Where fuel-saving products actually intervene
Everything sold to a fleet as a fuel saver acts at one of three points: in the fuel, in the lubricant, or in the air the engine breathes. The point of intervention decides what the product can plausibly do, how quickly, and — most usefully for a buyer — what kind of evidence should exist for it.
This page compares mechanisms, suppliers and evidence standards — not performance. We have measured none of the products named here, and we hold ourselves to the rule that no figure appears on this site without a source we can produce. That is why exactly one third-party number appears below: the one trial in this field that an independent classification society supervised.
What each category is trying to change
| Category | Acts on | Mechanism it relies on | Evidence to demand |
|---|---|---|---|
| Fuel additives | The fuel, at every fill | Detergency — keeping injectors and combustion surfaces free of deposits, so the engine returns towards its designed condition. | Deposit measurements before and after, not only a consumption figure. A dirty engine that is cleaned improves; a clean one has less to gain, so ask what state the test engines were in. |
| Oil additives | The lubricant | Friction reduction between moving parts. The fuel effect is indirect: less energy lost to friction means less fuel for the same work. | Friction or wear measurement alongside the fuel figure, and confirmation that the additive is compatible with the oil specification your engines are approved for. |
| Intake-air treatment | The air path, at the filter | Conditioning the air entering the cylinder so combustion is more complete — the fuel and the lubricant are untouched. | Fuel measured over months rather than days, and ideally an inspection of what changed inside the engine. Because nothing is added to the fuel, a durable effect has to show up in operating data, not in a bench snapshot. |
| Hardware retrofits | The engine or driveline | Physical change — aerodynamics, tyres, hybridisation, aftertreatment. The effect is engineered rather than chemical. | Type approval and manufacturer sign-off. This is the only category where a documented specification usually exists, and it is also the only one that carries capital cost and downtime. |
These are not mutually exclusive. A fleet can run a fuel additive, a specified low-friction oil and an intake-air treatment at the same time — which is also why any single trial has to hold the others constant if the result is to mean anything.
The suppliers a fleet is realistically offered
Naming who operates at each point of intervention is more useful to a buyer than ranking them. Below is who a fleet is likely to encounter, ordered by where the product acts. It is not a ranking and it carries no performance figures: we have measured none of these products, and the categorisation follows each manufacturer’s own public product description.
| Acts on | Companies with a public product range | What that means for how you buy it |
|---|---|---|
| The fuel | SulNOx · Fuel Ox · EnerBurn · Wilhelmsen · Innospec · Infineum · BASF · Afton Chemical · Lubrizol · Chevron Oronite · LIQUI MOLY | Dosed continuously, at every fill or every bunkering. The cost recurs for as long as you use it, and stopping generally ends the effect. |
| The lubricant | Hot Shot’s Secret · Archoil · XADO · Bardahl · Xenum | Added at each oil change, so the application interval is set by your service schedule rather than by fuelling. |
| The intake air | NanoEFX | Applied to the air filter. Nothing recurring in the tank; the interval follows filter servicing. |
| Hardware in the fuel line | Rentar · Fitch | Installed once, with a capital cost and downtime instead of a recurring dose. The only category where a replacement interval, not a treat rate, is the relevant number. |
Two things the table does not show
- Several of the largest names do not sell to you. BASF, Afton Chemical, Lubrizol and Chevron Oronite supply refineries and fuel marketers rather than fleets. Their technology may already be in the fuel you buy, without you ever having chosen it — which is worth knowing before you attribute a change to something you did.
- Company size is not evidence. A global chemical group and a small specialist can both be right or both be unproven. The next section is about the only thing that separates them from a buyer’s point of view.
Compiled from the manufacturers’ own public product descriptions, retrieved 16 August 2026. The list covers companies with a publicly documented product range in this field; it is not exhaustive, and inclusion is not a comparison.
What a properly evidenced figure looks like here
One published result in this field sets a standard the rest of us can be measured against — and it is not ours.
Wilhelmsen tested its FuelPower Catalyst, a marine fuel treatment, in a controlled shop trial on a MAN 6S35ME-B9 two-stroke engine. The China Classification Society supervised the work from test preparation through execution and data validation; the fuel was analysed in a laboratory accredited by the China National Accreditation Service for Conformity Assessment. The result was a reduction in specific fuel oil consumption of 1.51% to 2.17% against the untreated baseline fuel, or 1.31% weighted, calculated to CB/T 3254.2-2013 and GB/T 6072.1-2008.
We are citing a competitor’s result approvingly, which is unusual in supplier material. We do it because it is the clearest available answer to the question a buyer actually needs answered — what does a properly evidenced figure in this category look like — and because naming that standard obliges us to say where our own numbers fall against it.
Three things about it are worth more than the number itself. An independent body supervised the measurement rather than receiving the results afterwards. The unit is unambiguous: specific fuel oil consumption is fuel per unit of work, so it cannot be confused with an efficiency figure, which would read larger for the same effect. And the result is small — which is what results tend to be when someone independent is watching.
Measured against that standard, most of our own numbers do not qualify
- Of the seventeen trial and operator values we hold, sixteen compare the same vehicles against their own earlier records — the same methodological class as most supplier-reported figures in this market, ours included.
- Exactly one ran treated vehicles against an untreated control group over the same period: Sanden Kotsu, two buses, −7.2% fuel consumption. That is the only figure of ours that was produced in the same way as the one above.
- The pattern is consistent and it does not flatter us: the more independent the supervision, the smaller the number — across the market, and inside our own dataset.
| How the comparison was made | Trials | Consumption reduction |
|---|---|---|
| Against an untreated control group, same period | 1 | −7.2% |
| Against the same period a year earlier | 4 | median −8.1% |
| Against the vehicles’ own earlier records | 12 | median −11.6% |
| All seventeen values together | 17 | median −10.7% |
Most of the trials in this market — ours among them — are of the third kind. That is not fraud; it is what is practical when a fleet cannot spare vehicles to run untreated. But it means the headline figure of almost any supplier in this field, including ours, should be read as the largest number the method can produce, not the effect you should budget for.
How we measure, and what each method can and cannot show →
Wilhelmsen figures as published by Wilhelmsen Ships Service, retrieved 16 August 2026; we have not independently verified the underlying trial and have no relationship with the company. Our own figures are recalculated from the raw records behind our evidence pages.
The mechanism tells you what to check
Each point of intervention has a characteristic weakness. Knowing it is more useful than comparing headline percentages, which are rarely measured the same way twice.
Anything acting on deposits
Improvement depends on how dirty the engine was to begin with. A striking result on a neglected fleet says little about a well-maintained one. Ask for the service history of the test vehicles — and treat a trial that cannot supply it as indicative only.
Anything acting on friction
The effect is real but bounded: friction is a modest share of the energy an engine loses, and most of it is already engineered against. Large fuel claims from a friction mechanism deserve more scepticism than small ones.
Anything acting on combustion air
The honest weakness of our own category: the mechanism is harder to observe directly than a cleaned injector or a measured friction coefficient. That is why we put weight on a third-party engine teardown after 1,875 operating hours, and why our field-testing method asks for three to six months of records rather than a week.
Anything at all
If a supplier cannot name a duty cycle where their product did nothing, they have not tested it widely enough — or they are not telling you. On telecom generators running at near-zero load, ours did nothing measurable, and that trial has its own page.
What follows from being an air-path product
NanoEFX is applied to the air filter. Nothing enters the fuel and nothing enters the oil, which has three consequences a buyer should weigh.
It does not interact with fuel or lubricant specifications. There is no additive in the tank to be compatible with a fuel standard, and no change to the oil your engine is approved for.
The effect is not immediate. On an engine test bench the treated engine reads worse at the first measurement and crosses below the untreated reference only after five to ten minutes. In fleet trials the pattern repeats over months: one truck we measured improved by 7.7%, then 9.7%, then 19.0% across three applications. A trial that measures once, early, will see the smallest number the product produces.
It is harder to evidence than a chemical mechanism. We compensate with volume and with publishing failures — twelve documented trials, 386 individual measurements, and the weak, inconclusive results in the same detail as the successes. That is the trade we make for a mechanism you cannot see in a beaker.