Author: Lampros Yfantis, Aimsum

Demand for e-commerce has skyrocketed over the last decade, and this growth trend is projected to continue. As increasing numbers of people choose to live in cities, this growth in demand for delivery services poses economic, operational and environmental challenges for last-mile urban logistics.

Is there a solution? the growing green technology and sustainability market is the key facilitator. However, one of the main challenges that such ecosystems are facing lies in the identification and establishment of new governance frameworks and the application of sustainable solutions and technologies.

To address this challenge and accelerate the much-needed shift to smart and sustainable last-mile logistics, the GREEN-LOG project investigates, establishes and deploys:

1. co-creative and collaborative policy, regulation and business frameworks,
2. AI- and simulation-based last-mile logistic planning solutions and

• Sustainable last-mile logistics technology, platform, and infrastructure deployments.

In GREEN-LOG, the development and deployment of a new Last-mile Logistics Planning service is enabled by Aimsun to solve important planning and operational problems for stakeholders in Barcelona, ES and Oxford, UK. This blog provides a first glimpse of the envisioned solution.

A new framework for sustainable last-mile logistics planning – the value of AI and simulation

The main goal of the GREEN-LOG Planning service is the analysis and assessment of last-mile delivery systems; either from the point of view of a public authority or logistics service providers. The service primarily aims to identify the optimal infrastructure and fleet operation configurations for last-mile logistics services, based on a multitude of service and system sustainability and efficiency indicators. The provision of this service relies on a joint but modular – i.e., flexibly integrated – prediction-optimization-simulation modelling solution framework (see Figure 1), which enables:

• Day-to-day parcel demand prediction
• Infrastructure and operations optimization
• Fleet operations simulation in realistic traffic conditions.

Within the Prediction service – led by the University of Wolverhampton –state-of-the-art AI-based parcel demand forecasting models are investigated (and benchmarked) by Aimsun based on historical data from logistics service providers.

Predicted demand can be used by the joint optimization-simulation framework in two different ways:

1. either for the generation of mid-term to long-term/mid-term “what-if” scenarios for assessing different infrastructure interventions and operation logics, or
2. for next-day (pre-day) fleet management in more short-term fleet management cases.

Now, considering that the main goal is to find the best design/setting for new interventions, it could be argued that mere optimization seems sufficient for the task at hand. Due to static and often deterministic travel time assumptions within such approaches– and especially in dense urban transport networks – optimal service performance margins are overestimated [2]; in simple terms: higher service and fleet efficiency and subsequently customer satisfaction than in reality. This is where the added value of joint service fleet and traffic simulations for urban networks kicks in; ensuring higher operation evaluation robustness, realism, and reliability. To this end Aimsun delivers in GREEN-LOG:

• an interoperability framework; developed to couple the multimodal service fleet and traffic simulation tool, Aimsun Ride [1] (hint: tool description in this article!), with an external optimization module developed by the Athens University of Economy and Business, and
• fleet, network and traffic models considering project requirements such as cycling and transit network representations for multimodal and multi-class last-mile logistic simulations.

Stay tuned for our next article with more information on the methods we will apply and the results from the GREEN-LOG Planning Service solutions!

References

1. Narayanan, S., Salanova Grau, J.M., Frederix, R., Tympakianaki, A., Masegosa, A.D. and Antoniou, C., 2023. Modeling of shared mobility services-An approach in between aggregate four-step and disaggregate agent-based approaches for strategic transport planning. Journal of Intelligent Transportation Systems, pp.1-18.
2. Wolf, F., Engelhardt, R., Zhang, Y., Dandl, F., Bogenberger, K., 2023. Effects of dynamic and stochastic travel times on the operation of mobility-on-demand
   arXiv preprint arXiv:2308.05535

Photo Credits: iSTOCK