Confusing problem: Intermittent Lean Condition Above 4000rpm

Intermittent Lean Condition at WOT Above 4000rpm - Renault Megane 2 RS

I’m chasing a strange intermittent issue on my Renault Megane 2 RS - running the OEM Sagem S3000 ECU.

I use the car predominantly for track purposes. So it’s usually always driving at WOT - except to/from the track.

What the issue is:

SOMETIMES,

• When the engine is under 4000 RPM, Lambda is under 0.80.
• As RPM increases (from 4000 RPM upward), Lambda progressively starts to lean out.
• Can reach 0.90-1.00 Lambda by the limiter.
• If continuing to drive like this, anti-knock correction increases and continues increasing under sustained load. The AFR rises first, then knock increases afterwards.
• Engine still feels strong.
• No misfires logged.
• No obvious drivability issues.

When this happens, I come into the pits. Then, when I go out again later (+- an hour later) it might, or might not, happen.

If it doesn’t happen, then:

• Regardless of RPM, Lambda operates between 0.72-0.80 throughout the session.
• Knock correction remains very low (under 1 deg) / not at all.
• Car feels no stronger or different than during the lean condition.

The car can be:

• Lean one session.
• Rich the next session.
• Lean again later. Or, rich, rich, lean.
• Or, any combination of this.

Behaviour doesn’t change mid-session (i.e. It doesn’t go rich → lean mid session). The car’s behaviour is established from when the car is started for the session.

What I’ve Checked:

• No misfire faults logged.
• Fuel pump is oversized for the application.
• Fuel tank level does not appear to correlate with the issue.
• Battery voltage appears normal.
• Pump voltage appears stable.
• Fuel pump current only varies slightly (roughly 12.3-12.7A at ~13.3V).
• IAT remains steady without much change.

I’ve tried resetting all of these (granted, at the same time) - full learning reset:

• Idle regulation reset.
• Adaptation reset.
• Torque meter reset.

No clear pattern.

Logging Observations:

Comparing lean and rich sessions:

• Air mass appears very similar.
• Boost pressure appears very similar.
• Injector pulse width appears broadly similar.
• Engine torque estimates appear broadly similar.

What I DON’T Have:

Unfortunately I do not have:

• Fuel pressure logging.
• Commanded fuel mass.
• Lambda target (my wideband AFR sensor for logging is an aftermarket installation).

• The ECU is commanding less fuel, or
• The ECU is commanding normal fuel but the engine is receiving less fuel.
• My Current Thinking

  Fuel delivery issue

• Fuel pressure variation.
• Regulator issue.
• Injector flow issue.
• Something the ECU cannot see.

ECU strategy/calculation issue

• Different fueling state.
• Different compensation state.
• Sensor plausibility issue.
• Something causing the ECU to calculate fueling differently despite similar logged operating conditions.