Construction is a transport reliant sector accounting for a large share of urban freight transport in European cities. Traditionally, construction planners have focused on on-site efficiency, often neglecting logistics to and from the site. Additionally, urban planners focus on the current and future city structure, neglecting the transitional phase during construction. Traffic planners in the municipality tend to solve temporary traffic management plans ad-hoc, only looking at one construction site at one specific time and space, instead of looking at how multiple temporary traffic management plans affect the urban area. Thus, creating a planning gap for how we get from the current city structure to the future city structure.

The purpose of this thesis is to investigate how traffic simulation and machine learning can support collaborative planning processes for construction transport in urban environments. To address this purpose, two research questions were formulated and answered through two studies: one qualitative and one quantitative. The first research question and corresponding study aim to understand how traffic simulations can support collaborative planning processes to reduce disruptions caused by construction projects in urban areas. The second research question and its associated study aim to examine how machine learning can support decision-making in construction transport planning.

The qualitative study explores the collaborative planning process of the Ostlänken megaproject in Norrköping, a new highspeed railway connecting Stockholm and Linköping. Earlier studies show that to tackle the issues caused by construction projects, urban planners and traffic planners in the municipality needs tools to assess multiple disturbances at once. A tool that is currently used by traffic planners is traffic simulations; however, previous studies show that the use of traffic simulations in construction focus on car mobility, neglecting pedestrians, cyclists and public transport users. The qualitative study resulted in two conference papers, Paper I and Paper II. The findings highlight that disturbances often stem from a project-centric focus and that municipalities need to coordinate planning at an early stage. The study introduces strategic traffic management planning as a way to coordinate temporary traffic management plans. Traffic simulations are a key component, supporting both the assessment of traffic disturbances and the evaluation of mitigation measures. However, current simulations used by municipalities often lack detailed hourly data on travel flows across all modes (cars, pedestrians, cyclists, public transport, and goods transport), limiting analyses to a daily level.

The quantitative study examines one of the least detailed components of traffic simulations, construction transport in urban environments. Currently, construction transports are either omitted or included only in aggregate form within other transport sectors. This study takes a different approach by estimating construction transport demand using machine learning with construction context data as input. The findings show a potential of machine learning models to predict construction transport demand, however, the high variance in the contextual data limits accuracy. To provide more effective decision support for urban and traffic planners, the study highlights the need for higher-quality data and standardization of existing datasets.