LUNAROPS · OPERATIONAL UPLINK 100% UPTIME 1,247d POSTS 893 JEFF.MOON@LUNAROPS.DEV UTC --:--:--

How a Modern ICE Engine Really Works

22 min read Cars & Vehicles
carsenginesthermodynamicsturbochargingmechanicalautomotive
Contents

A modern naturally-aspirated 2.5-liter four-cylinder matches a 1995 BMW M3 inline-six on peak horsepower, returns 35 mpg combined doing it, runs on 87-octane regular, idles smoother than that M3 ever did, and converts roughly 40% of the chemical energy in its fuel into work at the crankshaft. Carbureted engines topped out around 25 to 28% brake thermal efficiency at their best operating point and spent most of their lives well below that. The 12-percentage-point improvement looks small on paper. It is the difference between a 1990 Civic and a 2025 RAV4 Hybrid. The way it was extracted is a decades-long campaign against four enemies every gasoline engine fights: knock, pumping losses, heat loss to the coolant, and exhaust enthalpy escaping out the tailpipe. Each buzzword on a modern engine spec sheet, direct injection, twin-scroll turbo, dual VVT, cooled EGR, Miller cycle, cylinder deactivation, is a counterattack against one of those four. This is what they actually do, why downsized turbo fours displaced V6s, and where the ceiling sits.

The textbook lies, but it’s the right place to start

Every introductory thermodynamics course teaches the air-standard Otto cycle: four reversible processes around the P-V diagram, with thermal efficiency

eta_otto = 1 - 1 / CR^(gamma - 1)

where CR is the geometric compression ratio and gamma is the ratio of specific heats, about 1.4 for air. Plug in numbers. At CR=8, eta is 56%. At CR=11, 61%. At CR=14, 65%. The textbook says: just keep raising compression and you keep winning. So why does a 2025 EcoBoost 2.3 run a 10.0:1 static compression ratio and a Toyota A25A-FXS run 14:1 only because it is an Atkinson hybrid that throws away cylinder fill?

Because the air-standard cycle is a fiction with five problems, every one of which a modern engine has to fight separately.

  1. Knock. The end-gas in the combustion chamber, the unburned mixture ahead of the flame front, gets compressed and heated by the advancing flame. If it autoignites before the flame arrives, you get a pressure spike with characteristic 6-to-9 kHz ringing that erodes piston ring lands and melts spark-plug electrodes. Knock is the single hardest constraint on a real gasoline engine. It is why a 1970s Pontiac 455 ran 8.2:1 compression on leaded premium and a 2024 Mazda Skyactiv-G runs 13:1 on regular by being cleverer in every other dimension.
  2. Pumping losses. A throttled SI engine at idle is a vacuum pump sucking against a closed butterfly. The pumping loop on a P-V diagram is negative work. At light load it can eat 10 to 20% of indicated power.
  3. Heat loss to the coolant. The combustion chamber is at 2000 K. The cylinder wall is at 400 K. Heat flux through the boundary layer is relentless and grows with surface-area-to-volume ratio, which is one reason small bores are inherently less efficient than big bores at the same compression.
  4. Exhaust enthalpy. The gas leaving the exhaust valve is still at roughly 800 to 1000 K and several bar. That is recoverable energy walking out the back of the car. Turbochargers recover some of it. Most still leaves.
  5. Mixture preparation, friction, and incomplete combustion. The real working fluid is not air, it is a partially mixed charge with residuals from the last cycle, it does not behave like an ideal gas, gamma falls as temperature rises, and pistons rub on cylinder walls.

Every modern engine technology is a targeted answer to one of those five. The list is short. The implementations are not.

Direct injection: the cheat code for compression ratio

Port fuel injection, which dominated from the mid-1980s through the late 2000s, sprays gasoline into the intake port upstream of the intake valve. The fuel has a long time to vaporize in the runner and arrives at the cylinder as a fairly homogeneous charge. Cheap, robust, clean intake valves because the gasoline literally washes them.

Direct injection (gasoline direct injection, GDI) sprays fuel under high pressure, typically 200 to 350 bar in current systems, straight into the combustion chamber on the intake or compression stroke. The fuel vaporizes inside the cylinder. Vaporization is endothermic. Gasoline’s heat of vaporization is about 350 kJ/kg. Spraying liquid fuel directly onto the inbound air drops the charge temperature by roughly 15 to 25 K, depending on load and injection timing. That is the cheat code.

A cooler charge at the start of compression means a cooler charge at the end of compression, which means more knock margin, which means you can run a higher geometric compression ratio on the same fuel. The Honda L15B7 in the Civic and Accord runs 10.6:1 on 87 octane and makes 192 hp at 1.5 liters of displacement. The pre-GDI K20 of the early 2000s ran 9.8:1 and made 160 hp from 2.0 liters. Same fuel, more compression, smaller engine, more power.

The downsides are real and well documented. With no fuel touching the intake valves, carbon deposits build up on the back of the valves over time and hurt flow. Manufacturers responded with several fixes: VW’s EA888 evo3 and Toyota’s D-4S combine port and direct injection so the port injectors periodically wash the valves. BMW’s high-pressure system recommends walnut- shell blasting at 60,000 to 80,000 miles. The other GDI problem is fine particulates, soot, generated when fuel droplets do not fully vaporize before combustion. PM emissions from GDI engines can rival diesels on a particle- number basis, which is why gasoline particulate filters (GPFs) are now mandatory on essentially every new GDI engine sold in Europe under Euro 6d and are being locked in further under Euro 7.

Variable valve timing: making one cam do many jobs

A fixed camshaft is a compromise. The valve events that make peak power at 6500 rpm are wrong for idle stability, wrong for low-end torque, wrong for catalyst light-off, and wrong for the Miller cycle. Variable valve timing gives the engine controller a third axis, alongside throttle and fuel quantity, to shape the in-cylinder process.

The earliest production systems were discrete. Honda’s original VTEC switched between two cam profiles at a threshold RPM. Modern systems are continuously variable on both intake and exhaust cams. The cam phaser is a hydraulic or, increasingly, electric actuator that rotates the cam relative to the timing chain through a few tens of crank degrees. Toyota’s Dual VVT-i, BMW’s double VANOS, Ford’s Ti-VCT, and Hyundai’s CVVD all do versions of this. Some go further: BMW Valvetronic and Fiat MultiAir vary valve lift itself, which lets the engine throttle on the intake valves instead of a butterfly and recovers most of the pumping loss at light load.

What VVT actually buys you:

  • Better volumetric efficiency across the RPM band. Phase the intake cam to keep the intake valve open later at high RPM to use intake-runner inertia.
  • Internal EGR for emissions and knock control. Hold the exhaust valve open slightly into the intake stroke. Hot residuals dilute the next charge and reduce NOx formation and knock tendency.
  • The Miller and Atkinson cycles, discussed next, both depend on shifting intake valve closing far from where a fixed cam could put it.
  • Cold-start catalyst heating. Retarded combustion plus carefully phased cams can dump heat into the catalyst in 10 to 20 seconds instead of a minute.

The Miller and Atkinson cycles: paying horsepower for efficiency

Atkinson’s original 1882 engine used a complicated linkage to make the expansion stroke physically longer than the compression stroke. The thermal benefit is direct: in a normal Otto cycle, the gas leaving the exhaust valve is at a few bar and hundreds of Kelvin above ambient, which is wasted enthalpy. If the expansion ratio is larger than the compression ratio, you let the gas push the piston farther and extract more of that enthalpy as work before opening the exhaust valve. The downside is also direct: you draw in less air per cycle, so peak power per liter drops.

In production engines, nobody builds an Atkinson linkage. They build an Otto engine and use VVT to delay intake valve closing until the piston has already started moving up on the compression stroke. The charge gets pushed back into the intake manifold until the valve finally closes. From that point on, you compress what remains. The effective compression ratio is lower than the expansion ratio. Thermodynamically, this is the same trick as Atkinson’s linkage.

The Miller cycle is the same idea but historically uses early intake-valve closing (EIVC) before BDC, often paired with forced induction so you can recover some of the lost power density. In practice, the Toyota Dynamic Force Atkinson hybrids close the intake valve late (LIVC); Mazda Skyactiv also uses late closure; Hyundai’s Smartstream G1.6 Atkinson hybrid does both depending on operating point with CVVD. The Miller-vs-Atkinson naming is now mostly historical.

Numbers. The Toyota A25A-FXS, the hybrid variant of the Dynamic Force 2.5, runs a geometric compression ratio of 14:1 but an effective compression ratio closer to 12:1 because of the late intake valve closing. EPA benchmarking measured peak brake thermal efficiency around 41%. The non-hybrid A25A-FKS uses a milder Atkinson schedule, runs 13:1, and peaks around 40%. For context, the EA888 evo4 in a current Audi A4 peaks around 36 to 37%, the EcoBoost 2.3 around 35 to 36%, and a typical port-injected NA V6 from 2005 was at 32%.

The penalty: the A25A-FXS specific output is about 50 kW/L, modest by modern standards. That is fine for a Camry Hybrid, where the electric motor fills in the torque hole below 2000 rpm. For a hybrid powertrain the narrow efficient operating band is a feature.

The four-stroke cycle, drawn

 INTAKE                COMPRESSION             POWER                   EXHAUST
 piston down           piston up               piston down             piston up
 intake valve OPEN     both CLOSED             both CLOSED             exhaust valve OPEN

  |   air+fuel  |       |    |                  | <-- flame |           |  -> out  |
  |     V       |       |    |                  | <-- front |           |          |
  |             |       | ^^ |                  |     |     |           | ^^^^^^^^^|
  |  __piston__ |       |____|                  |__piston__ |           |__piston__|

 P-V diagram (Otto vs Atkinson, qualitative):

    P |
      |  *.
      |    *.            <- power stroke (expansion)
      |      *.
      |        *.        Otto: expansion ends here, hot gas vented
      |          *.      Atkinson: expansion continues to lower pressure
      |            *.  <-- more work extracted
      |              *_____________
      |__________________________________  V
        TDC                          BDC      BDC_extended

The shaded extra area under the curve, between the Otto exhaust point and the Atkinson exhaust point, is the recovered work.

Turbocharging: stealing the energy in the exhaust

A turbocharger is a centrifugal compressor on the intake side mechanically linked, via a common shaft, to a radial-inflow turbine on the exhaust side. Hot, high-pressure exhaust gas spins the turbine, the turbine spins the compressor, the compressor packs more air into the cylinders, and the engine makes more power than its displacement should allow. Modern automotive turbos spin from idle up to 200,000 to 280,000 rpm and live on a hydrodynamic or, in BMW’s B58, a ball-bearing center cartridge cooled and lubricated by engine oil and, in most recent designs, engine coolant.

The full airflow path of a modern turbo gasoline engine:

ambient air
    |
    v
+---------+   +-----------+   +-------------+   +-----------+
| airbox  |-->| compressor|-->| intercooler |-->| throttle  |
| filter  |   | (turbo)   |   | (air-water  |   | (DBW)     |
+---------+   +-----------+   |  or air-air)|   +-----------+
                              +-------------+        |
                                                     v
                                              +--------------+
                                              | intake       |
                                              | manifold     |
                                              +--------------+
                                                     |
                                                     v
                                            +-------------------+
                                            | intake valves     |
                                            | -> CYLINDER       |
                                            | -> exhaust valves |
                                            +-------------------+
                                                     |
                                                     v
                                              +------------+
                                              | exhaust    |
                                              | manifold   |
                                              | (twin-     |
                                              |  scroll)   |
                                              +------------+
                                                     |
                                                     v
                                            +-----------------+
                                            | turbine wheel   |
                                            | (drives        |
                                            |  compressor)    |
                                            +-----------------+
                                                     |
                                                     v
                                            +-----------------+
                                            | catalyst (TWC)  |
                                            | + GPF           |
                                            +-----------------+
                                                     |
                                                     v
                                                  tailpipe

The reason downsized turbo fours wiped out the small V6 in mainstream cars between roughly 2010 and 2020 is not romance, it is regulation and physics.

CAFE (Corporate Average Fuel Economy) and the EU’s CO2 targets are measured on cycle test fuel consumption. Cycle test fuel consumption is dominated by the part-load operating region: 30 to 50 kW road load at cruise. A naturally aspirated 3.5-liter V6 producing 280 hp does that 40 kW cruise at maybe 20% of its peak power, deep in its inefficient pumping-loss-dominated low-load region. Its specific brake fuel consumption in that operating point might be 280 g/kWh. A turbo 2.0 producing the same peak 280 hp does the same 40 kW cruise much closer to its sweet spot, partly throttled but with less pumping loss because the throttle is more open at a given air mass, and might hit 240 g/kWh. The peak power is the same; the area under the daily-driving operating point is materially better.

What changed between the early turbos (think 1980s Saab) and modern ones is transient response. The two big enablers were twin-scroll turbines, which separate exhaust pulses from paired cylinders into two volutes that hit the turbine wheel without interfering with each other, and shrinking the turbine inertia so it spins up on a few-hundred-millisecond timescale. The Ford EcoBoost 2.3 in the Mustang and Bronco, the VW EA888 evo4 in the Golf GTI, and the BMW B58 in everything from the Z4 to the X3 M40i all use twin-scroll. Mercedes’ M139 in the C 63 went further with an electrically assisted turbo that uses a small motor on the shaft to eliminate lag entirely below 2000 rpm.

Twin turbo six-cylinder engines like the B58 are interesting because they were not killed by the turbo four. They survived because BMW kept finding ways to make the inline-six smoother and more efficient than any four. The B58 in current B58TU1 form produces 382 hp from 3.0 liters and runs an 11:1 compression ratio on 91 octane. Its specific output of 127 kW/L is V8 territory from the late 2000s.

Knock under boost is where it gets nasty. Boost pressure raises the charge density at the start of compression, which raises everything downstream: end-of-compression pressure, end-of-compression temperature, peak combustion temperature, end-gas temperature. The knock margin collapses. Engines manage this with three tools in combination: high-flow GDI for evaporative cooling, aggressive cooled EGR (re-routing 10 to 20% of cooled exhaust gas back into the intake to dilute the charge and lower peak combustion temperatures), and reactive knock control.

Knock detection: per-cylinder, per-cycle, on the cheap

A knock sensor is a piezoelectric accelerometer bolted to the block. When end-gas autoignites, the resulting pressure oscillation rings the cylinder at frequencies determined by cylinder bore and sound speed, typically between 6 and 9 kHz for an 80-to-95-mm bore. The ECU windows the sensor signal to the few crank degrees after each cylinder’s combustion, bandpasses to the knock frequency, and compares amplitude to a noise baseline. If amplitude exceeds threshold, the ECU retards ignition timing by some small amount, typically 1.5 to 3 crank degrees, on that specific cylinder for the next few cycles, then walks timing back to MBT (maximum brake torque) one small step at a time looking for the new knock threshold.

This closed loop runs constantly. It is why a 2025 turbo four can advertise “premium recommended, regular acceptable” and not melt itself on 87 octane. The ECU finds whatever MBT the fuel can support and lives there. The cost is some peak power on cheap gas. On a Honda L15B7, that is perhaps 4 to 6 hp out of 192. On a BMW B58 it might be 25 hp out of 382.

Knock control is also why pre-ignition (LSPI, low-speed pre-ignition) became a public scandal around 2015 with early downsized GDI turbos. LSPI is not the same as knock: it is an autoignition event triggered before the spark, by an oil droplet or carbon deposit acting as a glow plug at high cylinder pressure and low engine speed. By the time the knock sensor hears it, the piston is already cracked. The fix was new GF-6 / SP oil specifications with detergent chemistries less prone to LSPI-inducing deposits.

A century of engine technology, on one table

Era / Engine Years Disp. Config Power CR Specific output Peak BTE Key tech
Chevy 350 SBC (carbureted) 1970-1980 5.7 L V8 ~165 hp 8.5:1 22 kW/L ~26% Carb, fixed cam, hydraulic lifters
BMW M50B25 1992-1995 2.5 L I6 189 hp 10.5:1 56 kW/L ~30% Port FI, single VANOS
Honda K20A2 (RSX Type S) 2002-2006 2.0 L I4 200 hp 11.0:1 75 kW/L ~32% Port FI, i-VTEC, NA
Toyota 2GR-FE 2005-2017 3.5 L V6 268 hp 10.8:1 57 kW/L ~33% Port FI, Dual VVT-i
Ford EcoBoost 2.3 2015- 2.3 L I4 T 310 hp 9.5:1 100 kW/L ~36% GDI, twin-scroll turbo, dual VVT
VW EA888 evo4 2019- 2.0 L I4 T 261 hp 11.7:1 97 kW/L ~37% GDI+PFI, VVT+lift, Miller cycle option
Honda L15B7 2016- 1.5 L I4 T 192 hp 10.6:1 95 kW/L ~36% GDI, single-scroll turbo, dual VVT
BMW B58TU1 2018- 3.0 L I6 T 382 hp 11.0:1 127 kW/L ~38% GDI, twin-scroll, Valvetronic, water injection
Toyota A25A-FKS 2017- 2.5 L I4 203 hp 13.0:1 60 kW/L ~40% D-4S (GDI+PFI), Dual VVT-iE, mild Atkinson
Toyota A25A-FXS (hybrid) 2017- 2.5 L I4 176 hp 14.0:1 50 kW/L ~41% D-4S, full Atkinson, cooled EGR
Mazda Skyactiv-G 2.5T 2018- 2.5 L I4 T 250 hp 10.5:1 74 kW/L ~37% GDI, twin-scroll, cooled EGR
Mazda Skyactiv-X (SPCCI) 2019-2025 2.0 L I4 186 hp 16.3:1 70 kW/L ~42% SPCCI, 48V mild hybrid, supercharger
Hyundai Smartstream G1.6 hybrid 2020- 1.6 L I4 138 hp 14.0:1 64 kW/L ~40% GDI, CVVD continuous duration, Atkinson

The rough pattern: pre-2000 mainstream gasoline engines spent their time between 25 and 32% peak BTE. Modern non-hybrid turbo gasoline is 35 to 38%. Modern hybrid-optimized Atkinson and SPCCI engines have crossed 40%.

Skyactiv-X and SPCCI: the last serious thermodynamic experiment

Mazda’s Skyactiv-X is worth a section because it is the only production gasoline engine to commercialize controlled compression ignition. Diesel runs by compression ignition: fuel injected into a heavily compressed air charge autoignites because the air is hot enough. Diesels are efficient partly because they run unthrottled (load is set by fuel quantity, not by strangling air) and partly because they run lean. They have the problem that NOx and soot trade off and require expensive aftertreatment.

The HCCI (homogeneous-charge compression ignition) dream was to do that trick with gasoline: lean, unthrottled, compression-ignited. The engineering problem was you could not control when ignition happened. Combustion phasing drifted with load, temperature, residuals. It worked beautifully on a dyno at one operating point and did nothing reliable across a real drive cycle.

Mazda’s SPCCI (spark-controlled compression ignition) is a clever cheat. The engine compresses a lean charge to a 16.3:1 geometric ratio, almost to the autoignition threshold. A spark plug fires at the right crank angle. The small flame kernel acts as a final pressure boost that pushes the surrounding unburned charge over the autoignition line. The compression-ignition portion of the burn is then phased by spark timing rather than by hoping.

A small Roots-type supercharger packs extra air to keep the charge lean, typically at lambda 2 or higher (twice the stoichiometric air ratio) in the SPCCI mode. A 48V belt-starter-generator and a small battery handle transients and recover braking energy. Skyactiv-X claimed peak BTE around 42%, in the same ballpark as the Toyota A25A-FXS.

The commercial result was tepid. Skyactiv-X added cost without delivering fuel-economy gains large enough to justify it in showrooms where the comparison was no longer a port-injected engine but a Toyota hybrid. Mazda quietly retired Skyactiv-X in most markets by 2025 and pivoted to Skyactiv-Z for 2027, which simplifies the combustion approach back toward conventional lambda-1 lean-burn-adjacent operation with better heat insulation in the chamber. The lesson was less about SPCCI being wrong than about a hybrid electrified powertrain being a cheaper way to capture the same drive-cycle efficiency.

Cylinder deactivation, the last few percent

If you cannot make a big engine think small, you can at least make it act small at light load. Cylinder deactivation closes the intake and exhaust valves and cuts fuel to selected cylinders below a load threshold. The deactivated cylinders become air springs that absorb and return work each cycle with low net loss. The active cylinders are loaded more heavily, which moves them away from the inefficient low-load region. Honda VCM and GM AFM/DFM are the prominent examples. GM’s DFM (Dynamic Fuel Management) on the 5.3 and 6.2 L V8 truck engines does this on a per-cylinder, per-cycle basis with 17 active firing patterns. The benefit is real but modest, 5 to 10% on EPA combined cycle, depending on how much time the engine spends in the deactivation window.

The technology comes with NVH (noise, vibration, harshness) headaches because four cylinders running on an eight-cylinder crank produces an uneven firing order. Active engine mounts and electronic noise cancellation through the audio system, which itself is a tell-tale sign of a deactivating engine, are part of the price.

The honest thermal-efficiency ceiling

How much further can the gasoline ICE go? The air-standard Otto at 14:1 compression suggests roughly 60% efficiency. Real engines are at 41%. The gap is mostly heat loss to the chamber walls and exhaust enthalpy. Lab demonstrators using exotic combustion modes, lean burn with high-energy ignition, prechamber jet ignition (similar to F1 engines), insulated chambers with thermal barrier coatings, have reached 45 to 50% on the dyno at specific operating points. The path from those numbers to a production passenger car engine that meets cost, durability, cold-start emissions, and NVH targets is long, and the prize at the end is a 2-to-4 percentage-point fuel-economy gain on the drive cycle, in a world where hybrid and BEV powertrains have already taken larger gains for less engineering risk.

For context: diesel passenger engines peak around 45 to 46% BTE and heavy-duty diesel goes higher, 48 to 50%, but at the cost of an aftertreatment system with DPF, DOC, SCR, and urea dosing. Marine slow- speed two-strokes like the Wartsila 14RT-flex96C cross 50% because they run at 100 rpm with huge swept volumes and minimal heat-loss surface area per unit charge. Stationary combined-cycle gas turbines pass 60% because they get to use the exhaust enthalpy in a steam bottoming cycle. None of that geometry helps a passenger car.

The real ceiling for a mass-market passenger-car gasoline ICE is somewhere between 42 and 45% BTE at the sweet spot, with the operating window of high efficiency narrow enough that you need a hybrid powertrain to exploit it across a real drive cycle. That is why every serious new ICE program is hybrid-first: the engine is designed for a narrow optimal band, and the electric motor handles everything outside that band.

The Toyota A25A platform is the cleanest example of this co-design. As a hybrid engine it runs in its 41% efficiency island while the eCVT and motor absorb the cycle-to-cycle demand variation. Hyundai’s Smartstream G1.6 with CVVD does the same. The Honda L15Y7 hybrid Atkinson is the same idea on 1.5 liters. The conventional drivetrain version of the same block, when there is one, runs a different cam map, lower geometric compression, and gives up two to four points of peak BTE in exchange for usable torque across the full RPM range.

What this connects to in the rest of the drivetrain

The torque converter and planetary automatic transmission, covered in How Automatic Transmissions Work, exists partly to keep a narrow-band engine in its sweet spot. An eCVT in a hybrid does the same thing differently. The thermodynamic framing of how an engine extracts work from a hot reservoir and dumps the rest to a cold one is the same Carnot bound that governs both The Refrigeration Cycle running backwards and your Heat Pump Carnot in Your Garage running forwards. A heat pump fights the Carnot limit to move heat at COP 3 or 4. An engine fights the same limit to extract work at 40% efficiency. Both lose. The numbers are just the cost of doing business with the second law.

Verdict

The modern internal combustion engine is one of the great hidden engineering achievements of the last twenty years. Direct injection bought four to five points of compression ratio. Dual continuously variable valve timing enabled Miller and Atkinson operation, internal EGR, and pumping-loss recovery. Twin-scroll turbocharging plus knock-controlled timing maps let a 2.0-liter four do the job of a 3.5-liter V6 with the same drivability and materially better cycle economy. Per-cylinder knock detection turned the octane number of the fuel from a design constraint into a runtime parameter. Atkinson and SPCCI engines have walked peak brake thermal efficiency from the 32% typical of a 2005 NA V6 to 41% in production today, and the lab is showing 45 to 50% under specific conditions.

The honest answer to “how far can ICE go” is: a few more points of peak efficiency, narrowed into a smaller operating window, that only makes sense when paired with electrification. Below about 50 kW continuous, the road forward is a small Atkinson four married to a motor and a battery. Above that, in heavy trucks, marine, off-highway equipment, and the long-haul fleets, diesel and gas turbines are not going anywhere soon for energy- density reasons. The pure naturally-aspirated gasoline V6 sedan engine of 2005, the thing the EcoBoost killed, is genuinely gone. So is the carbureted big-block. What replaced them is a smaller, denser, smarter engine that does more work per liter of fuel than any internal combustion engine has ever done in a passenger car, and that is busy redesigning itself to be the better half of a hybrid powertrain.

Sources

Comments