Fuel is the largest line in the variable cost of a Thai truck, and it is one a fleet owner can still move without buying anything. It is also the line where the most confident numbers get quoted. "Eco-driving saves 17%" is true in one Thai trial and about three times too high in another.
This article is about what the Thai evidence actually shows, what to measure before you spend a baht, and how to test a saving on your own trucks so you know whether it is real.
The short answer
- Fuel and lubricants are about 49% of the variable cost of a Thai road freight operator, so a few percent on that line matters.
- Two Thai driver training trials, by the same research group, found 5.6% and 17.6% less fuel per kilometre. Plan on the low one until your own test says otherwise.
- A truck's fuel number is four things mixed together: the driver, the vehicle's condition, the load and the route. A league table of drivers that ignores load and route punishes whoever got the hilly lane.
- Measure per lane and per load state first. You cannot tell whether anything worked without a baseline of that kind.
Why the number needs unpicking first
A kilometres-per-litre figure for one truck in one week is the result of all of these at once.
| What moves it | Evidence | Size |
|---|---|---|
| Load | 400 Thai fuel tankers: 3.11 km per litre loaded, 4.31 empty | A loaded truck burns 39% more fuel per kilometre than an empty one (0.322 against 0.232 litres) |
| Route | The same tankers loaded, by region: 2.95 km per litre in the north to 3.58 in the west | The best region is 21% better on km per litre than the worst, before any driver differs |
| Driver | Trips grouped by harsh accelerations and brakings per 100 km: 4.45 down to 3.05 km per litre | The smoothest group burns about 32% less per kilometre than the harshest |
| Vehicle | Handbook list: slipping clutch, blocked air filter, fuel leak, soft tyres, binding brakes | No Thai figure found |
Three things to take from that.
- A fall in km per litre is not automatically a driver problem. If the truck ran loaded on a hilly lane this week and empty on a flat one last week, the figure moved with no change in anyone's behaviour.
- The driver effect is large but is not all training. The harsh-event groups are an association in observed trips. Smooth drivers may also get flatter routes and better trucks. Nobody assigned drivers to routes at random.
- Empty running looks good on this measure and is bad for you. An empty truck gets 4.31 km per litre, which makes its fuel per kilometre look fine. What it is not doing is earning. Fuel per kilometre cannot see that, so it needs its own measure.
Step 1: build a baseline you can trust
Before any training or equipment, get these four things from your own records for at least four weeks.
- Litres in, kilometres out, per truck, per week. Use full-tank to full-tank fills so the litres are real. A card statement or a gauge reading is not a fill.
- Split every trip by loaded and empty. Two numbers per truck, not one.
- Tag the lane. The same origin and destination, so a truck on the hilly lane is compared with itself.
- Separate price from consumption. The handbook's weekly fuel report does this by asking four questions when spend rises: did the price move, did the distance move, did litres per kilometre move, or was it recorded wrongly? Only the third is the vehicle or the driver.
If you already run a tracker, what a tracker can and cannot show you says which of these it gives you and which it does not. Fuel needs a fill record or an engine data feed alongside the kilometres.
The arithmetic of what a saving is worth
Take the pre-training average from the 2024 trial, 0.2401 litres per kilometre, on a truck that runs 100,000 km a year, at an assumed 40 baht a litre. Replace all three numbers with yours.
- Fuel a year: 100,000 x 0.2401 = 24,010 litres, or 960,400 baht
- At the 5.6% saving: 1,337 litres, about 53,480 baht a truck a year
- At 10%: 2,401 litres, 96,040 baht
- At the 17.6% saving: 4,219 litres, 168,760 baht
The span from lowest to highest is more than three times. That is the reason to run a test before putting any of it in a budget.
Step 2: know what the two Thai trials do and do not prove
| 2022 trial | 2024 trial | |
|---|---|---|
| Drivers | 100 | 20 |
| How fuel was measured | One 10 km run on a test track, before and after | A driver logbook of fuel use, followed one month later |
| Reported result | 3.56 to 3.77 km per litre, called 5.89% | 0.2401 to 0.1979 litres per km, 17.57% |
| Fuel saved per kilometre | 5.6% | 17.6% |
| Control group | None reported | None, as the author notes |
Two cautions before you quote either.
- The 5.89% is not a fuel saving. It is the rise in km per litre. The saving in litres is smaller, because 1 - 3.56/3.77 is 5.57%. Dividing by the new number rather than the old one is the usual way these get overstated. Check which way round any "% better" claim was calculated.
- Neither trial shows the saving lasts. One month is the longest follow-up, and the 2024 paper lists the short follow-up and small sample as its own limits. A driver who is being watched in a trial is not the same as a driver on a Friday night run.
One more detail worth keeping. In the 2024 trial, the one truck that did not reach a 10% saving was 15 years old, which the paper puts down to engine wear. Training moves the driver's share. It does not move the truck's.
Step 3: what the evidence supports changing
Ordered by how well it is backed for a Thai fleet.
Harsh acceleration and braking
This is the strongest Thai signal, from the on-board data on 8,794 tanker trips. The more harsh accelerations per 100 km, the lower the km per litre:
| Harsh accelerations per 100 km | Harsh brakings per 100 km | km per litre | Litres per km |
|---|---|---|---|
| 21.5 or fewer | 75.5 or fewer | 4.453 | 0.225 |
| 21.5 or fewer | more than 75.5 | 4.079 | 0.245 |
| 21.5 to 105 | not split | 3.440 | 0.291 |
| more than 105 | not split | 3.048 | 0.328 |
The gap from the first row to the last is 32% in litres per kilometre. Harsh brakings alone cost about 8% (4.453 against 4.079). The study also found that which behaviour mattered most depended on the terrain: acceleration on plains and hills, braking in heavy traffic, average speed on open roads. That is a reason to set targets per lane, not one fleet-wide number.
Tankers are heavy articulated vehicles, so treat the sizes as indicative for a rigid ten-wheeler. The direction is not in doubt.
Idling, speed and tyres
The Canadian long-haul figures give the direction and an order of magnitude. They are not Thai measurements.
- Idling burns about 3 litres an hour. At 40 baht that is about 120 baht an hour of a truck not moving. A dock queue is exactly where this hides.
- Speed above 90 km/h costs about 10% more fuel for each further 10 km/h.
- Tyre pressure: up to about 4% more fuel for each 10 psi under-inflated. It is the cheapest item here, since it needs a gauge and a weekly habit.
Vehicle condition
The handbook lists slipping clutches, blocked air filters, fuel leaks, soft tyres and binding brakes as ways a truck burns fuel it should not. There is no Thai figure for the size, so the useful move is to look for it rather than assume it: a truck whose km per litre sits below its sister trucks on the same lane and load, week after week, is a maintenance question before it is a driver question.
Step 4: run it as a test, not a campaign
If the saving is worth 50,000 to 170,000 baht a truck a year, it is worth a proper test of a few weeks.
- Pick one lane and one load state, with at least six trucks on it if you have them.
- Hold four weeks of baseline using the method above.
- Train or equip half the trucks, not all. The untouched half is your control, and it is what the Thai trials lacked.
- Compare like with like: the same lane, loaded against loaded, the same fill method.
- Wait three months before you believe it. Read the second check, not the first.
- Only then write it into the budget, at the lower of the result and the 5.6% figure. Building the fuel line from litres keeps price and consumption on separate lines so a saving shows up where it happened.
What to do this week
- Pull twelve weeks of fuel fills and kilometres for each truck, and note which were loaded
- Calculate litres per kilometre per truck per lane, loaded and empty separately
- Rank trucks against the same lane, not against the fleet
- Find the two or three worst against their own lane and check the truck before blaming the driver
- Add a tyre pressure check to the weekly routine and record it
- Count idling hours at the docks you use most, and price them at 3 litres an hour
- Do not set a driver bonus on km per litre alone. It rewards the lane, not the driving, and how you pay drivers already changes what they do