Route optimization in TRANSCEND: what AI sequencing actually changes for a lorry

Route optimization has a credibility problem, because most of it is demonstrated on vans delivering parcels. A 40 t combination is a different problem: it cannot use half the roads, it cannot legally work a fifteen-hour day, and a missed delivery window costs more than the kilometre it saved. This is what optimization means when the constraints are real, what it changes in numbers on a multi-drop day, and how to measure it in the first month.

What does route optimization mean for a lorry?

For a van, optimization is mostly about order and distance. For a lorry it is about satisfying constraints simultaneously:

• Physical restrictions — height, weight, axle load, width, length, and in some countries ADR tunnel categories;
• Legal restrictions — HGV bans on certain roads and city centres, weekend and night driving bans;
• Driving and rest time — a plan that requires 10.5 hours of driving is not a plan;
• Delivery windows — the dock that only takes goods between 08:00 and 11:00 sets the shape of the day;
• Vehicle capability — tail lift, refrigeration, ADR equipment, capacity in pallets and in tonnes.

An engine that ignores any of these produces a beautiful route that cannot be driven. That is why the routing profile matters more than the optimizer.

How does TRANSCEND sequence a day of jobs?

The order of operations, simplified:

1. Build the constraint set from the vehicle profile and the driver's remaining hours.
2. Route legally with the truck profile, so every candidate sequence is measured on roads the lorry may use.
3. Score sequences against windows, driving time, toll cost and fuel consumption rather than distance alone.
4. Reprice the winner: tolls by axle class, fuel at the consumption implied by gradient and load.
5. Re-optimize when traffic, a cancellation or a closed dock changes the inputs.

Step three is the difference between "shortest" and "cheapest". The shortest sequence is often not the one with the lowest total cost, because a free road that climbs a mountain can burn more diesel than a tolled motorway that runs flat, and a sequence that avoids a €45 toll by adding an hour can lose money.

What does it change in numbers?

Indicative 2026 results on typical regional days, comparing a planner-built route with the optimized one. Consumption at 28–31 l/100 km, diesel €1.48/l, driver and fixed cost €34/h.

Read the last row against the first. On a day of short urban legs the sequencing dominates and the saving is 12–18%; on a long trunk run with three drops at the end, the trunk is fixed and the optimizer can only touch the tail. Anyone promising 20% on international line-haul is selling something else.

A month of 20 working days at the Madrid rate is roughly 960 km and €400 of fuel saved — more than the software costs at any single-vehicle plan.

What does a worked day look like?

Take the Madrid scenario. Eight drops around the A-2 and A-4 corridors, one vehicle, windows between 08:00 and 17:00, 22 pallets, 18 t.

The planner's route follows the order the orders arrived: a first drop in Alcalá, then back west to Coslada, then east again to Torrejón, then south. The optimized route groups by corridor: Alcalá, Torrejón, San Fernando, Coslada, then the southern cluster.

The difference is 48 km — 18% — and 55 minutes. Fuel saved is about 13.4 litres, €20. The 55 minutes matter more: they turn a 17:20 finish into 16:25, which keeps the day inside the driving limits and removes the overtime that appears three times a week.

How does it interact with tachograph hours?

Optimization without driving-time awareness produces plans that fail at 15:00. TRANSCEND carries remaining driving time, break requirements and accumulated weekly hours into the plan, so a sequence that would breach the rules is rejected rather than discovered on the road.

The practical consequences: fewer unplanned nights out, fewer infringements, and realistic ETAs that customers can be told in the morning instead of in the afternoon.

What optimization will not fix?

Worth saying plainly, because expectations decide whether a project succeeds:

• a wrong vehicle profile — height, weight or axles — poisons every route it touches;
• bad addresses, which produce a geocode error rather than a routing problem;
• capacity that does not fit the demand, where the answer is a second vehicle, not a better sequence;
• customer windows that are simply incompatible with each other.

How do you measure it in the first month?

Pick one corridor or one delivery round and keep the current plan for a week as a baseline. Then compare:

• kilometres driven per drop;
• fuel litres per 100 km, from the card statement rather than the estimate;
• on-time delivery percentage;
• overtime hours;
• toll spend per load, once the toll engine is producing cost per load.

If the optimized plan is not better on at least three of those five, the constraint set is wrong somewhere — usually windows entered as wishes rather than as facts.

FAQ

How much can route optimization really save on lorry work? Typically 10–18% of kilometres on multi-drop urban and regional rounds, and 2–6% on long trunk runs where the corridor is fixed, at indicative 2026 consumption and prices.

Does it work with delivery windows? Yes, windows are a hard constraint: a sequence that cannot meet them is rejected, not merely penalized.

Does it account for tolls? Yes, sequences are priced with tolls by axle class and with fuel at the consumption implied by the route's gradient, so the cheapest total wins rather than the shortest distance.

Can it handle ADR and refrigerated loads? The vehicle profile carries ADR category and equipment, and tunnel restrictions and vehicle capability are part of the constraint set.

Where do I see the optimizer on my own data? Run two of your regular rounds in the live demo. Technical detail is on how it works, plans on pricing.

Optimize the day, not the map

A shorter line on a map is worth nothing if it breaches driving time or misses the dock. TRANSCEND sequences legally, prices tolls and fuel on the result, and hands back a plan a driver can actually run. Book a live demo or check pricing.