The first
1996 · Mitsubishi Carisma GDi
Mitsubishi's Gasoline Direct Injection on the Carisma was the first mass-production petrol direct injection system. The engine, an 1834cc four-cylinder, sprayed fuel directly into the combustion chamber rather than into the intake port — the same way diesels had been doing for fifty years, but at much higher pressure (50-100 bar versus a port-injection system's 4-5 bar) and with much more precise timing.
What that bought you was the ability to run a stratified-charge combustion mode at part throttle. The fuel injection happened on the compression stroke, creating a rich pocket of mixture near the spark plug surrounded by lean air; ignition started in the rich pocket and spread out. Effective air-fuel ratios in stratified mode could go as high as 40:1 (versus 14.7:1 stoichiometric), and the result was 15-20% better fuel economy than a conventional port-injected engine on the same combustion cycle.
The Carisma GDi was a commercial flop in Europe (Mitsubishi sold ~30,000 in three years) but the technology was the right one. By 1999 Toyota had its D-4 stratified-charge system. Audi launched FSI on the Lupo in 2000. Volkswagen brought stratified-charge GDI mainstream with the Mk5 Golf FSI in 2003.
The unintended problem
Stratified-charge combustion didn't survive contact with EU emissions regulations. The reason is unglamorous: lean-burn combustion at 40:1 produces high NOx emissions because combustion temperatures are higher and the exhaust gas has too much oxygen for a conventional three-way catalyst to reduce NOx effectively. Mitsubishi, Toyota, and Volkswagen all had to fit lean-NOx traps — additional catalytic converters that worked specifically in lean conditions — to meet Euro 4. By Euro 5 (2009) the lean-NOx trap was needed to be regenerated periodically by running rich, which negated most of the fuel-economy gain.
By 2012 every European manufacturer had quietly abandoned stratified-charge GDI. What survived — and what's now in every petrol car you can buy — is homogeneous-charge GDI. Same hardware (high-pressure pump, common rail, in-cylinder injectors) but the engine runs at conventional 14.7:1 air-fuel ratios at all loads. The benefits are different: you get cleaner combustion, the ability to compress the mixture harder (higher compression ratios — 11.5:1 on a turbocharged petrol where you used to need 9:1 with port injection), and finer control over emissions. The fuel-economy gain is real but smaller — around 5-10% rather than 20%.
The version that made it stick
2003 · Volkswagen Golf Mk5 FSI / TFSI
Volkswagen's Mk5 Golf launched with FSI (Fuel Stratified Injection) on naturally-aspirated 1.6 and 2.0 engines, and TFSI (turbocharged) on the GTI's 2.0. The TFSI 2.0 making 197 bhp from 2.0 litres was the breakthrough — it proved you could combine direct injection with turbocharging on a mass-market engine and the fuel economy advantage stayed real even at high loads.
VW called the technology TSI from 2005 onwards (the F was dropped — the engines were no longer running stratified). By 2010 every Volkswagen Group petrol engine over 1.4 litres was direct-injected. By 2015, every European mass-market petrol engine was direct-injected, full stop. The hold-outs were old-design engines being phased out (Honda's K20 and K24 stayed port-injected through 2015; the Mazda MZR Skyactiv-G stayed port-injected through 2018) but they were the exceptions, and they're all gone now.
Direct injection was the first piece of mass-market diesel technology to migrate to petrol engines. The second was particulate filters — and that arrived because of the same problems direct injection caused.
The catastrophic side effect
Direct-injection petrol engines produce a fundamentally different combustion product than port-injected ones. In a port-injected engine, fuel sprays into the intake manifold and washes the back of the intake valve as it goes past, keeping the valve clean. In a direct-injection engine, fuel never touches the back of the intake valve. So:
Carbon build-up on intake valves. Crankcase oil vapours and EGR-recirculated exhaust gas deposit hard, glassy carbon on the back of the intake valve. The deposit thickness reaches a few millimetres after 50,000-80,000 miles, restricting airflow, causing rough idle, misfires, and eventually engine damage.
The standard cleaning method is "walnut blasting" — disassembling the intake manifold, packing each port with masking tape, and using a media-blast machine with crushed walnut shells to scour the valve back to bare metal. It costs £400-700 per engine and is needed every 60-100k miles on a typical TFSI/TSI engine. Owners of pre-2015 GDI cars (Audi A4 B8, BMW 335i, Mercedes C250) often discovered this only when symptoms appeared — and the cost of remedial work on a 100k-mile premium saloon could easily exceed the car's value.
The fix, applied universally from around 2014, is dual injection — a port injector plus the direct injector. The port injector sprays at low loads to keep valves clean; the direct injector takes over at higher loads to deliver the efficiency benefits. Toyota was first to ship dual-injection in volume on the 2014 Corolla 1.8L. By 2018 BMW's B48 (2.0 turbo), Mazda's Skyactiv-X (2019), Ford's third-gen EcoBoost, and most VW Group EA888 evo4 engines were dual-injection. The carbon build-up problem was effectively solved on new engines.
But the 2008-2015 generation of single-direct-injection petrol engines is now the worst-aging cohort of petrol cars in the used market. They're the cars that age into £400 service bills around 80,000 miles.
The other unintended effect
Particulates. Direct-injection petrol engines produce solid particulate emissions in roughly the same range as a 1990s diesel — fine soot particles in the 50-200 nm size range. Port-injected engines don't, because the fuel and air mix in the intake manifold and burn cleanly.
The EU mandated petrol particulate filters (GPFs, structurally similar to the diesel particulate filter) on all new direct-injection petrol cars from September 2018. By 2026 every new petrol car has a GPF in the exhaust. The filter has the same regeneration cycle as a diesel — burning off accumulated particulates by raising exhaust gas temperatures periodically — and it has the same urban-driving problem: short trips don't produce enough exhaust heat to regenerate the filter, leading to clogging.
The GPF on a Polo or Fiesta isn't yet causing the dealership-survival problems that the diesel DPF caused on early-2010s cars, mostly because petrol GPFs are smaller and clog more slowly. But the warranty data is starting to show. Ford's 1.0 EcoBoost has a known GPF issue at 60-80k miles. The VW EA211 evo on early MQB cars has had a similar issue. The next decade of used-petrol-car reliability will be partly defined by GPF replacement costs.
The benefits that survived
Despite the collateral damage, direct injection delivered real improvements. The current generation of small-displacement turbocharged petrol engines (1.5-2.0 litres) makes power and torque figures that would have required 2.5-3.0 litres of normal aspiration in the 1990s. A 2025 Volkswagen Golf 1.5 TSI evo2 makes 148 bhp and 184 lb-ft of torque from 1,500 rpm. A 1995 Golf VR6 made 174 bhp and 173 lb-ft from a 2.8-litre normally-aspirated six.
Fuel economy is similarly improved — that 1.5 TSI evo2 returns 50+ mpg combined, where the VR6 was lucky to see 32 mpg. CO2 emissions per km are down by roughly 30% across the same comparison.
The reason those numbers are achievable is direct injection plus turbocharging plus variable valve timing plus dual-injection plus 16:1 compression ratios — five technologies that are all interlinked, all developed in roughly the same window (2003-2018), and all contributing to the same goal of making smaller engines do larger engines' work.
The mid-career obsolescence
The complication is that the technology that took fifteen years to mature is now reaching the end of its useful life as the new-car-of-choice. EVs have no use for direct injection. Hybrid petrols still use it, but the cycles are gentler — the petrol engine in a 2025 Toyota Prius or RAV4 PHEV runs in a much narrower load range than the petrol engine in a Mk5 Golf GTI did, so deposit build-up is slower and emissions are easier to manage.
By 2030, direct-injection petrol engines will be a declining technology — improved, refined, but increasingly relegated to plug-in hybrids and the cheapest segment of the market. By 2040, in the UK at least, they'll be in a minority of cars on the road.
What's interesting is that the basic invention — high-pressure injection of liquid fuel into a combustion chamber, with electronic timing — was never actually proprietary. Bosch sold the hardware; manufacturers tuned it. The technology that defined a generation of petrol engines was, in commercial terms, a single supplier's product. When direct injection finally fades, the engineers who developed it will mostly have moved on to electric drive units. Bosch will still be supplying the rest of the active-safety hardware in cars that don't have an engine at all.
That's probably the right outcome. Direct injection was a stop-gap — a way to keep the petrol engine competitive against the diesel in the 2000s, and against the EV in the 2010s. It bought the petrol engine an extra fifteen years of dominance. That's not bad for a technology that no one outside Bosch and a handful of OEMs ever really understood.