweeps from the leading edge of the chamber to the trailing edge; expanding gas pushes the rotor face outward, which (via the eccentric) drives the central shaft. Combustion is slow compared to a piston engine — the chamber is roughly 8 cm long and the flame must traverse it.
• Exhaust. The leading apex uncovers the exhaust port; the volume shrinks as the rotor advances; spent gas evacuates. Wankels famously have no exhaust valve overlap with intake (the geometry separates the ports), so no scavenging is possible — which is good for emissions but means the engine can't use exhaust pulse tuning to boost mid-range torque.

Worked example — sizing a Mazda 13B

The 13B (twin-rotor, 1.3 litres displacement total) is the canonical Wankel. Its dimensions are public. Let's work through the kinematics.

Geometry:
  Generating radius R    = 105 mm
  Eccentricity e         = 15 mm
  Rotor face width B     = 80 mm
  Compression ratio      = 9.0:1 (naturally aspirated)
  Effective displacement = 654 cc per rotor → 1308 cc twin

Manufacturer's rating (1992 FD RX-7, 13B-REW twin-turbo):
  Power      = 206 kW @ 6500 rpm
  Torque     = 314 Nm @ 5000 rpm
  Red line   = 8000 rpm

Power-impulse rate:
  1 combustion per shaft turn × 2 rotors = 2 impulses / shaft turn
  At 6500 rpm:  6500 / 60 × 2 = 217 power impulses / second
  Equivalent piston engine:  4-cylinder 4-stroke at same rpm
                            = 6500 / 60 × 4 / 2 = 217 / s   (identical!)

Specific output:
  206 kW / 1.308 L = 157 kW/L   (twin-turbo)
  175 kW / 1.308 L = 134 kW/L   (Renesis NA, RX-8 2003)
  Compare:  Honda K20A i-VTEC 2.0L NA = 110 kW/L

Fuel economy (BSFC):
  13B-REW under load:  290 g/kW·h
  Honda B16B NA:       240 g/kW·h
  Penalty:             ~20 % more fuel for same work

Two things stand out. First, the rotor's power-impulse rate is identical to a comparable 4-cylinder piston engine at the same rpm — they are kinematically the same. Second, the specific power per litre is roughly double what a naturally aspirated piston engine of the era could deliver. That's the entire commercial case for the Wankel — you can pack RX-7-class performance into an engine bay one-third smaller than the equivalent V6, which is exactly what Mazda did.

The apex-seal problem in detail

Three apex seals (one at each rotor tip), six side seals (running along each side of the rotor between apexes), and three corner seals (where apex meets side seal) together do the job a set of piston rings does on a reciprocating engine. The apex seals are the killer.

Each apex seal is a steel or carbon-impregnated strip roughly 3 mm wide, 8 mm tall, and the rotor's full thickness (80 mm on a 13B). It sits in a groove machined into the rotor tip with a spring underneath pressing it outward. Combustion-chamber pressure also enters a relief behind the seal, augmenting the spring force when load is highest. The seal slides at the rotor-tip velocity past the chrome- or ceramic-coated housing inner wall — about 16 m/s at 9,000 rpm.

The wear mechanisms that limit life are:

• Adhesive wear. Local micro-welding between seal and housing when the lubricating oil film thins. Cured by adding small amounts of two-stroke-style oil to the intake charge (Mazda's premix or OMP — Oil Metering Pump — system).
• Abrasive wear. Carbon particles from incomplete combustion act as grinding paste. Cured by direct injection (Renesis 13B has port injection but later experimental versions used DI) and rich-burn calibration.
• Apex-seal flutter. At high rpm the seal can momentarily lift off the housing wall as gas pressure behind it cycles. Cured by stiffer springs and lower seal mass — carbon fibre or PEEK seals.
• Spring fatigue. The spring under each seal cycles at engine rpm. After 100,000+ km of fatigue the spring relaxes and seal contact pressure drops. Cured by twin springs or wave-spring designs.
• Thermal distortion. The housing's lobe centre runs hotter than the side housings; differential thermal expansion creates oval distortion of the trochoid. The seal must conform. Cured by ceramic housings (experimental only) or careful coolant routing.

Mazda's evolution of apex-seal life is a microcosm of the rotary's history. The 1967 10A used flat steel apex seals lasting 50,000 km. The 1992 13B-REW used four-piece tipped seals with carbon-impregnated apex caps lasting 150,000 km. The 2003 Renesis used ceramic-tipped seals lasting 200,000 km under good maintenance. None ever matched a piston engine's typical 300,000 km piston-ring life, which is the entire reliability-perception problem of the rotary.

Variants and notable configurations

Variant	Rotors	Displacement	Output	Application
NSU KKM 502	1	500 cc	~30 kW	1964 NSU Spider — first production Wankel
Mazda 10A	2	982 cc	82 kW	1967 Cosmo Sport
Mazda 12A	2	1146 cc	96 kW	1978 RX-7 FB, peripheral port
Mazda 13B-REW	2	1308 cc	206 kW	1992 RX-7 FD — twin sequential turbo
Mazda Renesis 13B	2	1308 cc	175 kW NA	2003 RX-8 — side intake ports
Mazda 26B	4	2616 cc	522 kW	1991 787B — Le Mans winner
NSU Ro80	2	995 cc	85 kW	1967 sedan — premature apex-seal failure killed NSU
Mercedes C111 III	4	3600 cc	257 kW	1970 concept — never produced
Mazda 8C	1	830 cc	55 kW gen	2023 MX-30 R-EV range-extender

Failure modes — what actually kills a rotary

• Apex-seal failure. Most common. Symptom: low compression on one face of one rotor (180–250 kPa instead of 700+). Diagnosis: compression test with rotary-specific tester. Cure: rotor rebuild, ~3000 USD.
• Side-seal failure. Less common but as terminal. Symptoms: oil burning, blue smoke under boost. Cure: full engine rebuild.
• Coolant-seal failure. O-rings between the side housings and rotor housings fail under thermal cycling. Symptom: coolant in oil, milky oil cap. Cure: rebuild, typically combined with apex-seal refresh.
• Stationary-gear fracture. The internal-pinion gear that constrains rotor rotation can crack at very high rpm. Symptom: catastrophic engine failure. Cure: replace the side housing.
• Carbon build-up in ports. The Renesis side ports plug with carbon over 50,000 km on cars driven gently, costing power and creating idle problems. Cure: periodic 'Italian tune-up' (high-rpm operation) or chemical decarbonisation.
• Catalyst meltdown. If the engine runs rich for extended periods (a common failure mode of older rotaries) the catalytic converter overheats and the substrate melts. Common 1980s–1990s RX-7 failure; cured on the RX-8 by tighter combustion control.

How the Wankel compares to a four-stroke piston

Property	Wankel	4-stroke piston (similar displacement)
Specific power (kW/L)	130–160 NA, 200+ turbo	60–100 NA, 130+ turbo
Moving parts (single-cyl/rotor)	~12	~40–50
BSFC (g/kW·h)	280–300	230–260
Smoothness	Excellent (no reciprocating mass)	Requires balance shafts
Red line (rpm)	8,000–9,000 (street); 10,500 (race)	6,500–7,500 typical
HC emissions	High (wall quench)	Lower
Apex/ring life	150,000–200,000 km	250,000–400,000 km
Cold-start emissions	Worse (cool walls)	Better
Power density (kW/kg)	1.5–2.0	0.7–1.2

Why Mazda brought it back — and what comes next

The MX-30 R-EV (2023) put a single-rotor 8C Wankel back in a Mazda showroom for the first time since 2012. The engine is a 0.83-litre single-rotor making 55 kW (74 hp), used not as a traction engine but as a series-hybrid range-extender feeding a generator. It runs at near-constant rpm and load — the worst conditions for a rotary's fuel economy (transient response) are eliminated. The smooth, compact, light Wankel is well-suited to that duty: it tucks into the wheel-arch space ahead of the front axle without intruding into the cabin, weighs ~15 kg less than a comparable piston engine, and produces no second-order vibration that would otherwise leak into the cabin floor.

Hydrogen rotary research at Mazda continues. The wide flammability range and high flame speed of hydrogen partially compensate for the long, thin combustion chamber that hurts gasoline rotaries; emissions become trivial (water vapour plus a little NOx); the apex-seal problem persists but is no worse. A future MX-30 R-EV variant running on hydrogen is openly speculated about and consistent with Mazda's public patent activity.

What the Wankel demonstrably is not: a viable replacement for the piston engine in mainstream automotive applications. The fuel-economy disadvantage and apex-seal life problem have never been solved at scale, and the regulatory and emissions targets of the 2020s do not tolerate either. What it is: an extraordinary special-purpose engine for niches where smoothness, power density, and packaging matter more than fuel economy — exactly the niches Mazda keeps finding for it.

Common pitfalls when learning the rotary

• Counting combustion events wrong. Three per rotor revolution, one per shaft revolution (3:1 gearing). The shaft turns three times faster than the rotor. A single-rotor 13B has the same power-impulse rate as a two-cylinder four-stroke.
• Thinking the rotor 'spins'. It orbits and rotates at the same time. The combination produces the trochoidal sweep — neither motion alone gives the working chambers.
• Treating it like a high-rpm piston. Wankels need the OMP oil injection working perfectly, premix fuel for serious abuse, and a 30-second cool-down after hard running. They cost more in consumables than a piston engine of equal output.
• Ignoring the housing-coating spec. Different generations used chrome, Nikasil, or steel TiN coatings. The wrong oil for the wrong coating accelerates wear. The owner's manual is not optional.
• Believing turbocharging fixes the BSFC problem. Forced induction increases peak power but does not improve BSFC — it makes the surface-to-volume problem worse by raising peak cylinder pressure. The 13B-REW is faster than the Renesis NA but uses more fuel.