Microsimulation of vehicle ownership decisions within an agent-based integrated urban model: An event-based hybrid of continuous and discrete simulation approach

This study presents vehicle ownership simulation (VOSim) within the STELARS (Simulator for Transportation, Energy, LAnd use for Regional System) framework. VOSim follows an event-based decision process adopting a hybrid of continuous and discrete time simulation techniques. In STELARS, each household agent subscribes to a list of events (e.g., childbirth) that makes the agent actively adjust his or her vehicle fleet. Being active, agents make two interconnected decisions: vehicle transaction and type choice. In the vehicle transaction stage, for households that never owned a vehicle, the timing of the first vehicle purchase decision is simulated. For households with vehicles, their decision to add, dispose, or replace a vehicle is simulated. In the vehicle type choice stage, an agent’s decision to choose vehicles by body, vintage, fuel, and technology type is simulated. Vehicle transaction is simulated as a continuous-time decision using a hazard-based model. Once the timing of the transaction is determined, the vehicle type choice simulation transitions into a discrete-time step. This paper reports VOSim predictions and multi-year validation for the Okanagan region in Canada for the 2011-2021 period. Multi-year validation results confirm a satisfactory accuracy level. Prediction results suggest that a higher proportion of first-time vehicle purchasers reside in areas with lower accessibility to transit. A higher share of suburban dwellers is predicted to own alternative fuel vehicles. Overall, the VOSim adds capacity to integrated urban models to simulate vehicle ownership using a behaviorally realistic simulation procedure and be sensitive to plans and policies through an equity lens, such as who makes the first vehicle purchase decision.