Title: Selecting and scheduling interrelated projects: application in urban road network investment

Authors: Elham Shayanfar; Paul M. Schonfeld

Addresses: Department of Civil and Environmental Engineering, University of Maryland, 3109 Jeong H. Kim Engineering Building, College Park, MD 20742, USA ' Department of Civil and Environmental Engineering, University of Maryland, 1173 Glenn Martin Hall, College Park, MD 20742, USA

Abstract: Decisions about the selection of projects, alternatives, investments, operating policies and their implementation schedules are major subjects in various fields including operations research, financial analysis, business management, engineering economy and transportation planning. In these various disciplines sufficiently good methods have been developed for planning and prioritising projects when interrelations among those projects are negligible. However, methods for analysing interrelated alternatives are still inadequate. We propose a combinatorial method for evaluating and scheduling interrelated roadnetwork projects. In particular, this paper demonstrates how a traffic assignment model can be combined effectively with a genetic algorithm (GA) in a multi-period analysis to select and schedule road network projects while capturing interactions among those projects. The goal is to determine which projects should be selected and when they should be funded in order to minimise the present value of total system cost over a planning horizon, subject to budget flow constraints.

Keywords: project selection and scheduling; genetic algorithm; GA; project interrelations; user equilibrium; project evaluation; system optimisation; planning and prioritising projects; minimising system cost.

DOI: 10.1504/IJLSM.2018.090472

International Journal of Logistics Systems and Management, 2018 Vol.29 No.4, pp.436 - 454

Received: 08 Jul 2016
Accepted: 28 Oct 2016
Published online: 19 Mar 2018 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article