What is low-speed pre-ignition (LSPI), and why did an engine-oil detergent get the blame?
Last verified: 2026-10-07
Suppose a boosted petrol engine arrives with a damaged spark-plug electrode and an eroded piston crown, and the owner says it only ever misbehaved when pulling hard from low revs. The suspect list forms quickly: bad fuel, wrong timing, thin oil. Low-speed pre-ignition, LSPI, is combustion that starts before the spark fires, in an engine working hard at low speed. The oil came under suspicion because engine tests kept finding more events with calcium-based detergents. Why that happens is still unproven.
How does a flame start before the spark?
A 2023 peer-reviewed review of the research describes four stages: "LSPI precursor initiation, establishment and propagation of a pre-ignited flame, autoignition of end-gases and development to a detonation".
Read that as a chain. Something ignites the charge too early. A flame spreads from it. The unburned gas ahead of the flame then autoignites. Sometimes that ends in a detonation.
The chain does not always run to the end. The review states: "LSPI is a prerequisite for super-knock to occur, but not all LSPI cycles develop into super-knock." One pressure trace it reproduces, an example of LSPI followed by super-knock, reaches "a maximum pressure of almost 200 bar". That is a single illustrated cycle, not a general figure.
The damage the review lists matches the scene above: a broken spark-plug electrode, a melted exhaust valve, an eroded piston crown and a fractured piston ring land.
The review places LSPI mainly in a region of 15 to 24 bar brake mean effective pressure and 1,500 to 2,000 rpm, and says that region is attainable in commercial direct-injection engines. It puts the events at "every 10,000-100,000 cycles under high-load and low-speed conditions". That is how rare a single event is under those conditions. It is not a failure rate for vehicles on the road.
Two hypotheses for the first stage coexist in the review:
- Oil-fuel droplets. Fuel spray reaches the cylinder liner and dilutes the oil film. The oil-fuel mixture collects in the top-ring crevice and is released into the chamber as droplets.
- Hot deposits or particles. Deposits detached by an earlier event can glow and ignite the charge on a following compression stroke.
The review puts it in one line: "Oil-fuel droplets that have been ejected from the top ring crevice and deposits can be the precursors". It adds that one event can release more precursors and so set off following cycles, which is why events tend to arrive in bursts.
Why did a cleaning additive end up on the suspect list?
Because in engine tests, calcium kept turning up on the wrong side of the results.
The review says "The harmful impact of calcium in detergents was already addressed in the early 1970s", in a study of deposit-initiated pre-ignition, and that later studies confirmed it. It reports a "steep increase in LSPI for a calcium content between 1000 and 2500 ppm". Treat that as a research observation, not a formulation target.
Here is the part that surprises most readers. The detergent's chemical family showed no significant differences in those studies: sulfonate, phenate and salicylate behaved alike, and total base number showed no significant impact. What the studies flagged was the metal, not the family it came in.
Meanwhile the usual workshop suspects barely registered. Research octane number "has not shown a notable impact on the LSPI frequency". Oil volatility shows "no significant effect", and a "reduction in oil viscosity does not cause a clear impact".
Does "calcium causes LSPI" survive a close reading?
Only in part. The flat version, that calcium is the cause, is how the finding tends to get repeated. The review supports something narrower.
In engine tests calcium is a promoter of LSPI, not a proven sole or sufficient cause.
Three limits sit inside the review itself.
First, the effect depends on the operating point. In one study, at a moderate load, varying calcium by more than an order of magnitude left LSPI "virtually the same". The same study found calcium promoted LSPI at a higher load. The review's authors call it the only exception they know of.
Second, nobody has shown why. The review says the reason why substituting magnesium for calcium reduces LSPI "still remains unknown". Several chemical theories exist. None is proven.
Third, the field itself is unfinished. The review says "many questions about causing mechanisms and the affecting parameters remain unanswered", and calls these "rare phenomena, difficult to observe optically". A claim that the mechanism is fully understood does not survive that sentence.
What does the evidence say about each suspect?
The table maps the suspects into classes. It reports what engine-test studies found, as summarised in the 2023 review, and how firmly.
| Suspect | What engine-test studies reported | How settled |
|---|---|---|
| Calcium detergent | Promotes LSPI; steep increase reported between 1000 and 2500 ppm calcium; in one study at moderate load a more-than-tenfold change in calcium left LSPI virtually the same, while at higher load calcium promoted it | Promoter reported across studies; reason unknown; not a proven sole cause |
| Detergent type (sulfonate, phenate, salicylate) and total base number | No significant differences between the three types; no significant impact of total base number | Reported as no significant effect |
| Magnesium detergent | Partial or complete substitution of calcium for magnesium widely reported to suppress LSPI; magnesium appears neutral; benefit appeared to hold through ageing | Widely reported; why the substitution works is unknown |
| Sodium detergent | Studies indicate sodium promotes LSPI, likely even more than calcium | Indicated by studies; the review's own word for the comparison is "likely" |
| ZDDP (zinc dithiophosphate) and molybdenum dithiocarbamate | Identified as quenchers of LSPI; the mitigating effect of molybdenum deteriorated with ageing | Identified; no ranking between them and no dose claim here |
| Base oil | In one study, with the same additive package, Group III and IV base oils showed lower LSPI frequency than Group I and II; the review itself says "Surprisingly" | One study; not yet explained by base-oil reactivity |
| Viscosity and volatility | Reduced viscosity: no clear impact; volatility: no significant effect | Reported as no clear or significant effect |
| Aged oil | Ageing can show a promoting trend; results vary by study | Not universal; one study found no ageing effect in short tests |
Scope: findings from engine-test studies summarised in a 2023 peer-reviewed review. The ppm range is a research observation, not a recommendation. Nothing in this table is a field failure rate.
Ageing is the row with a time dimension. In one study the same molybdenum-containing oil was aged for 30,000 km in two different cars: one tripled the event frequency against fresh oil, and the other did not degrade. In another, an LSPI-resistant formulation aged for 10,000 miles showed "twice as many" events as a fresh one. The review notes that little research combines formulation and ageing, which leaves the question of how an oil behaves late in its life less settled than how it behaves new.
How much of the problem sits outside the oil?
A good part of it. The oil is one variable among several, and the review gives fuel and engine hardware sections of their own.
On fuel, octane is the suspect that does least of what is expected of it. Some studies indicate a higher average knock intensity after an event with low-octane fuel, so octane bears on how violent an event is. On how often events occur, the review reports no notable impact, apart from a slight indirect reduction it links to the spark timing a higher-octane fuel allows. Fuel volatility and heavier aromatics matter more. In one study, a detergent additive package in the petrol, whose solvent made up less than 0.4% of the blend, more than doubled LSPI events against the unadditised fuel. That is a fuel finding, not an oil one.
The engine has its own say. Injection strategy and spray targeting away from the liner reduce LSPI. Higher boost increased it: in one study, raising manifold air pressure by 0.2 bar gave "twice as many LSPI cycles". Coolant temperature, piston-ring tension and crevice geometry all matter. Engines differ individually, and in one study an engine's LSPI activity fell quickly during early testing, even after break-in, then kept declining more slowly.
If damage is found and the oil is the only thing that has been checked, the checking has covered one variable of several.
Which claim does a licence pass actually support?
Less than the label suggests. The test is ASTM D8291, titled "Standard Test Method for Evaluation of Performance of Automotive Engine Oils in the Mitigation of Low-Speed, Preignition in the Sequence IX Gasoline Turbocharged Direct-Injection, Spark-Ignition Engine". A Southwest Research Institute data sheet names the test fixture as a "Ford 2.0L Ecoboost inline four-cylinder engine as found in the 2012 Ford Explorer".
The run is long: "The test consists of 4 iterations. Each iteration is 175,000 ignition cycles from each cylinder with the first 170,000 valid cycles evaluated". One specified test fuel is used.
An event is counted statistically. In-cylinder pressure sensors record each cycle's peak pressure and the crank angle at which 2% of the fuel mass has burned. A cycle that is an outlier on both, very high pressure and very early burn, is deemed an LSPI event.
In API 1509, the ILSAC GF-7A table sets a maximum of 5 for the average number of events across four iterations and a maximum of 8 events in any one iteration. A separate row, for aged-oil LSPI prevention under ASTM D8291 Appendix X2, carries the same two limits. The GF-7B and API SQ tables list both tests as well. The API SP table lists Sequence IX but no aged-oil row. The comparison is covered in API SP vs SQ engine oil explained, and the SP category itself in what API SP is.
Now the documented silences. The licence document gives no event rate for vehicles in service. A pass is a result on one engine and one fuel, not a guarantee about any other engine, fuel or calibration. Nothing in these sources says an older or naturally aspirated engine "needs" an LSPI-tested oil. And the review gives no single cause.
A claim of meeting a specification is also not an approval from an engine maker. That distinction is covered in meets specification vs OEM-approved.
What should a specifier require in writing from a supplier?
The following is practical guidance, not a quotation from any standard.
When an oil is offered with an LSPI claim, three lines make the claim checkable:
- Which category and which table. Whether the claim rests on API SP, API SQ, ILSAC GF-7A or GF-7B, or an older add-on class, and the licence it is made under.
- Whether the aged-oil test is included. The licence document treats Sequence IX and Sequence IX Aged (Appendix X2) as separate rows. A document that mentions only one has answered half the question.
- The test report, not the claim. The event counts per iteration, with the test method named, so the figures can be read against the limits of 5 and 8.
A specifier can also ask whether the supplier will state which detergent metals the formulation uses. The review shows why that detail would be informative: calcium, magnesium and sodium sit at different points on the evidence above. Whether to disclose it is the supplier's decision, and the answer, or its absence, shows how far the claim can be checked.
For sourcing, petrol engine oils listed by suppliers on the platform can be put to an RFQ with these three items named as requirements. Altonex Global is a venue; the documents come from the supplier.