Tuning Wiki

Transient fuelling is the correction an engine controller applies while the throttle is moving. On a port-injected petrol engine part of every injection lands on the port wall and the back of the inlet valve and reaches the cylinder later, as it evaporates. That stored film is the main reason the mixture goes lean when the throttle opens and rich when it closes, even when the steady-state fuel map is correct. The remaining error comes from the air side: manifold filling and the lag of the load signal.

Two families of correction exist. Classical acceleration enrichment adds fuel in proportion to how fast the throttle or manifold pressure is changing. Model-based compensation estimates the film itself, usually with the two-parameter X-tau model, and injects whatever is needed to keep the fuel entering the cylinder on target. Both are calibrated from wideband lambda logs, and both depend on the steady-state fuelling being right first. Injector types are covered on petrol fuel injection.

The fuel film and the X-tau model

The model comes from C. F. Aquino's 1981 SAE paper on a 5-litre Ford central-injection engine. With the steady-state metering adjusted to be error-free, the air-fuel ratio excursions that remained during one-second throttle ramps were attributed to wall wetting, manifold air charging and injector phasing. The published abstract describes wall wetting as a linear first-order process with two parameters, a boiling time constant and a percentage of fuel that impacts the wall, and reports a good fit when the impaction percentage was allowed to vary with throttle angle (SAE 810494; abstract only, the full paper is paywalled).

In the form used by engine controllers:

  • X is the fraction of each injection that joins the film. The rest, 1 − X, goes straight into the cylinder.
  • tau is the time constant with which the film evaporates into the air stream.

The MegaSquirt-II implementation writes this per injection event: the film gains X times the injected fuel and loses its mass divided by tau (expressed in engine events), and the injected quantity is solved so that the direct part plus the film's contribution equals the fuel the cylinder needs (MegaSquirt-II X-Tau documentation).

A worked example from Adaptronic's ECU documentation shows the size of the effect. With X at 30%, an engine needing 10 mg of fuel per induction stroke at idle has 3 mg entering the film and 3 mg leaving it each cycle. If demand steps to 40 mg and the ECU simply injects 40 mg, 12 mg goes to the film, only 3 mg comes back, and the cylinder receives 31 mg: a lean mixture. Delivering the full 40 mg on that cycle needs about 53 mg injected (Adaptronic: transient throttle conditions). The extra fuel does not make the mixture richer than target. It fills the film to its new equilibrium. When the throttle closes the process reverses: a film sized for the higher fuel flow keeps evaporating into a smaller air flow, and the mixture goes rich until the film has shrunk.

What changes X and tau

FactorEffectEvidence
Coolant and port temperatureA cold port holds more fuel and releases it more slowly. Adaptronic's guidance is X of about 15% hot and 25% cold on production engines; the MegaSquirt documentation says tau during warm-up can run to many seconds.ECU-maker guidance, not measured data for a named engine
Engine speedtau falls as speed rises. Adaptronic quotes 0.15 to 0.4 s at idle as typical.ECU-maker guidance
Manifold pressureA second-order effect. A University of Michigan and Ford puddle model makes the wall-impacting fraction a function of coolant temperature, manifold pressure and fuel composition.Peer-reviewed model validated against limited test data
Fuel volatility and ethanol contentFilm behaviour depends on the fuel's composition, so compensation tuned for petrol does not transfer to E85.Same model, compared with warm-up tip-in and tip-out tests on E0, E22 and E85
Injector targetingA spray pattern that hits the port divider or wall raises X; Adaptronic reports up to about 40% with poorly matched aftermarket injectors.Vendor claim, unverified

The two ECU-maker documents disagree on manifold pressure. MegaSquirt's states that tau rises with manifold pressure; Adaptronic's states that tau does not change appreciably with it and that pressure slightly affects X. Neither publishes measurements. The points on which the sources agree are that temperature dominates X, that speed dominates tau, and that pressure is the smaller term, so the pressure correction is best left near neutral until logs show a need for it (Ahn, Stefanopoulou and Jankovic, 2009).

Classical acceleration enrichment

Speeduino and MegaSquirt add fuel from the rate of change of throttle position (TPSdot, in % per second) or of manifold pressure (MAPdot, in kPa per second). Speeduino's documentation compares the function to a carburettor's accelerator pump and gives 50 to 1000 %/s as the normal range of TPSdot. Its settings are a trigger threshold, a curve of percentage enrichment against TPSdot, a duration in milliseconds, a taper that reduces enrichment to zero between two engine speeds, and a cold adjustment scaled with coolant temperature (Speeduino: acceleration enrichment). The speed taper and the cold adjustment are empirical stand-ins for the fall of tau with speed and the rise of X on a cold port.

DIYAutoTune's MegaSquirt guidance recommends throttle-based enrichment wherever a usable variable throttle sensor is fitted and keeps the manifold-pressure mode for engines without one. The threshold is set as low as possible without triggering at steady throttle or idle; false triggering at idle usually indicates sensor noise or wiring faults (DIYAutoTune: acceleration enrichment tuning).

Model-based compensation

Production controllers generally use a feed-forward compensator that inverts a model of the film, because lambda feedback is too slow to correct a transient: the Michigan and Ford authors describe this as the usual design for port-injected engines. The MegaSquirt documentation traces the approach to Ford's central-injection calibration. In MegaSquirt-II the X-tau correction has separate acceleration and deceleration tables against engine speed and coolant-temperature corrections for X and tau, and it runs alongside the TPSdot and MAPdot enrichments, which have to be reduced as the model takes over. Bosch ME7 carries its own warm-up and wall-film functions; see ME7 fueling and MAF calibration and ME7 cold start and throttle transition fueling.

Deceleration enleanment and overrun fuel cut

On a closing throttle the correction is a fuel reduction. MegaSquirt's simple form is a "decel fuel amount" percentage: 100% means no reduction, and DIYAutoTune's experience is that manual-gearbox cars tolerate about 90 to 95% while automatics may accept about 70%, with bucking or bogging on deceleration as the sign of too much reduction (tuner guidance, not a specification). The X-tau implementation uses its deceleration tables instead.

Overrun fuel cut is a separate function that stops injection altogether. Speeduino's cut-off applies when engine speed is above a set value, throttle is below a threshold and coolant is above a minimum temperature, after a delay, with hysteresis on engine speed. By the film model, the film depletes during the cut and has to be rebuilt when fuelling resumes, so a brief lean reading on re-entry points to transient compensation before it points to the fuel map.

Cold engines

Cold running is where transient compensation matters most and where a warm calibration fails first. More of each injection stays on the wall and it evaporates slowly, so the same throttle movement needs more added fuel for longer. Simple systems handle this with the cold multiplier on acceleration enrichment; model-based systems apply coolant-temperature corrections to X and tau. Tip-in tests therefore need repeating during warm-up, not only on a hot engine.

Direct injection

A direct-injection engine sprays fuel into the combustion chamber, so there is no port film between the injector and the cylinder. Fuel can still strike the piston or cylinder wall, where it may not burn properly and raises hydrocarbon emissions on wall-guided designs (Wikipedia: Gasoline direct injection). Engines with both port and direct injectors keep a port film whenever the port injectors are in use.

Reading a wideband log

  • Lean spike immediately after the throttle opens: too little added fuel. Raise the enrichment at that TPSdot, or X in a film model.
  • Correct at first, then briefly lean: the enrichment ends too soon. Speeduino's advice is to lengthen the enrichment time in steps of 10 to 20 ms; in a film model, lengthen tau.
  • Rich, then lean: Adaptronic's reading is that X is too high and tau too short.
  • Correct, then rich: tau is too long.
  • Rich spike on lift-off: not enough deceleration reduction.

The lambda trace lags the event that caused it. The delay is the time for exhaust gas to reach the sensor plus the sensor's response, and it shortens as exhaust mass flow rises. Practitioner estimates range from roughly 6 to 10 engine revolutions to about 2.4 s at an 850 rpm idle; these are forum figures for individual installations, not specifications. The delay can be measured by stepping the fuelling 5 to 10% at fixed speed with closed-loop correction off and timing the response. Align the lambda trace with throttle position before deciding which part of a transient was lean. Sensor placement is covered on air-fuel ratio and wideband lambda sensors.

Tuning order

  1. Tune steady-state fuelling first. Transient corrections are applied on top of the base fuel calculation, so an error there appears as a false transient error.
  2. Check the load signal. A noisy throttle sensor triggers enrichment falsely, and a heavily filtered manifold pressure signal produces a lean tip-in that no film setting will cure.
  3. Set the trigger threshold from logs of idle and steady running.
  4. Tune slow throttle movements first, then progressively faster ones, with the engine fully warm.
  5. Repeat during warm-up to set the cold corrections.
  6. Set deceleration reduction and the overrun cut last.
  7. After an injector change, rescale the enrichment and expect the film fraction to change with the new spray pattern.

Not established

  • Aquino's parameter values and equations, which are not in the published abstract.
  • Measured X and tau for any named current production engine.
  • The sign and size of the manifold-pressure effect on tau, on which the two ECU-maker documents disagree.
  • Whether production direct-injection controllers retain a film-compensation term for piston and liner wetting.

Sources

Continue the conversation

If this topic interests you, consider joining the following forums. They host material referenced in this article.