Leaner Lambda Values when using Direct Injection?

I was reading on another forum that DI allows you to get away with leaner Lambda targets in turbo charged applications than you would with port injection. If this is true, can someone explain why?

It would seem to me that you'd still want the cylinder cooling and knock prevention that a richer .77 - .81 lambda provides regardless of how the fuel is getting into the cylinder.

Thoughts?

DI is a whole new ball game when it comes to AFR's. The AFR that you'll be reading in the exhaust may appear to be leaner than you'd usually target but that is because the air/fuel mixture in the chamber is delivered differently, the fuel (in some cases) is delivered in a funnel shape directly under the spark plug and the piston top is shaped differently which concentrates the mixture in the same area, this concentrated mixture is what combusts but there are areas of the combustion chamber that has a leaner mixture and sometimes fresh air. once the combustion process is finished and the exhaust valves open the combustion gases and leaner pockets mix giving an overall leaner AFR.

Wiki covers it pretty well:

The engine management system continually chooses among three combustion modes: ultra lean burn, stoichiometric, and full power output. Each mode is characterized by the air-fuel ratio. The stoichiometric air-fuel ratio for gasoline is 14.7:1 by weight (mass), but ultra lean mode can involve ratios as high as 65:1 (or even higher in some engines, for very limited periods). These mixtures are much leaner than in a conventional engine and reduce fuel consumption considerably.

Ultra lean burn or stratified charge mode is used for light-load running conditions, at constant or reducing road speeds, where no acceleration is required. The fuel is not injected at the intake stroke but rather at the latter stages of the compression stroke. The combustion takes place in a cavity on the piston's surface which has a toroidal or an ovoidal shape, and is placed either in the center (for central injector), or displaced to one side of the piston that is closer to the injector. The cavity creates the swirl effect so that the small amount of air-fuel mixture is optimally placed near the spark plug. This stratified charge is surrounded mostly by air and residual gases, which keeps the fuel and the flame away from the cylinder walls. Decreased combustion temperature allows for lowest emissions and heat losses and increases air quantity by reducing dilation, which delivers additional power. This technique enables the use of ultra-lean mixtures that would be impossible with carburetors or conventional fuel injection.

Stoichiometric mode is used for moderate load conditions. Fuel is injected during the intake stroke, creating a homogeneous fuel-air mixture in the cylinder. From the stoichiometric ratio, an optimum burn results in a clean exhaust emission, further cleaned by the catalytic converter.

Full power mode is used for rapid acceleration and heavy loads (as when climbing a hill). The air-fuel mixture is homogeneous and the ratio is slightly richer than stoichiometric, which helps prevent pinging. The fuel is injected during the intake stroke.

Thanks Chris. Light load and cruising I get as far as running a leaner mixture to reduce fuel consumption, but under heavy/WOT type loads, why wouldn't the Lambda targets of .77 - .81 no longer apply. Heat is heat regardless of how the fuel is getting into the cylinder to generate that heat no?

The AFR's measured don't reflect the in cylinder combustion mixture like port injection any more, the mixture now is directed and controlled unlike port injection where it was a spray and pray in comparison.

The concentrated mixture under the spark plug will still be rich on full throttle, but because of the targeted fuel delivery there will still be areas of the chamber that won't be filled with fuel meaning that the AFRs measured at the exhaust will read lean but the combustion mixture hasn't been.

There is also other areas of the tuning that need serious attention with DI compared to PI, LSPI for one.

The configuration that I'm running is a combination of DI and port injection. There's a 65/35 split on the fueling under load (DI / PI respectively). I suppose that complicates things slightly.