Parking Facility

Each parking location in the network is represented as an individual agent in POLARIS. Parking facilities are loaded from the supply database and are spatially linked to the road network through links, activity locations, and zones. The parking agent manages capacity, reservations, pricing, and the lifecycle of individual parking events.

Parking Types

POLARIS supports the following parking facility types, each read from the Parking table in the supply database:

Type	Description
Garage	Structured parking garage (also used for airport parking)
Lot	Surface parking lot (also used for park-and-ride / transit lots)
Street	On-street curbside parking (unmetered)
Meter	On-street metered parking
Location	Generic parking associated with a specific activity location
External	Parking at external stations
Sink	Sink parking used for special modeling purposes
Loading Dock	Freight loading dock (B2B heavy-duty trucks)
Loading Zone	Freight loading zone (B2C light-duty trucks)

Loading Dock and Loading Zone are freight-specific parking types. They do not have walk links assigned during network initialization and are searched separately from passenger parking. For details on how freight vehicles use these facilities, see Freight Vehicle Parking.

Each parking facility stores:

• Capacity: The total number of vehicle spaces available.

• Zone: The traffic analysis zone the parking facility belongs to.

• Links: The drive-access link(s) and, optionally, a walk-access link for multimodal routing.

• Activity location: The activity location the parking is associated with.

• E-scooter availability: The number of shared e-scooters docked at the facility, enabling micro-mobility access/egress.

Capacity Management

Parking capacity is tracked in real time. Two lists are maintained:

1. Parking list: Vehicles currently physically parked at the facility.

2. Reservation list: Vehicles that have reserved a spot but have not yet arrived.

Instantaneous capacity checks compare the current parking list size to the total number of spaces. For reservation-based capacity checks, a temporal sweep algorithm is used: reservation time windows are compared against existing reservations and currently parked vehicles to find the peak simultaneous usage at any point in the requested time window. A reservation is only accepted if peak usage does not exceed the facility’s capacity.

Pricing Rules

Each parking facility can have one or more pricing rules, each defining:

Field	Description
rule_type	The granularity of pricing: Free, Flat, Minute, Hourly, Daily, Monthly, or Yearly
rule_priority	Priority for rule selection
min_cost	Minimum cost charged regardless of duration
min_duration	Minimum duration required to use this rule
max_duration	Maximum duration that can be covered by this rule

Prices are time-of-day dependent, stored as a lookup by rule ID and simulation time. The cost calculation algorithm greedily allocates the parking duration across applicable rules, choosing the cheapest option at each step. For example, a 10-hour stay might be covered by an hourly rule for the first 3 hours and a daily flat rate for the remainder.

A facility is classified as free if the cost for a 24-hour stay is zero.

Park and Unpark Operations

The figure below illustrates the complete lifecycle of a vehicle through the parking facility—from reservation to parking, through the parked state, and finally unparking with record output. It also depicts the temporal sweep algorithm used for capacity checks.

Parking a Vehicle

When a vehicle arrives at a parking facility:

1. The system checks whether the vehicle holds an existing reservation. If so, the reservation is converted to an active parking event and the vehicle is marked as parked.

2. If no reservation exists, instantaneous capacity is checked. If space is available, a new parking info record is created and the vehicle is marked as parked.

3. If e-scooters are available at the facility and the person agent is eligible, an e-scooter is borrowed for access/egress to the final destination. The e-scooter count is decremented accordingly.

Unparking a Vehicle

When a vehicle departs a parking facility:

1. The vehicle is located in the parking list.

2. If an e-scooter was borrowed, it is returned and the facility’s e-scooter count is restored.

3. A parking record is finalized with the actual exit time, computed cost, and e-scooter usage, and is pushed to the demand database for output.

4. The vehicle status is updated to indicate it has left the network.

If the actual parking duration exceeds the anticipated duration by more than 20%, a debug warning is logged for overstaying.

End-of-Day Cleanup

At the end of the simulation day, all vehicles still parked are automatically unparked, their records are finalized, and any outstanding reservations are flagged with a warning.

Reservation System

Vehicles (both personal and TNC) can reserve a parking spot ahead of arrival. A reservation specifies:

• The vehicle requesting the spot

• The anticipated start time

• The anticipated duration

The reservation is accepted only if the facility has temporal capacity (verified by the sweep algorithm described above) and the estimated cost is below a threshold. Reservations are stored in both a list and a time-sorted multimap for efficient overlap checking.

When the vehicle arrives and parks, the reservation is consumed and removed from both data structures.

Spatial Indexing

Parking facilities are indexed spatially using an R-tree, organized by parking type. This enables efficient nearest-neighbor queries used by both the pre-trip parking choice model and the en-route parking search. The network provides nearest-parking-search capabilities that return sorted lists of nearby parking facilities and their distances from a given location or link.

Activity locations also maintain pre-computed lists of nearby free and paid parking facilities with their distances, enabling fast lookups during the en-route parking search.