Calibration and validation of parking search-time function

Abstract

Large off-street car parks are traditionally modelled as self-standing traffic zones representing origins/destinations in standard network assignment models. However, such a treatment precludes the drivers from choosing alternative car parks as it assumes the car parks are their final destinations. This paper discusses the feasibility of bringing car park choice and the effects of capacity within a traditional network assignment model. The search time within car parks depends on the car park occupancy and can be represented by a flow-delay type function on the car park occupancy/capacity. This research calibrates the search-time function based on practically observed occupancy and search time at two city-centre car parks in Leeds, England. The analysis follows a simple fixed search-time method as well as a sophisticated variable search-time method. The results are validated against the observed occupancies at the car parks. A car park specific constant was introduced to account for the unobserved preference for a given type of parking facility. In a multiperiod assignment, when car park occupancies are passed on dynamically, both fixed and variable search-time approaches are seen as an improvement over the standard approach, with the variable search time outperforming the fixed time approach.