Nowadays, many cities are intending to reduce the use of private vehicles. Governments are incorporating new mobility services and are adapting their parking policies to promote a more sustainable mobility, as both strategies are believed to have the potential to reduce private vehicle use. To understand the effects of these strategies, one needs to be able to model complex travel behaviour up to a very high level of detail. Owing to their flexibility, robustness and ability to model travel activity behaviour on an individual level, activity based travel demand models (ABM) offer a highly suitable methodology for this purpose. In this paper, we employ this methodology to perform a case study in a metropolitan region in the Netherlands which surrounds and includes the cities of Rotterdam and The Hague. This region is of vital economic importance and has a very developed and dense road network. The population of this region is growing, which motivates the ambition to improve its accessibility and move towards sustainable mobility. Therefore, the findings of this study are important to similar regions seeking to do this as well. After setting up a suitable, calibrated ABM able to perform a comprehensive study on the effects of new mobility services and parking policy adaptations in the above-mentioned region, we design seven scenarios to give quantitative answers to policy-related questions on how altering features can reduce the extent to which private vehicles are used for travelling. These features include the availability of mobility hubs (hubs on neighbourhood level where sustainable travel modes are linked), the availability of car/bike sharing services, the availability of ‘Mobility as a Service’ (MaaS) subscriptions, the amount of parking capacity in the region and the parking costs. We also study what the impact would be of an improved public transport service with lowered public transport travel times to and from the city centers, and the impact of an improved cycling network infrastructure with significantly lowered travel times for bike and e-bike travellers. Based on the case study, we find that the introduction of mobility hubs alone has limited impact. However, combining this with making sharing services available to the public through MaaS subscriptions, there is a potential to reduce the number of car trips significantly, while the number of trips undertaken by a more sustainable (shared) e-bike increases as well as the number of so-called multi-modal mode trips (trips undertaken by a combination of various modes). Furthermore, improving the public transport service and micromobility network further increases the potential of mobility hubs in terms of making mobility more sustainable. The case study also shows that limiting parking capacity and increasing parking costs in the city centers is especially helpful for the reduction of vehicle use, leading to an improved car flow.

Activity-based travel demand models provide a high level of detail when modeling complex travel behavior. Since stochastic simulation is used, however, this high level may induce large random fluctuations in the output, necessitating many model reruns to produce reliable output. This may become prohibitive in terms of computation time when comparing travel behavior between multiple scenarios, in which case each scenario requires its own simulation. To alleviate this issue, we study the use of common random numbers, which is a technique that reuses the same random numbers for choices made by travelers between scenarios. This ensures that any observed difference in output across scenarios cannot be attributed to mutual differences in drawn random numbers, eliminating an important source of random fluctuation. We demonstrate by a numerical study that common random numbers can greatly reduce the number of runs needed, and thus also the required computation time, to obtain reliable output.

Mobility as a Service (MaaS) and new mobility concepts mutually inspire each other, provide alternatives for the private car-oriented transport system as we know it, and will offer more mobility choices in a single journey than ever. This multitude of mobility choices however poses challenges in modeling the travelers’ mode choices in travel demand prediction models. To address these challenges, this paper develops a multimodal tour-based mode choice model as part of an activity-based demand model. By explicitly modeling access and egress modes, this choice model creates multimodal mode chain sets on a tour level based on restrictions with respect to personal vehicle ownership, MaaS subscription ownership and vehicle states, and subsequently makes mode choices for every traveler. For the creation of these mode chain sets, we introduce the concept of mode categorization. Seven mode categories are proposed, which include both private and shared mobility concepts. This categorization makes sure that modes are mutually sufficiently different in nature, so that reasonably unbiased mode chain choices can be made. Furthermore, the reduction to seven categories enables the study of large scenarios, while the introduced categories still represent new and already existing modes well. The potential of the model is illustrated by simulating travel demand in the Metropolitan region Rotterdam-The Hague. The results show that our model is capable of making plausible mode choices in the presence of MaaS and new mobility concepts, and can be used to assess the impact of mobility hubs where access and egress mode choice is important.

This paper studies a Gaussian process approximation for a class of stochastic traffic flow models. It can be used to efficiently and accurately evaluate the joint (in the spatial and temporal sense) distribution of vehicle-density distributions in road traffic networks of arbitrary topology. The Gaussian approximation follows, via a scaling-limit argument, from a Markovian model that is consistent with discrete-space kinematic wave models. We describe in detail how this formal result can be converted into a computational procedure. The performance of our approach is demonstrated through a series of experiments that feature various realistic scenarios. Moreover, we discuss the computational complexity of our approach by assessing how computation times depend on the network size. We also argue that the (debatable) assumption that the vehicles’ headways are exponentially distributed does not negatively impact the accuracy of our approximation.