Abstract Manager

Background and Objectives

Urban microsimulation models mostly rely on the 'typical weekday' travel data for forecasting and simulating travel demand. Such travel surveys collect single-day diaries due to the cost and participant fatigue associated with multi-day data collection. However, the concept of a typical weekday no longer aligns with current realities due to substantial post-pandemic changes in travel behavior. To address these concerns, this research proposes a simulation-based optimization framework for generating weekly travel activity schedules based on self-reported participation frequency of activities such as in-home and out-of-home work activities.

Methods

The integrated transportation, land use, and energy (iTLE) model is an operational microsimulation model for Halifax, Canada. The iTLE's activity scheduler uses a Markov-Chain-Monte-Carlo approach to generate individual’s activity schedules and durations for a typical weekday. In this paper, this module is revamped for generating weekly individual activity schedules. This research develops an Integer Programming-based optimization model that satisfies constraints such as the anticipated frequency of a specific activity and time allocations for a day. This study follows the Genetic Algorithm concept to create candidate solutions to the optimization problem aiming to minimize the deviation between actual versus simulated frequency of activities over the week.

Results

The weekly travel activity simulation model has been initially applied to generate weekly work activity distribution which will be further extended to generate shopping, entertainment, other in-home, and out-of-home activities. Significant differences have been observed between in-home and out-of-home work activity patterns simulated for five weekdays. A Similarity Index (SI) test is conducted on simulated data to understand the extent of daily variation in activity participation. A relatively small SI value indicating a significant variation over the weekdays is obtained for in-home work activity. Results also showed noticeable differences in the activity start hours for in-home and out-of-home work activities, with distinct morning peaks for out-of-home work, whereas in-home work started even at later periods of the day.

Conclusions

It is well established from the existing literature that activity participation and time allocation can greatly differ across the days of the weeks. However, this aspect is rarely explored in the existing literature. Although multi-day surveys are ideal for assessing temporal activity variability but face limitations like high costs and low response rates. This study aimed to make the most of a single-day travel survey by proposing a simulation-based optimization framework by generating weekly pseudo diaries with minimal deviation from observed travel patterns.