What does the research say about optimising log truck routes?

The most detailed recent study reports large gains, but read it carefully. A July 2026 preprint, not yet peer reviewed, compared optimised weekly plans with the plans one forest company actually ran over 50 weeks, and reports a 35% cut in total distance. The company's roughly 299 trucks belonged to about 59 independent contractors, and its planning was manual, with "almost no systematic backhauling". In our reading, planning across those fleets as one is the likely source of much of the gain. The paper doesn't split the saving by cause, though, and its baseline is a real week with real disruptions.

What the usual answer says

The usual answer is that route optimisation, using operations research and GPS data, cuts transport costs by 5 to 20% through better scheduling, backhauling and fewer empty kilometres. The 2026 figure sits well above that range. What matters is why, and how much of it would survive in practice.

What the 2026 study did

Abdellaoui, Benabbou, El Hallaoui, Aubé and Amazouz (arXiv 2607.16978) built a weekly log hauling model from two years of one company's operational data. The network covered 2,379 locations: 50 mills, 2,270 forest blocks and 59 contractor facilities. The fleet "comprises 299 trucks operated by 59 contractors, with a weekly average of 64 active trucks and 26 contractors". The model plans loaded and empty trips and schedules trucks so they don't queue at loaders.

The authors then "compared the transportation plans actually executed by our industrial partner with those generated by the optimization tool" over 50 weeks. Their summary table reports a 35% reduction in total distance travelled and planning time under 20 minutes a week.

What it describes, and what it doesn't test

In their discussion of managerial implications, the authors describe the partner's planning. It "is currently handled manually, relying mostly on repetitive back-and-forth trips with little optimization and almost no systematic backhauling, except when it is trivial". Planning "is decentralized, with each contractor managing its own fleet independently", and they write that this approach "prevents the exploitation of substantial cost-reduction opportunities".

That is context, not a measurement. The optimiser changes several things at once: which trucks carry which loads, backhauls, queueing at loaders, and which blocks supply which mills. The paper doesn't say how much of the 35% each accounts for.

It also doesn't say what information the optimised plans had. The executed plans were made in real time. The authors note that weekly plans face "frequent last-minute changes, including demand updates, equipment availability, and operational disruptions". If the comparison plans were built knowing how the week turned out, part of the gap is hindsight.

The money figures are harder still. The summary table's cost saving, in the local currency, is several hundred times the partner's annual transport budget as the text states it. We set the money aside and use only the distance figure.

The trucking structure behind it

A 2025 review in the International Journal of Forest Engineering by Derochers, Conrad, Bolding, Smidt and Da Silva describes a different trucking market from the preprint's. Log trucks there "were approximately 5 years older than other commercial trucks and operated in smaller fleets", of 5 to 10 trucks. Citing a regional price service, it reports that in the third quarter of 2024 transport made up "over 40% of the delivered cost of pulpwood and 26% of sawtimber products". It lists "limited information related to log truck transportation costs" among the gaps in knowledge. Small fleets appear in both markets, at about five trucks per contractor in the preprint's case.

Reading it through a data lens

The model needed every block, mill, contractor base, loader and truck in one network, with service times and demand. In our experience those facts sit in scale tickets, contractor invoices, harvest plans and dispatch calls, not in one place. Putting them together is the first job, and it has uses before any optimiser runs.

One such use is a baseline. From last month's scale tickets, each with its block, mill, truck and time, the legs each truck drove can be inferred and loaded distance set against total distance. The same records show how often one contractor's truck unloaded near a block another contractor was serving that day. That measures the room for backhauling in an operation's own network, rather than in a preprint's.

What we cannot take from it

This is one company, in one preprint, with a hindsight baseline and inconsistent cost figures. We could not open the earlier published studies of pooled transport between forest companies to set it beside, so we don't compare it with them here. Where contractors won't share plans or carry another company's loads, the plan can't be carried out. And where trucks are paid by the load and distance, a shorter route changes contractors' income. The authors suggest part of the savings "could be reinvested in increasing the remuneration rates of transportation contractors".

Quarri for operations teams is built for the people running the crews, the lines and the yard.

Quarri is an AI-native data platform for the timber supply chain. It connects buying, production, sales and inventory for forest management, sawmill, wood products and pulp, paper and packaging operators.