Measurement methodology
A fuel trial is easy to run and hard to run well. This page sets out how a NanoEFX field trial is measured, why the effect is measured over months rather than days, what an accredited laboratory measurement adds — and the five things that separate a credible trial from a persuasive one.
How a trial on your own fleet is run
This is the protocol we ask operators to follow. It is designed for real vehicles in real service, not for laboratory conditions.
Record the vehicle first
Type of vehicle, fuel, engine, age and odometer reading. The method is designed around vehicles of roughly 6–10 years and around 100,000 km or more, and around vehicles that have not been treated with other fuel additives.
Establish a baseline before anything is applied
Record fuel and distance for two weeks to one month before the first application, using the full-tank method: fill to full, note the odometer, run the vehicle, and record consumption at every refuelling. Keep the driver, route type and typical speed consistent.
Measure emissions before, if emissions matter to you
Take the vehicle to a workshop or test station with an exhaust gas analyser and record the reading before the first application, so there is a comparable measurement afterwards.
Apply to the air filter
Remove and clean the filter, shake the bottle, spray both sides, let it absorb for several minutes and refit. Then run the engine up through the rev range to clear loosened deposits. No part is modified and nothing is added to fuel or oil.
Apply three times and keep records for at least three months
The protocol calls for three applications, with records kept for a minimum of three months. These are trial intervals, deliberately closer together than reapplication in normal service, so that the trial covers the build-up across all three applications rather than only the first: for diesel vehicles roughly every 6,000–10,000 km, for petrol roughly every 5,000–7,000 km. Once a fleet is in routine service the interval is wider and set by duty cycle — for cars, trucks and buses typically every ~5,000–12,000 km, as stated on the product page.
Assess over sufficient time — the signal can appear sooner than you'd expect
Emissions tend to respond relatively quickly. Fuel-efficiency changes can appear within weeks of the first application: our own trials have recorded a first measurable signal in as little as 18 days. In some documented trials the measured effect increased further across subsequent applications — at Toyama Transport the first reading was 7.7%, rising to 19.0% across three applications. A credible fleet result should still be assessed over sufficient distance, refuelling cycles and operating time to distinguish a treatment signal from normal variation.
This field method is explicitly not a controlled-environment protocol. For controlled measurement it points to the method used by the Japan Automobile Transport Technology Association — which is how the accredited emissions figures below were produced.
What a laboratory adds
One measurement in our records was made by an accredited testing organisation under controlled conditions. It is small, it is old, and it is the only one of its kind we have — so we treat it as the reference point for emissions rather than as a headline.
| Measured | Before | After | Change |
|---|---|---|---|
| CO (g/km) | 0.483 | 0.336 | −30.4% |
| HC (g/km) | 0.090 | 0.069 | −23.3% |
| NOx (g/km) | 0.053 | 0.045 | −15.1% |
| CO₂ (g/km) | 253.8 | 252.7 | −0.4% |
- Why no fuel figure
- Only 88 km separated the before and after measurements. That is ample for an exhaust gas reading and far too short for a meaningful fuel-consumption figure, so we do not quote one from this test.
- The NOx ceiling
- −15.1% is the highest NOx reduction we can evidence from an accredited measurement anywhere in our records. If you see a larger NOx claim attached to this product, ask for the measurement behind it.
- Also on record
- A diesel van that failed an accredited emissions test in the Philippines passed after treatment, and passed again a month later. A pass or fail is a coarser measure than a gas analysis — but it is a third party’s verdict, not ours.
What a controlled test bench shows about run-in time
A controlled engine test bench recorded a pattern some of our field trials also show: at the very first reading the treated engine measured worse, before moving ahead as running time accumulated. This is one documented pattern, not proof of a single universal onset time — other documented trials have shown a positive signal within weeks. Either way, a single early reading can miss the real result; that is why we measure over a defined period rather than at one moment.
Five questions to ask of any fuel-saving claim
Including ours. If a supplier cannot answer these about their own data, the number they are quoting is decoration.
Is there a control group?
Untreated vehicles running the same routes over the same period are the only way to separate the product from the season, the weather and the traffic. One trial in our records has one — the Sanden Kotsu bus trial, at 7.2%. That is one data point, so it does not prove that control groups always yield lower figures; what it does show is that the trials without one scatter across the whole range, which is the part you cannot plan a budget around.
What is the baseline, and when was it?
A March–May baseline compared with a June–August trial period carries a seasonal difference before the product does anything. Comparing the same calendar months a year apart, as in the Alfresa and boiler trials, removes most of that.
Did the drivers know?
Drivers who know a fuel-saving product is being trialled tend to drive differently. In the Waste Services trial they were deliberately not informed — which is one of the reasons we rate that dataset highly despite its other limitations.
Is the quoted figure the whole trial or the best part of it?
A trial with three measurement points has three numbers and one aggregate. Quoting the highest is legitimate arithmetic and misleading evidence. Every trial page on this site shows all measurement points and quotes the aggregate.
Are the failures published too?
A dataset in which nothing ever gets worse has been filtered. Our field data includes vehicles that got worse, months with no change, and a generator trial with a small, inconclusive result — because that is what real measurement looks like.
What we do not claim
Not a fixed percentage
Results depend on engine, duty cycle, baseline and season. We publish ranges and individual trials, and we do not promise a number in advance of your own measurement.
Not a certification
NanoEFX is not an emissions certification and does not make a vehicle compliant with any standard. It complements your compliance measures; it does not replace them.
Not guaranteed on NOx
NOx response is engine- and control-system dependent. Our accredited measurement shows −15.1% on one petrol vehicle. We do not extrapolate that to diesel fleets.
Not a fixed onset time
Emissions may respond quickly, and some documented fuel-efficiency signals have appeared within weeks — but we do not promise an onset time. A credible fleet result still needs sufficient distance, refuelling cycles and operating time to separate a treatment signal from normal variation, so a very short trial risks the wrong conclusion in either direction.