Modelling Park & Ride Trips and Parking

Developed by: Minnerva Ltd, Miles Logie
Client: Various
Country: United Kingdom and Ireland
Agent: Minnerva Ltd, website www.minnerva.co.uk
Project year: 2009

Park and Ride (P&R) travel is a topic of particular interest to many transport policy makers that has been addressed in various ways by modellers, but normally as an appended element rather than the more integrated approach now offered by OmniTRANS version 5. The OmniTRANS approach treats P&R in terms of travel that is accessing the public transport system by car rather than on foot as for the usual public transport trips. This provides a consistent modelling framework but it also includes the necessary allowance for the fact that access and egress modes from public transport differ for P&R, with outbound trips accessing by car and egressing on foot, and the reverse for return trips. There is also a significant difference concerning the point at which P&R trips can join the public transport system, which might be relatively near the beginning or the end of trips. Hence OmniTRANS provides additional controls to allow efficient consideration in the modelling of P&R access options over a much greater geographical range than is required for access on foot.

P&R sites are identified as public transport stops with a user-defined characteristic ‘type' that is used both to control displays and inclusion into access choice sets. As much rail travel is accessed via car trips, rail stations are typically included as P&R sites as well as formal P&R sites served by buses.

Although P&R is modelled within the public transport assignment component of OmniTRANS, OtTransit, it is effectively a distinct travel mode and trip matrices are required for each of the P&R outbound and return elements.

Minnerva has applied OmniTRANS P&R modelling to models in the UK (Reading) and in Ireland (Cork) and which have included demand and network modelling that have generated the required P&R matrices. These models have also paid attention to issues of modelling parking in general, not just at P&R sites, specifically to consider the effects of limited parking capacities.

The demand modelling follows an incremental choice modelling form, so requires a base year view of P&R trips. The choice modelling splits trips into non-motorised (walk and cycle), only public transport, and some use of the car. This latter category is then split into ‘car-all-the-way' and P&R. [The choice modelling is shown schematically in Figure 1.] The demand modelling first generates outbound home-based trips and subsequently derived the return portions at different times of day, so this provides an automatic means of differentiating the outbound and return portions of P&R travel.

The choice modelling uses travel costs generated from the network modelling, including separate sets for highway, public transport, and P&R in the form of ‘cost skim matrices'. This means that changes in travel costs in any of these can result in changes to the number of P&R travellers.

Modelling Parking

The approach to the modelling of parking has been to derive a cost function that corresponds to ‘search times' spent looking for a parking space. For moderate levels of utilisation (ratio of use to capacity) search times are low, but increase significantly once usage approaches car park capacity. Detailed modelling of car parks can be complex on account of the time-dependencies as vehicles arrive, accumulate, and leave at different rates during the course of the day. For the purpose of demand modelling only the average utilisation levels in a modelling period (e.g. AM peak, inter-peak, and PM peak periods) is considered. Demand is not constrained to car park capacity limits, but over-capacity demand results in large search times that feed through to the demand model and promote choices that do not involve excessive use of the parking site.

This approach to parking is applied at a zonal level and to P&R sites. In the case of zones the parking arrivals are given by car arrivals to the zones; these can be ordinary study area zones, including on-street parking, or may correspond to large car parks. (A parking component of the choice model calculates whether parking is in ordinary zones or in alternative sites including car park zones.) The search times are added to the other travel costs for reaching a zone, that is, added to the relevant columns of the generalised cost matrix.

 For modelling P&R parking matters are more complex as the parking occurs within a trip rather than at the end. The arrivals at the P&R car parks are therefore obtained by summing trips on the car access links; the inclusion of search times is more difficult because P&R sites are used by trips going to many destinations, but not all P&R travellers to those destinations will necessarily use the same P&R site. Search times are therefore associated with boarding penalties, which can then be converted to skim costs using the standard OtTransit procedures. These elements require accessing and, for search times, updating OmniTRANS database tables, which are done using the OmniTRANS OtTable functionality.

The ability of OmniTRANS to make the modelling of P&R travel and parking an integral element of a general multi-modal modelling system has provided a valuable means of keeping transport modelling aligned with the policy interests of those seeking to improve transport systems.

For more information, please contact Erik de Romph.