Decision support for the operational planning phase of the transit network planning problem

The planning problem underlying the design of a transit system is known in the literature as the transit network planning problem (TNPP). Due to the high degree of complexity of decisions pertaining to the transit network planning process, the TNPP has traditionally been partitioned into strategic, tactical, operational, and control phases. The operational planning phase of the TNPP requires the solution of complex, interdependent sub-problems, such as the vehicle scheduling problem (VSP), the crew scheduling problem (CSP), and the crew rostering problem (CRP). While various mathematical models are available for integrating these problems, the general failure to incorporate staff transportation requirements into these integrated models is a significant gap in the existing body of knowledge. This gap is addressed in the current dissertation by formally establishing a novel, integrated problem called the vehicle crew rostering problem with staff transportation (VCRPST). A novel decision support framework (DSF) is proposed in this dissertation for aiding planners during the operational planning phase of attempts at solving instances of the TNPP. The DSF is designed to provide a structured and adaptable guide for assisting transport planners with the complex decision-making processes involved in holistically integrating vehicle scheduling, crew scheduling, crew rostering, and staff transport decisions during the operational planning phase of the TNPP. The framework is modular in nature, and consists of a database component, a model component, a solution component, and a user-interface component. A novel mathematical model and an accompanying two-phase solution approach are proposed for the VCRPST. The mathematical model innovatively employs a time-space network representation to capture the interdependencies between vehicle and crew assignments over a multi-day rostering period. Recognising the computational intractability of this integrated model, the twophase solution approach involves first employing a column generation scheme to solve the daily VCSPST during a scheduling phase. The high-quality set of feasible duties generated during this first phase is then exploited to solve the full VCRPST model during a subsequent rostering phase. The framework and computerised implementations of its components are finally applied to three case studies involving real-world VCRPST data provided by an industry partner, in order to demonstrate the practical applicability of the dissertation contributions. The results demonstrate that the solutions generated by the DSF are capable of achieving significant operational improvements and cost savings over solutions produced by a manual scheduling process.

Grantor dc:publisher

Stellenbosch : Stellenbosch University

Year dc:date.issued

2026

Author dc:creator

• Steenkamp, Pieter Jacobus

Advisor dc:contributor.advisor

• Van Vuuren, J. H.

Language dc:language.iso

en

Repository record dc:identifier.uri

https://scholar.sun.ac.za/handle/10019.1/135816

OAI identifier oai:identifier

oai:scholar.sun.ac.za:10019.1/135816