What impact does online grocery shopping have on urban transport? - Simulating location choice, passenger and freight transport in Berlin

Purpose: Due to the increasing volume of online shopping, urban transport has been changing tremendously during the past years. The question remains how this will influence the environment in terms of emissions. Answering this question becomes even more relevant today with increasing e-grocery purchases which can be expected to aggravate the problem since grocery shopping induces a considerable share of urban transport, either by logistic companies, retailers or consumers. Experts in the field find generally ambiguous effects of e-commerce on transport with many of them concluding that home delivery has a high potential to reduce emissions even for groceries. However, only few studies considered passenger and freight transport at once and in a modeling or simulation framework. Since many of these studies originated in logistics research, they neglect the following drivers: substitution between online-shopping and other activities, emissions caused by chilling food, travel-reducing effects such as trip-chaining and effects of different modes. Thus, we test a methodology that integrates passenger and freight transport models in order to estimate CO2-emissions for both types of transport and calculating the balance between them. Research approach: We introduce a framework of several interlinked simulation models ranging from location choice to tour planning. From the application to the city of Berlin, Germany, we derive resulting emissions comparing scenarios for the year 2030 with population growth and with and without home-delivery of groceries. In the first step a simple location model predicts the number of small and large grocery stores by zone for each scenario which are used by the model SYNTHESIZER to create a synthetic population of grocery stores additionally taking into account a 10 % reduction of the number of stores due to competition, and the same distribution of stores across zones as in 2015. In the third step the agent-based travel demand model TAPAS generates passenger shopping trips for both scenarios. E-commerce is simulated here by randomly selecting and deleting 3 % of short-term shopping trips. The addresses of households pertaining to these trips and the synthetic addresses of small and large grocery stores are used in the fourth model which creates the delivery tours from shops to households for the e-commerce scenario. Finally, we calculate emissions for passenger and freight transport trips and assign them to the causing zone. Findings and originality: We find for the home delivery scenario that freight transport optimizes routes and is thus more efficient than private shopping trips by car reducing VKT in total, it generates more emissions particularly due to refrigeration and is thus not sustainable. Inner-city areas are more affected than zones at the edge of the city. Research impact: Our study contributes to the research in the field by showing a methodology that combines several simulation models to simulate the impact of grocery online-shopping on the environment in terms of emissions. Furthermore, we consider for the first time emissions caused by secondary effects such as refrigeration which implies a total increase in emissions when introducing home delivery contrasting the results of many similar studies. Practical impact: Assumptions in our methodology are easily adjustable and thus allow for a range of scenarios. Applying such a model enables policy-makers to test different measures in order to deal with the challenges increasing e-tailing of groceries poses to urban planning