Railway renewal and maintenance cost estimating

• Main Author: Ling, Daniel
• Other Authors: Roy, Rajkumar ; Shehab, Essam
• Published: Cranfield University 2005
• Subjects: 625.100681
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.485088

The aim of this thesis is to present a structured methodology which estimates Railway Infrastructure renewal and maintenance costs when there is a lack of quantitative cost data at the early stages of the project life cycle. Furthermore, this thesis presents renewal and maintenance infrastructure cost estimating issues and investigates current Railway renewal and maintenance cost estimating practice using an industrial case study approach. A flexible design using a case study strategy is described as the most appropriate approach to the successful completion of this study. Industrial case studies using workshops and interview techniques are the primary sources of data whereas literature is used as the secondary sources of data. Following the identification of Railway renewal and maintenance cost estimating issues, a further review of literature leads to the development of a hypothesis. In order to investigate the hypothesis a structured cost estimating methodology is developed which comprises four main stages: creating a project structure that composes the goal, project criteria and alternatives; collecting the necessary data in the form of pairwise comparisons made by a domain expert; producing alternative weights using a geometric mean; and finally employing an algorithmic method using the produced alternative weights and the known cost of one alternative per criteria. The model was implemented within a prototype software tool. This provided a means to validate the proposed model using three industrial case studies. These results provide evidence that the application of a pairwise comparisons based methodology to Railway renewal and maintenance cost estimating problems can provide beneficial. The results indicated that twelve of the fifteen estimates produced by the model were within the expected accuracy and therefore on most occasions prove the hypothesis to be true.