1. Problem statement and contribution E-commerce is a rapidly growing and evolving sector. The sector is however struggling with its organization of the last-mile deliveries in order that it meets the sustainability requirements, both economically and environmentally. Different delivery methods translate into different environmental impacts. Several studies have established the advantages of collection points in terms of air pollution and/or CO2 emissions in comparison to home delivery (McLeod et al., 2006; McLeod and Cherrett, 2009; Song et al., 2009; Edwards et al., 2010; Giuffrida et al., 2016; Carotenuto et al., 2018). Zhang et al. (2018) and Verlinde et al. (2019) reach similar conclusions but only given a specific time of consumer behavior, identifying consumers’ collection trips as determining factor in comparing the environmental impact. Yet, these studies all apply for urban areas. However, consumers of all places (also suburban and rural) engage in online retailing. Moreover, web-shops hardly focus on urban inhabitants solely and often take a national approach. Given that the results of available research is bound to the local/urban context, their conclusions cannot be generalized or transferred to rural or suburban areas. 2. Methodology Following our objective to address this gap, we investigated the environmental impact associated with four last mile delivery methods from the perspective of non-food retail products in Belgium. We compared deliveries originating from a dedicated distribution center to homes (direct with 3PL or via dedicated local distribution centers) and collection points (direct or via store supply) in terms transport-related external costs for CO2 emissions, air polluting emissions, accidents, noise nuisance, infrastructure and congestion and assess the differences in impact between consumers’ residences, specifically urban, urbanized and rural areas. Both logistics flows and customer movements to the collection points were considered. To do so we applied the agent-based simulation model TRABAM (Mommens, 2019). The model uses MATSim and extends the Freight Extension (Schröder et al., 2012). 3. Results The results indicate on the one hand that homedeliveries via a well-established 3PL (with high daily volume i.e. 250.000 parcels) is the most sustainable scenario for urban, urbanized and rural areas. Deliveries to collection points are in none of the considered scenarios more sustainable than homedeliveries via 3PL, this is due to the customer movements. Yet it has to be noted that the difference between homedeliveries via 3PL and deliveries to collection points via store supply is very small to almost neglectable for the urban area. The difference becomes bigger for urbanized area and is the biggest for the rural area. Organizing a proper homedelivery system is not a sustainable option for the considered case. On the other hand, significant differences (between 10 and 15%) in terms of sustainability were found within scenario’s between urban, urbanized and rural areas. Both indicate that the environment (urban, urbanized, rural) should be considered as a parameter.

In this paper we present a methodology to more accurately calculate emissions in regional freight transportation models by taking the loading rate of vehicles into account in the emission factors. This is done by incorporating the shipment-based demand data from a multi-agent simulation framework into the route assignment. In a case study we show that disregarding the vehicle loading rate can lead to underestimations of the amount of pollutant gasses. The application of this methodology in combination with a multi-agent simulation model allows to simulate impacts of changes in logistic behavior, such as vehicle type use or the scheduling of round tours, on emissions more accurately.

The paper presents a model to estimate the demand of parcels by disaggregating nation-wide commodity flows. Secondly, the model generates parcel delivery tours to quantify transport-related effects. This model is applied to simulate parcel deliveries using cargo bikes. It consists of a two-step process: first, micro depots located close to the demand are fed with vans; second, parcels from micro depots to the customers are distributed by cargo bikes. The model simulates different shares of cargo bikes vs. motorized vans to deliver the same demand. We also studied the effect of micro depot densities and different parcel demand intensities in the same catchment area. Lastly, we compared the cargo bike tours at locations with different demand densities (parcels/km2). The results find beneficial effects of cargo bikes when the demand density and the share of cargo bikes is high. Under these conditions, the total vehicle-kilometer traveled and motorized vehicle emissions can be reduced.