A Framework for a Joint Mode Choice ? Household Car Allocation Model
E Petersen, Parsons Brinckerhoff, US; P Vovsha, PB Consult, US
This paper describes a new model developed for regional travel demand forecasting that introduces explicit competition between drivers in the household over available vehicles in order to satisfy the overall household constraints on vehicles.
Despite the many advances made in activity-based models, certain gaps exist in the overall framework, particularly when looking at models that have been developed for practical implementation. One of the important issues is that activity-based models developed in the U. S. lack an explicit linkage between household auto ownership and individual-level mode choice. Auto ownership is unquestionably a key variable in mode choice. Without explicit control over intra-household car-allocation, however, an activity-based model might inadvertently violate internal household consistency by generating more competing tours than autos yet still assign each tour to a drive alternative during the mode choice step. This potential for overestimating single-occupancy vehicle (SOV) tours will have been corrected either within the estimation process or during validation of mode choice. We expect to find that, in the absence of a car-allocation model, household members that do not face auto shortages will come up somewhat short in terms of generating SOV tours at the aggregate level, while members of vehicle insufficient households will be producing too many SOV tours.
In reality, the number of household vehicles acts as a constraint upon the number of competing auto tours that can be implemented by the household members at the same time. In a household with limited auto availability, household members should be observed utilizing transit, non-motorized modes or sharing cars more frequently. Thus, mode choice made by different household members for different tours should not be modeled independently. Introducing constraints through an explicit modeling of intra-household car-allocation mechanism improves behavioral realism of the modeled mode choices and may have a significant impact on evaluation of projects and policy measures associated with transit improvements, congestion pricing schemes, and car sharing promotions.
The activity-based microsimulation modeling framework used in the New York and Columbus, Ohio models is flexible enough to support explicit modeling of the intra-household car-allocation mechanism and linking mode choices made by different household members through a car-availability constraint. Within this general modeling framework, it is assumed that fully and partially joint tours (including escorting travel arrangements) are explicitly modeled at the activity generation stage. At the car-allocation stage we consider the amount of joint travel fixed and do not further reschedule household members? tours, since the modeled tours should not be more efficient than those observed in the survey data. By adding a household level car allocation model towards the end of the overall model stream, we still generate the total household activities in a manner consistent with the previous modeling efforts, but we introduce explicit competition between drivers in the household over available vehicles in order to satisfy the overall household constraints on vehicles. We estimated this new type of model on the basis of a travel diary-type survey for the Atlanta, Georgia metropolitan region. We intend to replace the typical mode choice module with the newly estimated joint mode choice ? car allocation model. This paper will include some comparisons between the output of these two types of mode choice models.
At first glance, a car allocation model does not appear to be necessary for U.S. metropolitan regions, since car ownership is close to the saturation point with many households owning more cars than drivers. A vehicle allocation model would appear more relevant for European or Asian cities. Nevertheless, even in the U.S., households living in the urban cores of metropolitan regions (where air quality is typically worst) typically demonstrate a ?car shortage? that must be considered to accurately portray their travel decisions. Second and perhaps more importantly, we have developed this framework for a joint car allocation ? mode choice model in such a way that specific cars may be associated with specific tours. It will eventually allow for explicit tracking of each vehicle throughout the modeled day. If these cars are tagged with information about their age and size, which is typically available in newer household travel surveys, traffic simulation and air quality modeling can be much more precise than it currently is. This paper will show how this car tagging might be done as an extension of this model, though the focus will be on implementing the joint mode choice-auto allocation model.
The model structure can be briefly summarized as follows: all estimated tours are gathered and sorted at the household level including individual and joint tours. For each tour, the competing tours where there is time window overlap are identified. The number of available cars is derived from reported car ownership. For each processed tour, the binary choice of auto versus non-auto mode is considered. The non-auto mode category includes various transit and non-motorized options. The critical linkage across tours is ensured by constructing the utility of the non-auto mode so that it includes the auto utilities for the competing tours. This composite non-auto utility is scaled based on the deficit of vehicles for the competing tours. If there are sufficient vehicles to meet all demand, then these competing utilities are scaled to 0. Thus, for households without car-allocation conflicts (households with high car ownership), mode choice will be processed independently for each tour without concern for competing tours? utilities. When there is meaningful competition over vehicles, each household member must take into account the utility of the other household members. After the first tour is modeled, the car deficit index is adjusted if necessary, and the process is repeated until all tours have been modeled. Household members completing a tour towards the bottom of the queue with a low utility, such as an individual discretionary tour, may find that all cars have been allocated and the drive-alone mode is unavailable. This structure allows for an integrative accounting of all vehicles and household tours with the further advantage that intra-household competition over vehicles has been explicitly added into the model framework. This work adds important insights into the intra-household structure of individual car preferences and car-allocation mechanisms into the mode choice modeling framework.
Association for European Transport