Parking in the City: The Model as a Tool for Policy Evaluation

Full text

1
Parking in the City: The Model as a Tool for Policy Evaluation
I. Benenson1, K. Martens2, S. Birfir1
1Deptartment of Geography and Human Environment, Tel Aviv University
Telephone: +97236409178
Fax:+97236406243
Email: bennya@post.tau.ac.il, sbirfir@yahoo.com
2Institute for Management Research, Radboud University Nijmegen, the Netherlands
Telephone: +31243612740
Fax: +31243611841
Email: k.martens@fm.ru.nl
1. The Goal
Parking policies have a strong impact on the functioning of cities. The introduction of a
new or changes in the existing parking policy requires a careful analysis and evaluation
of these impacts in light of policy goals. To do that we need a model of parking in the
city which could serve as a tool for systematic analysis of the impacts of various policy
scenarios.
Surprisingly, quantitative data about parking in the city are rare and models play a
limited role in the analysis of urban parking policies, with few exceptions (e.g. Shiftan
2001). Much of the modelling literature regarding parking is theoretical in nature and has
not been applied to real-world situations (e.g. Voith 1998; Petiot 2004; Lam 2006). Most
policy-oriented work, in turn, hardly makes use of the potential offered by state-of-the-art
modelling techniques (e.g. Ferguson 2003; Marsden 2006). Against this background, we
propose using an agent-based model (Benenson and Torrens, 2004) to simulate urban
parking policy scenarios and analyze their impacts from user and public policy
perspective.
In practical cases, policy-makers may have many reasons to introduce new parking
policies, such as the wish to reduce parking search times, to improve parking availability
for visitors, or to guarantee parking for local residents. Whatever the policy goals, in all
cases in which the ratio between demand and supply for parking approaches or exceeds
one, it will be extremely difficult to forecast the impacts of new policies, without testing
these policies at the spatial and temporal resolution at which they will be implemented.
Classical models based on averages will not do under such circumstances. In other words,
we need a spatially explicit agent-based dynamic model of parking in order to analyze,
and ultimately tackle, parking problems in current highly motorized societies.
2. A Model of Parking in the City
The proposed model aims helping planners and decision-makers to formulate and
compare parking policies. The model has been built using a Geosimulation approach
(Benenson and Torrens, 2004). In this approach, real-world entities are directly
represented as inanimate and animate model objects, which “behave”, that is, change
their properties and location in space. The inanimate objects directly represent the
features belonging to the layers of a high-resolution GIS of urban infrastructure. The only

2
animated objects in our case are car drivers, and their behavioural rules describe all stages
of driving: driving towards the area in which parking search starts, parking search, and
leaving the study area after parking. However, the model focuses on parking search. The
model enables the formulation of parking constraints and enforcement levels and its
outcomes can be aggregated over ensembles of individual drivers delineated by areas and
time periods, according to the interests of the policy-maker.
2.1. Static model objects
To adequately represent the parking process, we build on the following components of an
urban GIS, which are available or can be constructed for most Israeli cities: (1) Road
network with data on number of lanes, traffic directions and on-street parking
permissions; (2) Houses (Destinations); (3) Off-street parking places; and (4) On-street
parking places (Figure 1). The attributes of on-street parking places are parking
permission, fees and, when available, the probability of a fine for illegal parking.
Figure 1: To represent a two-way traffic street, the
centreline is duplicated, and each copy is
employed for representing one direction. Parking
places are built by dividing the segments into 4m
fragments.
2.2. Animated driver agents
The essence of the agents’ representation in a Geosimulation model is their behaviour. In
case of drivers, the complete description of the behaviour should include behaviour
during: (1) driving towards the destination, (2) parking search, (3) parking, and (4)
driving out. The model is thus built in two versions. The “full” one accounts for the entire
driving process, starting from the moment the car enters the system till the moment it
leaves through one of the exit points. In this paper we focus on the second stage, and
consider the “parking only” version, in which drivers “land” at the outer boundary of the
parking search area, immediately start their parking search, and disappear from the
system just after leaving the parking place.
2.3. Performance indicators
The object-based nature of the model makes it possible to follow every driver and, thus,
enables direct estimation of the performance of the parking policy from the driver’s and
the policy-maker’s point of view.
Drivers’ view: Given the set of targets, time interval, and group of drivers we
estimate distributions of:
- Parking search time;
10 m

3
- Distance between parking place and destination;
- Overall/hourly payment.
Policy-maker’s view: The policy-maker observes (but not necessarily accounts for)
drivers’ indicators. In addition, the policy maker accounts for the following collective
characteristics of the parking situation:
- Fraction of occupied parking places, and its changes over time;
- Number of cars searching for parking place, and its changes over time;
- Parking turnover (number of cars using a parking place during a time interval);
- Revenues from on-street and off-street parking.
2.4. Technical characteristics of the model
The model is implemented as a VBA ArcGIS application and can work with a practically
unlimited number of simultaneously parking drivers. Model parameters and results at
resolution of cars and parking places are managed with the SQL Server and, thus, policy
performance indicators can be constructed for various groups of drivers, sets of
destinations and time intervals without re-running the simulation.
3. Application of the model
Within the framework of the project, various policy scenarios will be tested to improve
the existing parking situation in the centre of Tel Aviv. As a first try-out, the model has
been applied to analyze the consequences of a local scenario, the construction of a multi-
level underground garage in a neighbourhood, where all places will be for sale to local
residents. The consequences of this local scenario have been studied for the Basel
neighbourhood, a densely built, mixed-use, neighbourhood, located in the old centre of
Tel Aviv. According to the GIS-based estimates, the demand for parking in this
neighbourhood amounts to about 1.10 cars per parking place. The municipality is
considering allowing the construction of an underground parking garage in the area of up
to 200 places, to reduce parking problems for local residents, who complain on a regular
basis about the lack of parking.
The model demonstrates that the main effect of local improvements in parking supply
lies in the reduction of the fraction of drivers who search for parking for a long period of
time. This finding suggests that, assuming no positive feedback loop in terms of increases
in car ownership, the additional supply could substantially reduce overall parking search
time, at least in the short run. Following the modelling results, if about 250 additional
parking places were to be added in the centre of each urban block of 500 by 500 meter in
the dense Tel-Aviv centre (about 1,000 parking places every 1 km2), the share of
residents searching for more than 10 minutes for a parking space would drop from 25% to
less than 10%, with evident consequences regarding air pollution, traffic congestion, and
public opinion. At the same time, even with such an additional supply, residents will
continue experiencing a lack of parking in Tel Aviv’s central area, i.e. they will still face
substantial average search time and walking distance between parking place and place of
residence. This, in turn, suggests that if the developer will be able to offer the parking
places in the new garage at a price attractive enough for local residents, they will be eager

4
to buy them. The decision about the size of the parking garage has thus been reduced to
an economic rather than a transport issue.
REFERENCES
Carrese, S., E. Negrenti and B. B. Belles (2004) Simulation of the Parking Phase for Urban Traffic
Emission Models. Paper presented at TRISTAN V - Triennial Symposium on Transportation Analysis,
Guadeloupe.
Benenson, I. and P. M. Torrens (2004) Geosimulation: Automata-Based Modelling of Urban
Phenomena, London, Wiley, 204 pp.
Ferguson, E. (2003) Zoning for parking as policy innovation. Transportation Quarterly, 57/2: 47-55.
Lam, W. H. K., Z.-C. Li, et al. (2006) Modelling time-dependent travel choice problems in road
networks with multiple user classes and multiple parking facilities. Transportation Research Part B:
Methodological, 40: 368-395
Marsden, G. (2006) The evidence base for parking policies: a review. Transport Policy, 13/6: 447-457
Petiot, R. (2004) Parking enforcement and travel demand management. Transport Policy, 11/4: 399-
411.
Shiftan, Y. and R. Burd-Eden (2001) Modelling response to parking policy. Transportation Research
Records, 1765: 27-34
Voith, R. (1998) Parking, transit, and employment in a Central Business District. Journal of Urban
Economics, 44: 43-58.