Waste Collection Routing Optimization: The Complete Guide

Key takeaways

• The cost of waste services is soaring. The full cost of public waste management in France reached €144 per inhabitant and €266 per tonne on 2022 data, a very sharp rise since the previous study (ADEME, 2025 benchmark, cited by ORD&EC).
• Every new waste flow multiplies routes. Since 2024, French local authorities must offer source separation of biowaste. For businesses, the “7-flow” decree requires separate sorting of paper, metal, plastic, glass, wood and, since 2025, textiles, plus mineral fractions and plaster on construction sites.
• An advanced optimization solver cuts operating costs by 15 to 20%, by acting simultaneously on clustering, sequencing and the placement of tipping trips and, in industrial collection, on the chaining of container rotations.
• Industrial collection is a problem in its own right. On-demand orders, skip and roll-off container rotations, access windows set by each customer site and hazardous waste subject to ADR: a tool designed for household kerbside collection is not enough.
• Moving to on-demand collection requires a solver. With fill-level sensors, routes are no longer fixed: they must be recomputed every day, which no manual planning can keep up with.

Contents

1. Why has waste collection route planning become so complex?
2. What is a waste collection route optimization solver and how does it work?
3. What features should a solver cover for household and industrial waste collection?
4. Why move from fixed-frequency collection to fill-level-driven collection?
5. Glossary
6. Frequently Asked Questions

1. Why has waste collection route planning become so complex?

Short answer: waste collection combines six sources of complexity that compound one another: a growing number of waste flows, vehicle capacity that forces round trips to disposal sites, safety rules that constrain the direction of travel, highly heterogeneous collection points, industrial collection driven by customer demand, and a tightening economic and regulatory framework.

More and more waste flows

Local authorities now manage five or six flows: residual household waste, packaging and paper, glass, biowaste, green waste and bulky waste. Each flow has its own frequency, container type, vehicle and disposal site.

Source separation of biowaste, mandatory in France since 1 January 2024, adds a low-tonnage but high-frequency flow. Conversely, falling residual waste volumes often allow its collection frequency to be reduced. Together, these two shifts mean redesigning every route, not just adding new ones.

Businesses are following the same trend. The French decree of 16 July 2021 requires producers of commercial and industrial waste to sort at source and have paper-cardboard, metal, plastic, glass and wood collected separately, with textiles added in 2025 and, on construction sites, mineral fractions and plaster (AIDA-Ineris). For a collection operator, a customer who used to fill a single skip of mixed general waste now fills several, each with its own rotation and disposal site.

Vehicle capacity and tipping trips

A refuse collection vehicle (RCV) fills up during its route. Once full, it must travel to a disposal site (transfer station, sorting centre, energy-from-waste plant, composting facility) before resuming collection. The number of tipping trips and when they happen weigh heavily on total mileage.

Two constraints make this calculation tricky:

• Fill rates depend on the flow and the season: green waste surges in spring, packaging after the holidays, and residual waste drops in summer in some towns while surging in tourist areas. On the industrial side, customer activity (production peaks, stocktaking, summer shutdowns, construction season) makes volumes fluctuate.
• Disposal sites have their own constraints: opening hours, daily intake capacity, queues at the weighbridge, and a disposal site imposed per flow, per contract or per material offtake agreement.

Crew safety and direction of travel

Waste collection is a high-accident activity. Recommendation R437 of the French national health insurance fund, adopted in 2008, urges local authorities and contractors to act on route plans to eliminate hazardous manoeuvres: reversing, two-sided collection and excessive speed.

For a planner, this translates into hard constraints:

• One-sided collection: each side of the street is collected separately, unless an exception applies
• No reversing to reach a collection point: dead ends without a turning area are collected another way
• Traffic direction, turn restrictions, road width and weight limits
• On industrial sites: internal traffic plans, loading and unloading safety protocols, protective equipment and site-specific authorizations

Highly heterogeneous collection points

A single operator combines several collection modes:

• Kerbside collection of individual bins, street by street (an arc routing problem)
• Communal bins in apartment blocks and wheeled bins at shops and offices
• Above-ground, semi-underground or underground drop-off points, emptied by crane vehicle
• Skips, roll-off containers and compactors at recycling centres, industrial sites and construction sites, collected by hook-lift or skip loader truck

Each mode is a different optimization problem: street segments in one case, isolated points in another, container rotations in the last. They often coexist on the same day, or even on the same route when businesses are collected alongside households.

Industrial collection, a demand-driven problem

Commercial and industrial waste collection follows a different logic from household collection:

• Orders rather than frequencies: some pickups are recurring, but many are triggered on customer demand, often for the same or the next day.
• Container rotations: a hook-lift truck carries only one container at a time: it drops off an empty skip, picks up the full one, empties it at the disposal site and returns. The order of operations (drop-off, pickup, exchange, relocation) and the availability of empty skips at the depot shape the entire day.
• Time windows and rules per customer: access hours, available docks, container location, contact to notify, no access during production.
• Hazardous waste: oils, solvents, infectious healthcare waste or batteries require vehicles and drivers certified for the transport of dangerous goods (ADR), loading incompatibility rules and traceability through a waste tracking form.
• Profitability per customer: every pickup is invoiced. The planner weighs commercial urgency, geographic grouping and the true cost of the rotation.

A tightening economic and regulatory framework

• Sharply rising costs: the full cost of the service reached €144 per inhabitant and €266 per tonne on 2022 data, a very steep increase over the previous study (ORD&EC, based on ADEME).
• Pay-as-you-throw pricing: it makes bin set-outs vary from week to week and routes less predictable.
• Fleet energy transition: electric and bio-CNG collection vehicles have range and charging constraints that become routing constraints.
• Low-emission zones (LEZ): they determine which vehicle can be assigned to each urban district.
• Crew recruitment: HGV drivers and collection operatives are hard to recruit; every route hour saved counts.

2. What is a waste collection route optimization solver and how does it work?

Definition: a waste collection route optimization solver is a computational engine that models the territory (road network, collection points, customer sites, depots, disposal sites), the fleet, the containers and the crews, then explores a huge number of clustering, sequencing and tipping-trip combinations to identify the lowest-cost plan that satisfies every constraint.

What a solver does that field experience cannot

An operations manager knows their territory better than anyone. They know which street is jammed on Tuesday mornings, which customer requires a visit before 7 a.m. and which crew knows which district. What they cannot do is compare millions of clustering options and visit sequences, flow by flow, while simultaneously accounting for traffic direction, collection side, vehicle capacity, customer time windows and disposal site opening hours.

This is why most household collection districts in use today are inherited: they were drawn once, then tweaked bit by bit with each new housing development, without ever being redesigned from scratch. In industrial collection, the problem is the opposite: the plan is rebuilt by hand every day from the orders received the day before, at the expense of geographic grouping.

Three families of problems to solve together

• Arc collection (kerbside): the vehicle must travel along street segments, on one side or both. The goal is to minimize deadheading, i.e. kilometres driven without collecting. In operations research, this is known as the capacitated arc routing problem (CARP).
• Node collection (drop-off points, commercial bins): the vehicle visits isolated points. This is a classic vehicle routing problem (VRP), with time windows, capacities and intermediate tipping trips.
• Container rotations (skips, roll-off containers, compactors): the vehicle carries one container at a time and chains drop-offs, pickups, exchanges and tipping trips. The goal is to minimize empty trips between customers, depot and disposal sites, while managing the stock of available empty skips. This VRP variant is known as the roll-on/roll-off routing problem.

A solver designed for waste collection must handle all three families in a single model: the same fleet and depot often serve the local authority in the morning and commercial customers in the afternoon.

What concrete gains for a collection service?

• Kilometres and route hours: less deadheading, better-placed tipping trips, better-balanced districts. Gains translate into fuel, vehicle wear and crew hours.
• Number of routes and vehicles: optimized clustering often eliminates one or more routes per day, or even defers the purchase of a new vehicle.
• Rotations per vehicle: in industrial collection, better chaining of drop-offs and pickups increases the number of rotations per hook-lift truck per day and reduces the number of idle skips.
• Safety: reversing, two-sided collection and street crossings are ruled out when the plan is built, not corrected after the fact.
• Workload fairness: routes of comparable duration and tonnage.
• Customer service: time windows and pickup lead times promised to businesses are respected, with fewer skips overflowing for lack of a rotation.
• Carbon footprint: every kilometre saved directly cuts fuel consumption and emissions, an indicator tracked in local authorities’ annual reports and in industrial customers’ CSR reporting.
• Decision-making capability: simulate before committing to a frequency change, a new waste flow, a switch to drop-off points, a tender response or the opening of a transfer station.

Why replanning speed matters as much as the initial plan

Household routes look stable, but conditions on the ground change constantly: roadworks, public holidays, a broken-down vehicle, a green waste peak, an absent driver. Industrial collection adds urgent orders received during the day. A fast solver can rebuild a viable plan in minutes to absorb the disruption or insert a request, instead of leaving the crew or dispatcher to improvise.

3. What features should a solver cover for household and industrial waste collection?

Short answer: a solver fit for waste collection must cover seven areas without workarounds: detailed road network modelling, waste flows and frequencies, capacities and tipping trips, commercial and industrial collection, safety and access constraints, fleet and crew management, and scenario simulation with its KPIs.

Road network and collection point modelling

• Arc collection (streets, left side / right side) and node collection (drop-off points, communal bins, customer sites) in a single model
• Traffic direction, turn restrictions, U-turns allowed or not
• Service time per point or per linear metre, differentiated by mode (bin, bag, drop-off point, skip, compactor)

Waste flows and frequencies

• Household multi-flow: residual waste, packaging, glass, biowaste, green waste, bulky waste
• Commercial multi-flow: paper-cardboard, metal, plastic, glass, wood, textiles, mineral fractions, plaster, biowaste, mixed general waste
• Frequency per district (weekly, twice weekly, fortnightly…) and fixed or flexible collection days
• Weekly balancing: spreading visits across the days of the week to smooth the workload
• Split-body vehicles collecting two flows in a single pass

Capacities, tipping trips and disposal sites

• Vehicle capacity by weight and volume, depending on the flow’s compaction ratio
• Multiple tipping trips during the route, choosing the most relevant disposal site
• Opening hours, daily capacity and waiting times at disposal sites
• Disposal site imposed per flow, per treatment contract or per material offtake agreement

Commercial and industrial collection

• On-demand orders: integration of pickup requests received the day before or the same day, with promised lead time and priority
• Recurring pickups: weekly or monthly subscriptions combined with one-off orders in the same plan
• Container operations: drop-off, pickup, full-for-empty exchange, relocation between sites, with container / vehicle compatibility (hook-lift, skip loader, crane, RCV with bin lifter)
• Container fleet management: stock of empty skips by depot and by type, the right container in the right place at the right time
• Customer site constraints: access windows, closing days, safety protocols, required authorizations, on-site manoeuvring time
• Hazardous waste: ADR-certified vehicles and drivers, loading incompatibilities, licensed disposal sites, regulatory traceability
• Mixed routes: commercial customers collected alongside households, or industrial customers served by the same fleet as the public contract

Safety and accessibility

• No reversing and no two-sided collection, with configurable exceptions
• Vehicle width, gross weight, height and turning radius per road and per site
• Time restrictions (schools, markets, pedestrian town centres, industrial zones closed at night) and LEZs
• Black spots reported by crews, integrated as constraints

Fleet and crews

• Heterogeneous fleet: RCVs of various sizes, small collection vehicles, crane vehicles, hook-lift trucks, skip loaders, ADR vehicles
• Range and charging of electric vehicles, bio-CNG refuelling
• Working time, breaks and shift spans, regulatory driving times, multiple depots
• Driver skills and authorizations (licences, ADR, access to specific sites)
• In-house / contractor / subcontracting trade-offs and clustering by contract lot

Simulation and monitoring

• Scenario simulation: frequency change, new biowaste flow, switch from kerbside to drop-off points, new neighbourhood, new disposal site, winning an industrial customer or a public contract
• KPIs: total kilometres and deadheading, route hours, number of tipping trips, rotations per vehicle, tonnage per route, cost per tonne or per pickup, on-time pickup rate, CO2 emissions, number of hazardous manoeuvres eliminated

4. Why move from fixed-frequency collection to fill-level-driven collection?

Short answer: a fixed-frequency route empties half-full containers and leaves others overflowing. Fill-level sensors, on drop-off points as well as on commercial compactors and skips, make it possible to collect only what needs collecting. But they turn a stable problem into a daily one: the list of points to visit changes every day, and only a solver can rebuild optimal routes at that pace.

The hidden cost of “blind” collection

According to Birdz, Veolia’s IoT subsidiary, containers are only around 45% full on routes without fill-level data, compared with an average trigger level of 85% when collection is sensor-driven (Journal du Net, 2021). In other words, a large share of visits mobilizes a vehicle and a crew for half-empty containers.

For industrial customers: from fixed appointments to pickups at the right time

In industrial collection, the equivalent of fixed frequency is the pickup scheduled on a fixed date, or the customer’s call when their skip is already overflowing. In the first case, the hook-lift truck often carries a half-empty container, even though a rotation costs almost the same full or empty. In the second, the pickup comes too late and disrupts the day.

Compactors and skips fitted with fill-level or pressure sensors make it possible to trigger the pickup at the right fill rate and anticipate it a few days ahead. The gain is measured in rotations avoided, tonnage carried per rotation and emergencies eliminated.

Why sensors alone are not enough

A sensor tells you which containers to collect. It does not tell you in what order, with which vehicle or when to empty the truck. Without an optimization engine, fill-level data often ends up as a list of alerts that operators squeeze into existing routes as best they can.

To capture the gain, the solver must:

• Select each day the points and containers to collect based on measured and forecast fill levels, anticipating overflow before the next possible route
• Decide whether to collect a container today at 70% because the vehicle is passing nearby, or leave it until tomorrow
• Rebuild routes, chain rotations and place tipping trips in minutes, every day
• Combine sensor-driven points, fixed-frequency points, kerbside collection and customer orders on the same day

Optimize the territory rather than each route

The same logic applies beyond sensors. Optimizing route by route reproduces inherited boundaries: former districts, former municipalities, former contract lots or each sales rep’s portfolio. A territory-wide approach makes it possible to:

• Redraw districts flow by flow, unconstrained by historical boundaries
• Pool depots, disposal sites and vehicles between household and industrial collection
• Pool resources between neighbouring inter-municipal authorities or after a merger
• Balance the workload across days of the week and between crews
• Choose the disposal site for each tipping trip based on the vehicle’s position, rather than a fixed assignment

This is what sets an optimization engine apart from simple waste management software: the ability to challenge the scope of each route, not just the order of visits.

Conclusion

Local authorities and collection operators face an unprecedented combination: sharply rising costs, waste flows multiplying with biowaste and “7-flow” sorting for businesses, safety requirements that rule out shortcuts, industrial customers expecting fast pickups, crews that are hard to recruit and a fleet in the middle of its energy transition.

In this context, route optimization is no longer a one-off re-clustering exercise but a permanent management lever, reinforced by the arrival of fill-level sensors. But not all tools are equal: what makes the difference is the ability to model the reality of household and industrial collection without workarounds, across the entire territory, while integrating with the tools already in place.

It is on these three dimensions — algorithmic performance, functional coverage tailored to waste collection and scalability — that Atoptima supports local authorities and collection operators.

Waste collection glossary

Frequently Asked Questions