To test the effect of extending the park and ride to 3,600 parking spaces, two different scenarios were evaluated at microscopic level to identify potential issues depending on which direction the vehicles accessed the car park.

Utilising a year’s data, typical day patterns were created using Aimsun’s Pattern Generator, used for demand generation and model calibration.

Modellers replicated a realistic toll pattern and the decision making of the individual vehicles.

This meant modelling 13 vehicle types, including toll-exempt vehicles, preregistered vehicles and new users, both of which can decide whether to use the fast lane or not depending on their perceived value of time with the option to park then continue by bus or rideshare, pick up passengers, or pay a toll.

Separate models were used for the initial fast lane decision, then for behaviour within the facility. A custom Aimsun AAPI was used to communicate with a Conduct+ system to calculate toll prices and HOV2+ HOV3+ conditions.

Vehicle occupancy was tracked for HOV2+ and HOV3+ behaviour while Car to Ride transfer time was modelled based on dynamic parking occupancy.

The model validation consisted of a complex combination of reproducing correct flow and speed on the detectors, toll data, travel time, the vehicle split on the Fast Lane (registered toll user, on the spot toll payment, toll exempt, bus, truck, etc.) and car park occupancy.