Development of a flood risk assessment framework for railway drainage infrastructure assets at component-level

Aljafari, Nour ORCID: 0000-0002-3498-7363 (2023). Development of a flood risk assessment framework for railway drainage infrastructure assets at component-level. University of Birmingham. Ph.D.

Aljafari2023PhD.pdf
Text
Restricted to Repository staff only until 29 December 2024.
Available under License All rights reserved.
Download (42MB)

Abstract

Inadequate performance of the railway drainage system has detrimental effects on the performance of the various other systems in a railway section including track superstructure, substructure, earthworks, electric, and operational equipment. Flooding incidents in railway drainage systems are exacerbated as severe rainfall events are becoming more frequent due to climate change. Drainage asset interventions need optimised to improve resilience of the railway to flooding. Given the time and money limitations, asset management decision-making needs to rely on targeted and evidence-based approaches for optimised resource allocations. Since the drainage infrastructure of many railways is old and often buried, there is a paucity of detailed inspection and maintenance history data and available data suffers from being subjective, inconsistent and incomplete. Such data introduces uncertainty into the decision-making process, which is aggravated by the uncertain future impacts of land-use and climate change. Therefore, there is a need for approaches that can deal with these uncertainties, such as risk management, to support established asset management techniques.

This research was conducted to provide railway drainage infrastructure managers with a decision-support framework for quantitative assessment of flood risk associated with drainage assets at a component level. The developed framework consists of four modules that utilise data-driven technologies to obtain elements required for a quantitative risk score. The first module looks at condition modelling of railway drainage assets using Machine learning (ML) algorithms. The second module proposes the use of two-dimensional hydraulic modelling of railway drainage assets for simulation of flooding scenarios. The third module quantifies flood damage through the development of an economic damage costing model which considers both tangible and intangible impacts of a flooding incident. The damage quantification entails an uncertainty analysis to account for uncertainties in costing and exposure parameters. The fourth module quantifies risk in monetary terms using probabilities of the flooding scenarios and their respective economic damage. Applicability of the proposed tools for each of the modules is demonstrated through three case studies. The first case study entailed training and testing of a variety of ML algorithms for modelling the structural and service condition of railway drainage pipes using asset records provided by the UK railway infrastructure manager, Network Rail. The best-performing model of the trained ML algorithms was the Boosted Trees algorithm which achieved 88% overall accuracy for structural condition prediction and 72% overall accuracy for service condition prediction. The second case study involved the use of a two-dimensional hydraulic modelling software to model flooding scenarios at a case-study railway section in the UK, with asset data provided by Network Rail. The flood exposure outputs for the various tested scenarios were used in the third case study to demonstrate the damage quantification module through running Monte Carlo simulations using the developed cost model at two locations within the studied railway section. The damage was presented with monetary damage charts for the various scenarios at either locations. The sensitivity of cost model was largest towards the parameters: Length of flooded track, number of passenger trains passing the site, train cancellation cost and renewal work possession time. Larger uncertainty in these parameters will result in larger uncertainty in the damage quantification and subsequently risk quantification. The damage quantification followed with a risk quantification for the two locations, reflecting the monetary risk level for each. The risk was demonstrated with uncertainty envelopes at the two locations. At one location the expected risk value was £1.21M and at the other it was £2.41M, although the second location entailed larger uncertainty in the quantified damage. This demonstrates the applicability of the framework to support operational level of drainage infrastructure management. It enables streamlining risk-informed decision-making for scheduling and prioritisation of proactive intervention activities. Thereby, it allows optimal resource allocations, building resilience within the railway drainage system against extreme rainfall events and evaluation of future flood risk mitigation measures.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):