TDIM. The system you are suggesting sounds unlikely. The main injection event would merely displace some of the gas remaining in the pre-chamber, leaving most of the fuel in the pre-chamber. The connection passages to the main chamber are quite narrow to generate high velocity and penetration to the chamber extremities. The result would be an excessively rich mixture in the pre-chamber. Even if it was ignitable, it would be impossible to control the pre-chamber mixture accurately enough under all operating conditions. Add to this the lack of auxiliary-charge precision - using the same injector to deliver full power main-charge and part load auxiliary-charge, a ratio of perhaps 500:1.
None of your stated "challenges" have not be addressed by current established technology, a lot from the Diesel world. I'll attempt to do so in turn.
"The main injection event would merely displace some of the gas remaining in the pre-chamber, leaving most of the fuel in the pre-chamber."
I disagree and can draw from my own experience from my PhD research in a divided-chamber SI engine - I did schlieren imaging and CFD simulations that showed that a substantial amount of fuel injected into this chamber leaves it into the main cylinder. I used only 10 bar injection pressure, imagine what 500 bar would do. In usual DISI engine practise with or without a divided chamber, fuel injection would take place during the intake stroke to give sufficient time for fuel evaporation and formation of a well homogenised mixture in the main cylinder. I believe F1 does the same. My engine was a 2-stroke, so fuel injection occurred after exhaust port closure in a window from 90° to 85° BTDC. When cylinder pressure rises during the compression stroke, mass transfer of the gas mixture in the main cylinder - appreciably lean by design - go back into the divided chamber to equalise the pressures and reduce the equivalence ratio down to the range for most rapid and robust inflammation.
The connection passages to the main chamber are quite narrow to generate high velocity and penetration to the chamber extremities.
Doesn't matter. There is a pressure gradient between the injected fuel and the prevailing pressures in the divided chamber and main cylinder, and mass transfer is driven by this gradient. High velocity, turbulent gas issuing out from the prechamber due to combustion-induced gas expansion is no different than going the other way back into it again due to differential pressure during the compression stroke, only a matter of degree.
The result would be an excessively rich mixture in the pre-chamber.
Not necessarily. If well optimised, as I'm sure F1 engineers expend a lot of effort doing, you can control the lambda in the prechamber quite precisely. In my own research engine, the prechamber is, as expected and as you correctly stated at the outset, composed almost entirely of pure fuel in the moments right after the end of injection - lambda approaches zero, As the compression stroke proceeds and lean mixture returns back into the divided chamber from the main cylinder and remixes turbulently with the super-rich contents in the former, I can target my equivalence ratio to pretty much whatever I want through injection timing and strategy, limited on the lower end only by what the lambda of the mixture in the main cylinder is coming back in from the main cylinder.
Even if it was ignitable, it would be impossible to control the pre-chamber mixture accurately enough under all operating conditions.
What makes it so impossible? Piezo injectors in Diesel engines are already delivering up to 7 separate injection events in a single cycle. Distinct injections can be controlled down to an order of 1°CA apart from the end of one to the start of the next. Granted, Diesels, run at much lower speeds than F1; let take one point at 3000 RPM vs. 15000 RPM -- assuming the same limiting response times, injections can be spaced in the order of just 5°CA apart in F1 engine speeds at max RPM.
Add to this the lack of auxiliary-charge precision - using the same injector to deliver full power main-charge and part load auxiliary-charge, a ratio of perhaps 500:1.
Addressed in my previous paragraphs, except I will add that injection quantities in Diesel piezo injectors have an incredible granularity of control that you may not be aware of. The pilot injection can go down to as little as 0.8 mg/injection, well less than the 5% of the total injected fuel quantity for the cycle (which can be anywhere from 20-50 mg/stroke in a typical Diesel engine at full load depending on per-cylinder output) quoted in the article I posted above. I understand what you're implying, but it is doubtful that the ratio is anywhere near 500:1. At its most simplistic, it's ~95:5 as per the article in a single cycle, but allowing for the range of operating loads, I get it. In Diesel engines, the oft-quoted injection pressure (e.g. 1800, 2000 bar, etc.) represents the peak. It doesn't operate at this pressure over the entire load and speed range. I don't suspect F1 engines run at the full limiting 500 bar from idle/overrun to max power either; maybe it does, but it doesn't matter and doesn't weaken the argument. Nevertheless, going back to Diesel engine practise, it is 100% reliant on fuel injection for load control, going from idle to upwards of 30 bar BMEP without any difficulty, and it has emissions regulations to worry about! The precision and control is only an encumbrance if you're stuck on the first point of the mass transfer interaction going on between the divided chamber and main cylinder.
Edited by TDIMeister, 16 November 2016 - 17:58.