Deliverable D3.2 presents the AUTOFLEX transport system, covering the network architecture, infrastructure placement, and performance evaluation that define how Oskar 2.0 and the supporting building blocks operate as an integrated zero-emission freight system across the Netherlands.
Background and Purpose
Building on Deliverable D3.1, which developed the Stow&Charge Hub (SCH), Temporary Port Terminal (TPT), and Mobile Distribution Centre (MDC) concepts, D3.2 addresses where this infrastructure should be placed, how many vessels should operate the network, and how the system performs in terms of cost, cargo volume, and energy use.
The work covers two tasks from the AUTOFLEX Grant Agreement. Task 3.4 develops the transport system architecture by identifying the most promising network configurations through mathematical optimisation and logistics simulation. Task 3.5 evaluates performance by simulating how Oskar 2.0 operates on the resulting routes, producing transport cost and energy estimates for each voyage.
Estimating Transport Demand
The deliverable develops a method for estimating freight demand between candidate terminal locations using NUTS 3 level data from the ETISplus dataset, covering 17 relevant regions across the Netherlands. Containerisation rates are applied by goods category: food products and textiles are rated High (100%), agriculture and wood products Medium (50%), and bulk goods such as coal, ores, and liquids are excluded. A modal shift potential factor is then applied to estimate the share of road freight that could realistically transfer to an AUTOFLEX service.
Demand estimates were calculated for 14 candidate locations: Rotterdam, Den Haag, Leiden, Delft, Amsterdam, Alphen aan den Rijn, Haarlem, Katwijk, Breda, Eindhoven, Utrecht, Roosendaal, ‘s-Hertogenbosch, and Moerdijk, with Rotterdam and ‘s-Hertogenbosch representing the highest freight volumes.
Designing the Network Architecture
An integer programming optimisation model was developed to determine the best configuration of the AUTOFLEX transport network, covering placement of Stow&Charge hubs and Temporary Port Terminals, fleet size, and route assignment. The objective is to maximise transport work shifted from road to inland waterways, weighted by the social impact of each corridor.
The model was solved for fleet sizes from 10 to 30 vessels in increments of five:
- A fleet of 10 vessels focuses on the highest-impact corridors between Rotterdam, Den Haag, Leiden, and Delft, delivering close to 600,000 impact-weighted TEU-kilometres per week
- Expanding to 20 vessels brings in the long-haul corridors to Breda, Eindhoven, and ‘s-Hertogenbosch
- By 30 vessels, all beneficial demand in the model is covered
- The first 10 vessels deliver nearly three times the impact-weighted output per vessel compared to the next 10
The four Stow&Charge hub locations confirmed for the network are Rotterdam, Amsterdam, Alphen aan den Rijn, and ‘s-Hertogenbosch, selected for their central position on the network and existing or near-term ZES battery charging capability.
Selecting Concrete TPT Locations
The deliverable moves from the candidate area analysis of D3.1 to the selection and detailed assessment of specific quay locations for each Temporary Port Terminal. A structured rating method is applied across nine criteria covering quay surface condition, lifting clearance, quay access, ship-terminal interface, hinterland connection, public interface, legal and contractual feasibility, and transport demand.
Preferred TPT locations selected across the seven modal shift clusters:
- Haarlem – Brownfield, southern Haarlem industrial area (score 2.75/3)
- Leiden/Katwijk – Joh Guyt at Heen Industrial Park, Maandagse Wetering canal, Katwijk aan Zee
- Den Haag – Bouwmaat XL at Binckhorsthaven (primary); former bulk terminal at Haven van Rijswijk (secondary)
- Delft – Western waterfront digger site at Delftse Schie
- Roosendaal – Quay along the Mark canal
- Breda – Southern Breda brownfield
- Eindhoven – Two locations: Ekkersijt industrial area and MIREC quay near city centre
Each selected location is assessed for how a TPT could be realised using the AUTOFLEX building block concepts, including cargo handling equipment, storage configuration, waterway interface, hinterland connection, and operating model.
Performance Evaluation with SIMPACT
The 10-vessel network was evaluated in SIMPACT, an agent-based logistics simulator developed by SINTEF Ocean that models each vessel, cargo unit, battery container, and terminal individually. SIMPACT incorporates bridge and lock delays from EuRIS waterway data, dynamic cargo handling times per port call, and high-fidelity energy consumption modelling based on scale model test data for Oskar 2.0.
The simulation covers a full year of operations across 10 routes. Key results for the 10-vessel network:
- Total annual transport work: 73 million tonne-kilometres, equivalent to approximately 97,000 truck trips eliminated per year
- Average vessel capacity utilisation of 80 to 100% on main corridors
- Battery container availability confirmed at all three Stow&Charge hubs throughout the simulated year with no vessel delays caused by energy availability
- Maximum lead times range from 2.2 to 4.2 days depending on origin-destination pair, with minimum lead times as low as 0.4 days on shorter corridors
Cost Performance
Transport cost was evaluated for two representative routes: Vessel 10 on the short, high-volume Rotterdam to Delft corridor and Vessel 9 on the longer Den Haag to Alphen aan den Rijn to Amsterdam route.
Short route (Vessel 10):
- Oskar 2.0: 14,560 containers per year at €132 per container (single route) or €125 per container (10-vessel network)
- Conventional diesel baseline: 7,440 containers per year at €151 per container
- Conventional electric baseline: 6,460 to 7,392 containers per year at €195 to €217 per container
Long route (Vessel 9):
- Oskar 2.0: €249 per container (single route) or €235 per container (10-vessel network)
- Conventional diesel baseline: €346 per container
- Conventional electric baseline: €458 per container
Across both routes, Oskar 2.0 operating autonomously around the clock moves significantly more containers per year at lower cost per container than any conventionally crewed alternative.
Investment and Operating Costs
The estimated total capital investment for one Oskar 2.0 vessel entering the network is €8.25 million, covering:
- Newbuild hull: €2.4M
- Autonomous technology package: €1.1M
- Two ZES battery containers: €2M
- Electric reach stacker: €1.14M
- Charging station with installation: €1.31M
- TPT permits and planning: €0.3M
Annual operating costs per vessel are estimated at €545,000, covering Remote Operations Centre service, vessel maintenance and insurance, TPT operations, and quay lease.
Outlook: Foundation for Work Package 5
The transport system results from D3.2 feed directly into AUTOFLEX Work Package 5, which will assess the broader environmental, social, and economic impacts of deploying the AUTOFLEX system at the regional and European level.
The results also highlight a key policy challenge: many CEMT II waterways in the Randstad pass under moveable bridges that must be opened to allow the AUTOFLEX vessel to pass. While waterborne transport relieves road congestion, more frequent bridge openings could partially offset this benefit. This is identified as a subject for the policy recommendations work of Work Package 5.
Partner Contributions
This deliverable was produced by contributions across the AUTOFLEX consortium:
SO – Lead author, optimisation model development, TPT location analysis, SIMPACT logistics simulation and cost evaluation
DFDS – Reviewer
