Fanø: emptying by data instead of gut feeling

Why fixed frequencies were impossible

Fanø has roughly 3,400 permanent residents and around 2,900 summer cottages. It operates six “super” recycling collection points and eight regular recycling stations. The regular stations mainly serve permanent residents and take cardboard, plastic and cans; the super points sit near the holiday cottages and also take organic and residual waste. All 82 containers are semi-buried and hold up to three cubic metres.

In areas with permanent residents, containers typically take between two weeks and a month to fill. In the summer cottage areas, the same containers fill in anywhere between one or two days and several weeks, depending on the weather. The senior engineer responsible described planning efficient waste collection in advance as almost impossible, and was blunt about the previous method: emptying happened by gut feeling.

The pressure was compounding. Increased online shopping and greater environmental awareness among residents and visitors had pushed recycling volumes at the collection points roughly 20% above the level assumed when the collection contract was tendered. The authority needed better operating economics and, simultaneously, to stop residents encountering overloaded containers.

What was deployed

Sensors were installed on all 82 containers, measuring fill level and passing the data to the municipality via the cloud. Signals are sent up to 60 times a day, giving minimal battery consumption, low cost and good coverage.

The workflow is simple and worth describing because it is what most UK authorities would actually run. On screen, the engineer sees every container with its fill level marked green, amber or red. He marks the containers to be emptied, creates a route plan, and sends it to the contractor, who receives it in the driver app. The contractor uses the same app to report when each container has been emptied. Fill history for each container is available, so he can judge how urgent an emptying is rather than treating every red flag identically.

An honest note about accuracy

Accuracy was not good at the beginning. Working with the supplier, the authority established that the sensor needed to be positioned slightly differently to avoid false signals. Once repositioned, when a sensor reported 90% full, it was 90% full. A built-in algorithm also learns continuously and becomes more accurate over time.

We include this deliberately. Mounting position is the most common cause of early inaccuracy in a sensor deployment, and it is entirely avoidable — which is why we set position per container type at installation and validate first readings physically before a container joins live round planning.

The results

• Emptyings reduced by 20% within the first two months, against a target of 25%.
• The target reduction would bring the number of emptyings back down to the level assumed in the original tender, despite volumes having grown 20% since.
• The authority expected the investment to pay for itself within a manageable period.
• Complaints about missed emptying became documentable — the municipality can now show whether a container was emptied or not.

Bringing the contractor in

Fanø's handling of its contractor is instructive, because the contractor's initial concern is universal. The lorry driver, the engineer noted, might worry about having fewer containers to empty — but he shouldn't. He stops spending time checking containers that turn out to be half full and which it makes no sense to empty and invoice, and that time can be sold to another customer. His administration also reduces, because invoicing data comes directly through the app.

Connecting the contractor to the system was planned as the project's next step, so that emptying orders arrive directly from the platform rather than being passed on by the municipality.