Advanced Route Planning: Optimising for Multiple Constraints

Capacity constraints

A vehicle routing problem with capacity limits (a Capacitated VRP, in operations-research terms) has to track how much a vehicle is carrying at every point along its route and never let it exceed the vehicle's stated capacity - whether that's measured in weight, volume, or a count of packages. This is why route optimisation for delivery fleets needs real vehicle data, not just a generic "truck": a cargo van and a heavy-duty truck have very different capacity ceilings, and a plan built without that distinction will either underload the truck or try to overload the van.

In practice this constraint changes which stops can be grouped onto the same run, not just the order they're visited in - a set of deliveries that's geographically efficient to combine might still need to split across two vehicles once you account for what each one can actually carry.

Time windows

A time window says a stop can only be serviced between two specific times - a customer who's only home in the evening, a business with a 9-to-5 receiving dock, a retail location that won't accept deliveries during opening rush. Adding time windows to a route turns a pure distance-minimisation problem into one where the sequence has to satisfy every window simultaneously, which can force a route to be less distance-efficient than the theoretical shortest path in exchange for actually being deliverable.

This is also where service time per stop matters: a delivery that takes two minutes to hand off and one that takes twenty minutes for an installation affect how many other time-windowed stops are reachable that day, even if they're geographically similar.

Priority and SLA-driven deliveries

Not every stop is equally flexible. Same-day orders, contractual service-level commitments, and high-value or perishable deliveries often need to be sequenced earlier in a route even when a pure distance calculation would put them later. Advanced route planning handles this with a priority weighting that biases the search toward servicing higher-priority stops sooner, trading some overall efficiency for meeting the commitments that actually matter to the business.

How these constraints interact

The hard part isn't handling any single constraint - it's that they interact. A high-priority stop might also have a narrow time window and be at the edge of a vehicle's remaining capacity. Multi-constraint route optimisation has to evaluate feasibility across all of these simultaneously, which is why it's a meaningfully different (and more computationally demanding) problem than simple distance minimisation, and why the quality of the underlying solver matters more as the constraint list grows. For how this plays out when payload is the binding limit, see route planning for drinks distributors and delivery planning for builders' merchants.

Practical takeaway

If your operation only ever deals with one constraint at a time, simple routing may be enough. But most real fleets are juggling capacity, time windows, and priority together on every route - and the businesses that model all three tend to end up with plans that are not just short, but actually executable by the driver who has to run them.