Inferring commercial vehicle activities in Gauteng, South Africa

Abstract

a b s t r a c tTo address the underreporting of freight from a transport geography point of view, we present a novel analysis of the time and spatial characteristics of disaggregated commercial vehicle activities. The activities were extracted from raw global positioning system (GPS) data collected in South Africa over a sixmonth period for more than 30,000 commercial vehicles. The analyses of the activity chains provide useful characteristics such as activity and chain durations, number of activities per chain, and the spatial extent of the activity chains. Key results indicate that about 60% of activity chains have between 5 and 15 activities per chain while 25% of the chains have 4 or less; 89% of the chains have a duration of 24 hours or less; and approximately 75% of all activities start between 08:00 and 17:00. The paper's contribution is twofold: it firstly demonstrates a methodology to extract and evaluate vehicle activities and activity chains from raw GPS data. Novel results and characteristics about transport geographies in Gauteng, the economic centre of South Africa, are presented. We also report on the sensitivity of the analyses to certain parameters. Secondly, we introduce new metrics to evaluate a geographical area's economic productivity based on commercial activity.

Large-Scale Agent-Based Combined Traffic Simulation of Private Cars and Commercial Vehicles

“…To determine if the simulated commercial activities occur at the right geographic location at the right time of day, activities were geographically aggregated to the GAP mesozone level (Figure 3a) so as to be able to compare it with the observed activities reported by Joubert and Axhausen (2009). Temporally, the events were aggregated to the hour of the day in which it occurred.…”

“…There are fewer short trips and more trips exceeding 20 minutes. Since only internal distribution trips were simulated, which accounted for only an estimated 35% of Gauteng commercial traffic (Joubert and Axhausen, 2009), we argue that the impact will become higher once through-traffic is incorporated too. Still, the inclusion of commercial vehicles in the agentbased simulation model do have the desired (expected) effect: slowing of traffic, and thus increased travel time.…”

“…From the original 31041 vehicles used by Joubert and Axhausen (2009) to describe the activity chains of commercial vehicles, we were only interested in those that spend 90% or more of their activities within the province of Gauteng. For those 5196 vehicles, activities and activity chains were extracted.…”

“…With the rich activity set, we were able to build conditional probabilities for four chain parameters: the start location, denoted by M l ; start time, M t ; chain duration, M d ; and the number of activities, Since, from Joubert and Axhausen (2009), nearly 50% of commercial vehicle activity chains observed has duration in excess of 24 hours, our generated chains had to be adapted to fit within the 24-hour simulation time window imposed on the simulation scenario. Figure 4 depicts the process followed in this paper.…”

“…In Joubert and Axhausen (2009) we laid the foundation for understanding commercial vehicle movement and activities. In this paper we build on that foundation, generating intra-provincial commercial vehicle chains for vehicles performing at least 90% of their activities within the study region in Gauteng, South Africa.…”