A new academic study from TU Delft takes a deep dive into one of Amsterdam’s most pressing urban challenges: how to keep goods flowing into the city without choking its historic streets and aging infrastructure.
The paper, by Bijvoet, Karademir, and Atasoy (2026), models a multimodal two-echelon distribution system for Amsterdam’s HoReCa sector (hotels, restaurants, and cafés) — integrating canal freight by vessel with last-mile delivery by cargo bike, e-bike, or on foot. Goods move from a distribution center outside the city to waterside transshipment points (satellites), then on to customers via low-emission street vehicles.
What makes the study stand out is its realism. Amsterdam’s canals are narrow, its bridges and quay walls crumbling, and its streets congested. The model explicitly accounts for space scarcity at satellites (only one vessel docking at a time), load-dependent transshipment times, and strict city access time windows — constraints often glossed over in academic literature.
Key findings from a scenario analysis of over 10,000 instances
- Longer access windows pay off. Extending city access from 4 to 8 hours enables multi-trip vessel operations, dramatically cutting the required fleet size.
- Parallel transshipments are a game-changer. Allowing multiple vehicles to unload simultaneously, combined with a flexible workforce, reduces the minimum number of satellites needed from 25 to just 6.
- Denser satellite networks shift traffic to the water. More transshipment points mean shorter street-level distances and more zero-emission walking deliveries — but require more urban space.
- Relaxing full-service targets saves infrastructure. Dropping service coverage from 100% to 90% yields significant savings in vessels and satellites, especially under tight access windows.
The study offers a practical framework for policymakers: access time rules, infrastructure investments, and service level targets should not be set in isolation; they interact, and the trade-offs are quantifiable.
Scope limitations
The model assumes a single distribution center, which is a significant simplification for a city like Amsterdam, where goods arrive from multiple origins. The authors acknowledge this, but it does limit the real-world applicability of the findings.
The case study covers hotels, restaurants, and cafés. This is a relatively homogeneous and predictable demand segment. It says little about waste, parcel delivery, retail supply, or construction logistics, which together account for a large share of Amsterdam’s freight movements and exhibit very different spatial and temporal patterns. The demand data is based on a specific HoReCa dataset for Amsterdam. Details on how representative this dataset is and how sensitive the results are to demand assumptions could have been explored more thoroughly.
The model uses fixed, deterministic inputs. Real urban logistics involve traffic delays, variable demand, and last-minute changes. The authors flag this as future work, but it means the operational schedules generated may be less robust than they appear.
One city, one canal network
Amsterdam’s waterway system is unusually extensive and well-suited to freight. The transferability of the findings to other European cities with fewer or less navigable waterways is not discussed, which limits the paper’s broader policy relevance.
With Amsterdam’s canal network already there, the question is no longer whether waterborne city logistics is feasible, but how to design it smartly.