The Jerusalem Activity-Based Model: Challenges and Innovations
P P Vovsha, J Freedman, Parsons Brinckerhoff, Inc, US; M Bradley, MB Research & Consulting, US; J Zmud, NuStats, US; Y Birotker, D Givon, A Dubov, Jerusalem Transport Masterplan Team, IL
The paper describes a new activity-based model currently being developed for the Jerusalem, Israel, Metropolitan Region. It incorporates many advanced features and innovative components compared to the activity-based models developed elsewhere.
The paper describes a new activity-based model that is currently being developed for the Jerusalem, Israel, Metropolitan Region. The design of the Jerusalem Activity-Based Model (JABM) incorporates many advanced features previously included in the CT-RAMP (Coordinated Travel ¡V Regional Activity-based Model Platform) system of activity-based models developed for US cities (including New York, Columbus, San Francisco Bay Area, and Atlanta), such as:
- A fully-disaggregate micro-simulation of individual households and persons,
- Using tours as the basic unit of travel,
- Consistent generation of daily activity-travel patterns and schedules,
- Enhanced temporal resolution (30 min),
- Explicitly modeled intra-household interactions and joint travel.
There are several components of JABM that are innovative compared to the activity-based models developed elsewhere. In particular, the following two innovative components are discussed in detail:
- A model for prediction of person / household mobility attributes,
- Constrained parking equilibrium model with individual parking location choice and simulation of parking search process.
Person and household mobility attributes relate to the medium-term choices that are conditional upon long-term choices (residential, workplace, and school location) but should logically precede short-term travel choices related to a particular day, tour, or trip. In most of the previously developed models, mobility attributes included car ownership only. In JABM this component is significantly expanded to include a wider range of interrelated person and household attributes:
- For each person:
o Possession of a driver license,
o Disability or limited mobility category,
o Transit pass, ticket discounts, and/or subsidy from the employer or school,
o Employer provided transportation for commuting,
o Employed provided or subsidized parking,
o School bus availability,
o Personal car availability model (car only available to one person) by car type,
o Car/cycle available from work / business,
- For each household:
o Household car/cycle ownership model (cars available to every household member) by car type,
o Usual car/cycle allocation to drivers in the household,
o Toll transponder.
The combination of these mobility-related attributes in a coherent choice framework allows for capturing trade-offs between different choices. In particular, car ownership is negatively correlated with transit attributes and employer-provided transportation while it is encouraged by subsidized parking, etc. The final purpose of this model is three-fold:
- Add behavioral realism and explanatory power to the subsequent models (in particular, mode choice) that would benefit from the additional variables,
- Create policy-sensitive variables that might be used for certain scenarios and policies (for example, peak-spreading through encouraging compressed work weeks),
- Make mobility attributes endogenous to the demand model system and sensitive to network level-of-service variables, which enhances the overall integrity of the model system.
One of significant advantages of an activity-based microsimulation structure is the ability to explicitly incorporate parking behavior that makes the model sensitive to constraints and policies associated with parking. By virtue of individual microsimulation with an enhanced temporal resolution, the model can portray the dynamics of parking in each traffic zone during the day.
The most important individual variables that relate to parking demand are tour destinations, arrival times, and planned activity durations (time for which the auto would occupy the parking space). All these variables are endogenous to JABM and available in the process of microsimulation. It requires additional information including person characteristics, associated willingness to pay, and possible eligibility for free parking, as well as the size of the travel party.
Parking supply is estimated by free and paid parking capacity in each zone as well as parking rates including the daily rate (relevant for long parking) and hourly rate (relevant for short parking). Parking constraints and variable rates requires a consideration of demand-supply equilibrium where some parking demands cannot be satisfied.
The equilibrium mechanism is implemented by means of the parking choice model that is applied in combination with two functional models to estimate the associated parking search time and track the actual parking availability at any point of time during the day. With this model, a driver does not necessarily park in the destination zone but can choose to park in some other zone (where parking is more available or cheaper) and then walk to the final destination. In the model application, it is applied successively for all tours with dynamic update of the parking availability. Associated parking search time is estimated as a function of the distance from the final destination and parking occupancy.
Association for European Transport