Tuning Wiki

Pre-ignition is combustion that starts before the spark. It is a different fault from knock, which is auto-ignition of the unburnt end gas after the spark has started a normal flame, although severe pre-ignition usually ends in very heavy knock. Two forms matter in practice. Surface or hot-spot pre-ignition comes from an overheated component such as a spark plug electrode and tends to run away. Low-speed pre-ignition (LSPI, also called stochastic pre-ignition) is a rare, random event in downsized turbocharged direct-injection petrol engines at low speed and high load, and is the starting point of the destructive event known as super-knock.

The research evidence is consistent on the main points: LSPI is confined to a region of roughly 1500–2000 rpm and 15–24 bar brake mean effective pressure, it is initiated mainly by oil-and-fuel droplets or detached deposits, it is not prevented by ignition retard or by higher octane fuel, and its frequency depends strongly on engine oil additive chemistry. Raising low-speed boost and torque moves an engine further into this region with no protection from knock control.

Pre-ignition, knock and super-knock

A 2023 open-access review of the field describes four stages: a precursor forms, it ignites the surrounding mixture before the spark, the early flame compresses the end gas until it auto-ignites, and that auto-ignition may develop into a detonation. Conventional knock is low in intensity and is removed by retarding the spark. Super-knock is not a sign of over-advanced timing, and not every pre-ignition cycle produces it: the outcome ranges from no knock to slight knock to super-knock, depending on when and where the early flame starts.

The pressures involved are of a different order from ordinary knock. The review's example cycle approaches 200 bar peak pressure and it cites measured pressure-wave amplitudes above 160 bar. Whereas knock causes gradual wear, a single super-knock cycle can cause abrupt failure. Reported damage includes fractured piston ring lands, broken spark plug electrodes, an exhaust valve melted after an electrode fragment lodged in its seat, erosion of the piston crown and cylinder head, and damaged ring coatings; oil-additive industry sources also report broken rings and bent connecting rods.

Surface pre-ignition

In surface ignition the flame starts in the boundary layer of a hot component. The earliest reported cases involved overheated spark plug electrodes, and it can occur at high engine speed in naturally aspirated engines when the plug heat range is too hot for the duty. Each pre-ignited cycle heats the hot spot further, so the fault can become continuous within a few cycles. Continuous pre-ignition at an exhaust valve has also been observed, and glowing combustion-chamber deposits can act in the same way. This form is managed with the correct plug heat range, sound valve seating and deposit control; see spark plugs for performance, burnt valves and carbon deposits and cleaning.

The review judges a hot surface an unlikely cause of pre-ignition in modern production turbocharged engines, because optical studies show the ignition location varying from event to event, and because one study found that plug heat range and electrode design did not change super-knock frequency. A colder plug is therefore not a remedy for LSPI, although it remains necessary against surface ignition on a highly boosted engine.

Low-speed pre-ignition

Operating region and frequency

Most published tests are run at 1500–2000 rpm at or near maximum load. The review gives an occurrence of one event per 10,000 to 100,000 cycles in that region. Events are not independent: one study found up to 75% of pre-ignitions in groups of two to seven cycles, often alternating with normal cycles. Transient operation appears more prone than steady running, and events have been reported at 3000–4000 rpm in on-road tests. Boost is the main driver: in one study a 10% rise in manifold pressure doubled the event count.

Mechanisms

Two precursors are supported by optical and parametric evidence, and both probably operate.

  • Oil-and-fuel droplets. Injected fuel that reaches the cylinder liner dilutes the oil film; the mixture collects in the piston top-land crevice and is thrown into the chamber, where it ignites more readily than the petrol-air charge. Evidence for this route is that injection strategy, fuel volatility and oil composition all change LSPI frequency.
  • Deposits and glowing particles. Solid particles, often shaken loose by the pressure oscillation of one event, survive the gas exchange, heat up over the next cycle and ignite the charge after that. This explains the clustered, alternating pattern.

The chemical route by which calcium promotes ignition has not been established.

Engine oil chemistry

Oil formulation has the largest documented effect of any variable outside the engine's own hardware and calibration.

AdditiveEffect on LSPI frequencyEvidence
Calcium detergentIncreases it, steeply between about 1000 and 2500 ppm calcium; oils not formulated for LSPI typically exceed 2000 ppmMeasured in many independent engine studies; detergent type (sulphonate, phenate, salicylate) makes no significant difference
Magnesium detergentNeutral; replacing part or all of the calcium with magnesium lowers the rateMeasured; an additive-supplier test found magnesium did not quench calcium-induced events, so the benefit comes from removing calcium
SodiumIncreases it, probably more than calciumMeasured in three studies
Zinc dialkyldithiophosphate (ZDDP)Reduces it; one supplier test recorded over 90% fewer events when phosphorus rose from 0.05% to 0.19%Measured; usable levels are limited by phosphorus limits that protect the catalyst
Molybdenum (MoDTC)Reduces it; one Southwest Research Institute study found the effect only at higher calcium levelsMeasured; the benefit is reported to fade as the oil ages, whereas the low-calcium benefit persists
Viscosity grade, volatilityNo clear effectStudies disagree, mostly without statistical significance

These findings produced an engine test and a series of specifications. The Sequence IX test (ASTM D8291) runs a 2012 Ford 2.0-litre turbocharged direct-injection engine at 1750 rpm and 269 N·m for four iterations of 175,000 cycles and counts pre-ignition events; a test laboratory lists the API SN Plus limit as fewer than five. API SN Plus was a supplementary category licensed from 1 May 2018 because ILSAC GF-6 was delayed. API SP and ILSAC GF-6A and GF-6B, introduced in May 2020, include LSPI protection, as do ACEA A7/B7 and C6. General Motors' dexos1 Gen 2 added an LSPI requirement, and dexos1 Gen 3, announced by GM in November 2021, tightened the limits of its stochastic pre-ignition test. An API or ILSAC category does not replace the vehicle manufacturer's own oil approval.

Fuel

Research octane number shows no significant correlation with LSPI frequency, but it affects what follows: in one study only 2% of pre-ignition cycles went on to auto-ignite on a 94 anti-knock index fuel against 56% on an 88 index fuel. Fuels with heavy, low-volatility fractions, notably C9 and heavier aromatics, increase LSPI because they wet the liner and stay in the crevice. Reported effects of ethanol vary between studies. Higher octane fuel therefore reduces the severity of an event but does not make a low-speed, high-boost calibration safe; see fuel octane.

Manufacturer countermeasures

  • Enrichment. Rich mixtures (λ 0.8–0.95) reduce LSPI and slightly lean mixtures increase it. Enriching the cycle immediately after a detected event almost eliminates the follow-on events, and the review notes that some ECUs respond to LSPI in this way. Continuous enrichment costs fuel and catalyst efficiency; see component protection and full-load lambda enrichment.
  • Injection strategy. Split injection and spray targeting that avoids the liner reduce events, by 50–90% in the studies reviewed. Late single injection during the intake stroke generally makes LSPI worse, and higher rail pressure has not been shown to help.
  • Gas exchange. Scavenging removes particles and hot residual gas. High exhaust back pressure sharply increases LSPI. Results for increased valve overlap are mixed.
  • Temperatures and hardware. Higher coolant temperature, around 95 °C, reduces liner wetting and events. Higher ring tension, piston crevice geometry, better oil separation in crankcase ventilation and cooled exhaust gas recirculation have all reduced LSPI in tests.
  • Detection. A knock sensor responds only to the knock that follows. Research engines detect the early flame from cylinder pressure or ion current, and a community cross-reference for Simos 18 lists a separate, earlier knock-sensor window for pre-ignition; see Simos 18 low-speed pre-ignition and knock.

Ignition retard is absent from this list. A 5° retard produced no significant change in LSPI in one study, which is expected for an event that starts before the spark.

Implications for tuning

  • The standard low-speed boost and torque limits of a downsized direct-injection engine sit beside the LSPI region. An additive supplier's test engine had to have its factory LSPI and knock-suppression strategies disabled to measure oil effects, which indicates that production calibrations contain such strategies. Raising low-speed boost, or removing a low-speed torque limit, increases exposure, and the research link between boost and event rate is steep.
  • A log with no knock retard does not show freedom from LSPI. At one event per tens of thousands of cycles, a few dynamometer pulls cannot sample the risk either.
  • Leaning the low-speed full-load mixture towards λ 1 or beyond, lowering coolant temperature, and hardware that raises exhaust back pressure at low speed all act in the direction the research associates with more events.
  • Oil that meets a current LSPI-tested specification alongside the manufacturer's approval, changed on schedule, is the best-evidenced owner-side measure. Mitigation from molybdenum is reported to fade with oil age.
  • Sustained full load at low engine speed, as when accelerating in a high gear, is the condition that matches the research region. See cylinder pressure and BMEP.

Not established

  • How calcium detergents promote ignition, and why magnesium does not.
  • LSPI rates, thresholds or dedicated protection maps for specific production engines and ECU software; published rates come from research engines.
  • The date and test detail of the dexos1 Gen 2 LSPI requirement, for which no General Motors document has been identified, and the content of the Gen 3 transient test.

Sources